| Message ID | 20230714150326.1152359-1-alessandro.carminati@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v2] scripts/link-vmlinux.sh: Add alias to duplicate symbols for kallsyms | expand |
On Fri, 14 Jul 2023 15:03:26 +0000 Alessandro Carminati <alessandro.carminati@gmail.com> wrote: > It is not uncommon for drivers or modules related to similar peripherals > to have symbols with the exact same name. > While this is not a problem for the kernel's binary itself, it becomes an > issue when attempting to trace or probe specific functions using > infrastructure like ftrace or kprobe. > > The tracing subsystem relies on the `nm -n vmlinux` output, which provides > symbol information from the kernel's ELF binary. However, when multiple > symbols share the same name, the standard nm output does not differentiate > between them. This can lead to confusion and difficulty when trying to > probe the intended symbol. > > ~ # cat /proc/kallsyms | grep " name_show" > ffffffff8c4f76d0 t name_show > ffffffff8c9cccb0 t name_show > ffffffff8cb0ac20 t name_show > ffffffff8cc728c0 t name_show > ffffffff8ce0efd0 t name_show > ffffffff8ce126c0 t name_show > ffffffff8ce1dd20 t name_show > ffffffff8ce24e70 t name_show > ffffffff8d1104c0 t name_show > ffffffff8d1fe480 t name_show > > **kas_alias** addresses this challenge by extending the symbol names with > unique suffixes during the kernel build process. > The newly created aliases for these duplicated symbols are unique names > that can be fed to the ftracefs interface. By doing so, it enables > previously unreachable symbols to be probed. > > ~ # cat /proc/kallsyms | grep " name_show" > ffffffff974f76d0 t name_show > ffffffff974f76d0 t name_show__alias__6340 > ffffffff979cccb0 t name_show > ffffffff979cccb0 t name_show__alias__6341 > ffffffff97b0ac20 t name_show > ffffffff97b0ac20 t name_show__alias__6342 > ffffffff97c728c0 t name_show > ffffffff97c728c0 t name_show__alias__6343 > ffffffff97e0efd0 t name_show > ffffffff97e0efd0 t name_show__alias__6344 > ffffffff97e126c0 t name_show > ffffffff97e126c0 t name_show__alias__6345 > ffffffff97e1dd20 t name_show > ffffffff97e1dd20 t name_show__alias__6346 > ffffffff97e24e70 t name_show > ffffffff97e24e70 t name_show__alias__6347 > ffffffff981104c0 t name_show > ffffffff981104c0 t name_show__alias__6348 > ffffffff981fe480 t name_show > ffffffff981fe480 t name_show__alias__6349 > ~ # echo "p:kprobes/evnt1 name_show__alias__6349" \ > > >/sys/kernel/tracing/kprobe_events > ~ # cat /sys/kernel/tracing/kprobe_events > p:kprobes/evnt1 name_show__alias__6349 > Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't give you any clue to what function you are actually attaching to. There's been other approaches that show module and/or file names. I know there's still some issues with getting those accepted, but I much rather have them than this! See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ -- Steve > Changes from v1: > - Integrated changes requested by Masami to exclude symbols with prefixes > "_cfi" and "_pfx". > - Introduced a small framework to handle patterns that need to be excluded > from the alias production. > - Excluded other symbols using the framework. > - Introduced the ability to discriminate between text and data symbols. > - Added two new config symbols in this version: CONFIG_KALLSYMS_ALIAS_DATA, > which allows data for data, and CONFIG_KALLSYMS_ALIAS_DATA_ALL, which > excludes all filters and provides an alias for each duplicated symbol. > > https://lore.kernel.org/all/20230711151925.1092080-1-alessandro.carminati@gmail.com/ > > Signed-off-by: Alessandro Carminati (Red Hat) <alessandro.carminati@gmail.com> > --- > init/Kconfig | 36 +++++ > scripts/Makefile | 4 + > scripts/kas_alias/Makefile | 4 + > scripts/kas_alias/duplicates_list.c | 70 +++++++++ > scripts/kas_alias/duplicates_list.h | 15 ++ > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++++++ > scripts/kas_alias/item_list.h | 26 ++++ > scripts/kas_alias/kas_alias.c | 186 ++++++++++++++++++++++ > scripts/link-vmlinux.sh | 6 +- > 9 files changed, 576 insertions(+), 1 deletion(-) > create mode 100644 scripts/kas_alias/Makefile > create mode 100644 scripts/kas_alias/duplicates_list.c > create mode 100644 scripts/kas_alias/duplicates_list.h > create mode 100644 scripts/kas_alias/item_list.c > create mode 100644 scripts/kas_alias/item_list.h > create mode 100644 scripts/kas_alias/kas_alias.c > > diff --git a/init/Kconfig b/init/Kconfig > index f7f65af4ee12..5ae7c613cdc4 100644 > --- a/init/Kconfig > +++ b/init/Kconfig > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > time constants, and no relocation pass is required at runtime to fix > up the entries based on the runtime load address of the kernel. > > +config KALLSYMS_ALIAS > + bool "Produces alias for duplicated symbols" if EXPERT > + depends on KALLSYMS > + help > + It is not uncommon for drivers or modules related to similar > + peripherals to have symbols with the exact same name. > + While this is not a problem for the kernel's binary itself, it > + becomes an issue when attempting to trace or probe specific > + functions using infrastructure like ftrace or kprobe. > + > + This option addresses this challenge by extending the symbol names > + with unique suffixes during the kernel build process. > + The newly created aliases for these duplicated symbols are unique > + names that can be fed to the ftrace sysfs interface. By doing so, it > + enables previously unreachable symbols to be probed. > + > +config CONFIG_KALLSYMS_ALIAS_DATA > + bool "Produces alias also for data" > + depends on KALLSYMS_ALIAS > + help > + Sometimes it can be useful to refer to data. In live patch scenarios, > + you may find yourself needing to use symbols that are shared with > + other functions. Since symbols face the same issue as functions, this > + option allows you to create aliases for data as well. > + > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > + bool "Removes all filter when producing data alias" > + depends on CONFIG_KALLSYMS_ALIAS_DATA > + help > + When selecting data aliases, not all symbols are included in the set > + This is because many symbols are unlikely to be used. If you choose > + to have an alias for all data symbols, be aware that it will > + significantly increase the size. > + > + If unsure, say N. > + > # end of the "standard kernel features (expert users)" menu > > # syscall, maps, verifier > diff --git a/scripts/Makefile b/scripts/Makefile > index 32b6ba722728..65fafe17cfe5 100644 > --- a/scripts/Makefile > +++ b/scripts/Makefile > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > # Let clean descend into subdirs > subdir- += basic dtc gdb kconfig mod > + > +# KALLSyms alias > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > + > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > new file mode 100644 > index 000000000000..523fa3441013 > --- /dev/null > +++ b/scripts/kas_alias/Makefile > @@ -0,0 +1,4 @@ > +# SPDX-License-Identifier: GPL-2.0 > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > + > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o > diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c > new file mode 100644 > index 000000000000..e7a3d2917937 > --- /dev/null > +++ b/scripts/kas_alias/duplicates_list.c > @@ -0,0 +1,70 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdint.h> > +#include <stdio.h> > +#include <string.h> > +#include <stdlib.h> > +#include <stdbool.h> > + > +#include "item_list.h" > +#include "duplicates_list.h" > + > +struct duplicate_item *find_duplicates(struct item *list) > +{ > + struct duplicate_item *current_duplicate = NULL; > + struct duplicate_item *duplicates = NULL; > + struct duplicate_item *new_duplicate; > + struct item *current_item = list; > + bool prev_was_duplicate = false; > + struct item *prev_item = NULL; > + > + while (current_item) { > + if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) || > + prev_was_duplicate) { > + if (!duplicates) { > + duplicates = malloc(sizeof(struct duplicate_item)); > + if (!duplicates) > + return NULL; > + > + duplicates->original_item = prev_item; > + duplicates->next = NULL; > + current_duplicate = duplicates; > + } else { > + new_duplicate = malloc(sizeof(struct duplicate_item)); > + if (!new_duplicate) { > + free_duplicates(&duplicates); > + return NULL; > + } > + > + new_duplicate->original_item = prev_item; > + new_duplicate->next = NULL; > + current_duplicate->next = new_duplicate; > + current_duplicate = new_duplicate; > + > + if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) && > + (prev_was_duplicate)) > + prev_was_duplicate = false; > + else > + prev_was_duplicate = true; > + } > + } > + > + prev_item = current_item; > + current_item = current_item->next; > + } > + > + return duplicates; > +} > + > +void free_duplicates(struct duplicate_item **duplicates) > +{ > + struct duplicate_item *duplicates_iterator = *duplicates; > + struct duplicate_item *app; > + > + while (duplicates_iterator) { > + app = duplicates_iterator; > + duplicates_iterator = duplicates_iterator->next; > + free(app); > + } > + > + *duplicates = NULL; > +} > diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h > new file mode 100644 > index 000000000000..76aa73e584bc > --- /dev/null > +++ b/scripts/kas_alias/duplicates_list.h > @@ -0,0 +1,15 @@ > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > +#ifndef DUPLICATES_LIST_H > +#define DUPLICATES_LIST_H > + > +#include "item_list.h" > + > +struct duplicate_item { > + struct item *original_item; > + struct duplicate_item *next; > +}; > + > +struct duplicate_item *find_duplicates(struct item *list); > +void free_duplicates(struct duplicate_item **duplicates); > + > +#endif > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > new file mode 100644 > index 000000000000..7c9d5aecca9a > --- /dev/null > +++ b/scripts/kas_alias/item_list.c > @@ -0,0 +1,230 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdio.h> > +#include <stdlib.h> > +#include <stdint.h> > +#include <string.h> > +#include <stdbool.h> > +#include <assert.h> > +#include "item_list.h" > + > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0) > + > +struct item *list_index[96] = {0}; > + > +void build_index(struct item *list) > +{ > + char current_first_letter = ' '; > + struct item *current = list; > + > + while (current) { > + if (current->symb_name[0] != current_first_letter) { > + current_first_letter = current->symb_name[0]; > + list_index[current_first_letter - 32] = current; > + } > + current = current->next; > + } > +} > + > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr) > +{ > + struct item *new_item; > + struct item *current; > + > + new_item = malloc(sizeof(struct item)); > + if (!new_item) > + return NULL; > + > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > + new_item->addr = addr; > + new_item->stype = stype; > + new_item->next = NULL; > + > + if (!(*list)) { > + *list = new_item; > + } else { > + current = *list; > + while (current->next) > + current = current->next; > + > + current->next = new_item; > + } > + return new_item; > +} > + > +void sort_list(struct item **list, int sort_by) > +{ > + struct item *current = *list; > + struct item *sorted = NULL; > + struct item *next_item; > + struct item *temp; > + > + if (!(*list) || !((*list)->next)) > + return; > + > + while (current) { > + next_item = current->next; > + if (!sorted || > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) || > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > + current->next = sorted; > + sorted = current; > + } else { > + temp = sorted; > + while (temp->next && > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) || > + CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) > + temp = temp->next; > + > + current->next = temp->next; > + temp->next = current; > + } > + current = next_item; > + } > + > + *list = sorted; > +} > + > +struct item *merge(struct item *left, struct item *right, int sort_by) > +{ > + struct item *current = NULL; > + struct item *result = NULL; > + > + if (!left) > + return right; > + if (!right) > + return left; > + > + if (sort_by == BY_NAME) { > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > + result = left; > + left = left->next; > + } else { > + result = right; > + right = right->next; > + } > + } else { > + if (sort_by == BY_ADDRESS) { > + if (left->addr <= right->addr) { > + result = left; > + left = left->next; > + } else { > + result = right; > + right = right->next; > + } > + } > + } > + > + current = result; > + > + while (left && right) { > + if (sort_by == BY_NAME) { > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > + current->next = left; > + left = left->next; > + } else { > + current->next = right; > + right = right->next; > + } > + } else { > + if (sort_by == BY_ADDRESS) { > + if (left->addr <= right->addr) { > + current->next = left; > + left = left->next; > + } else { > + current->next = right; > + right = right->next; > + } > + } > + } > + > + current = current->next; > + } > + > + if (left) { > + current->next = left; > + } else { > + if (right) > + current->next = right; > + } > + > + return result; > +} > + > +struct item *merge_sort(struct item *head, int sort_by) > +{ > + struct item *right; > + struct item *slow; > + struct item *fast; > + struct item *left; > + > + if (!head || !head->next) > + return head; > + > + slow = head; > + fast = head->next; > + > + while (fast && fast->next) { > + slow = slow->next; > + fast = fast->next->next; > + } > + > + left = head; > + right = slow->next; > + slow->next = NULL; > + > + left = merge_sort(left, sort_by); > + right = merge_sort(right, sort_by); > + > + return merge(left, right, sort_by); > +} > + > +void sort_list_m(struct item **head, int sort_by) > +{ > + if (!(*head) || !((*head)->next)) > + return; > + > + *head = merge_sort(*head, sort_by); > +} > + > +int insert_after(struct item *list, const uint64_t search_addr, > + const char *name, uint64_t addr, char stype) > +{ > + struct item *new_item; > + struct item *current; > + int ret = 0; > + > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list; > + while (current) { > + if (current->addr == search_addr) { > + new_item = malloc(sizeof(struct item)); > + if (!new_item) > + return ret; > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > + new_item->addr = addr; > + new_item->stype = stype; > + new_item->next = current->next; > + current->next = new_item; > + ret = 1; > + break; > + } > + current = current->next; > + } > + return ret; > +} > + > +void free_items(struct item **head) > +{ > + struct item *app, *item_iterator = *head; > + > + while (item_iterator) { > + app = item_iterator; > + item_iterator = item_iterator->next; > + free(app); > + } > + *head = NULL; > +} > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > new file mode 100644 > index 000000000000..b4891cb088ee > --- /dev/null > +++ b/scripts/kas_alias/item_list.h > @@ -0,0 +1,26 @@ > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > +#ifndef ITEM_LIST_H > +#define ITEM_LIST_H > +#include <stdint.h> > + > +#define MAX_NAME_SIZE 256 > +#define BY_ADDRESS 1 > +#define BY_NAME 2 > + > +struct item { > + char symb_name[MAX_NAME_SIZE]; > + uint64_t addr; > + char stype; > + struct item *next; > +}; > + > +void build_index(struct item *list); > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr); > +void sort_list(struct item **list, int sort_by); > +struct item *merge(struct item *left, struct item *right, int sort_by); > +struct item *merge_sort(struct item *head, int sort_by); > +void sort_list_m(struct item **head, int sort_by); > +int insert_after(struct item *list, const uint64_t search_addr, > + const char *name, uint64_t addr, char stype); > +void free_items(struct item **head); > +#endif > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > new file mode 100644 > index 000000000000..eb873b368023 > --- /dev/null > +++ b/scripts/kas_alias/kas_alias.c > @@ -0,0 +1,186 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdio.h> > +#include <stdlib.h> > +#include <stdint.h> > +#include <string.h> > +#include <stdbool.h> > +#include <stdarg.h> > +#include <regex.h> > + > +#include "item_list.h" > +#include "duplicates_list.h" > + > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \ > + (((s)->stype) == 'd') || (((s)->stype) == 'D') || \ > + (((s)->stype) == 'r') || (((s)->stype) == 'R')) > +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > +#else > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > +#endif > +#define FNOMATCH 0 > +#define FMATCH 1 > +#define EREGEX 2 > + > +const char *ignore_regex[] = { > + "^__cfi_.*$", // __cfi_ preamble > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > + "^_*TRACE_SYSTEM.*$", > + "^__already_done\\.[0-9]+$", // Call a function once data > + "^___tp_str\\.[0-9]+$", > + "^___done\\.[0-9]+$", > + "^__print_once\\.[0-9]+$", > + "^_rs\\.[0-9]+$", > + "^__compound_literal\\.[0-9]+$", > + "^___once_key\\.[0-9]+$", > + "^__func__\\.[0-9]+$", > + "^__msg\\.[0-9]+$", > + "^CSWTCH\\.[0-9]+$", > + "^__flags\\.[0-9]+$", > + "^__wkey.*$", > + "^__mkey.*$", > + "^__key.*$", > +#endif > + "^__pfx_.*$" // NOP-padding > +}; > + > +int suffix_serial; > + > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > +{ > + va_list args; > + > + va_start(args, fmt); > + if (verbose) > + printf(fmt, args); > + > + va_end(args); > +} > + > +static void create_suffix(const char *name, char *output_suffix) > +{ > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > +} > + > +static int filter_symbols(char *symbol, const char **ignore_list, int regex_no) > +{ > + regex_t regex; > + int res, i; > + > + for (i = 0; i < regex_no; i++) { > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > + if (res) > + return -EREGEX; > + > + res = regexec(®ex, symbol, 0, NULL, 0); > + regfree(®ex); > + switch (res) { > + case 0: > + return FMATCH; > + case REG_NOMATCH: > + break; > + default: > + return -EREGEX; > + } > + > + } > + > + return FNOMATCH; > +} > + > +int main(int argc, char *argv[]) > +{ > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > + struct duplicate_item *duplicate_iterator; > + struct duplicate_item *duplicate; > + struct item *head = {NULL}; > + bool need_2_process = true; > + struct item *last = {NULL}; > + struct item *current; > + int verbose_mode = 0; > + uint64_t address; > + FILE *fp; > + int res; > + > + if (argc < 2 || argc > 3) { > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > + return 1; > + } > + > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > + verbose_mode = 1; > + > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > + > + fp = fopen(argv[1], "r"); > + if (!fp) { > + printf("Can't open input file.\n"); > + return 1; > + } > + > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > + if (strstr(sym_name, "__alias__1") != NULL) { > + if (verbose_mode && need_2_process) > + printf("Already processed\n"); > + need_2_process = false; > + } > + last = add_item(&last, sym_name, t, address); > + if (!last) { > + printf("Error in allocate memory\n"); > + free_items(&head); > + return 1; > + } > + > + if (!head) > + head = last; > + } > + > + fclose(fp); > + > + if (need_2_process) { > + verbose_msg(verbose_mode, "Sorting nm data\n"); > + sort_list_m(&head, BY_NAME); > + verbose_msg(verbose_mode, "Scanning nm data for duplicates\n"); > + duplicate = find_duplicates(head); > + if (!duplicate) { > + printf("Error in duplicates list\n"); > + return 1; > + } > + > + verbose_msg(verbose_mode, "Applying suffixes\n"); > + build_index(head); > + duplicate_iterator = duplicate; > + while (duplicate_iterator) { > + res = filter_symbols(duplicate_iterator->original_item->symb_name, > + ignore_regex, sizeof(ignore_regex) / > + sizeof(ignore_regex[0])); > + if ((res != FMATCH) && > + SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { > + if (res < 0) > + return 1; > + > + create_suffix(duplicate_iterator->original_item->symb_name, > + new_name); > + if (!insert_after(head, duplicate_iterator->original_item->addr, > + new_name, duplicate_iterator->original_item->addr, > + duplicate_iterator->original_item->stype)) > + return 1; > + } > + > + duplicate_iterator = duplicate_iterator->next; > + } > + > + sort_list_m(&head, BY_ADDRESS); > + } > + current = head; > + while (current) { > + printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name); > + current = current->next; > + } > + > + free_items(&head); > + free_duplicates(&duplicate); > + > + return 0; > +} > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > index a432b171be82..4e3dea3ac2df 100755 > --- a/scripts/link-vmlinux.sh > +++ b/scripts/link-vmlinux.sh > @@ -161,7 +161,11 @@ kallsyms() > fi > > info KSYMS ${2} > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > + ALIAS=".alias" > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > + fi > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > } > > # Perform one step in kallsyms generation, including temporary linking of
On Mon, Jul 17, 2023 at 10:52:40AM -0400, Steven Rostedt wrote: > Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't > give you any clue to what function you are actually attaching to. Agreed. > There's > been other approaches that show module and/or file names. I know there's > still some issues with getting those accepted, but I much rather have them > than this! > > See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ Yes, please coordinate with Nick and review each other's work, now we have two separate efforts with different reasons but hopefully we'll come back with one unified solution. Please Cc live-patching also, as they had suggested before just to provide the file filename + line number, that'll make it even more valuable. Luis
On Mon, 17 Jul 2023 09:50:25 -0700 Luis Chamberlain <mcgrof@kernel.org> wrote: > On Mon, Jul 17, 2023 at 10:52:40AM -0400, Steven Rostedt wrote: > > Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't > > give you any clue to what function you are actually attaching to. > > Agreed. +1 :) > > There's > > been other approaches that show module and/or file names. I know there's > > still some issues with getting those accepted, but I much rather have them > > than this! > > > > See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ > > Yes, please coordinate with Nick and review each other's work, now we > have two separate efforts with different reasons but hopefully we'll > come back with one unified solution. > > Please Cc live-patching also, as they had suggested before just to > provide the file filename + line number, that'll make it even more > valuable. I want to involve BTF, that is currently only support looking up function by name. This alias is currently decouple from the BTF, but BTF name is also need to be updated so that we can lookup BTF from the alias name. Thank you,
Hi Steven, Il giorno lun 17 lug 2023 alle ore 16:52 Steven Rostedt <rostedt@goodmis.org> ha scritto: > > On Fri, 14 Jul 2023 15:03:26 +0000 > Alessandro Carminati <alessandro.carminati@gmail.com> wrote: > > > It is not uncommon for drivers or modules related to similar peripherals > > to have symbols with the exact same name. > > While this is not a problem for the kernel's binary itself, it becomes an > > issue when attempting to trace or probe specific functions using > > infrastructure like ftrace or kprobe. > > > > The tracing subsystem relies on the `nm -n vmlinux` output, which provides > > symbol information from the kernel's ELF binary. However, when multiple > > symbols share the same name, the standard nm output does not differentiate > > between them. This can lead to confusion and difficulty when trying to > > probe the intended symbol. > > > > ~ # cat /proc/kallsyms | grep " name_show" > > ffffffff8c4f76d0 t name_show > > ffffffff8c9cccb0 t name_show > > ffffffff8cb0ac20 t name_show > > ffffffff8cc728c0 t name_show > > ffffffff8ce0efd0 t name_show > > ffffffff8ce126c0 t name_show > > ffffffff8ce1dd20 t name_show > > ffffffff8ce24e70 t name_show > > ffffffff8d1104c0 t name_show > > ffffffff8d1fe480 t name_show > > > > **kas_alias** addresses this challenge by extending the symbol names with > > unique suffixes during the kernel build process. > > The newly created aliases for these duplicated symbols are unique names > > that can be fed to the ftracefs interface. By doing so, it enables > > previously unreachable symbols to be probed. > > > > ~ # cat /proc/kallsyms | grep " name_show" > > ffffffff974f76d0 t name_show > > ffffffff974f76d0 t name_show__alias__6340 > > ffffffff979cccb0 t name_show > > ffffffff979cccb0 t name_show__alias__6341 > > ffffffff97b0ac20 t name_show > > ffffffff97b0ac20 t name_show__alias__6342 > > ffffffff97c728c0 t name_show > > ffffffff97c728c0 t name_show__alias__6343 > > ffffffff97e0efd0 t name_show > > ffffffff97e0efd0 t name_show__alias__6344 > > ffffffff97e126c0 t name_show > > ffffffff97e126c0 t name_show__alias__6345 > > ffffffff97e1dd20 t name_show > > ffffffff97e1dd20 t name_show__alias__6346 > > ffffffff97e24e70 t name_show > > ffffffff97e24e70 t name_show__alias__6347 > > ffffffff981104c0 t name_show > > ffffffff981104c0 t name_show__alias__6348 > > ffffffff981fe480 t name_show > > ffffffff981fe480 t name_show__alias__6349 > > ~ # echo "p:kprobes/evnt1 name_show__alias__6349" \ > > > >/sys/kernel/tracing/kprobe_events > > ~ # cat /sys/kernel/tracing/kprobe_events > > p:kprobes/evnt1 name_show__alias__6349 > > > > Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't > give you any clue to what function you are actually attaching to. There's > been other approaches that show module and/or file names. I know there's > still some issues with getting those accepted, but I much rather have them > than this! > > See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ > Thanks for sharing your thoughts on the current symbol naming convention and Nick Alcock's approach. I see your point about "alias<some-random-number>" not providing much context about the attached function. However, my intention was only to create a means to probe all functions. While other approaches, such as including module and file names, are more informative, it appears that these proposals are encountering certain challenges in gaining acceptance. Also, they do not fully address the specific problem I wanted to tackle. In my upcoming proposal, I would like to maintain the conservative approach I have followed so far while still achieving my original objective of making all functions reachable from kprobe. To date, I have been working on addressing all the necessary aspects from outside the kernel source code. Essentially, I take the output of 'nm' and refine it to provide more meaningful information. I am working in a V3, I am considering using `addr2line` to determine the origin of each symbol. Initially, I avoided it because the configuration dependency it introduces, specifically requiring DWARF information such as CONFIG_DEBUG_INFO_DWARF4 or CONFIG_DEBUG_INFO_DWARF5. However, after reading Luis' comment suggesting the use of 'filename + line number', I realized that relying on DWARF information might be the only way to achieve a satisfying result. I will try that and see if it works... unless you know it will not already :-) Actually, I did a fast try using DWARF and addr2line, the strategy I'd like to use in the v3, and the output looks like this: what do you guys think? $ ../kas_alias/main .tmp_vmlinux.kallsyms2.syms | grep name_show ffffffff8100e5e0 t __pfx_pmu_name_show ffffffff8100e5f0 t pmu_name_show ffffffff810ed8e0 t __pfx_name_show ffffffff810ed8f0 t name_show ffffffff810ed8f0 t name_show@_kernel_irq_irqdesc_c_232 ffffffff810ed950 t __pfx_chip_name_show ffffffff810ed960 t chip_name_show ffffffff810ed960 t chip_name_show@_kernel_irq_irqdesc_c_168 ffffffff814cae80 t __pfx_name_show ffffffff814cae90 t name_show ffffffff814cae90 t name_show@_drivers_pnp_card_c_186 ffffffff816006d0 t __pfx_name_show ffffffff816006e0 t name_show ffffffff816006e0 t name_show@_drivers_gpu_drm_i915_gt_sysfs_engines_c_26 ffffffff81765460 t __pfx_name_show ffffffff81765470 t name_show ffffffff81765470 t name_show@_drivers_base_power_wakeup_stats_c_93 ffffffff8189b340 t __pfx_name_show [...] I want to emphasize that my goal is to find a solution that satisfies everyone involved and addresses the concerns raised. Bests Alessandro > -- Steve > > > > Changes from v1: > > - Integrated changes requested by Masami to exclude symbols with prefixes > > "_cfi" and "_pfx". > > - Introduced a small framework to handle patterns that need to be excluded > > from the alias production. > > - Excluded other symbols using the framework. > > - Introduced the ability to discriminate between text and data symbols. > > - Added two new config symbols in this version: CONFIG_KALLSYMS_ALIAS_DATA, > > which allows data for data, and CONFIG_KALLSYMS_ALIAS_DATA_ALL, which > > excludes all filters and provides an alias for each duplicated symbol. > > > > https://lore.kernel.org/all/20230711151925.1092080-1-alessandro.carminati@gmail.com/ > > > > Signed-off-by: Alessandro Carminati (Red Hat) <alessandro.carminati@gmail.com> > > --- > > init/Kconfig | 36 +++++ > > scripts/Makefile | 4 + > > scripts/kas_alias/Makefile | 4 + > > scripts/kas_alias/duplicates_list.c | 70 +++++++++ > > scripts/kas_alias/duplicates_list.h | 15 ++ > > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++++++ > > scripts/kas_alias/item_list.h | 26 ++++ > > scripts/kas_alias/kas_alias.c | 186 ++++++++++++++++++++++ > > scripts/link-vmlinux.sh | 6 +- > > 9 files changed, 576 insertions(+), 1 deletion(-) > > create mode 100644 scripts/kas_alias/Makefile > > create mode 100644 scripts/kas_alias/duplicates_list.c > > create mode 100644 scripts/kas_alias/duplicates_list.h > > create mode 100644 scripts/kas_alias/item_list.c > > create mode 100644 scripts/kas_alias/item_list.h > > create mode 100644 scripts/kas_alias/kas_alias.c > > > > diff --git a/init/Kconfig b/init/Kconfig > > index f7f65af4ee12..5ae7c613cdc4 100644 > > --- a/init/Kconfig > > +++ b/init/Kconfig > > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > > time constants, and no relocation pass is required at runtime to fix > > up the entries based on the runtime load address of the kernel. > > > > +config KALLSYMS_ALIAS > > + bool "Produces alias for duplicated symbols" if EXPERT > > + depends on KALLSYMS > > + help > > + It is not uncommon for drivers or modules related to similar > > + peripherals to have symbols with the exact same name. > > + While this is not a problem for the kernel's binary itself, it > > + becomes an issue when attempting to trace or probe specific > > + functions using infrastructure like ftrace or kprobe. > > + > > + This option addresses this challenge by extending the symbol names > > + with unique suffixes during the kernel build process. > > + The newly created aliases for these duplicated symbols are unique > > + names that can be fed to the ftrace sysfs interface. By doing so, it > > + enables previously unreachable symbols to be probed. > > + > > +config CONFIG_KALLSYMS_ALIAS_DATA > > + bool "Produces alias also for data" > > + depends on KALLSYMS_ALIAS > > + help > > + Sometimes it can be useful to refer to data. In live patch scenarios, > > + you may find yourself needing to use symbols that are shared with > > + other functions. Since symbols face the same issue as functions, this > > + option allows you to create aliases for data as well. > > + > > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > > + bool "Removes all filter when producing data alias" > > + depends on CONFIG_KALLSYMS_ALIAS_DATA > > + help > > + When selecting data aliases, not all symbols are included in the set > > + This is because many symbols are unlikely to be used. If you choose > > + to have an alias for all data symbols, be aware that it will > > + significantly increase the size. > > + > > + If unsure, say N. > > + > > # end of the "standard kernel features (expert users)" menu > > > > # syscall, maps, verifier > > diff --git a/scripts/Makefile b/scripts/Makefile > > index 32b6ba722728..65fafe17cfe5 100644 > > --- a/scripts/Makefile > > +++ b/scripts/Makefile > > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > > > # Let clean descend into subdirs > > subdir- += basic dtc gdb kconfig mod > > + > > +# KALLSyms alias > > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > + > > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > > new file mode 100644 > > index 000000000000..523fa3441013 > > --- /dev/null > > +++ b/scripts/kas_alias/Makefile > > @@ -0,0 +1,4 @@ > > +# SPDX-License-Identifier: GPL-2.0 > > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > + > > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o > > diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c > > new file mode 100644 > > index 000000000000..e7a3d2917937 > > --- /dev/null > > +++ b/scripts/kas_alias/duplicates_list.c > > @@ -0,0 +1,70 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdint.h> > > +#include <stdio.h> > > +#include <string.h> > > +#include <stdlib.h> > > +#include <stdbool.h> > > + > > +#include "item_list.h" > > +#include "duplicates_list.h" > > + > > +struct duplicate_item *find_duplicates(struct item *list) > > +{ > > + struct duplicate_item *current_duplicate = NULL; > > + struct duplicate_item *duplicates = NULL; > > + struct duplicate_item *new_duplicate; > > + struct item *current_item = list; > > + bool prev_was_duplicate = false; > > + struct item *prev_item = NULL; > > + > > + while (current_item) { > > + if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) || > > + prev_was_duplicate) { > > + if (!duplicates) { > > + duplicates = malloc(sizeof(struct duplicate_item)); > > + if (!duplicates) > > + return NULL; > > + > > + duplicates->original_item = prev_item; > > + duplicates->next = NULL; > > + current_duplicate = duplicates; > > + } else { > > + new_duplicate = malloc(sizeof(struct duplicate_item)); > > + if (!new_duplicate) { > > + free_duplicates(&duplicates); > > + return NULL; > > + } > > + > > + new_duplicate->original_item = prev_item; > > + new_duplicate->next = NULL; > > + current_duplicate->next = new_duplicate; > > + current_duplicate = new_duplicate; > > + > > + if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) && > > + (prev_was_duplicate)) > > + prev_was_duplicate = false; > > + else > > + prev_was_duplicate = true; > > + } > > + } > > + > > + prev_item = current_item; > > + current_item = current_item->next; > > + } > > + > > + return duplicates; > > +} > > + > > +void free_duplicates(struct duplicate_item **duplicates) > > +{ > > + struct duplicate_item *duplicates_iterator = *duplicates; > > + struct duplicate_item *app; > > + > > + while (duplicates_iterator) { > > + app = duplicates_iterator; > > + duplicates_iterator = duplicates_iterator->next; > > + free(app); > > + } > > + > > + *duplicates = NULL; > > +} > > diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h > > new file mode 100644 > > index 000000000000..76aa73e584bc > > --- /dev/null > > +++ b/scripts/kas_alias/duplicates_list.h > > @@ -0,0 +1,15 @@ > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > +#ifndef DUPLICATES_LIST_H > > +#define DUPLICATES_LIST_H > > + > > +#include "item_list.h" > > + > > +struct duplicate_item { > > + struct item *original_item; > > + struct duplicate_item *next; > > +}; > > + > > +struct duplicate_item *find_duplicates(struct item *list); > > +void free_duplicates(struct duplicate_item **duplicates); > > + > > +#endif > > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > > new file mode 100644 > > index 000000000000..7c9d5aecca9a > > --- /dev/null > > +++ b/scripts/kas_alias/item_list.c > > @@ -0,0 +1,230 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdio.h> > > +#include <stdlib.h> > > +#include <stdint.h> > > +#include <string.h> > > +#include <stdbool.h> > > +#include <assert.h> > > +#include "item_list.h" > > + > > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0) > > + > > +struct item *list_index[96] = {0}; > > + > > +void build_index(struct item *list) > > +{ > > + char current_first_letter = ' '; > > + struct item *current = list; > > + > > + while (current) { > > + if (current->symb_name[0] != current_first_letter) { > > + current_first_letter = current->symb_name[0]; > > + list_index[current_first_letter - 32] = current; > > + } > > + current = current->next; > > + } > > +} > > + > > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr) > > +{ > > + struct item *new_item; > > + struct item *current; > > + > > + new_item = malloc(sizeof(struct item)); > > + if (!new_item) > > + return NULL; > > + > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > + new_item->addr = addr; > > + new_item->stype = stype; > > + new_item->next = NULL; > > + > > + if (!(*list)) { > > + *list = new_item; > > + } else { > > + current = *list; > > + while (current->next) > > + current = current->next; > > + > > + current->next = new_item; > > + } > > + return new_item; > > +} > > + > > +void sort_list(struct item **list, int sort_by) > > +{ > > + struct item *current = *list; > > + struct item *sorted = NULL; > > + struct item *next_item; > > + struct item *temp; > > + > > + if (!(*list) || !((*list)->next)) > > + return; > > + > > + while (current) { > > + next_item = current->next; > > + if (!sorted || > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) || > > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > > + current->next = sorted; > > + sorted = current; > > + } else { > > + temp = sorted; > > + while (temp->next && > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) || > > + CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) > > + temp = temp->next; > > + > > + current->next = temp->next; > > + temp->next = current; > > + } > > + current = next_item; > > + } > > + > > + *list = sorted; > > +} > > + > > +struct item *merge(struct item *left, struct item *right, int sort_by) > > +{ > > + struct item *current = NULL; > > + struct item *result = NULL; > > + > > + if (!left) > > + return right; > > + if (!right) > > + return left; > > + > > + if (sort_by == BY_NAME) { > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > + result = left; > > + left = left->next; > > + } else { > > + result = right; > > + right = right->next; > > + } > > + } else { > > + if (sort_by == BY_ADDRESS) { > > + if (left->addr <= right->addr) { > > + result = left; > > + left = left->next; > > + } else { > > + result = right; > > + right = right->next; > > + } > > + } > > + } > > + > > + current = result; > > + > > + while (left && right) { > > + if (sort_by == BY_NAME) { > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > + current->next = left; > > + left = left->next; > > + } else { > > + current->next = right; > > + right = right->next; > > + } > > + } else { > > + if (sort_by == BY_ADDRESS) { > > + if (left->addr <= right->addr) { > > + current->next = left; > > + left = left->next; > > + } else { > > + current->next = right; > > + right = right->next; > > + } > > + } > > + } > > + > > + current = current->next; > > + } > > + > > + if (left) { > > + current->next = left; > > + } else { > > + if (right) > > + current->next = right; > > + } > > + > > + return result; > > +} > > + > > +struct item *merge_sort(struct item *head, int sort_by) > > +{ > > + struct item *right; > > + struct item *slow; > > + struct item *fast; > > + struct item *left; > > + > > + if (!head || !head->next) > > + return head; > > + > > + slow = head; > > + fast = head->next; > > + > > + while (fast && fast->next) { > > + slow = slow->next; > > + fast = fast->next->next; > > + } > > + > > + left = head; > > + right = slow->next; > > + slow->next = NULL; > > + > > + left = merge_sort(left, sort_by); > > + right = merge_sort(right, sort_by); > > + > > + return merge(left, right, sort_by); > > +} > > + > > +void sort_list_m(struct item **head, int sort_by) > > +{ > > + if (!(*head) || !((*head)->next)) > > + return; > > + > > + *head = merge_sort(*head, sort_by); > > +} > > + > > +int insert_after(struct item *list, const uint64_t search_addr, > > + const char *name, uint64_t addr, char stype) > > +{ > > + struct item *new_item; > > + struct item *current; > > + int ret = 0; > > + > > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list; > > + while (current) { > > + if (current->addr == search_addr) { > > + new_item = malloc(sizeof(struct item)); > > + if (!new_item) > > + return ret; > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > + new_item->addr = addr; > > + new_item->stype = stype; > > + new_item->next = current->next; > > + current->next = new_item; > > + ret = 1; > > + break; > > + } > > + current = current->next; > > + } > > + return ret; > > +} > > + > > +void free_items(struct item **head) > > +{ > > + struct item *app, *item_iterator = *head; > > + > > + while (item_iterator) { > > + app = item_iterator; > > + item_iterator = item_iterator->next; > > + free(app); > > + } > > + *head = NULL; > > +} > > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > > new file mode 100644 > > index 000000000000..b4891cb088ee > > --- /dev/null > > +++ b/scripts/kas_alias/item_list.h > > @@ -0,0 +1,26 @@ > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > +#ifndef ITEM_LIST_H > > +#define ITEM_LIST_H > > +#include <stdint.h> > > + > > +#define MAX_NAME_SIZE 256 > > +#define BY_ADDRESS 1 > > +#define BY_NAME 2 > > + > > +struct item { > > + char symb_name[MAX_NAME_SIZE]; > > + uint64_t addr; > > + char stype; > > + struct item *next; > > +}; > > + > > +void build_index(struct item *list); > > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr); > > +void sort_list(struct item **list, int sort_by); > > +struct item *merge(struct item *left, struct item *right, int sort_by); > > +struct item *merge_sort(struct item *head, int sort_by); > > +void sort_list_m(struct item **head, int sort_by); > > +int insert_after(struct item *list, const uint64_t search_addr, > > + const char *name, uint64_t addr, char stype); > > +void free_items(struct item **head); > > +#endif > > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > > new file mode 100644 > > index 000000000000..eb873b368023 > > --- /dev/null > > +++ b/scripts/kas_alias/kas_alias.c > > @@ -0,0 +1,186 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdio.h> > > +#include <stdlib.h> > > +#include <stdint.h> > > +#include <string.h> > > +#include <stdbool.h> > > +#include <stdarg.h> > > +#include <regex.h> > > + > > +#include "item_list.h" > > +#include "duplicates_list.h" > > + > > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \ > > + (((s)->stype) == 'd') || (((s)->stype) == 'D') || \ > > + (((s)->stype) == 'r') || (((s)->stype) == 'R')) > > +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > > +#else > > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > > +#endif > > +#define FNOMATCH 0 > > +#define FMATCH 1 > > +#define EREGEX 2 > > + > > +const char *ignore_regex[] = { > > + "^__cfi_.*$", // __cfi_ preamble > > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > > + "^_*TRACE_SYSTEM.*$", > > + "^__already_done\\.[0-9]+$", // Call a function once data > > + "^___tp_str\\.[0-9]+$", > > + "^___done\\.[0-9]+$", > > + "^__print_once\\.[0-9]+$", > > + "^_rs\\.[0-9]+$", > > + "^__compound_literal\\.[0-9]+$", > > + "^___once_key\\.[0-9]+$", > > + "^__func__\\.[0-9]+$", > > + "^__msg\\.[0-9]+$", > > + "^CSWTCH\\.[0-9]+$", > > + "^__flags\\.[0-9]+$", > > + "^__wkey.*$", > > + "^__mkey.*$", > > + "^__key.*$", > > +#endif > > + "^__pfx_.*$" // NOP-padding > > +}; > > + > > +int suffix_serial; > > + > > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > > +{ > > + va_list args; > > + > > + va_start(args, fmt); > > + if (verbose) > > + printf(fmt, args); > > + > > + va_end(args); > > +} > > + > > +static void create_suffix(const char *name, char *output_suffix) > > +{ > > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > > +} > > + > > +static int filter_symbols(char *symbol, const char **ignore_list, int regex_no) > > +{ > > + regex_t regex; > > + int res, i; > > + > > + for (i = 0; i < regex_no; i++) { > > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > > + if (res) > > + return -EREGEX; > > + > > + res = regexec(®ex, symbol, 0, NULL, 0); > > + regfree(®ex); > > + switch (res) { > > + case 0: > > + return FMATCH; > > + case REG_NOMATCH: > > + break; > > + default: > > + return -EREGEX; > > + } > > + > > + } > > + > > + return FNOMATCH; > > +} > > + > > +int main(int argc, char *argv[]) > > +{ > > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > > + struct duplicate_item *duplicate_iterator; > > + struct duplicate_item *duplicate; > > + struct item *head = {NULL}; > > + bool need_2_process = true; > > + struct item *last = {NULL}; > > + struct item *current; > > + int verbose_mode = 0; > > + uint64_t address; > > + FILE *fp; > > + int res; > > + > > + if (argc < 2 || argc > 3) { > > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > > + return 1; > > + } > > + > > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > > + verbose_mode = 1; > > + > > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > > + > > + fp = fopen(argv[1], "r"); > > + if (!fp) { > > + printf("Can't open input file.\n"); > > + return 1; > > + } > > + > > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > > + if (strstr(sym_name, "__alias__1") != NULL) { > > + if (verbose_mode && need_2_process) > > + printf("Already processed\n"); > > + need_2_process = false; > > + } > > + last = add_item(&last, sym_name, t, address); > > + if (!last) { > > + printf("Error in allocate memory\n"); > > + free_items(&head); > > + return 1; > > + } > > + > > + if (!head) > > + head = last; > > + } > > + > > + fclose(fp); > > + > > + if (need_2_process) { > > + verbose_msg(verbose_mode, "Sorting nm data\n"); > > + sort_list_m(&head, BY_NAME); > > + verbose_msg(verbose_mode, "Scanning nm data for duplicates\n"); > > + duplicate = find_duplicates(head); > > + if (!duplicate) { > > + printf("Error in duplicates list\n"); > > + return 1; > > + } > > + > > + verbose_msg(verbose_mode, "Applying suffixes\n"); > > + build_index(head); > > + duplicate_iterator = duplicate; > > + while (duplicate_iterator) { > > + res = filter_symbols(duplicate_iterator->original_item->symb_name, > > + ignore_regex, sizeof(ignore_regex) / > > + sizeof(ignore_regex[0])); > > + if ((res != FMATCH) && > > + SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { > > + if (res < 0) > > + return 1; > > + > > + create_suffix(duplicate_iterator->original_item->symb_name, > > + new_name); > > + if (!insert_after(head, duplicate_iterator->original_item->addr, > > + new_name, duplicate_iterator->original_item->addr, > > + duplicate_iterator->original_item->stype)) > > + return 1; > > + } > > + > > + duplicate_iterator = duplicate_iterator->next; > > + } > > + > > + sort_list_m(&head, BY_ADDRESS); > > + } > > + current = head; > > + while (current) { > > + printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name); > > + current = current->next; > > + } > > + > > + free_items(&head); > > + free_duplicates(&duplicate); > > + > > + return 0; > > +} > > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > > index a432b171be82..4e3dea3ac2df 100755 > > --- a/scripts/link-vmlinux.sh > > +++ b/scripts/link-vmlinux.sh > > @@ -161,7 +161,11 @@ kallsyms() > > fi > > > > info KSYMS ${2} > > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > > + ALIAS=".alias" > > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > > + fi > > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > > } > > > > # Perform one step in kallsyms generation, including temporary linking of >
On Tue, 18 Jul 2023 18:06:36 +0200 Alessandro Carminati <alessandro.carminati@gmail.com> wrote: > Hi Steven, > > > > > Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't > > give you any clue to what function you are actually attaching to. There's > > been other approaches that show module and/or file names. I know there's > > still some issues with getting those accepted, but I much rather have them > > than this! > > > > See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ > > > > Thanks for sharing your thoughts on the current symbol naming convention and > Nick Alcock's approach. > I see your point about "alias<some-random-number>" not providing much context > about the attached function. > However, my intention was only to create a means to probe all functions. We already have something that differentiates those functions. We have the address. For example, we have: ~# grep calipso_sock_getattr /proc/kallsyms ffffffff9b540a70 t calipso_sock_getattr ffffffff9b55cef0 T calipso_sock_getattr I can then add the second one to kprobe events: ~# echo 'p:sock_getattr 0xffffffff9b55cef0 sk=%di' > /sys/kernel/tracing/kprobe_events As we have this work around, I don't want to add something that gives us not much better. > > While other approaches, such as including module and file names, are more > informative, it appears that these proposals are encountering certain > challenges in gaining acceptance. Also, they do not fully address the > specific problem I wanted to tackle. > > In my upcoming proposal, I would like to maintain the conservative approach I > have followed so far while still achieving my original objective of making all > functions reachable from kprobe. They should already be, as you can attach kprobes via the address and not just the name. > > To date, I have been working on addressing all the necessary aspects from > outside the kernel source code. Essentially, I take the output of 'nm' and > refine it to provide more meaningful information. > > I am working in a V3, I am considering using `addr2line` to > determine the origin of each symbol. Initially, I avoided it because the > configuration dependency it introduces, specifically requiring DWARF > information such as CONFIG_DEBUG_INFO_DWARF4 or > CONFIG_DEBUG_INFO_DWARF5. > > However, after reading Luis' comment suggesting the use of > 'filename + line number', I realized that relying on DWARF information might > be the only way to achieve a satisfying result. > I will try that and see if it works... unless you know it will not already :-) Do you only need DWARF for the build process, or need to include it into kernel memory? > > Actually, I did a fast try using DWARF and addr2line, the strategy I'd like to > use in the v3, and the output looks like this: what do you guys think? > > $ ../kas_alias/main .tmp_vmlinux.kallsyms2.syms | grep name_show > ffffffff8100e5e0 t __pfx_pmu_name_show > ffffffff8100e5f0 t pmu_name_show > ffffffff810ed8e0 t __pfx_name_show > ffffffff810ed8f0 t name_show > ffffffff810ed8f0 t name_show@_kernel_irq_irqdesc_c_232 > ffffffff810ed950 t __pfx_chip_name_show > ffffffff810ed960 t chip_name_show > ffffffff810ed960 t chip_name_show@_kernel_irq_irqdesc_c_168 > ffffffff814cae80 t __pfx_name_show > ffffffff814cae90 t name_show > ffffffff814cae90 t name_show@_drivers_pnp_card_c_186 > ffffffff816006d0 t __pfx_name_show > ffffffff816006e0 t name_show > ffffffff816006e0 t name_show@_drivers_gpu_drm_i915_gt_sysfs_engines_c_26 > ffffffff81765460 t __pfx_name_show > ffffffff81765470 t name_show > ffffffff81765470 t name_show@_drivers_base_power_wakeup_stats_c_93 > ffffffff8189b340 t __pfx_name_show > [...] At build time you'll add the names and then include the above? -- Steve > > I want to emphasize that my goal is to find a solution that satisfies everyone > involved and addresses the concerns raised. > > Bests > Alessandro > > > -- Steve > > > > > > > Changes from v1: > > > - Integrated changes requested by Masami to exclude symbols with prefixes > > > "_cfi" and "_pfx". > > > - Introduced a small framework to handle patterns that need to be excluded > > > from the alias production. > > > - Excluded other symbols using the framework. > > > - Introduced the ability to discriminate between text and data symbols. > > > - Added two new config symbols in this version: CONFIG_KALLSYMS_ALIAS_DATA, > > > which allows data for data, and CONFIG_KALLSYMS_ALIAS_DATA_ALL, which > > > excludes all filters and provides an alias for each duplicated symbol. > > > > > > https://lore.kernel.org/all/20230711151925.1092080-1-alessandro.carminati@gmail.com/ > > > > > > Signed-off-by: Alessandro Carminati (Red Hat) <alessandro.carminati@gmail.com> > > > --- > > > init/Kconfig | 36 +++++ > > > scripts/Makefile | 4 + > > > scripts/kas_alias/Makefile | 4 + > > > scripts/kas_alias/duplicates_list.c | 70 +++++++++ > > > scripts/kas_alias/duplicates_list.h | 15 ++ > > > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++++++ > > > scripts/kas_alias/item_list.h | 26 ++++ > > > scripts/kas_alias/kas_alias.c | 186 ++++++++++++++++++++++ > > > scripts/link-vmlinux.sh | 6 +- > > > 9 files changed, 576 insertions(+), 1 deletion(-) > > > create mode 100644 scripts/kas_alias/Makefile > > > create mode 100644 scripts/kas_alias/duplicates_list.c > > > create mode 100644 scripts/kas_alias/duplicates_list.h > > > create mode 100644 scripts/kas_alias/item_list.c > > > create mode 100644 scripts/kas_alias/item_list.h > > > create mode 100644 scripts/kas_alias/kas_alias.c > > > > > > diff --git a/init/Kconfig b/init/Kconfig > > > index f7f65af4ee12..5ae7c613cdc4 100644 > > > --- a/init/Kconfig > > > +++ b/init/Kconfig > > > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > > > time constants, and no relocation pass is required at runtime to fix > > > up the entries based on the runtime load address of the kernel. > > > > > > +config KALLSYMS_ALIAS > > > + bool "Produces alias for duplicated symbols" if EXPERT > > > + depends on KALLSYMS > > > + help > > > + It is not uncommon for drivers or modules related to similar > > > + peripherals to have symbols with the exact same name. > > > + While this is not a problem for the kernel's binary itself, it > > > + becomes an issue when attempting to trace or probe specific > > > + functions using infrastructure like ftrace or kprobe. > > > + > > > + This option addresses this challenge by extending the symbol names > > > + with unique suffixes during the kernel build process. > > > + The newly created aliases for these duplicated symbols are unique > > > + names that can be fed to the ftrace sysfs interface. By doing so, it > > > + enables previously unreachable symbols to be probed. > > > + > > > +config CONFIG_KALLSYMS_ALIAS_DATA > > > + bool "Produces alias also for data" > > > + depends on KALLSYMS_ALIAS > > > + help > > > + Sometimes it can be useful to refer to data. In live patch scenarios, > > > + you may find yourself needing to use symbols that are shared with > > > + other functions. Since symbols face the same issue as functions, this > > > + option allows you to create aliases for data as well. > > > + > > > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > + bool "Removes all filter when producing data alias" > > > + depends on CONFIG_KALLSYMS_ALIAS_DATA > > > + help > > > + When selecting data aliases, not all symbols are included in the set > > > + This is because many symbols are unlikely to be used. If you choose > > > + to have an alias for all data symbols, be aware that it will > > > + significantly increase the size. > > > + > > > + If unsure, say N. > > > + > > > # end of the "standard kernel features (expert users)" menu > > > > > > # syscall, maps, verifier > > > diff --git a/scripts/Makefile b/scripts/Makefile > > > index 32b6ba722728..65fafe17cfe5 100644 > > > --- a/scripts/Makefile > > > +++ b/scripts/Makefile > > > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > > > > > # Let clean descend into subdirs > > > subdir- += basic dtc gdb kconfig mod > > > + > > > +# KALLSyms alias > > > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > + > > > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > > > new file mode 100644 > > > index 000000000000..523fa3441013 > > > --- /dev/null > > > +++ b/scripts/kas_alias/Makefile > > > @@ -0,0 +1,4 @@ > > > +# SPDX-License-Identifier: GPL-2.0 > > > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > + > > > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o > > > diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c > > > new file mode 100644 > > > index 000000000000..e7a3d2917937 > > > --- /dev/null > > > +++ b/scripts/kas_alias/duplicates_list.c > > > @@ -0,0 +1,70 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdint.h> > > > +#include <stdio.h> > > > +#include <string.h> > > > +#include <stdlib.h> > > > +#include <stdbool.h> > > > + > > > +#include "item_list.h" > > > +#include "duplicates_list.h" > > > + > > > +struct duplicate_item *find_duplicates(struct item *list) > > > +{ > > > + struct duplicate_item *current_duplicate = NULL; > > > + struct duplicate_item *duplicates = NULL; > > > + struct duplicate_item *new_duplicate; > > > + struct item *current_item = list; > > > + bool prev_was_duplicate = false; > > > + struct item *prev_item = NULL; > > > + > > > + while (current_item) { > > > + if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) || > > > + prev_was_duplicate) { > > > + if (!duplicates) { > > > + duplicates = malloc(sizeof(struct duplicate_item)); > > > + if (!duplicates) > > > + return NULL; > > > + > > > + duplicates->original_item = prev_item; > > > + duplicates->next = NULL; > > > + current_duplicate = duplicates; > > > + } else { > > > + new_duplicate = malloc(sizeof(struct duplicate_item)); > > > + if (!new_duplicate) { > > > + free_duplicates(&duplicates); > > > + return NULL; > > > + } > > > + > > > + new_duplicate->original_item = prev_item; > > > + new_duplicate->next = NULL; > > > + current_duplicate->next = new_duplicate; > > > + current_duplicate = new_duplicate; > > > + > > > + if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) && > > > + (prev_was_duplicate)) > > > + prev_was_duplicate = false; > > > + else > > > + prev_was_duplicate = true; > > > + } > > > + } > > > + > > > + prev_item = current_item; > > > + current_item = current_item->next; > > > + } > > > + > > > + return duplicates; > > > +} > > > + > > > +void free_duplicates(struct duplicate_item **duplicates) > > > +{ > > > + struct duplicate_item *duplicates_iterator = *duplicates; > > > + struct duplicate_item *app; > > > + > > > + while (duplicates_iterator) { > > > + app = duplicates_iterator; > > > + duplicates_iterator = duplicates_iterator->next; > > > + free(app); > > > + } > > > + > > > + *duplicates = NULL; > > > +} > > > diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h > > > new file mode 100644 > > > index 000000000000..76aa73e584bc > > > --- /dev/null > > > +++ b/scripts/kas_alias/duplicates_list.h > > > @@ -0,0 +1,15 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef DUPLICATES_LIST_H > > > +#define DUPLICATES_LIST_H > > > + > > > +#include "item_list.h" > > > + > > > +struct duplicate_item { > > > + struct item *original_item; > > > + struct duplicate_item *next; > > > +}; > > > + > > > +struct duplicate_item *find_duplicates(struct item *list); > > > +void free_duplicates(struct duplicate_item **duplicates); > > > + > > > +#endif > > > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > > > new file mode 100644 > > > index 000000000000..7c9d5aecca9a > > > --- /dev/null > > > +++ b/scripts/kas_alias/item_list.c > > > @@ -0,0 +1,230 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <stdint.h> > > > +#include <string.h> > > > +#include <stdbool.h> > > > +#include <assert.h> > > > +#include "item_list.h" > > > + > > > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > > > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > > > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > > > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0) > > > + > > > +struct item *list_index[96] = {0}; > > > + > > > +void build_index(struct item *list) > > > +{ > > > + char current_first_letter = ' '; > > > + struct item *current = list; > > > + > > > + while (current) { > > > + if (current->symb_name[0] != current_first_letter) { > > > + current_first_letter = current->symb_name[0]; > > > + list_index[current_first_letter - 32] = current; > > > + } > > > + current = current->next; > > > + } > > > +} > > > + > > > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr) > > > +{ > > > + struct item *new_item; > > > + struct item *current; > > > + > > > + new_item = malloc(sizeof(struct item)); > > > + if (!new_item) > > > + return NULL; > > > + > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > + new_item->addr = addr; > > > + new_item->stype = stype; > > > + new_item->next = NULL; > > > + > > > + if (!(*list)) { > > > + *list = new_item; > > > + } else { > > > + current = *list; > > > + while (current->next) > > > + current = current->next; > > > + > > > + current->next = new_item; > > > + } > > > + return new_item; > > > +} > > > + > > > +void sort_list(struct item **list, int sort_by) > > > +{ > > > + struct item *current = *list; > > > + struct item *sorted = NULL; > > > + struct item *next_item; > > > + struct item *temp; > > > + > > > + if (!(*list) || !((*list)->next)) > > > + return; > > > + > > > + while (current) { > > > + next_item = current->next; > > > + if (!sorted || > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) || > > > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > > > + current->next = sorted; > > > + sorted = current; > > > + } else { > > > + temp = sorted; > > > + while (temp->next && > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) || > > > + CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) > > > + temp = temp->next; > > > + > > > + current->next = temp->next; > > > + temp->next = current; > > > + } > > > + current = next_item; > > > + } > > > + > > > + *list = sorted; > > > +} > > > + > > > +struct item *merge(struct item *left, struct item *right, int sort_by) > > > +{ > > > + struct item *current = NULL; > > > + struct item *result = NULL; > > > + > > > + if (!left) > > > + return right; > > > + if (!right) > > > + return left; > > > + > > > + if (sort_by == BY_NAME) { > > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > > + result = left; > > > + left = left->next; > > > + } else { > > > + result = right; > > > + right = right->next; > > > + } > > > + } else { > > > + if (sort_by == BY_ADDRESS) { > > > + if (left->addr <= right->addr) { > > > + result = left; > > > + left = left->next; > > > + } else { > > > + result = right; > > > + right = right->next; > > > + } > > > + } > > > + } > > > + > > > + current = result; > > > + > > > + while (left && right) { > > > + if (sort_by == BY_NAME) { > > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > > + current->next = left; > > > + left = left->next; > > > + } else { > > > + current->next = right; > > > + right = right->next; > > > + } > > > + } else { > > > + if (sort_by == BY_ADDRESS) { > > > + if (left->addr <= right->addr) { > > > + current->next = left; > > > + left = left->next; > > > + } else { > > > + current->next = right; > > > + right = right->next; > > > + } > > > + } > > > + } > > > + > > > + current = current->next; > > > + } > > > + > > > + if (left) { > > > + current->next = left; > > > + } else { > > > + if (right) > > > + current->next = right; > > > + } > > > + > > > + return result; > > > +} > > > + > > > +struct item *merge_sort(struct item *head, int sort_by) > > > +{ > > > + struct item *right; > > > + struct item *slow; > > > + struct item *fast; > > > + struct item *left; > > > + > > > + if (!head || !head->next) > > > + return head; > > > + > > > + slow = head; > > > + fast = head->next; > > > + > > > + while (fast && fast->next) { > > > + slow = slow->next; > > > + fast = fast->next->next; > > > + } > > > + > > > + left = head; > > > + right = slow->next; > > > + slow->next = NULL; > > > + > > > + left = merge_sort(left, sort_by); > > > + right = merge_sort(right, sort_by); > > > + > > > + return merge(left, right, sort_by); > > > +} > > > + > > > +void sort_list_m(struct item **head, int sort_by) > > > +{ > > > + if (!(*head) || !((*head)->next)) > > > + return; > > > + > > > + *head = merge_sort(*head, sort_by); > > > +} > > > + > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > + const char *name, uint64_t addr, char stype) > > > +{ > > > + struct item *new_item; > > > + struct item *current; > > > + int ret = 0; > > > + > > > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list; > > > + while (current) { > > > + if (current->addr == search_addr) { > > > + new_item = malloc(sizeof(struct item)); > > > + if (!new_item) > > > + return ret; > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > + new_item->addr = addr; > > > + new_item->stype = stype; > > > + new_item->next = current->next; > > > + current->next = new_item; > > > + ret = 1; > > > + break; > > > + } > > > + current = current->next; > > > + } > > > + return ret; > > > +} > > > + > > > +void free_items(struct item **head) > > > +{ > > > + struct item *app, *item_iterator = *head; > > > + > > > + while (item_iterator) { > > > + app = item_iterator; > > > + item_iterator = item_iterator->next; > > > + free(app); > > > + } > > > + *head = NULL; > > > +} > > > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > > > new file mode 100644 > > > index 000000000000..b4891cb088ee > > > --- /dev/null > > > +++ b/scripts/kas_alias/item_list.h > > > @@ -0,0 +1,26 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef ITEM_LIST_H > > > +#define ITEM_LIST_H > > > +#include <stdint.h> > > > + > > > +#define MAX_NAME_SIZE 256 > > > +#define BY_ADDRESS 1 > > > +#define BY_NAME 2 > > > + > > > +struct item { > > > + char symb_name[MAX_NAME_SIZE]; > > > + uint64_t addr; > > > + char stype; > > > + struct item *next; > > > +}; > > > + > > > +void build_index(struct item *list); > > > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr); > > > +void sort_list(struct item **list, int sort_by); > > > +struct item *merge(struct item *left, struct item *right, int sort_by); > > > +struct item *merge_sort(struct item *head, int sort_by); > > > +void sort_list_m(struct item **head, int sort_by); > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > + const char *name, uint64_t addr, char stype); > > > +void free_items(struct item **head); > > > +#endif > > > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > > > new file mode 100644 > > > index 000000000000..eb873b368023 > > > --- /dev/null > > > +++ b/scripts/kas_alias/kas_alias.c > > > @@ -0,0 +1,186 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <stdint.h> > > > +#include <string.h> > > > +#include <stdbool.h> > > > +#include <stdarg.h> > > > +#include <regex.h> > > > + > > > +#include "item_list.h" > > > +#include "duplicates_list.h" > > > + > > > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > > > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \ > > > + (((s)->stype) == 'd') || (((s)->stype) == 'D') || \ > > > + (((s)->stype) == 'r') || (((s)->stype) == 'R')) > > > +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > > > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > > > +#else > > > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > > > +#endif > > > +#define FNOMATCH 0 > > > +#define FMATCH 1 > > > +#define EREGEX 2 > > > + > > > +const char *ignore_regex[] = { > > > + "^__cfi_.*$", // __cfi_ preamble > > > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > + "^_*TRACE_SYSTEM.*$", > > > + "^__already_done\\.[0-9]+$", // Call a function once data > > > + "^___tp_str\\.[0-9]+$", > > > + "^___done\\.[0-9]+$", > > > + "^__print_once\\.[0-9]+$", > > > + "^_rs\\.[0-9]+$", > > > + "^__compound_literal\\.[0-9]+$", > > > + "^___once_key\\.[0-9]+$", > > > + "^__func__\\.[0-9]+$", > > > + "^__msg\\.[0-9]+$", > > > + "^CSWTCH\\.[0-9]+$", > > > + "^__flags\\.[0-9]+$", > > > + "^__wkey.*$", > > > + "^__mkey.*$", > > > + "^__key.*$", > > > +#endif > > > + "^__pfx_.*$" // NOP-padding > > > +}; > > > + > > > +int suffix_serial; > > > + > > > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > > > +{ > > > + va_list args; > > > + > > > + va_start(args, fmt); > > > + if (verbose) > > > + printf(fmt, args); > > > + > > > + va_end(args); > > > +} > > > + > > > +static void create_suffix(const char *name, char *output_suffix) > > > +{ > > > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > > > +} > > > + > > > +static int filter_symbols(char *symbol, const char **ignore_list, int regex_no) > > > +{ > > > + regex_t regex; > > > + int res, i; > > > + > > > + for (i = 0; i < regex_no; i++) { > > > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > > > + if (res) > > > + return -EREGEX; > > > + > > > + res = regexec(®ex, symbol, 0, NULL, 0); > > > + regfree(®ex); > > > + switch (res) { > > > + case 0: > > > + return FMATCH; > > > + case REG_NOMATCH: > > > + break; > > > + default: > > > + return -EREGEX; > > > + } > > > + > > > + } > > > + > > > + return FNOMATCH; > > > +} > > > + > > > +int main(int argc, char *argv[]) > > > +{ > > > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > > > + struct duplicate_item *duplicate_iterator; > > > + struct duplicate_item *duplicate; > > > + struct item *head = {NULL}; > > > + bool need_2_process = true; > > > + struct item *last = {NULL}; > > > + struct item *current; > > > + int verbose_mode = 0; > > > + uint64_t address; > > > + FILE *fp; > > > + int res; > > > + > > > + if (argc < 2 || argc > 3) { > > > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > > > + return 1; > > > + } > > > + > > > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > > > + verbose_mode = 1; > > > + > > > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > > > + > > > + fp = fopen(argv[1], "r"); > > > + if (!fp) { > > > + printf("Can't open input file.\n"); > > > + return 1; > > > + } > > > + > > > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > > > + if (strstr(sym_name, "__alias__1") != NULL) { > > > + if (verbose_mode && need_2_process) > > > + printf("Already processed\n"); > > > + need_2_process = false; > > > + } > > > + last = add_item(&last, sym_name, t, address); > > > + if (!last) { > > > + printf("Error in allocate memory\n"); > > > + free_items(&head); > > > + return 1; > > > + } > > > + > > > + if (!head) > > > + head = last; > > > + } > > > + > > > + fclose(fp); > > > + > > > + if (need_2_process) { > > > + verbose_msg(verbose_mode, "Sorting nm data\n"); > > > + sort_list_m(&head, BY_NAME); > > > + verbose_msg(verbose_mode, "Scanning nm data for duplicates\n"); > > > + duplicate = find_duplicates(head); > > > + if (!duplicate) { > > > + printf("Error in duplicates list\n"); > > > + return 1; > > > + } > > > + > > > + verbose_msg(verbose_mode, "Applying suffixes\n"); > > > + build_index(head); > > > + duplicate_iterator = duplicate; > > > + while (duplicate_iterator) { > > > + res = filter_symbols(duplicate_iterator->original_item->symb_name, > > > + ignore_regex, sizeof(ignore_regex) / > > > + sizeof(ignore_regex[0])); > > > + if ((res != FMATCH) && > > > + SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { > > > + if (res < 0) > > > + return 1; > > > + > > > + create_suffix(duplicate_iterator->original_item->symb_name, > > > + new_name); > > > + if (!insert_after(head, duplicate_iterator->original_item->addr, > > > + new_name, duplicate_iterator->original_item->addr, > > > + duplicate_iterator->original_item->stype)) > > > + return 1; > > > + } > > > + > > > + duplicate_iterator = duplicate_iterator->next; > > > + } > > > + > > > + sort_list_m(&head, BY_ADDRESS); > > > + } > > > + current = head; > > > + while (current) { > > > + printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name); > > > + current = current->next; > > > + } > > > + > > > + free_items(&head); > > > + free_duplicates(&duplicate); > > > + > > > + return 0; > > > +} > > > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > > > index a432b171be82..4e3dea3ac2df 100755 > > > --- a/scripts/link-vmlinux.sh > > > +++ b/scripts/link-vmlinux.sh > > > @@ -161,7 +161,11 @@ kallsyms() > > > fi > > > > > > info KSYMS ${2} > > > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > > > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > > > + ALIAS=".alias" > > > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > > > + fi > > > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > > > } > > > > > > # Perform one step in kallsyms generation, including temporary linking of > >
[Reply concocted by hand out of lore.kernel.org archives because this entire thread has seemingly been mis-sorted somewhere into the stygian blackness of my mail spool and I can't find it anywhere: apologies if I messed anything up] On Mon, Jul 17, 2023 at 10:52:40AM -0400, Steven Rostedt wrote: >> Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't >> give you any clue to what function you are actually attaching to. > > Agreed. Agreed, and not *only* because I have a patch series that I think does better. I'm also a consumer of this stuff (that's why I wrote the patch) and I'm afraid I couldn't do anything with a bunch of random numbers. What user interface are we supposed to show people? All it tells you is that two symbols are different, not which is which or where they come from. This seems to be useful for the one case of "look up a symbol out of a trace log and then filter by it or re-trace using it": no other use case seems possible, because a number is just so little information. I'm also hoping that people can say *before they run a trace* "I know I'm looking at a problem with cgroups, I want to look at blah@cgroup/cgroup.o" without having to know beforehand that this is blah__alias_19404. Actually, you probably *can* get something else out of plain numbers -- by looking at the aliased symbol name you got back, then looking at /proc/kallmodsyms to see what other symbols it's textually close to, guessing what object file they're in from their names and then grepping around for what module that might be related to. i.e. doing by eye-and-grep what my patch series is doing automatically. A further problem: doing this using aliases in particular requires modifying the tracers to ensure that they always report the aliased symbol, not the non-aliased one at the same address. As it is, I suspect a lot of them will either report whichever symbol comes first, or pick one at random, or even try to disambiguate them themselves a second time, leading to blah__alias_19404@something (double disambiguation) or in some cases they'll simply hang. You might be lucky and see them pick the longest aliased name, or whicheer one comes last, I suppose. I think all the hanging tracers have been fixed in the last couple of years but they certainly *used* to exist. This is of course a bug in those tools, but it is easily overlookable when only a few symbols are aliases: less so after this patch. Nonetheless, it is unarguable that this series probably requires auditing (all?) kallsyms users, at least to be sure they don't misbehave. (Mine doesn't need any of that, but only because I'm using a new /proc file for that exact reason. People migrating to it would also need adjustment.) I know I've got too many concerns in one mail by now, but I'm *also* a bit concerned about the space requirements of this: kallsyms's name compression won't help compress all those textual numbers much. It's certainly higher space requirements than my approach: it's just more hidden because it's lumped into the existing kallsyms tables. Alessandro: I'll have a look through your patch: after dumping on it like this I think I have to. Alternate approaches are always worth looking at :) >> There's >> been other approaches that show module and/or file names. I know there's > > still some issues with getting those accepted, but I much rather have them > > than this! > > > > See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ > > Yes, please coordinate with Nick and review each other's work, now we > have two separate efforts with different reasons but hopefully we'll > come back with one unified solution. Yes indeed! I am getting back to this very soon (a few days), after a prolonged hiatus getting USDT probing working in DTrace V2. Sorry about that. (I suspect you wanted a break from my cover letters too!) I think what my series really needs is someone capable of writing less convoluted English, because my cover letter skills are still not brilliant. > Please Cc live-patching also, as they had suggested before just to > provide the file filename + line number, that'll make it even more > valuable. That would be amazing! Nothing obvious occurs to me though. My series could easily give objfile + address offset for every symbol with little more overhead than we have now (we'd need to encode more partial objfile names, that's all, low tens of KiB) -- the problem is translating that address offset into a line number without a cripplingly expensive storage-per-symbol lookup table. I mean if storing the size of each symbol is too expensive (and it is, costing hundreds of KiB), storing an offset->line number mapping probably would be too! I'll think on it, maybe a trick to avoid representing every offset individually will occur to me. (If any occurs to anyone else, I'd be fascinated.)
From: Nick Alcock <nick.alcock@oracle.com> Date: Wed, 19 Jul 2023 12:12:06 +0100 > [Reply concocted by hand out of lore.kernel.org archives because this > entire thread has seemingly been mis-sorted somewhere into the stygian > blackness of my mail spool and I can't find it anywhere: apologies if I > messed anything up] > > On Mon, Jul 17, 2023 at 10:52:40AM -0400, Steven Rostedt wrote: >>> Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't >>> give you any clue to what function you are actually attaching to. >> >> Agreed. > > Agreed, and not *only* because I have a patch series that I think does > better. I'm also a consumer of this stuff (that's why I wrote the patch) > and I'm afraid I couldn't do anything with a bunch of random numbers. > What user interface are we supposed to show people? All it tells you is > that two symbols are different, not which is which or where they come > from. > > This seems to be useful for the one case of "look up a symbol out of a > trace log and then filter by it or re-trace using it": no other use case > seems possible, because a number is just so little information. I'm also > hoping that people can say *before they run a trace* "I know I'm looking > at a problem with cgroups, I want to look at blah@cgroup/cgroup.o" > without having to know beforehand that this is blah__alias_19404. > > Actually, you probably *can* get something else out of plain numbers -- > by looking at the aliased symbol name you got back, then looking at > /proc/kallmodsyms to see what other symbols it's textually close to, > guessing what object file they're in from their names and then grepping > around for what module that might be related to. i.e. doing by > eye-and-grep what my patch series is doing automatically. > > > A further problem: doing this using aliases in particular requires > modifying the tracers to ensure that they always report the aliased > symbol, not the non-aliased one at the same address. As it is, I suspect > a lot of them will either report whichever symbol comes first, or pick > one at random, or even try to disambiguate them themselves a second > time, leading to blah__alias_19404@something (double disambiguation) or > in some cases they'll simply hang. You might be lucky and see them pick > the longest aliased name, or whicheer one comes last, I suppose. > > I think all the hanging tracers have been fixed in the last couple of > years but they certainly *used* to exist. This is of course a bug in > those tools, but it is easily overlookable when only a few symbols are > aliases: less so after this patch. Nonetheless, it is unarguable that > this series probably requires auditing (all?) kallsyms users, at least > to be sure they don't misbehave. (Mine doesn't need any of that, but > only because I'm using a new /proc file for that exact reason. People > migrating to it would also need adjustment.) > > I know I've got too many concerns in one mail by now, but I'm *also* a > bit concerned about the space requirements of this: kallsyms's name > compression won't help compress all those textual numbers much. It's > certainly higher space requirements than my approach: it's just more > hidden because it's lumped into the existing kallsyms tables. > > > Alessandro: I'll have a look through your patch: after dumping on it > like this I think I have to. Alternate approaches are always worth > looking at :) > >>> There's >>> been other approaches that show module and/or file names. I know there's >>> still some issues with getting those accepted, but I much rather have them >>> than this! >>> >>> See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ >> >> Yes, please coordinate with Nick and review each other's work, now we >> have two separate efforts with different reasons but hopefully we'll Three efforts[0] :D Mine went unnoticed unfortunately, so I switched to other projects then. My idea was to give relative path from the kernel root to the objfile, as we have a good bunch of non-unique "filename + symbol name" pairs. >> come back with one unified solution. > > Yes indeed! I am getting back to this very soon (a few days), after a > prolonged hiatus getting USDT probing working in DTrace V2. Sorry about > that. (I suspect you wanted a break from my cover letters too!) > > I think what my series really needs is someone capable of writing less > convoluted English, because my cover letter skills are still not > brilliant. > >> Please Cc live-patching also, as they had suggested before just to >> provide the file filename + line number, that'll make it even more >> valuable. > > That would be amazing! Nothing obvious occurs to me though. > > My series could easily give objfile + address offset for every symbol > with little more overhead than we have now (we'd need to encode more > partial objfile names, that's all, low tens of KiB) -- the problem is > translating that address offset into a line number without a cripplingly > expensive storage-per-symbol lookup table. I mean if storing the size of > each symbol is too expensive (and it is, costing hundreds of KiB), > storing an offset->line number mapping probably would be too! > > I'll think on it, maybe a trick to avoid representing every offset > individually will occur to me. (If any occurs to anyone else, I'd be > fascinated.) > [0] https://lore.kernel.org/linux-kbuild/20220818115306.1109642-1-alexandr.lobakin@intel.com Thanks, Olek
On Tue, 18 Jul 2023 18:06:36 +0200 Alessandro Carminati <alessandro.carminati@gmail.com> wrote: > Hi Steven, > > Il giorno lun 17 lug 2023 alle ore 16:52 Steven Rostedt > <rostedt@goodmis.org> ha scritto: > > > > On Fri, 14 Jul 2023 15:03:26 +0000 > > Alessandro Carminati <alessandro.carminati@gmail.com> wrote: > > > > > It is not uncommon for drivers or modules related to similar peripherals > > > to have symbols with the exact same name. > > > While this is not a problem for the kernel's binary itself, it becomes an > > > issue when attempting to trace or probe specific functions using > > > infrastructure like ftrace or kprobe. > > > > > > The tracing subsystem relies on the `nm -n vmlinux` output, which provides > > > symbol information from the kernel's ELF binary. However, when multiple > > > symbols share the same name, the standard nm output does not differentiate > > > between them. This can lead to confusion and difficulty when trying to > > > probe the intended symbol. > > > > > > ~ # cat /proc/kallsyms | grep " name_show" > > > ffffffff8c4f76d0 t name_show > > > ffffffff8c9cccb0 t name_show > > > ffffffff8cb0ac20 t name_show > > > ffffffff8cc728c0 t name_show > > > ffffffff8ce0efd0 t name_show > > > ffffffff8ce126c0 t name_show > > > ffffffff8ce1dd20 t name_show > > > ffffffff8ce24e70 t name_show > > > ffffffff8d1104c0 t name_show > > > ffffffff8d1fe480 t name_show > > > > > > **kas_alias** addresses this challenge by extending the symbol names with > > > unique suffixes during the kernel build process. > > > The newly created aliases for these duplicated symbols are unique names > > > that can be fed to the ftracefs interface. By doing so, it enables > > > previously unreachable symbols to be probed. > > > > > > ~ # cat /proc/kallsyms | grep " name_show" > > > ffffffff974f76d0 t name_show > > > ffffffff974f76d0 t name_show__alias__6340 > > > ffffffff979cccb0 t name_show > > > ffffffff979cccb0 t name_show__alias__6341 > > > ffffffff97b0ac20 t name_show > > > ffffffff97b0ac20 t name_show__alias__6342 > > > ffffffff97c728c0 t name_show > > > ffffffff97c728c0 t name_show__alias__6343 > > > ffffffff97e0efd0 t name_show > > > ffffffff97e0efd0 t name_show__alias__6344 > > > ffffffff97e126c0 t name_show > > > ffffffff97e126c0 t name_show__alias__6345 > > > ffffffff97e1dd20 t name_show > > > ffffffff97e1dd20 t name_show__alias__6346 > > > ffffffff97e24e70 t name_show > > > ffffffff97e24e70 t name_show__alias__6347 > > > ffffffff981104c0 t name_show > > > ffffffff981104c0 t name_show__alias__6348 > > > ffffffff981fe480 t name_show > > > ffffffff981fe480 t name_show__alias__6349 > > > ~ # echo "p:kprobes/evnt1 name_show__alias__6349" \ > > > > >/sys/kernel/tracing/kprobe_events > > > ~ # cat /sys/kernel/tracing/kprobe_events > > > p:kprobes/evnt1 name_show__alias__6349 > > > > > > > Honestly, I think the "_alias_<some-random-number>" is useless. It doesn't > > give you any clue to what function you are actually attaching to. There's > > been other approaches that show module and/or file names. I know there's > > still some issues with getting those accepted, but I much rather have them > > than this! > > > > See: https://lore.kernel.org/all/20221205163157.269335-1-nick.alcock@oracle.com/ > > > > Thanks for sharing your thoughts on the current symbol naming convention and > Nick Alcock's approach. > I see your point about "alias<some-random-number>" not providing much context > about the attached function. > However, my intention was only to create a means to probe all functions. FYI: you can use "_text+OFFSET" to probe any function by kprobe event. Actually `perf probe` already does that. > While other approaches, such as including module and file names, are more > informative, it appears that these proposals are encountering certain > challenges in gaining acceptance. Also, they do not fully address the > specific problem I wanted to tackle. > > In my upcoming proposal, I would like to maintain the conservative approach I > have followed so far while still achieving my original objective of making all > functions reachable from kprobe. > > To date, I have been working on addressing all the necessary aspects from > outside the kernel source code. Essentially, I take the output of 'nm' and > refine it to provide more meaningful information. > > I am working in a V3, I am considering using `addr2line` to > determine the origin of each symbol. Initially, I avoided it because the > configuration dependency it introduces, specifically requiring DWARF > information such as CONFIG_DEBUG_INFO_DWARF4 or > CONFIG_DEBUG_INFO_DWARF5. Yeah, addr2line will help you to find line number and filename easily. I think that is a good way to do that. > However, after reading Luis' comment suggesting the use of > 'filename + line number', I realized that relying on DWARF information might > be the only way to achieve a satisfying result. > I will try that and see if it works... unless you know it will not already :-) > > Actually, I did a fast try using DWARF and addr2line, the strategy I'd like to > use in the v3, and the output looks like this: what do you guys think? > > $ ../kas_alias/main .tmp_vmlinux.kallsyms2.syms | grep name_show > ffffffff8100e5e0 t __pfx_pmu_name_show > ffffffff8100e5f0 t pmu_name_show > ffffffff810ed8e0 t __pfx_name_show > ffffffff810ed8f0 t name_show > ffffffff810ed8f0 t name_show@_kernel_irq_irqdesc_c_232 > ffffffff810ed950 t __pfx_chip_name_show > ffffffff810ed960 t chip_name_show > ffffffff810ed960 t chip_name_show@_kernel_irq_irqdesc_c_168 > ffffffff814cae80 t __pfx_name_show > ffffffff814cae90 t name_show > ffffffff814cae90 t name_show@_drivers_pnp_card_c_186 > ffffffff816006d0 t __pfx_name_show > ffffffff816006e0 t name_show > ffffffff816006e0 t name_show@_drivers_gpu_drm_i915_gt_sysfs_engines_c_26 > ffffffff81765460 t __pfx_name_show > ffffffff81765470 t name_show > ffffffff81765470 t name_show@_drivers_base_power_wakeup_stats_c_93 > ffffffff8189b340 t __pfx_name_show > [...] Looks good to me. Thank you! > > I want to emphasize that my goal is to find a solution that satisfies everyone > involved and addresses the concerns raised. > > Bests > Alessandro > > > -- Steve > > > > > > > Changes from v1: > > > - Integrated changes requested by Masami to exclude symbols with prefixes > > > "_cfi" and "_pfx". > > > - Introduced a small framework to handle patterns that need to be excluded > > > from the alias production. > > > - Excluded other symbols using the framework. > > > - Introduced the ability to discriminate between text and data symbols. > > > - Added two new config symbols in this version: CONFIG_KALLSYMS_ALIAS_DATA, > > > which allows data for data, and CONFIG_KALLSYMS_ALIAS_DATA_ALL, which > > > excludes all filters and provides an alias for each duplicated symbol. > > > > > > https://lore.kernel.org/all/20230711151925.1092080-1-alessandro.carminati@gmail.com/ > > > > > > Signed-off-by: Alessandro Carminati (Red Hat) <alessandro.carminati@gmail.com> > > > --- > > > init/Kconfig | 36 +++++ > > > scripts/Makefile | 4 + > > > scripts/kas_alias/Makefile | 4 + > > > scripts/kas_alias/duplicates_list.c | 70 +++++++++ > > > scripts/kas_alias/duplicates_list.h | 15 ++ > > > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++++++ > > > scripts/kas_alias/item_list.h | 26 ++++ > > > scripts/kas_alias/kas_alias.c | 186 ++++++++++++++++++++++ > > > scripts/link-vmlinux.sh | 6 +- > > > 9 files changed, 576 insertions(+), 1 deletion(-) > > > create mode 100644 scripts/kas_alias/Makefile > > > create mode 100644 scripts/kas_alias/duplicates_list.c > > > create mode 100644 scripts/kas_alias/duplicates_list.h > > > create mode 100644 scripts/kas_alias/item_list.c > > > create mode 100644 scripts/kas_alias/item_list.h > > > create mode 100644 scripts/kas_alias/kas_alias.c > > > > > > diff --git a/init/Kconfig b/init/Kconfig > > > index f7f65af4ee12..5ae7c613cdc4 100644 > > > --- a/init/Kconfig > > > +++ b/init/Kconfig > > > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > > > time constants, and no relocation pass is required at runtime to fix > > > up the entries based on the runtime load address of the kernel. > > > > > > +config KALLSYMS_ALIAS > > > + bool "Produces alias for duplicated symbols" if EXPERT > > > + depends on KALLSYMS > > > + help > > > + It is not uncommon for drivers or modules related to similar > > > + peripherals to have symbols with the exact same name. > > > + While this is not a problem for the kernel's binary itself, it > > > + becomes an issue when attempting to trace or probe specific > > > + functions using infrastructure like ftrace or kprobe. > > > + > > > + This option addresses this challenge by extending the symbol names > > > + with unique suffixes during the kernel build process. > > > + The newly created aliases for these duplicated symbols are unique > > > + names that can be fed to the ftrace sysfs interface. By doing so, it > > > + enables previously unreachable symbols to be probed. > > > + > > > +config CONFIG_KALLSYMS_ALIAS_DATA > > > + bool "Produces alias also for data" > > > + depends on KALLSYMS_ALIAS > > > + help > > > + Sometimes it can be useful to refer to data. In live patch scenarios, > > > + you may find yourself needing to use symbols that are shared with > > > + other functions. Since symbols face the same issue as functions, this > > > + option allows you to create aliases for data as well. > > > + > > > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > + bool "Removes all filter when producing data alias" > > > + depends on CONFIG_KALLSYMS_ALIAS_DATA > > > + help > > > + When selecting data aliases, not all symbols are included in the set > > > + This is because many symbols are unlikely to be used. If you choose > > > + to have an alias for all data symbols, be aware that it will > > > + significantly increase the size. > > > + > > > + If unsure, say N. > > > + > > > # end of the "standard kernel features (expert users)" menu > > > > > > # syscall, maps, verifier > > > diff --git a/scripts/Makefile b/scripts/Makefile > > > index 32b6ba722728..65fafe17cfe5 100644 > > > --- a/scripts/Makefile > > > +++ b/scripts/Makefile > > > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > > > > > # Let clean descend into subdirs > > > subdir- += basic dtc gdb kconfig mod > > > + > > > +# KALLSyms alias > > > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > + > > > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > > > new file mode 100644 > > > index 000000000000..523fa3441013 > > > --- /dev/null > > > +++ b/scripts/kas_alias/Makefile > > > @@ -0,0 +1,4 @@ > > > +# SPDX-License-Identifier: GPL-2.0 > > > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > + > > > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o > > > diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c > > > new file mode 100644 > > > index 000000000000..e7a3d2917937 > > > --- /dev/null > > > +++ b/scripts/kas_alias/duplicates_list.c > > > @@ -0,0 +1,70 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdint.h> > > > +#include <stdio.h> > > > +#include <string.h> > > > +#include <stdlib.h> > > > +#include <stdbool.h> > > > + > > > +#include "item_list.h" > > > +#include "duplicates_list.h" > > > + > > > +struct duplicate_item *find_duplicates(struct item *list) > > > +{ > > > + struct duplicate_item *current_duplicate = NULL; > > > + struct duplicate_item *duplicates = NULL; > > > + struct duplicate_item *new_duplicate; > > > + struct item *current_item = list; > > > + bool prev_was_duplicate = false; > > > + struct item *prev_item = NULL; > > > + > > > + while (current_item) { > > > + if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) || > > > + prev_was_duplicate) { > > > + if (!duplicates) { > > > + duplicates = malloc(sizeof(struct duplicate_item)); > > > + if (!duplicates) > > > + return NULL; > > > + > > > + duplicates->original_item = prev_item; > > > + duplicates->next = NULL; > > > + current_duplicate = duplicates; > > > + } else { > > > + new_duplicate = malloc(sizeof(struct duplicate_item)); > > > + if (!new_duplicate) { > > > + free_duplicates(&duplicates); > > > + return NULL; > > > + } > > > + > > > + new_duplicate->original_item = prev_item; > > > + new_duplicate->next = NULL; > > > + current_duplicate->next = new_duplicate; > > > + current_duplicate = new_duplicate; > > > + > > > + if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) && > > > + (prev_was_duplicate)) > > > + prev_was_duplicate = false; > > > + else > > > + prev_was_duplicate = true; > > > + } > > > + } > > > + > > > + prev_item = current_item; > > > + current_item = current_item->next; > > > + } > > > + > > > + return duplicates; > > > +} > > > + > > > +void free_duplicates(struct duplicate_item **duplicates) > > > +{ > > > + struct duplicate_item *duplicates_iterator = *duplicates; > > > + struct duplicate_item *app; > > > + > > > + while (duplicates_iterator) { > > > + app = duplicates_iterator; > > > + duplicates_iterator = duplicates_iterator->next; > > > + free(app); > > > + } > > > + > > > + *duplicates = NULL; > > > +} > > > diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h > > > new file mode 100644 > > > index 000000000000..76aa73e584bc > > > --- /dev/null > > > +++ b/scripts/kas_alias/duplicates_list.h > > > @@ -0,0 +1,15 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef DUPLICATES_LIST_H > > > +#define DUPLICATES_LIST_H > > > + > > > +#include "item_list.h" > > > + > > > +struct duplicate_item { > > > + struct item *original_item; > > > + struct duplicate_item *next; > > > +}; > > > + > > > +struct duplicate_item *find_duplicates(struct item *list); > > > +void free_duplicates(struct duplicate_item **duplicates); > > > + > > > +#endif > > > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > > > new file mode 100644 > > > index 000000000000..7c9d5aecca9a > > > --- /dev/null > > > +++ b/scripts/kas_alias/item_list.c > > > @@ -0,0 +1,230 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <stdint.h> > > > +#include <string.h> > > > +#include <stdbool.h> > > > +#include <assert.h> > > > +#include "item_list.h" > > > + > > > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > > > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > > > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > > > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0) > > > + > > > +struct item *list_index[96] = {0}; > > > + > > > +void build_index(struct item *list) > > > +{ > > > + char current_first_letter = ' '; > > > + struct item *current = list; > > > + > > > + while (current) { > > > + if (current->symb_name[0] != current_first_letter) { > > > + current_first_letter = current->symb_name[0]; > > > + list_index[current_first_letter - 32] = current; > > > + } > > > + current = current->next; > > > + } > > > +} > > > + > > > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr) > > > +{ > > > + struct item *new_item; > > > + struct item *current; > > > + > > > + new_item = malloc(sizeof(struct item)); > > > + if (!new_item) > > > + return NULL; > > > + > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > + new_item->addr = addr; > > > + new_item->stype = stype; > > > + new_item->next = NULL; > > > + > > > + if (!(*list)) { > > > + *list = new_item; > > > + } else { > > > + current = *list; > > > + while (current->next) > > > + current = current->next; > > > + > > > + current->next = new_item; > > > + } > > > + return new_item; > > > +} > > > + > > > +void sort_list(struct item **list, int sort_by) > > > +{ > > > + struct item *current = *list; > > > + struct item *sorted = NULL; > > > + struct item *next_item; > > > + struct item *temp; > > > + > > > + if (!(*list) || !((*list)->next)) > > > + return; > > > + > > > + while (current) { > > > + next_item = current->next; > > > + if (!sorted || > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) || > > > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > > > + current->next = sorted; > > > + sorted = current; > > > + } else { > > > + temp = sorted; > > > + while (temp->next && > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) || > > > + CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) > > > + temp = temp->next; > > > + > > > + current->next = temp->next; > > > + temp->next = current; > > > + } > > > + current = next_item; > > > + } > > > + > > > + *list = sorted; > > > +} > > > + > > > +struct item *merge(struct item *left, struct item *right, int sort_by) > > > +{ > > > + struct item *current = NULL; > > > + struct item *result = NULL; > > > + > > > + if (!left) > > > + return right; > > > + if (!right) > > > + return left; > > > + > > > + if (sort_by == BY_NAME) { > > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > > + result = left; > > > + left = left->next; > > > + } else { > > > + result = right; > > > + right = right->next; > > > + } > > > + } else { > > > + if (sort_by == BY_ADDRESS) { > > > + if (left->addr <= right->addr) { > > > + result = left; > > > + left = left->next; > > > + } else { > > > + result = right; > > > + right = right->next; > > > + } > > > + } > > > + } > > > + > > > + current = result; > > > + > > > + while (left && right) { > > > + if (sort_by == BY_NAME) { > > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > > + current->next = left; > > > + left = left->next; > > > + } else { > > > + current->next = right; > > > + right = right->next; > > > + } > > > + } else { > > > + if (sort_by == BY_ADDRESS) { > > > + if (left->addr <= right->addr) { > > > + current->next = left; > > > + left = left->next; > > > + } else { > > > + current->next = right; > > > + right = right->next; > > > + } > > > + } > > > + } > > > + > > > + current = current->next; > > > + } > > > + > > > + if (left) { > > > + current->next = left; > > > + } else { > > > + if (right) > > > + current->next = right; > > > + } > > > + > > > + return result; > > > +} > > > + > > > +struct item *merge_sort(struct item *head, int sort_by) > > > +{ > > > + struct item *right; > > > + struct item *slow; > > > + struct item *fast; > > > + struct item *left; > > > + > > > + if (!head || !head->next) > > > + return head; > > > + > > > + slow = head; > > > + fast = head->next; > > > + > > > + while (fast && fast->next) { > > > + slow = slow->next; > > > + fast = fast->next->next; > > > + } > > > + > > > + left = head; > > > + right = slow->next; > > > + slow->next = NULL; > > > + > > > + left = merge_sort(left, sort_by); > > > + right = merge_sort(right, sort_by); > > > + > > > + return merge(left, right, sort_by); > > > +} > > > + > > > +void sort_list_m(struct item **head, int sort_by) > > > +{ > > > + if (!(*head) || !((*head)->next)) > > > + return; > > > + > > > + *head = merge_sort(*head, sort_by); > > > +} > > > + > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > + const char *name, uint64_t addr, char stype) > > > +{ > > > + struct item *new_item; > > > + struct item *current; > > > + int ret = 0; > > > + > > > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list; > > > + while (current) { > > > + if (current->addr == search_addr) { > > > + new_item = malloc(sizeof(struct item)); > > > + if (!new_item) > > > + return ret; > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > + new_item->addr = addr; > > > + new_item->stype = stype; > > > + new_item->next = current->next; > > > + current->next = new_item; > > > + ret = 1; > > > + break; > > > + } > > > + current = current->next; > > > + } > > > + return ret; > > > +} > > > + > > > +void free_items(struct item **head) > > > +{ > > > + struct item *app, *item_iterator = *head; > > > + > > > + while (item_iterator) { > > > + app = item_iterator; > > > + item_iterator = item_iterator->next; > > > + free(app); > > > + } > > > + *head = NULL; > > > +} > > > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > > > new file mode 100644 > > > index 000000000000..b4891cb088ee > > > --- /dev/null > > > +++ b/scripts/kas_alias/item_list.h > > > @@ -0,0 +1,26 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef ITEM_LIST_H > > > +#define ITEM_LIST_H > > > +#include <stdint.h> > > > + > > > +#define MAX_NAME_SIZE 256 > > > +#define BY_ADDRESS 1 > > > +#define BY_NAME 2 > > > + > > > +struct item { > > > + char symb_name[MAX_NAME_SIZE]; > > > + uint64_t addr; > > > + char stype; > > > + struct item *next; > > > +}; > > > + > > > +void build_index(struct item *list); > > > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr); > > > +void sort_list(struct item **list, int sort_by); > > > +struct item *merge(struct item *left, struct item *right, int sort_by); > > > +struct item *merge_sort(struct item *head, int sort_by); > > > +void sort_list_m(struct item **head, int sort_by); > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > + const char *name, uint64_t addr, char stype); > > > +void free_items(struct item **head); > > > +#endif > > > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > > > new file mode 100644 > > > index 000000000000..eb873b368023 > > > --- /dev/null > > > +++ b/scripts/kas_alias/kas_alias.c > > > @@ -0,0 +1,186 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <stdint.h> > > > +#include <string.h> > > > +#include <stdbool.h> > > > +#include <stdarg.h> > > > +#include <regex.h> > > > + > > > +#include "item_list.h" > > > +#include "duplicates_list.h" > > > + > > > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > > > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \ > > > + (((s)->stype) == 'd') || (((s)->stype) == 'D') || \ > > > + (((s)->stype) == 'r') || (((s)->stype) == 'R')) > > > +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > > > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > > > +#else > > > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > > > +#endif > > > +#define FNOMATCH 0 > > > +#define FMATCH 1 > > > +#define EREGEX 2 > > > + > > > +const char *ignore_regex[] = { > > > + "^__cfi_.*$", // __cfi_ preamble > > > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > + "^_*TRACE_SYSTEM.*$", > > > + "^__already_done\\.[0-9]+$", // Call a function once data > > > + "^___tp_str\\.[0-9]+$", > > > + "^___done\\.[0-9]+$", > > > + "^__print_once\\.[0-9]+$", > > > + "^_rs\\.[0-9]+$", > > > + "^__compound_literal\\.[0-9]+$", > > > + "^___once_key\\.[0-9]+$", > > > + "^__func__\\.[0-9]+$", > > > + "^__msg\\.[0-9]+$", > > > + "^CSWTCH\\.[0-9]+$", > > > + "^__flags\\.[0-9]+$", > > > + "^__wkey.*$", > > > + "^__mkey.*$", > > > + "^__key.*$", > > > +#endif > > > + "^__pfx_.*$" // NOP-padding > > > +}; > > > + > > > +int suffix_serial; > > > + > > > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > > > +{ > > > + va_list args; > > > + > > > + va_start(args, fmt); > > > + if (verbose) > > > + printf(fmt, args); > > > + > > > + va_end(args); > > > +} > > > + > > > +static void create_suffix(const char *name, char *output_suffix) > > > +{ > > > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > > > +} > > > + > > > +static int filter_symbols(char *symbol, const char **ignore_list, int regex_no) > > > +{ > > > + regex_t regex; > > > + int res, i; > > > + > > > + for (i = 0; i < regex_no; i++) { > > > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > > > + if (res) > > > + return -EREGEX; > > > + > > > + res = regexec(®ex, symbol, 0, NULL, 0); > > > + regfree(®ex); > > > + switch (res) { > > > + case 0: > > > + return FMATCH; > > > + case REG_NOMATCH: > > > + break; > > > + default: > > > + return -EREGEX; > > > + } > > > + > > > + } > > > + > > > + return FNOMATCH; > > > +} > > > + > > > +int main(int argc, char *argv[]) > > > +{ > > > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > > > + struct duplicate_item *duplicate_iterator; > > > + struct duplicate_item *duplicate; > > > + struct item *head = {NULL}; > > > + bool need_2_process = true; > > > + struct item *last = {NULL}; > > > + struct item *current; > > > + int verbose_mode = 0; > > > + uint64_t address; > > > + FILE *fp; > > > + int res; > > > + > > > + if (argc < 2 || argc > 3) { > > > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > > > + return 1; > > > + } > > > + > > > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > > > + verbose_mode = 1; > > > + > > > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > > > + > > > + fp = fopen(argv[1], "r"); > > > + if (!fp) { > > > + printf("Can't open input file.\n"); > > > + return 1; > > > + } > > > + > > > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > > > + if (strstr(sym_name, "__alias__1") != NULL) { > > > + if (verbose_mode && need_2_process) > > > + printf("Already processed\n"); > > > + need_2_process = false; > > > + } > > > + last = add_item(&last, sym_name, t, address); > > > + if (!last) { > > > + printf("Error in allocate memory\n"); > > > + free_items(&head); > > > + return 1; > > > + } > > > + > > > + if (!head) > > > + head = last; > > > + } > > > + > > > + fclose(fp); > > > + > > > + if (need_2_process) { > > > + verbose_msg(verbose_mode, "Sorting nm data\n"); > > > + sort_list_m(&head, BY_NAME); > > > + verbose_msg(verbose_mode, "Scanning nm data for duplicates\n"); > > > + duplicate = find_duplicates(head); > > > + if (!duplicate) { > > > + printf("Error in duplicates list\n"); > > > + return 1; > > > + } > > > + > > > + verbose_msg(verbose_mode, "Applying suffixes\n"); > > > + build_index(head); > > > + duplicate_iterator = duplicate; > > > + while (duplicate_iterator) { > > > + res = filter_symbols(duplicate_iterator->original_item->symb_name, > > > + ignore_regex, sizeof(ignore_regex) / > > > + sizeof(ignore_regex[0])); > > > + if ((res != FMATCH) && > > > + SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { > > > + if (res < 0) > > > + return 1; > > > + > > > + create_suffix(duplicate_iterator->original_item->symb_name, > > > + new_name); > > > + if (!insert_after(head, duplicate_iterator->original_item->addr, > > > + new_name, duplicate_iterator->original_item->addr, > > > + duplicate_iterator->original_item->stype)) > > > + return 1; > > > + } > > > + > > > + duplicate_iterator = duplicate_iterator->next; > > > + } > > > + > > > + sort_list_m(&head, BY_ADDRESS); > > > + } > > > + current = head; > > > + while (current) { > > > + printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name); > > > + current = current->next; > > > + } > > > + > > > + free_items(&head); > > > + free_duplicates(&duplicate); > > > + > > > + return 0; > > > +} > > > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > > > index a432b171be82..4e3dea3ac2df 100755 > > > --- a/scripts/link-vmlinux.sh > > > +++ b/scripts/link-vmlinux.sh > > > @@ -161,7 +161,11 @@ kallsyms() > > > fi > > > > > > info KSYMS ${2} > > > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > > > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > > > + ALIAS=".alias" > > > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > > > + fi > > > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > > > } > > > > > > # Perform one step in kallsyms generation, including temporary linking of > >
On 19 Jul 2023, Alexander Lobakin verbalised: > From: Nick Alcock <nick.alcock@oracle.com> > Date: Wed, 19 Jul 2023 12:12:06 +0100 >>> Yes, please coordinate with Nick and review each other's work, now we >>> have two separate efforts with different reasons but hopefully we'll > > Three efforts[0] :D Mine went unnoticed unfortunately, so I switched to > other projects then. It's odd, nobody seems to have noticed these until recently and now suddenly people are crawling out of the woodwork wanting unique addresses :) maybe the ambiguous ones are just getting commonplace enough that they're biting people more often? > My idea was to give relative path from the kernel root to the objfile, > as we have a good bunch of non-unique "filename + symbol name" pairs. I considered that, but unfortunately that has two problems to a raging perfectionist like me: - the objfile probably won't exist except if you're actually doing kernel development, since kernel build trees are big enough that a lot of people delete them after building or ship kernels to other machines: if someone else built your kernel (overwhelmingly common among non-kernel-devs) the objfiles are sure to be absent. (But an option to not truncate the names when you know they won't be absent might be a good idea, though this pushes space requirements up by hundreds of kilobytes so it should probably be off by default.) - even giving a path to the kernel module on disk (much lower resolution and vulnerable to ambiguity again) is unreliable because there's absolutely no guarantee that any given process can see any of them: they might be in a different fs namespace or the modules might only be present in an initramfs (hell, I even know setups which *compile* the modules needed for rootfs mounting in the initramfs! Yes this is borderline insane, yes it happens). More commonly, they might be compressed using any of a number of compressors, changing the name, and the kernel has no idea which compressor might have been used (not unless you want it to go and look, and, well, wandering around over the fs hunting down .ko.* files from kernelspace to get their names right is *not* my idea of a good time! It's hard enough to get that right from userspace, honestly, even with kmod helping.) The most you could do would be to provide a key you could use with kmod to dig the real modules out from userspace. Partial names are as good as anything for that :) So all the objfile names are, when it comes down to it, is names with no intrinsic meaning: even if they're filenames of some kind, tools can't rely on being able to access those files. (For my most common use case, using a tracer on an enterprise-built production kernel, they'd almost never be able to.) So you might as well treat the objfile names as arbitrary string keys that might be a memory-jogger for humans, which means you can chop boring bits off them to save space :)
From: Nick Alcock <nick.alcock@oracle.com> Date: Thu, 20 Jul 2023 14:00:46 +0100 > On 19 Jul 2023, Alexander Lobakin verbalised: "verbalised", I like that one :D > >> From: Nick Alcock <nick.alcock@oracle.com> >> Date: Wed, 19 Jul 2023 12:12:06 +0100 >>>> Yes, please coordinate with Nick and review each other's work, now we >>>> have two separate efforts with different reasons but hopefully we'll >> >> Three efforts[0] :D Mine went unnoticed unfortunately, so I switched to >> other projects then. > > It's odd, nobody seems to have noticed these until recently and now > suddenly people are crawling out of the woodwork wanting unique > addresses :) maybe the ambiguous ones are just getting commonplace > enough that they're biting people more often? I read kbuild, modules and livepatching lists time to time. That means, some small patch with no discussion can be easily missed by me, but not some big series or when there's a lot of comments. But it was strange and funny back then, in the FG-KASLR thread some devs asked me to implement unique kallsyms before moving forward ("symbol position" means nothing when this "position" is random each boot), I posted that RFC I linked in the previous reply and then... silence. "We want it so much so that we don't care" :D > >> My idea was to give relative path from the kernel root to the objfile, >> as we have a good bunch of non-unique "filename + symbol name" pairs. > > I considered that, but unfortunately that has two problems to a raging > perfectionist like me: > > - the objfile probably won't exist except if you're actually doing > kernel development, since kernel build trees are big enough that a > lot of people delete them after building or ship kernels to other > machines: if someone else built your kernel (overwhelmingly common > among non-kernel-devs) the objfiles are sure to be absent. (But an > option to not truncate the names when you know they won't be absent > might be a good idea, though this pushes space requirements up by > hundreds of kilobytes so it should probably be off by default.) > > - even giving a path to the kernel module on disk (much lower > resolution and vulnerable to ambiguity again) is unreliable because > there's absolutely no guarantee that any given process can see any of > them: they might be in a different fs namespace or the modules might > only be present in an initramfs (hell, I even know setups which > *compile* the modules needed for rootfs mounting in the initramfs! > Yes this is borderline insane, yes it happens). More commonly, they > might be compressed using any of a number of compressors, changing > the name, and the kernel has no idea which compressor might have been > used (not unless you want it to go and look, and, well, wandering > around over the fs hunting down .ko.* files from kernelspace to get > their names right is *not* my idea of a good time! It's hard enough > to get that right from userspace, honestly, even with kmod helping.) > > The most you could do would be to provide a key you could use with > kmod to dig the real modules out from userspace. Partial names are as > good as anything for that :) > > So all the objfile names are, when it comes down to it, is names with no > intrinsic meaning: even if they're filenames of some kind, tools can't > rely on being able to access those files. (For my most common use case, > using a tracer on an enterprise-built production kernel, they'd almost > never be able to.) > > So you might as well treat the objfile names as arbitrary string keys > that might be a memory-jogger for humans, which means you can chop > boring bits off them to save space :) Oh, maybe I phrased it badly, I was saving C source file path actually, not objfile :D Like: 0xdeadbeef net/core/dev.c:napi_complete_done Thanks, Olek
On Thu, 20 Jul 2023 14:00:46 +0100 Nick Alcock <nick.alcock@oracle.com> wrote: > On 19 Jul 2023, Alexander Lobakin verbalised: > > > From: Nick Alcock <nick.alcock@oracle.com> > > Date: Wed, 19 Jul 2023 12:12:06 +0100 > >>> Yes, please coordinate with Nick and review each other's work, now we > >>> have two separate efforts with different reasons but hopefully we'll > > > > Three efforts[0] :D Mine went unnoticed unfortunately, so I switched to > > other projects then. > > It's odd, nobody seems to have noticed these until recently and now > suddenly people are crawling out of the woodwork wanting unique > addresses :) maybe the ambiguous ones are just getting commonplace > enough that they're biting people more often? Usually, the ambiguous symbols are used as internal functions and are easily changed by kernel update. Thus it is only used for debugging. On the other hand, exposed symbols are considered as more stable (It's not really that stable.) so users tend to use that. BTW, note that `perf probe` and kprobe-events already supported that by '_text+OFFSET' style to point those functions (`perf probe` convert the given 'function@file-path' place to '_text+OFFSET' using DWARF and ELF). BPF doesn't because it only supports "function name". (I'm not sure how Dtrace support it) If we really consider to improve BPF trace to trace such internal functions, I think you should consider to reuse perf-probe's code to find actual address and convert it to '_text+OFFSET' style to specify the probe point. I think this still useful if user can identify the traced symbol from the source code line, easily without DWARF analysis. And BPF also need to support "SYMBOL+OFFSET" style probe points. > > > My idea was to give relative path from the kernel root to the objfile, > > as we have a good bunch of non-unique "filename + symbol name" pairs. > > I considered that, but unfortunately that has two problems to a raging > perfectionist like me: > > - the objfile probably won't exist except if you're actually doing > kernel development, since kernel build trees are big enough that a > lot of people delete them after building or ship kernels to other > machines: if someone else built your kernel (overwhelmingly common > among non-kernel-devs) the objfiles are sure to be absent. (But an > option to not truncate the names when you know they won't be absent > might be a good idea, though this pushes space requirements up by > hundreds of kilobytes so it should probably be off by default.) As I said, these internal symbol tracing is usually only for debugging the kernel. So I think this is not so problematic. Thank you, > > - even giving a path to the kernel module on disk (much lower > resolution and vulnerable to ambiguity again) is unreliable because > there's absolutely no guarantee that any given process can see any of > them: they might be in a different fs namespace or the modules might > only be present in an initramfs (hell, I even know setups which > *compile* the modules needed for rootfs mounting in the initramfs! > Yes this is borderline insane, yes it happens). More commonly, they > might be compressed using any of a number of compressors, changing > the name, and the kernel has no idea which compressor might have been > used (not unless you want it to go and look, and, well, wandering > around over the fs hunting down .ko.* files from kernelspace to get > their names right is *not* my idea of a good time! It's hard enough > to get that right from userspace, honestly, even with kmod helping.) > > The most you could do would be to provide a key you could use with > kmod to dig the real modules out from userspace. Partial names are as > good as anything for that :) > > So all the objfile names are, when it comes down to it, is names with no > intrinsic meaning: even if they're filenames of some kind, tools can't > rely on being able to access those files. (For my most common use case, > using a tracer on an enterprise-built production kernel, they'd almost > never be able to.) > > So you might as well treat the objfile names as arbitrary string keys > that might be a memory-jogger for humans, which means you can chop > boring bits off them to save space :)
On Fri, Jul 21, 2023 at 10:09:31AM +0900, Masami Hiramatsu wrote: > On Thu, 20 Jul 2023 14:00:46 +0100 > Nick Alcock <nick.alcock@oracle.com> wrote: > > > On 19 Jul 2023, Alexander Lobakin verbalised: > > > > > From: Nick Alcock <nick.alcock@oracle.com> > > > Date: Wed, 19 Jul 2023 12:12:06 +0100 > > >>> Yes, please coordinate with Nick and review each other's work, now we > > >>> have two separate efforts with different reasons but hopefully we'll > > > > > > Three efforts[0] :D Mine went unnoticed unfortunately, so I switched to > > > other projects then. > > > > It's odd, nobody seems to have noticed these until recently and now > > suddenly people are crawling out of the woodwork wanting unique > > addresses :) maybe the ambiguous ones are just getting commonplace > > enough that they're biting people more often? > > Usually, the ambiguous symbols are used as internal functions and > are easily changed by kernel update. Thus it is only used for debugging. > On the other hand, exposed symbols are considered as more stable (It's > not really that stable.) so users tend to use that. In the use case of DTrace, tracing internal functions is certainly something that is used, be it for debugging/performance analysis, for gaining better understanding what internal operations are triggerd by userspace actions, etc. > BTW, note that `perf probe` and kprobe-events already supported that by > '_text+OFFSET' style to point those functions (`perf probe` convert the > given 'function@file-path' place to '_text+OFFSET' using DWARF and ELF). > BPF doesn't because it only supports "function name". (I'm not sure how > Dtrace support it) DTrace supports arbitrary address probing based on an offset from a symbol. > If we really consider to improve BPF trace to trace such internal functions, > I think you should consider to reuse perf-probe's code to find actual > address and convert it to '_text+OFFSET' style to specify the probe point. > > I think this still useful if user can identify the traced symbol from the > source code line, easily without DWARF analysis. And BPF also need to > support "SYMBOL+OFFSET" style probe points. While identifying a traced symbol is important, the use cases I am seeing are the other way around... being able to specify a symbol name (with whatever additional elements needed to make it unique) as target for placing a probe. And doing so without needing to depend on DWARF data or other rather large collection of debug data. Especially on production systems, requiring the typically very large debuginfo style data to be installed is not acceptable for the use cases I encounter. > > > My idea was to give relative path from the kernel root to the objfile, > > > as we have a good bunch of non-unique "filename + symbol name" pairs. > > > > I considered that, but unfortunately that has two problems to a raging > > perfectionist like me: > > > > - the objfile probably won't exist except if you're actually doing > > kernel development, since kernel build trees are big enough that a > > lot of people delete them after building or ship kernels to other > > machines: if someone else built your kernel (overwhelmingly common > > among non-kernel-devs) the objfiles are sure to be absent. (But an > > option to not truncate the names when you know they won't be absent > > might be a good idea, though this pushes space requirements up by > > hundreds of kilobytes so it should probably be off by default.) > > As I said, these internal symbol tracing is usually only for debugging > the kernel. So I think this is not so problematic. > > Thank you, > > > > > - even giving a path to the kernel module on disk (much lower > > resolution and vulnerable to ambiguity again) is unreliable because > > there's absolutely no guarantee that any given process can see any of > > them: they might be in a different fs namespace or the modules might > > only be present in an initramfs (hell, I even know setups which > > *compile* the modules needed for rootfs mounting in the initramfs! > > Yes this is borderline insane, yes it happens). More commonly, they > > might be compressed using any of a number of compressors, changing > > the name, and the kernel has no idea which compressor might have been > > used (not unless you want it to go and look, and, well, wandering > > around over the fs hunting down .ko.* files from kernelspace to get > > their names right is *not* my idea of a good time! It's hard enough > > to get that right from userspace, honestly, even with kmod helping.) > > > > The most you could do would be to provide a key you could use with > > kmod to dig the real modules out from userspace. Partial names are as > > good as anything for that :) > > > > So all the objfile names are, when it comes down to it, is names with no > > intrinsic meaning: even if they're filenames of some kind, tools can't > > rely on being able to access those files. (For my most common use case, > > using a tracer on an enterprise-built production kernel, they'd almost > > never be able to.) > > > > So you might as well treat the objfile names as arbitrary string keys > > that might be a memory-jogger for humans, which means you can chop > > boring bits off them to save space :) > > > -- > Masami Hiramatsu (Google) <mhiramat@kernel.org>
Hello Steven, Alexander, and Nick,
Since now I realize that a lot of people are working on my very same argument,
I'm writing you this mail just with the intent of not leaving my statements
floating in the air, and help the cause by sharing my experience, as little it
is, to solve this problem.
The addr2line strategy, I tried to implement, allows me to achieve the level
of accuracy that Luis mentioned in his comment about live-patching suggestions.
The addr2line-based approach is feasible, however, to make this approach
operable, I had to ensure that vmlinux contains the necessary debug
information.
Let me clarify.
During the kernel build, I do need the vmlinux to contain debug information.
The approach I pursued parses the nm output and, when it detects a duplicate
symbol, it needs to query vmlinux using addr2line to resolve the address
where the symbol is reported to be.
The whole process takes place during the building phase.
I've completed the work to integrate the new alias within the compilation
phase, and I wanted to share with you how it looks now.
```
~ # uname -a
Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 GNU/Linux
~ # cat /proc/kallsyms | grep gic_mask_irq
ffffacf2eb64dae4 t gic_mask_irq
ffffacf2eb64dae4 t gic_mask_irq@_drivers_irqchip_irq-gic_c_167
ffffacf2eb650960 t gic_mask_irq
ffffacf2eb650960 t gic_mask_irq@_drivers_irqchip_irq-gic-v3_c_404
~ #
```
Despite this appearing to have a sufficient level of accuracy, when I tested
it, I realized something I really did not expect. Multiple instances of the
same function exist, and when I say "same function," I don't just mean the
name, but also the function's body. I apologize for stating something that
may sound obvious to you, but I found this really unexpected.
```
~ # uname -a
Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 GNU/Linux
~ # cat /proc/kallsyms | grep -E " [ttT] " | cut -d" " -f3|sort| uniq -d| grep @_
BIT_initDStream@_lib_zstd_common_bitstream_h_259
__dev_hold@_include_linux_netdevice_h_4059
__dev_put@_include_linux_netdevice_h_4048
__flush_tlb_range@_arch_arm64_include_asm_tlbflush_h_291
__flush_tlb_range_constprop_0@_arch_arm64_include_asm_tlbflush_h_291
__kern_my_cpu_offset@_arch_arm64_include_asm_percpu_h_40
__kvm_nvhe_$x@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22
__kvm_nvhe___kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34
__kvm_nvhe_hyp_page_count@_arch_arm64_kvm_hyp_include_nvhe_memory_h_47
__kvm_nvhe_hyp_phys_to_virt@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22
__kvm_nvhe_hyp_virt_to_phys@_arch_arm64_kvm_hyp_include_nvhe_memory_h_27
__kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34
__nodes_weight_constprop_0@_include_linux_nodemask_h_239
__pi_$x@_scripts_dtc_libfdt_libfdt_h_145
__pi_fdt32_ld@_scripts_dtc_libfdt_libfdt_h_145
__preempt_count_dec_and_test@_arch_arm64_include_asm_current_h_19
acpi_dev_filter_resource_type_cb@_include_linux_acpi_h_520
add_quirk@_drivers_mmc_core_card_h_158
bpf_enable_instrumentation@_include_linux_bpf_h_1937
btf_id_cmp_func@_include_linux_btf_h_469
copy_from_sockptr_offset_constprop_0@_include_linux_sockptr_h_44
cpucap_multi_entry_cap_matches@_arch_arm64_include_asm_cpufeature_h_395
cpufreq_register_em_with_opp@_include_linux_cpufreq_h_1228
cpumask_weight@_include_linux_cpumask_h_691
cpumask_weight_constprop_0@_include_linux_cpumask_h_690
css_put@_include_linux_cgroup_refcnt_h_77
dma_cookie_status@_drivers_dma_dmaengine_h_74
dst_discard@_include_net_dst_h_392
dst_output@_include_net_dst_h_457
elf_core_dump@_fs_binfmt_elf_c_2027
fixup_use_fwh_lock@_drivers_mtd_chips_fwh_lock_h_102
flush_tlb_mm@_arch_arm64_include_asm_tlbflush_h_250
fwh_lock_varsize@_drivers_mtd_chips_fwh_lock_h_81
fwh_unlock_varsize@_drivers_mtd_chips_fwh_lock_h_92
fwh_xxlock_oneblock@_drivers_mtd_chips_fwh_lock_h_31
get_net_ns@_include_net_net_namespace_h_231
inline_map_read_isra_0@_include_linux_mtd_map_h_393
inline_map_write_part_0@_include_linux_mtd_map_h_426
io_run_task_work@_io_uring_io_uring_h_282
ipv6_portaddr_hash_isra_0@_include_net_ipv6_h_732
irq_find_host@_include_linux_irqdomain_h_322
jhash@_include_linux_jhash_h_76
jhash_constprop_0@_include_linux_jhash_h_76
ktime_get_boottime@_include_linux_timekeeping_h_94
ktime_get_real@_include_linux_timekeeping_h_78
load_elf_binary@_fs_binfmt_elf_c_824
load_elf_phdrs@_fs_binfmt_elf_c_468
may_use_simd@_arch_arm64_include_asm_alternative-macros_h_232
netif_tx_disable@_include_linux_netdevice_h_4486
of_parse_phandle@_include_linux_of_h_946
of_parse_phandle_constprop_0@_include_linux_of_h_943
padzero@_fs_binfmt_elf_c_142
percpu_down_read_constprop_0@_include_linux_percpu-rwsem_h_47
percpu_ref_get_many@_include_linux_percpu-refcount_h_199
percpu_ref_get_many_constprop_0@_include_linux_percpu-refcount_h_198
percpu_ref_put_many_constprop_0@_include_linux_percpu-refcount_h_326
percpu_up_read@_include_linux_percpu-rwsem_h_98
percpu_up_read_constprop_0@_include_linux_percpu-rwsem_h_97
pinconf_generic_dt_node_to_map_all@_include_linux_pinctrl_pinconf-generic_h_223
pinconf_generic_dt_node_to_map_group@_include_linux_pinctrl_pinconf-generic_h_207
pinconf_generic_dt_node_to_map_pin@_include_linux_pinctrl_pinconf-generic_h_215
platform_device_register_resndata_constprop_0@_include_linux_platform_device_h_125
ptrauth_keys_install_user@_arch_arm64_include_asm_alternative-macros_h_232
readl@_arch_arm64_include_asm_io_h_75
reqsk_put@_include_net_request_sock_h_133
rht_key_get_hash_constprop_0_isra_0@_include_linux_rhashtable_h_133
rht_key_get_hash_isra_0@_include_linux_rhashtable_h_125
role_show@_include_linux_device_h_763
sdhci_and_cqhci_reset@_drivers_mmc_host_sdhci-cqhci_h_16
set_active_memcg_part_0@_include_linux_sched_mm_h_410
set_brk@_fs_binfmt_elf_c_114
set_is_seen@_arch_arm64_include_asm_current_h_19
set_lookup@_arch_arm64_include_asm_current_h_19
sha1_base_init@_include_crypto_sha1_base_h_22
sha224_base_init@_include_crypto_sha256_base_h_23
sha256_base_init@_include_crypto_sha256_base_h_31
spi_sync_transfer_constprop_0@_include_linux_spi_spi_h_1339
spi_write@_include_linux_spi_spi_h_1363
tlb_flush@_arch_arm64_include_asm_tlb_h_54
trace_xhci_dbg_quirks@_drivers_usb_host_xhci-trace_h_48
udp_lib_close@_include_net_udp_h_197
udp_lib_hash@_include_net_udp_h_189
virtio_net_hdr_to_skb_constprop_0@_include_linux_virtio_net_h_48
writenote@_fs_binfmt_elf_c_1479
zero_user_segments@_include_linux_highmem_h_271
zero_user_segments_constprop_0@_include_linux_highmem_h_268
```
The previous evidence demonstrates that adding a tag with a file and line
number does not make a symbol unique.
Here, I inspect the function body to verify if the functions are indeed
the same.
```
Welcome to Buildroot
buildroot login: root
~ # cat /proc/kallsyms | grep set_brk
ffffa8a9d1cf1d2c t set_brk
ffffa8a9d1cf1d2c t set_brk@_fs_binfmt_elf_c_114
ffffa8a9d1cf4454 t set_brk
ffffa8a9d1cf4454 t set_brk@_fs_binfmt_elf_c_114
~ # QEMU: Terminated
$ cat System.map | grep set_brk
ffff8000082f1d2c t set_brk
ffff8000082f4454 t set_brk
~ $ r2 vmlinux
Warning: run r2 with -e bin.cache=true to fix relocations in disassembly
-- Reduce the delta where flag resolving by address is used with cfg.delta
[0xffff800008000000]> s 0xffff8000082f1d2c
[0xffff8000082f1d2c]> aa
[x] Analyze all flags starting with sym. and entry0 (aa)
[x] Analyze all functions arguments/locals
[0xffff8000082f1d2c]> pdf
; CALL XREFS from sym.load_elf_binary @ 0xffff8000082f352c(x), 0xffff8000082f38d4(x), 0xffff8000082f3a68(x)
┌ 112: sym.set_brk (int64_t arg1, int64_t arg2, int64_t arg3);
│ ; arg int64_t arg1 @ x0
│ ; arg int64_t arg2 @ x1
│ ; arg int64_t arg3 @ x2
│ ; var int64_t var_10h @ sp+0x10
│ 0xffff8000082f1d2c 3f2303d5 paciasp ; binfmt_elf.c:114
│ 0xffff8000082f1d30 fd7bbea9 stp x29, x30, [sp, -0x20]!
│ 0xffff8000082f1d34 00fc3f91 add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1
│ 0xffff8000082f1d38 fd030091 mov x29, sp ; binfmt_elf.c:114
│ 0xffff8000082f1d3c 21fc3f91 add x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2
│ 0xffff8000082f1d40 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; arg1
│ 0xffff8000082f1d44 f30b00f9 str x19, [sp, 0x10] ; binfmt_elf.c:116
│ 0xffff8000082f1d48 33cc7492 and x19, x1, 0xfffffffffffff000 ; arg2
│ 0xffff8000082f1d4c 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1
│ ┌─< 0xffff8000082f1d50 63010054 b.lo 0xffff8000082f1d7c
│ ┌──> 0xffff8000082f1d54 014138d5 mrs x1, sp_el0 ; binfmt_elf.c:128
│ ╎│ 0xffff8000082f1d58 22fc41f9 ldr x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217
│ ╎│ 0xffff8000082f1d5c 00008052 mov w0, 0 ; binfmt_elf.c:129
│ ╎│ 0xffff8000082f1d60 539400f9 str x19, [x2, 0x128] ; binfmt_elf.c:128
│ ╎│ 0xffff8000082f1d64 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217
│ ╎│ 0xffff8000082f1d68 339000f9 str x19, [x1, 0x120] ; binfmt_elf.c:128
│ ╎│ 0xffff8000082f1d6c f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:129 ; 5
│ ╎│ 0xffff8000082f1d70 fd7bc2a8 ldp x29, x30, [sp], 0x20
│ ╎│ 0xffff8000082f1d74 bf2303d5 autiasp
│ ╎│ 0xffff8000082f1d78 c0035fd6 ret
│ ╎└─> 0xffff8000082f1d7c 42007e92 and x2, x2, 4 ; binfmt_elf.c:123 ; arg3
│ ╎ 0xffff8000082f1d80 610200cb sub x1, x19, x0 ; arg1
│ ╎ 0xffff8000082f1d84 cae3fc97 bl sym.vm_brk_flags
│ └──< 0xffff8000082f1d88 60feff34 cbz w0, 0xffff8000082f1d54 ; binfmt_elf.c:125
│ 0xffff8000082f1d8c f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:130 ; 5
│ 0xffff8000082f1d90 fd7bc2a8 ldp x29, x30, [sp], 0x20
│ 0xffff8000082f1d94 bf2303d5 autiasp
â”” 0xffff8000082f1d98 c0035fd6 ret
[0xffff8000082f1d2c]> s 0xffff8000082f4454
[0xffff8000082f4454]> pdf
; CALL XREFS from sym.load_elf_binary_1 @ 0xffff8000082f5cc0(x), 0xffff8000082f5e5c(x), 0xffff8000082f6648(x)
┌ 112: sym.set_brk_1 (int64_t arg1, int64_t arg2, int64_t arg3);
│ ; arg int64_t arg1 @ x0
│ ; arg int64_t arg2 @ x1
│ ; arg int64_t arg3 @ x2
│ ; var int64_t var_10h @ sp+0x10
│ 0xffff8000082f4454 3f2303d5 paciasp ; binfmt_elf.c:114
│ 0xffff8000082f4458 fd7bbea9 stp x29, x30, [sp, -0x20]!
│ 0xffff8000082f445c 00fc3f91 add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1
│ 0xffff8000082f4460 fd030091 mov x29, sp ; binfmt_elf.c:114
│ 0xffff8000082f4464 21fc3f91 add x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2
│ 0xffff8000082f4468 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; arg1
│ 0xffff8000082f446c f30b00f9 str x19, [sp, 0x10] ; binfmt_elf.c:116
│ 0xffff8000082f4470 33cc7492 and x19, x1, 0xfffffffffffff000 ; arg2
│ 0xffff8000082f4474 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1
│ ┌─< 0xffff8000082f4478 63010054 b.lo 0xffff8000082f44a4
│ ┌──> 0xffff8000082f447c 014138d5 mrs x1, sp_el0 ; binfmt_elf.c:128
│ ╎│ 0xffff8000082f4480 22fc41f9 ldr x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217
│ ╎│ 0xffff8000082f4484 00008052 mov w0, 0 ; binfmt_elf.c:129
│ ╎│ 0xffff8000082f4488 539400f9 str x19, [x2, 0x128] ; binfmt_elf.c:128
│ ╎│ 0xffff8000082f448c 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217
│ ╎│ 0xffff8000082f4490 339000f9 str x19, [x1, 0x120] ; binfmt_elf.c:128
│ ╎│ 0xffff8000082f4494 f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:129 ; 5
│ ╎│ 0xffff8000082f4498 fd7bc2a8 ldp x29, x30, [sp], 0x20
│ ╎│ 0xffff8000082f449c bf2303d5 autiasp
│ ╎│ 0xffff8000082f44a0 c0035fd6 ret
│ ╎└─> 0xffff8000082f44a4 42007e92 and x2, x2, 4 ; binfmt_elf.c:123 ; arg3
│ ╎ 0xffff8000082f44a8 610200cb sub x1, x19, x0 ; arg1
│ ╎ 0xffff8000082f44ac 00dafc97 bl sym.vm_brk_flags
│ └──< 0xffff8000082f44b0 60feff34 cbz w0, 0xffff8000082f447c ; binfmt_elf.c:125
│ 0xffff8000082f44b4 f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:130 ; 5
│ 0xffff8000082f44b8 fd7bc2a8 ldp x29, x30, [sp], 0x20
│ 0xffff8000082f44bc bf2303d5 autiasp
â”” 0xffff8000082f44c0 c0035fd6 ret
[0xffff8000082f4454]>
```
While I can speculate on a reasonable explanation for why I see symbols coming
from header files being duplicated – when a header file is included in a C
file and it produces a function that the compiler does not want to inline, it
is replicated every time the file is included – I have no convincing
explanation as to why the fs/binfmt.c functions exist in more than one
instance.
```
elf_core_dump@_fs_binfmt_elf_c_2027
load_elf_binary@_fs_binfmt_elf_c_824
load_elf_phdrs@_fs_binfmt_elf_c_468
padzero@_fs_binfmt_elf_c_142
set_brk@_fs_binfmt_elf_c_114
writenote@_fs_binfmt_elf_c_1479
```
As a final note, please understand that my patch was not intended to undermine
anyone's work. I simply encountered a problem that I wanted to help solve.
Attached to this message is my code, in case anyone wants to replicate it.
I would appreciate being kept in the loop, as I genuinely want to assist in
fixing this problem.
BR
Alessandro
---
init/Kconfig | 36 ++++
scripts/Makefile | 4 +
scripts/kas_alias/Makefile | 4 +
scripts/kas_alias/a2l.c | 268 ++++++++++++++++++++++++++++
scripts/kas_alias/a2l.h | 32 ++++
scripts/kas_alias/duplicates_list.c | 70 ++++++++
scripts/kas_alias/duplicates_list.h | 15 ++
scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++
scripts/kas_alias/item_list.h | 26 +++
scripts/kas_alias/kas_alias.c | 217 ++++++++++++++++++++++
scripts/link-vmlinux.sh | 11 +-
11 files changed, 910 insertions(+), 3 deletions(-)
create mode 100644 scripts/kas_alias/Makefile
create mode 100644 scripts/kas_alias/a2l.c
create mode 100644 scripts/kas_alias/a2l.h
create mode 100644 scripts/kas_alias/duplicates_list.c
create mode 100644 scripts/kas_alias/duplicates_list.h
create mode 100644 scripts/kas_alias/item_list.c
create mode 100644 scripts/kas_alias/item_list.h
create mode 100644 scripts/kas_alias/kas_alias.c
diff --git a/init/Kconfig b/init/Kconfig
index f7f65af4ee12..bc69fcd9cbc8 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE
time constants, and no relocation pass is required at runtime to fix
up the entries based on the runtime load address of the kernel.
+config KALLSYMS_ALIAS
+ bool "Produces alias for duplicated symbols" if EXPERT
+ depends on KALLSYMS && (DEBUG_INFO_DWARF4 || DEBUG_INFO_DWARF5)
+ help
+ It is not uncommon for drivers or modules related to similar
+ peripherals to have symbols with the exact same name.
+ While this is not a problem for the kernel's binary itself, it
+ becomes an issue when attempting to trace or probe specific
+ functions using infrastructure like ftrace or kprobe.
+
+ This option addresses this challenge by extending the symbol names
+ with unique suffixes during the kernel build process.
+ The newly created aliases for these duplicated symbols are unique
+ names that can be fed to the ftrace sysfs interface. By doing so, it
+ enables previously unreachable symbols to be probed.
+
+config CONFIG_KALLSYMS_ALIAS_DATA
+ bool "Produces alias also for data"
+ depends on KALLSYMS_ALIAS
+ help
+ Sometimes it can be useful to refer to data. In live patch scenarios,
+ you may find yourself needing to use symbols that are shared with
+ other functions. Since symbols face the same issue as functions, this
+ option allows you to create aliases for data as well.
+
+config CONFIG_KALLSYMS_ALIAS_DATA_ALL
+ bool "Removes all filter when producing data alias"
+ depends on CONFIG_KALLSYMS_ALIAS_DATA
+ help
+ When selecting data aliases, not all symbols are included in the set
+ This is because many symbols are unlikely to be used. If you choose
+ to have an alias for all data symbols, be aware that it will
+ significantly increase the size.
+
+ If unsure, say N.
+
# end of the "standard kernel features (expert users)" menu
# syscall, maps, verifier
diff --git a/scripts/Makefile b/scripts/Makefile
index 32b6ba722728..65fafe17cfe5 100644
--- a/scripts/Makefile
+++ b/scripts/Makefile
@@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux
# Let clean descend into subdirs
subdir- += basic dtc gdb kconfig mod
+
+# KALLSyms alias
+subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias
+
diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile
new file mode 100644
index 000000000000..e1fde69232b4
--- /dev/null
+++ b/scripts/kas_alias/Makefile
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0
+hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias
+
+kas_alias-objs := duplicates_list.o item_list.o kas_alias.o a2l.o
diff --git a/scripts/kas_alias/a2l.c b/scripts/kas_alias/a2l.c
new file mode 100644
index 000000000000..a9692ac30180
--- /dev/null
+++ b/scripts/kas_alias/a2l.c
@@ -0,0 +1,268 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <string.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+#include "a2l.h"
+
+int addr2line_pid = -1;
+int a2l_in[2];
+int a2l_out[2];
+char line[MAX_BUF];
+char vmlinux_path[MAX_BUF];
+char addr2line_cmd[MAX_CMD_LEN];
+FILE *a2l_stdin, *a2l_stdout;
+
+static char *normalize_path(const char *input_path, char *output_path)
+{
+ char *prev_token = NULL;
+ char *delimiter = "/";
+ char inbuf[MAX_BUF];
+ char *token;
+ char *pos;
+
+ memset(inbuf, 0, MAX_BUF);
+ *output_path = '\0';
+ strncpy(inbuf, input_path, MAX_BUF);
+ if (!input_path || !output_path || strlen(input_path) == 0)
+ return NULL;
+
+ token = strtok(inbuf, delimiter);
+ while (token) {
+ if (strcmp(token, "..") == 0 && prev_token) {
+ pos = strrchr(output_path, '/');
+ if (pos)
+ *pos = '\0';
+
+ } else if (strcmp(token, ".") != 0) {
+ strcat(output_path, "/");
+ strcat(output_path, token);
+ }
+
+ prev_token = token;
+ token = strtok(NULL, delimiter);
+ }
+
+ return output_path;
+}
+
+static void path_of(const char *full_path, char *path)
+{
+ const char *last_slash = strrchr(full_path, '/');
+ size_t path_length;
+ char cwd[MAX_BUF];
+
+ if (!last_slash) {
+ if (getcwd(cwd, sizeof(cwd)))
+ strcpy(path, cwd);
+ else
+ strcpy(path, ".");
+ } else {
+ path_length = last_slash - full_path;
+ strncpy(path, full_path, path_length);
+ path[path_length] = '\0';
+ }
+}
+
+static bool file_exists(const char *file_path)
+{
+ FILE *file;
+
+ file = fopen(file_path, "r");
+ if (file) {
+ fclose(file);
+ return true;
+ }
+ return false;
+}
+
+int addr2line_init(const char *cmd, const char *vmlinux)
+{
+ if ((!file_exists(cmd)) || (!file_exists(vmlinux))) {
+ printf("file not found\n");
+ return 0;
+ }
+
+ path_of(vmlinux, vmlinux_path);
+ if (pipe(a2l_in) == -1) {
+ printf("Failed to create pipe\n");
+ return 0;
+ }
+
+ if (pipe(a2l_out) == -1) {
+ printf("Failed to create pipe\n");
+ return 0;
+ }
+
+ addr2line_pid = fork();
+ if (addr2line_pid == -1) {
+ printf("Failed to fork process\n");
+ close(a2l_in[P_READ]);
+ close(a2l_in[P_WRITE]);
+ close(a2l_out[P_READ]);
+ close(a2l_out[P_WRITE]);
+ return 0;
+ }
+
+ if (addr2line_pid == 0) {
+ dup2(a2l_in[P_READ], 0);
+ dup2(a2l_out[P_WRITE], 1);
+ close(a2l_in[P_WRITE]);
+ close(a2l_out[P_READ]);
+
+ execlp(cmd, cmd, ADDR2LINE_ARGS, vmlinux, NULL);
+
+ printf("Failed to execute addr2line command\n");
+ exit(1);
+ } else {
+ close(a2l_in[P_READ]);
+ close(a2l_out[P_WRITE]);
+ }
+
+ a2l_stdin = fdopen(a2l_in[P_WRITE], "w");
+ if (!a2l_stdin) {
+ printf("Failed to open pipe a2l_in\n");
+ return 0;
+ }
+
+ a2l_stdout = fdopen(a2l_out[P_READ], "r");
+ if (!a2l_stdout) {
+ printf("Failed to open pipe a2l_out\n");
+ fclose(a2l_stdin);
+ return 0;
+ }
+
+ return 1;
+}
+
+const char *remove_subdir(const char *home, const char *f_path)
+{
+ int i = 0;
+
+ while (*(home + i) == *(f_path + i))
+ i++;
+
+ return (strlen(home) != i) ? NULL : f_path + i;
+}
+
+char *addr2line_get_lines(uint64_t address)
+{
+ char buf[MAX_BUF];
+
+ fprintf(a2l_stdin, "%08lx\n", address);
+ fflush(a2l_stdin);
+
+ if (!fgets(line, sizeof(line), a2l_stdout)) {
+ printf("Failed to read lines from addr2line\n");
+ return NULL;
+ }
+
+ if (!fgets(line, sizeof(line), a2l_stdout)) {
+ printf("Failed to read lines from addr2line\n");
+ return NULL;
+ }
+
+ line[strcspn(line, "\n")] = '\0';
+ strncpy(buf, line, MAX_BUF);
+ return normalize_path(buf, line);
+}
+
+int addr2line_cleanup(void)
+{
+ int status;
+
+ if (addr2line_pid != -1) {
+ kill(addr2line_pid, SIGKILL);
+ waitpid(addr2line_pid, &status, 0);
+ fclose(a2l_stdin);
+ fclose(a2l_stdout);
+ addr2line_pid = -1;
+ }
+
+ return 1;
+}
+
+static char *find_executable(const char *command)
+{
+ char *path_env = getenv("PATH");
+ char *executable_path;
+ char *path_copy;
+ char *path;
+ int n;
+
+ if (!path_env)
+ return NULL;
+
+ path_copy = strdup(path_env);
+ if (!path_copy)
+ return NULL;
+
+ path = strtok(path_copy, ":");
+ while (path) {
+ n = snprintf(0, 0, "%s/%s", path, command);
+ executable_path = (char *)malloc(n + 1);
+ snprintf(executable_path, n + 1, "%s/%s", path, command);
+ if (access(executable_path, X_OK) == 0) {
+ free(path_copy);
+ return executable_path;
+ }
+
+ path = strtok(NULL, ":");
+ free(executable_path);
+ executable_path = NULL;
+ }
+
+ free(path_copy);
+ if (executable_path)
+ free(executable_path);
+ return NULL;
+}
+
+const char *get_addr2line(int mode)
+{
+ char *buf = "";
+
+ switch (mode) {
+ case A2L_CROSS:
+ buf = getenv("CROSS_COMPILE");
+ memcpy(addr2line_cmd, buf, strlen(buf));
+ case A2L_DEFAULT:
+ memcpy(addr2line_cmd + strlen(buf), ADDR2LINE, strlen(ADDR2LINE));
+ buf = find_executable(addr2line_cmd);
+ if (buf) {
+ memcpy(addr2line_cmd, buf, strlen(buf));
+ free(buf);
+ }
+ return addr2line_cmd;
+ case A2L_LLVM:
+ default:
+ return NULL;
+ }
+}
+
+char *get_vmlinux(char *input)
+{
+ const char *match_string1 = ".syms";
+ const char *match_string2 = ".tmp_vmlinux.kallsyms";
+ char *result = NULL;
+ char *match_pos;
+
+ match_pos = strstr(input, match_string1);
+ if (!match_pos)
+ return NULL;
+
+ match_pos = strstr(input, match_string2);
+ if (!match_pos)
+ return NULL;
+
+ result = strdup(input);
+ match_pos = strstr(result, match_string1);
+ *match_pos = '\0';
+ return result;
+}
diff --git a/scripts/kas_alias/a2l.h b/scripts/kas_alias/a2l.h
new file mode 100644
index 000000000000..ca6419229dde
--- /dev/null
+++ b/scripts/kas_alias/a2l.h
@@ -0,0 +1,32 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef A2L_H
+#define A2L_H
+#include <stdint.h>
+
+#define ADDR2LINE "addr2line"
+#define ADDR2LINE_ARGS "-fe"
+//#define VMLINUX "vmlinux"
+#define MAX_BUF 4096
+#define MAX_CMD_LEN 256
+#define P_READ 0
+#define P_WRITE 1
+#define A2L_DEFAULT 1
+#define A2L_CROSS 2
+#define A2L_LLVM 3
+#define A2L_MAKE_VALUE 2
+
+extern int addr2line_pid;
+extern int a2l_in[2];
+extern int a2l_out[2];
+extern char line[MAX_BUF];
+extern char vmlinux_path[MAX_BUF];
+extern char addr2line_cmd[MAX_CMD_LEN];
+
+int addr2line_init(const char *cmd, const char *vmlinux);
+char *addr2line_get_lines(uint64_t address);
+int addr2line_cleanup(void);
+const char *remove_subdir(const char *home, const char *f_path);
+const char *get_addr2line(int mode);
+char *get_vmlinux(char *input);
+
+#endif
diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c
new file mode 100644
index 000000000000..e7a3d2917937
--- /dev/null
+++ b/scripts/kas_alias/duplicates_list.c
@@ -0,0 +1,70 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <stdbool.h>
+
+#include "item_list.h"
+#include "duplicates_list.h"
+
+struct duplicate_item *find_duplicates(struct item *list)
+{
+ struct duplicate_item *current_duplicate = NULL;
+ struct duplicate_item *duplicates = NULL;
+ struct duplicate_item *new_duplicate;
+ struct item *current_item = list;
+ bool prev_was_duplicate = false;
+ struct item *prev_item = NULL;
+
+ while (current_item) {
+ if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) ||
+ prev_was_duplicate) {
+ if (!duplicates) {
+ duplicates = malloc(sizeof(struct duplicate_item));
+ if (!duplicates)
+ return NULL;
+
+ duplicates->original_item = prev_item;
+ duplicates->next = NULL;
+ current_duplicate = duplicates;
+ } else {
+ new_duplicate = malloc(sizeof(struct duplicate_item));
+ if (!new_duplicate) {
+ free_duplicates(&duplicates);
+ return NULL;
+ }
+
+ new_duplicate->original_item = prev_item;
+ new_duplicate->next = NULL;
+ current_duplicate->next = new_duplicate;
+ current_duplicate = new_duplicate;
+
+ if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) &&
+ (prev_was_duplicate))
+ prev_was_duplicate = false;
+ else
+ prev_was_duplicate = true;
+ }
+ }
+
+ prev_item = current_item;
+ current_item = current_item->next;
+ }
+
+ return duplicates;
+}
+
+void free_duplicates(struct duplicate_item **duplicates)
+{
+ struct duplicate_item *duplicates_iterator = *duplicates;
+ struct duplicate_item *app;
+
+ while (duplicates_iterator) {
+ app = duplicates_iterator;
+ duplicates_iterator = duplicates_iterator->next;
+ free(app);
+ }
+
+ *duplicates = NULL;
+}
diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h
new file mode 100644
index 000000000000..76aa73e584bc
--- /dev/null
+++ b/scripts/kas_alias/duplicates_list.h
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef DUPLICATES_LIST_H
+#define DUPLICATES_LIST_H
+
+#include "item_list.h"
+
+struct duplicate_item {
+ struct item *original_item;
+ struct duplicate_item *next;
+};
+
+struct duplicate_item *find_duplicates(struct item *list);
+void free_duplicates(struct duplicate_item **duplicates);
+
+#endif
diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c
new file mode 100644
index 000000000000..48f2e525592a
--- /dev/null
+++ b/scripts/kas_alias/item_list.c
@@ -0,0 +1,230 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <string.h>
+#include <stdbool.h>
+#include <assert.h>
+#include "item_list.h"
+
+#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \
+ ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr)
+#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \
+ ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0)
+
+struct item *list_index[96] = {0};
+
+void build_index(struct item *list)
+{
+ char current_first_letter = ' ';
+ struct item *current = list;
+
+ while (current) {
+ if (current->symb_name[0] != current_first_letter) {
+ current_first_letter = current->symb_name[0];
+ list_index[current_first_letter - 32] = current;
+ }
+ current = current->next;
+ }
+}
+
+struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr)
+{
+ struct item *new_item;
+ struct item *current;
+
+ new_item = malloc(sizeof(struct item));
+ if (!new_item)
+ return NULL;
+
+ strncpy(new_item->symb_name, name, MAX_NAME_SIZE);
+ new_item->symb_name[MAX_NAME_SIZE - 1] = '\0';
+ new_item->addr = addr;
+ new_item->stype = stype;
+ new_item->next = NULL;
+
+ if (!(*list)) {
+ *list = new_item;
+ } else {
+ current = *list;
+ while (current->next)
+ current = current->next;
+
+ current->next = new_item;
+ }
+ return new_item;
+}
+
+void sort_list(struct item **list, int sort_by)
+{
+ struct item *current = *list;
+ struct item *sorted = NULL;
+ struct item *next_item;
+ struct item *temp;
+
+ if (!(*list) || !((*list)->next))
+ return;
+
+ while (current) {
+ next_item = current->next;
+ if (!sorted ||
+ (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) ||
+ CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) {
+ current->next = sorted;
+ sorted = current;
+ } else {
+ temp = sorted;
+ while (temp->next &&
+ (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) ||
+ CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=)))
+ temp = temp->next;
+
+ current->next = temp->next;
+ temp->next = current;
+ }
+ current = next_item;
+ }
+
+ *list = sorted;
+}
+
+struct item *merge(struct item *left, struct item *right, int sort_by)
+{
+ struct item *current = NULL;
+ struct item *result = NULL;
+
+ if (!left)
+ return right;
+ if (!right)
+ return left;
+
+ if (sort_by == BY_NAME) {
+ if (strcmp(left->symb_name, right->symb_name) <= 0) {
+ result = left;
+ left = left->next;
+ } else {
+ result = right;
+ right = right->next;
+ }
+ } else {
+ if (sort_by == BY_ADDRESS) {
+ if (left->addr <= right->addr) {
+ result = left;
+ left = left->next;
+ } else {
+ result = right;
+ right = right->next;
+ }
+ }
+ }
+
+ current = result;
+
+ while (left && right) {
+ if (sort_by == BY_NAME) {
+ if (strcmp(left->symb_name, right->symb_name) <= 0) {
+ current->next = left;
+ left = left->next;
+ } else {
+ current->next = right;
+ right = right->next;
+ }
+ } else {
+ if (sort_by == BY_ADDRESS) {
+ if (left->addr <= right->addr) {
+ current->next = left;
+ left = left->next;
+ } else {
+ current->next = right;
+ right = right->next;
+ }
+ }
+ }
+
+ current = current->next;
+ }
+
+ if (left) {
+ current->next = left;
+ } else {
+ if (right)
+ current->next = right;
+ }
+
+ return result;
+}
+
+struct item *merge_sort(struct item *head, int sort_by)
+{
+ struct item *right;
+ struct item *slow;
+ struct item *fast;
+ struct item *left;
+
+ if (!head || !head->next)
+ return head;
+
+ slow = head;
+ fast = head->next;
+
+ while (fast && fast->next) {
+ slow = slow->next;
+ fast = fast->next->next;
+ }
+
+ left = head;
+ right = slow->next;
+ slow->next = NULL;
+
+ left = merge_sort(left, sort_by);
+ right = merge_sort(right, sort_by);
+
+ return merge(left, right, sort_by);
+}
+
+void sort_list_m(struct item **head, int sort_by)
+{
+ if (!(*head) || !((*head)->next))
+ return;
+
+ *head = merge_sort(*head, sort_by);
+}
+
+int insert_after(struct item *list, const uint64_t search_addr,
+ const char *name, uint64_t addr, char stype)
+{
+ struct item *new_item;
+ struct item *current;
+ int ret = 0;
+
+ current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list;
+ while (current) {
+ if (current->addr == search_addr) {
+ new_item = malloc(sizeof(struct item));
+ if (!new_item)
+ return ret;
+ strncpy(new_item->symb_name, name, MAX_NAME_SIZE);
+ new_item->symb_name[MAX_NAME_SIZE - 1] = '\0';
+ new_item->addr = addr;
+ new_item->stype = stype;
+ new_item->next = current->next;
+ current->next = new_item;
+ ret = 1;
+ break;
+ }
+ current = current->next;
+ }
+ return ret;
+}
+
+void free_items(struct item **head)
+{
+ struct item *app, *item_iterator = *head;
+
+ while (item_iterator) {
+ app = item_iterator;
+ item_iterator = item_iterator->next;
+ free(app);
+ }
+ *head = NULL;
+}
diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h
new file mode 100644
index 000000000000..b4891cb088ee
--- /dev/null
+++ b/scripts/kas_alias/item_list.h
@@ -0,0 +1,26 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+#ifndef ITEM_LIST_H
+#define ITEM_LIST_H
+#include <stdint.h>
+
+#define MAX_NAME_SIZE 256
+#define BY_ADDRESS 1
+#define BY_NAME 2
+
+struct item {
+ char symb_name[MAX_NAME_SIZE];
+ uint64_t addr;
+ char stype;
+ struct item *next;
+};
+
+void build_index(struct item *list);
+struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr);
+void sort_list(struct item **list, int sort_by);
+struct item *merge(struct item *left, struct item *right, int sort_by);
+struct item *merge_sort(struct item *head, int sort_by);
+void sort_list_m(struct item **head, int sort_by);
+int insert_after(struct item *list, const uint64_t search_addr,
+ const char *name, uint64_t addr, char stype);
+void free_items(struct item **head);
+#endif
diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c
new file mode 100644
index 000000000000..532aeb39f851
--- /dev/null
+++ b/scripts/kas_alias/kas_alias.c
@@ -0,0 +1,217 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <string.h>
+#include <stdbool.h>
+#include <stdarg.h>
+#include <regex.h>
+
+#include "item_list.h"
+#include "duplicates_list.h"
+#include "a2l.h"
+
+#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T'))
+#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \
+ (((s)->stype) == 'd') || (((s)->stype) == 'D') || \
+ (((s)->stype) == 'r') || (((s)->stype) == 'R'))
+#ifdef CONFIG_KALLSYMS_ALIAS_DATA
+#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s))
+#else
+#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s)
+#endif
+#define FNOMATCH 0
+#define FMATCH 1
+#define EREGEX 2
+
+const char *ignore_regex[] = {
+ "^__cfi_.*$", // __cfi_ preamble
+#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL
+ "^_*TRACE_SYSTEM.*$",
+ "^__already_done\\.[0-9]+$", // Call a function once data
+ "^___tp_str\\.[0-9]+$",
+ "^___done\\.[0-9]+$",
+ "^__print_once\\.[0-9]+$",
+ "^_rs\\.[0-9]+$",
+ "^__compound_literal\\.[0-9]+$",
+ "^___once_key\\.[0-9]+$",
+ "^__func__\\.[0-9]+$",
+ "^__msg\\.[0-9]+$",
+ "^CSWTCH\\.[0-9]+$",
+ "^__flags\\.[0-9]+$",
+ "^__wkey.*$",
+ "^__mkey.*$",
+ "^__key.*$",
+#endif
+ "^__pfx_.*$" // NOP-padding
+};
+
+int suffix_serial;
+
+static inline void verbose_msg(bool verbose, const char *fmt, ...)
+{
+ va_list args;
+
+ va_start(args, fmt);
+ if (verbose)
+ printf(fmt, args);
+
+ va_end(args);
+}
+
+static void create_suffix(const char *name, char *output_suffix)
+{
+ sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++);
+}
+
+static void create_file_suffix(const char *name, uint64_t address, char *output_suffix, char *cwd)
+{
+ const char *f_path;
+ char *buf;
+ int i = 0;
+
+ buf = addr2line_get_lines(address);
+ f_path = remove_subdir(cwd, buf);
+ if (f_path) {
+ sprintf(output_suffix, "%s@%s", name, f_path);
+ while (*(output_suffix + i) != '\0') {
+ switch (*(output_suffix + i)) {
+ case '/':
+ case ':':
+ case '.':
+ *(output_suffix + i) = '_';
+ break;
+ default:
+ }
+ i++;
+ }
+ } else {
+ create_suffix(name, output_suffix);
+ }
+}
+
+static int filter_symbols(char *symbol, const char **ignore_list, int regex_no)
+{
+ regex_t regex;
+ int res, i;
+
+ for (i = 0; i < regex_no; i++) {
+ res = regcomp(®ex, ignore_list[i], REG_EXTENDED);
+ if (res)
+ return -EREGEX;
+
+ res = regexec(®ex, symbol, 0, NULL, 0);
+ regfree(®ex);
+ switch (res) {
+ case 0:
+ return FMATCH;
+ case REG_NOMATCH:
+ break;
+ default:
+ return -EREGEX;
+ }
+ }
+
+ return FNOMATCH;
+}
+
+int main(int argc, char *argv[])
+{
+ char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15];
+ struct duplicate_item *duplicate_iterator;
+ struct duplicate_item *duplicate;
+ struct item *head = {NULL};
+ bool need_2_process = true;
+ struct item *last = {NULL};
+ struct item *current;
+ int verbose_mode = 0;
+ uint64_t address;
+ FILE *fp;
+ int res;
+
+ if (argc < 2 || argc > 3) {
+ printf("Usage: %s <nmfile> [-verbose]\n", argv[0]);
+ return 1;
+ }
+
+ if (argc == 3 && strcmp(argv[2], "-verbose") == 0)
+ verbose_mode = 1;
+
+ verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]);
+
+ fp = fopen(argv[1], "r");
+ if (!fp) {
+ printf("Can't open input file.\n");
+ return 1;
+ }
+
+ if (!addr2line_init(get_addr2line(A2L_DEFAULT), get_vmlinux(argv[1])))
+ return 1;
+
+ while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) {
+ if (strstr(sym_name, "@_")) {
+ if (verbose_mode && need_2_process)
+ printf("Already processed\n");
+ need_2_process = false;
+ }
+ last = add_item(&last, sym_name, t, address);
+ if (!last) {
+ printf("Error in allocate memory\n");
+ free_items(&head);
+ return 1;
+ }
+
+ if (!head)
+ head = last;
+ }
+
+ fclose(fp);
+
+ if (need_2_process) {
+ verbose_msg(verbose_mode, "Sorting nm data\n");
+ sort_list_m(&head, BY_NAME);
+ verbose_msg(verbose_mode, "Scanning nm data for duplicates\n");
+ duplicate = find_duplicates(head);
+ if (!duplicate) {
+ printf("Error in duplicates list\n");
+ return 1;
+ }
+
+ verbose_msg(verbose_mode, "Applying suffixes\n");
+ build_index(head);
+ duplicate_iterator = duplicate;
+ while (duplicate_iterator) {
+ res = filter_symbols(duplicate_iterator->original_item->symb_name,
+ ignore_regex, sizeof(ignore_regex) /
+ sizeof(ignore_regex[0]));
+ if (res != FMATCH &&
+ SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) {
+ if (res < 0)
+ return 1;
+
+ create_file_suffix(duplicate_iterator->original_item->symb_name,
+ duplicate_iterator->original_item->addr,
+ new_name, vmlinux_path);
+ if (!insert_after(head, duplicate_iterator->original_item->addr,
+ new_name, duplicate_iterator->original_item->addr,
+ duplicate_iterator->original_item->stype))
+ return 1;
+ }
+
+ duplicate_iterator = duplicate_iterator->next;
+ }
+
+ sort_list_m(&head, BY_ADDRESS);
+ }
+ current = head;
+ while (current) {
+ printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name);
+ current = current->next;
+ }
+
+ free_items(&head);
+ free_duplicates(&duplicate);
+ addr2line_cleanup();
+ return 0;
+}
diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh
index a432b171be82..cacf60b597ce 100755
--- a/scripts/link-vmlinux.sh
+++ b/scripts/link-vmlinux.sh
@@ -89,8 +89,9 @@ vmlinux_link()
ldflags="${ldflags} ${wl}--script=${objtree}/${KBUILD_LDS}"
- # The kallsyms linking does not need debug symbols included.
- if [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then
+ # The kallsyms linking does not need debug symbols included, unless the KALLSYMS_ALIAS.
+ if [ ! is_enabled CONFIG_KALLSYMS_ALIAS ] && \
+ [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then
ldflags="${ldflags} ${wl}--strip-debug"
fi
@@ -161,7 +162,11 @@ kallsyms()
fi
info KSYMS ${2}
- scripts/kallsyms ${kallsymopt} ${1} > ${2}
+ if is_enabled CONFIG_KALLSYMS_ALIAS; then
+ ALIAS=".alias"
+ scripts/kas_alias/kas_alias ${1} >${1}${ALIAS}
+ fi
+ scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2}
}
# Perform one step in kallsyms generation, including temporary linking of
Hello, I apologize for being noisy today. In an effort to be collaborative, I would like to share my thoughts on why I see duplicate symbols in fs/binfmt_elf.c. Il giorno ven 21 lug 2023 alle ore 11:22 Alessandro Carminati <alessandro.carminati@gmail.com> ha scritto: > > Hello Steven, Alexander, and Nick, > > Since now I realize that a lot of people are working on my very same argument, > I'm writing you this mail just with the intent of not leaving my statements > floating in the air, and help the cause by sharing my experience, as little it > is, to solve this problem. > > The addr2line strategy, I tried to implement, allows me to achieve the level > of accuracy that Luis mentioned in his comment about live-patching suggestions. > The addr2line-based approach is feasible, however, to make this approach > operable, I had to ensure that vmlinux contains the necessary debug > information. > Let me clarify. > During the kernel build, I do need the vmlinux to contain debug information. > The approach I pursued parses the nm output and, when it detects a duplicate > symbol, it needs to query vmlinux using addr2line to resolve the address > where the symbol is reported to be. > The whole process takes place during the building phase. > I've completed the work to integrate the new alias within the compilation > phase, and I wanted to share with you how it looks now. > > ``` > ~ # uname -a > Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 GNU/Linux > ~ # cat /proc/kallsyms | grep gic_mask_irq > ffffacf2eb64dae4 t gic_mask_irq > ffffacf2eb64dae4 t gic_mask_irq@_drivers_irqchip_irq-gic_c_167 > ffffacf2eb650960 t gic_mask_irq > ffffacf2eb650960 t gic_mask_irq@_drivers_irqchip_irq-gic-v3_c_404 > ~ # > ``` > > Despite this appearing to have a sufficient level of accuracy, when I tested > it, I realized something I really did not expect. Multiple instances of the > same function exist, and when I say "same function," I don't just mean the > name, but also the function's body. I apologize for stating something that > may sound obvious to you, but I found this really unexpected. > > ``` > ~ # uname -a > Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 GNU/Linux > ~ # cat /proc/kallsyms | grep -E " [ttT] " | cut -d" " -f3|sort| uniq -d| grep @_ > BIT_initDStream@_lib_zstd_common_bitstream_h_259 > __dev_hold@_include_linux_netdevice_h_4059 > __dev_put@_include_linux_netdevice_h_4048 > __flush_tlb_range@_arch_arm64_include_asm_tlbflush_h_291 > __flush_tlb_range_constprop_0@_arch_arm64_include_asm_tlbflush_h_291 > __kern_my_cpu_offset@_arch_arm64_include_asm_percpu_h_40 > __kvm_nvhe_$x@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22 > __kvm_nvhe___kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34 > __kvm_nvhe_hyp_page_count@_arch_arm64_kvm_hyp_include_nvhe_memory_h_47 > __kvm_nvhe_hyp_phys_to_virt@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22 > __kvm_nvhe_hyp_virt_to_phys@_arch_arm64_kvm_hyp_include_nvhe_memory_h_27 > __kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34 > __nodes_weight_constprop_0@_include_linux_nodemask_h_239 > __pi_$x@_scripts_dtc_libfdt_libfdt_h_145 > __pi_fdt32_ld@_scripts_dtc_libfdt_libfdt_h_145 > __preempt_count_dec_and_test@_arch_arm64_include_asm_current_h_19 > acpi_dev_filter_resource_type_cb@_include_linux_acpi_h_520 > add_quirk@_drivers_mmc_core_card_h_158 > bpf_enable_instrumentation@_include_linux_bpf_h_1937 > btf_id_cmp_func@_include_linux_btf_h_469 > copy_from_sockptr_offset_constprop_0@_include_linux_sockptr_h_44 > cpucap_multi_entry_cap_matches@_arch_arm64_include_asm_cpufeature_h_395 > cpufreq_register_em_with_opp@_include_linux_cpufreq_h_1228 > cpumask_weight@_include_linux_cpumask_h_691 > cpumask_weight_constprop_0@_include_linux_cpumask_h_690 > css_put@_include_linux_cgroup_refcnt_h_77 > dma_cookie_status@_drivers_dma_dmaengine_h_74 > dst_discard@_include_net_dst_h_392 > dst_output@_include_net_dst_h_457 > elf_core_dump@_fs_binfmt_elf_c_2027 > fixup_use_fwh_lock@_drivers_mtd_chips_fwh_lock_h_102 > flush_tlb_mm@_arch_arm64_include_asm_tlbflush_h_250 > fwh_lock_varsize@_drivers_mtd_chips_fwh_lock_h_81 > fwh_unlock_varsize@_drivers_mtd_chips_fwh_lock_h_92 > fwh_xxlock_oneblock@_drivers_mtd_chips_fwh_lock_h_31 > get_net_ns@_include_net_net_namespace_h_231 > inline_map_read_isra_0@_include_linux_mtd_map_h_393 > inline_map_write_part_0@_include_linux_mtd_map_h_426 > io_run_task_work@_io_uring_io_uring_h_282 > ipv6_portaddr_hash_isra_0@_include_net_ipv6_h_732 > irq_find_host@_include_linux_irqdomain_h_322 > jhash@_include_linux_jhash_h_76 > jhash_constprop_0@_include_linux_jhash_h_76 > ktime_get_boottime@_include_linux_timekeeping_h_94 > ktime_get_real@_include_linux_timekeeping_h_78 > load_elf_binary@_fs_binfmt_elf_c_824 > load_elf_phdrs@_fs_binfmt_elf_c_468 > may_use_simd@_arch_arm64_include_asm_alternative-macros_h_232 > netif_tx_disable@_include_linux_netdevice_h_4486 > of_parse_phandle@_include_linux_of_h_946 > of_parse_phandle_constprop_0@_include_linux_of_h_943 > padzero@_fs_binfmt_elf_c_142 > percpu_down_read_constprop_0@_include_linux_percpu-rwsem_h_47 > percpu_ref_get_many@_include_linux_percpu-refcount_h_199 > percpu_ref_get_many_constprop_0@_include_linux_percpu-refcount_h_198 > percpu_ref_put_many_constprop_0@_include_linux_percpu-refcount_h_326 > percpu_up_read@_include_linux_percpu-rwsem_h_98 > percpu_up_read_constprop_0@_include_linux_percpu-rwsem_h_97 > pinconf_generic_dt_node_to_map_all@_include_linux_pinctrl_pinconf-generic_h_223 > pinconf_generic_dt_node_to_map_group@_include_linux_pinctrl_pinconf-generic_h_207 > pinconf_generic_dt_node_to_map_pin@_include_linux_pinctrl_pinconf-generic_h_215 > platform_device_register_resndata_constprop_0@_include_linux_platform_device_h_125 > ptrauth_keys_install_user@_arch_arm64_include_asm_alternative-macros_h_232 > readl@_arch_arm64_include_asm_io_h_75 > reqsk_put@_include_net_request_sock_h_133 > rht_key_get_hash_constprop_0_isra_0@_include_linux_rhashtable_h_133 > rht_key_get_hash_isra_0@_include_linux_rhashtable_h_125 > role_show@_include_linux_device_h_763 > sdhci_and_cqhci_reset@_drivers_mmc_host_sdhci-cqhci_h_16 > set_active_memcg_part_0@_include_linux_sched_mm_h_410 > set_brk@_fs_binfmt_elf_c_114 > set_is_seen@_arch_arm64_include_asm_current_h_19 > set_lookup@_arch_arm64_include_asm_current_h_19 > sha1_base_init@_include_crypto_sha1_base_h_22 > sha224_base_init@_include_crypto_sha256_base_h_23 > sha256_base_init@_include_crypto_sha256_base_h_31 > spi_sync_transfer_constprop_0@_include_linux_spi_spi_h_1339 > spi_write@_include_linux_spi_spi_h_1363 > tlb_flush@_arch_arm64_include_asm_tlb_h_54 > trace_xhci_dbg_quirks@_drivers_usb_host_xhci-trace_h_48 > udp_lib_close@_include_net_udp_h_197 > udp_lib_hash@_include_net_udp_h_189 > virtio_net_hdr_to_skb_constprop_0@_include_linux_virtio_net_h_48 > writenote@_fs_binfmt_elf_c_1479 > zero_user_segments@_include_linux_highmem_h_271 > zero_user_segments_constprop_0@_include_linux_highmem_h_268 > ``` > > The previous evidence demonstrates that adding a tag with a file and line > number does not make a symbol unique. > > Here, I inspect the function body to verify if the functions are indeed > the same. > > ``` > Welcome to Buildroot > buildroot login: root > ~ # cat /proc/kallsyms | grep set_brk > ffffa8a9d1cf1d2c t set_brk > ffffa8a9d1cf1d2c t set_brk@_fs_binfmt_elf_c_114 > ffffa8a9d1cf4454 t set_brk > ffffa8a9d1cf4454 t set_brk@_fs_binfmt_elf_c_114 > ~ # QEMU: Terminated > $ cat System.map | grep set_brk > ffff8000082f1d2c t set_brk > ffff8000082f4454 t set_brk > ~ $ r2 vmlinux > Warning: run r2 with -e bin.cache=true to fix relocations in disassembly > -- Reduce the delta where flag resolving by address is used with cfg.delta > [0xffff800008000000]> s 0xffff8000082f1d2c > [0xffff8000082f1d2c]> aa > [x] Analyze all flags starting with sym. and entry0 (aa) > [x] Analyze all functions arguments/locals > [0xffff8000082f1d2c]> pdf > ; CALL XREFS from sym.load_elf_binary @ 0xffff8000082f352c(x), 0xffff8000082f38d4(x), 0xffff8000082f3a68(x) > ┌ 112: sym.set_brk (int64_t arg1, int64_t arg2, int64_t arg3); > │ ; arg int64_t arg1 @ x0 > │ ; arg int64_t arg2 @ x1 > │ ; arg int64_t arg3 @ x2 > │ ; var int64_t var_10h @ sp+0x10 > │ 0xffff8000082f1d2c 3f2303d5 paciasp ; binfmt_elf.c:114 > │ 0xffff8000082f1d30 fd7bbea9 stp x29, x30, [sp, -0x20]! > │ 0xffff8000082f1d34 00fc3f91 add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1 > │ 0xffff8000082f1d38 fd030091 mov x29, sp ; binfmt_elf.c:114 > │ 0xffff8000082f1d3c 21fc3f91 add x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2 > │ 0xffff8000082f1d40 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; arg1 > │ 0xffff8000082f1d44 f30b00f9 str x19, [sp, 0x10] ; binfmt_elf.c:116 > │ 0xffff8000082f1d48 33cc7492 and x19, x1, 0xfffffffffffff000 ; arg2 > │ 0xffff8000082f1d4c 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1 > │ ┌─< 0xffff8000082f1d50 63010054 b.lo 0xffff8000082f1d7c > │ ┌──> 0xffff8000082f1d54 014138d5 mrs x1, sp_el0 ; binfmt_elf.c:128 > │ ╎│ 0xffff8000082f1d58 22fc41f9 ldr x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217 > │ ╎│ 0xffff8000082f1d5c 00008052 mov w0, 0 ; binfmt_elf.c:129 > │ ╎│ 0xffff8000082f1d60 539400f9 str x19, [x2, 0x128] ; binfmt_elf.c:128 > │ ╎│ 0xffff8000082f1d64 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217 > │ ╎│ 0xffff8000082f1d68 339000f9 str x19, [x1, 0x120] ; binfmt_elf.c:128 > │ ╎│ 0xffff8000082f1d6c f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:129 ; 5 > │ ╎│ 0xffff8000082f1d70 fd7bc2a8 ldp x29, x30, [sp], 0x20 > │ ╎│ 0xffff8000082f1d74 bf2303d5 autiasp > │ ╎│ 0xffff8000082f1d78 c0035fd6 ret > │ ╎└─> 0xffff8000082f1d7c 42007e92 and x2, x2, 4 ; binfmt_elf.c:123 ; arg3 > │ â•Ž 0xffff8000082f1d80 610200cb sub x1, x19, x0 ; arg1 > │ â•Ž 0xffff8000082f1d84 cae3fc97 bl sym.vm_brk_flags > │ └──< 0xffff8000082f1d88 60feff34 cbz w0, 0xffff8000082f1d54 ; binfmt_elf.c:125 > │ 0xffff8000082f1d8c f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:130 ; 5 > │ 0xffff8000082f1d90 fd7bc2a8 ldp x29, x30, [sp], 0x20 > │ 0xffff8000082f1d94 bf2303d5 autiasp > â”” 0xffff8000082f1d98 c0035fd6 ret > [0xffff8000082f1d2c]> s 0xffff8000082f4454 > [0xffff8000082f4454]> pdf > ; CALL XREFS from sym.load_elf_binary_1 @ 0xffff8000082f5cc0(x), 0xffff8000082f5e5c(x), 0xffff8000082f6648(x) > ┌ 112: sym.set_brk_1 (int64_t arg1, int64_t arg2, int64_t arg3); > │ ; arg int64_t arg1 @ x0 > │ ; arg int64_t arg2 @ x1 > │ ; arg int64_t arg3 @ x2 > │ ; var int64_t var_10h @ sp+0x10 > │ 0xffff8000082f4454 3f2303d5 paciasp ; binfmt_elf.c:114 > │ 0xffff8000082f4458 fd7bbea9 stp x29, x30, [sp, -0x20]! > │ 0xffff8000082f445c 00fc3f91 add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1 > │ 0xffff8000082f4460 fd030091 mov x29, sp ; binfmt_elf.c:114 > │ 0xffff8000082f4464 21fc3f91 add x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2 > │ 0xffff8000082f4468 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; arg1 > │ 0xffff8000082f446c f30b00f9 str x19, [sp, 0x10] ; binfmt_elf.c:116 > │ 0xffff8000082f4470 33cc7492 and x19, x1, 0xfffffffffffff000 ; arg2 > │ 0xffff8000082f4474 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1 > │ ┌─< 0xffff8000082f4478 63010054 b.lo 0xffff8000082f44a4 > │ ┌──> 0xffff8000082f447c 014138d5 mrs x1, sp_el0 ; binfmt_elf.c:128 > │ ╎│ 0xffff8000082f4480 22fc41f9 ldr x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217 > │ ╎│ 0xffff8000082f4484 00008052 mov w0, 0 ; binfmt_elf.c:129 > │ ╎│ 0xffff8000082f4488 539400f9 str x19, [x2, 0x128] ; binfmt_elf.c:128 > │ ╎│ 0xffff8000082f448c 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217 > │ ╎│ 0xffff8000082f4490 339000f9 str x19, [x1, 0x120] ; binfmt_elf.c:128 > │ ╎│ 0xffff8000082f4494 f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:129 ; 5 > │ ╎│ 0xffff8000082f4498 fd7bc2a8 ldp x29, x30, [sp], 0x20 > │ ╎│ 0xffff8000082f449c bf2303d5 autiasp > │ ╎│ 0xffff8000082f44a0 c0035fd6 ret > │ ╎└─> 0xffff8000082f44a4 42007e92 and x2, x2, 4 ; binfmt_elf.c:123 ; arg3 > │ â•Ž 0xffff8000082f44a8 610200cb sub x1, x19, x0 ; arg1 > │ â•Ž 0xffff8000082f44ac 00dafc97 bl sym.vm_brk_flags > │ └──< 0xffff8000082f44b0 60feff34 cbz w0, 0xffff8000082f447c ; binfmt_elf.c:125 > │ 0xffff8000082f44b4 f30b40f9 ldr x19, [var_10h] ; binfmt_elf.c:130 ; 5 > │ 0xffff8000082f44b8 fd7bc2a8 ldp x29, x30, [sp], 0x20 > │ 0xffff8000082f44bc bf2303d5 autiasp > â”” 0xffff8000082f44c0 c0035fd6 ret > [0xffff8000082f4454]> > ``` > > While I can speculate on a reasonable explanation for why I see symbols coming > from header files being duplicated – when a header file is included in a C > file and it produces a function that the compiler does not want to inline, it > is replicated every time the file is included – I have no convincing > explanation as to why the fs/binfmt.c functions exist in more than one > instance. > > ``` > elf_core_dump@_fs_binfmt_elf_c_2027 > load_elf_binary@_fs_binfmt_elf_c_824 > load_elf_phdrs@_fs_binfmt_elf_c_468 > padzero@_fs_binfmt_elf_c_142 > set_brk@_fs_binfmt_elf_c_114 > writenote@_fs_binfmt_elf_c_1479 > ``` > After examining the contents of built-in.a, I have come up with a new interpretation of what I am observing. According to built-in.a, System.map, and vmlinux there are two symbols named set_brk. ``` ~ $ cat System.map | grep set_brk ffff8000082f1d2c t set_brk ffff8000082f4454 t set_brk ~ $ ~ $ nm -n vmlinux | grep set_brk ffff8000082f1d2c t set_brk ffff8000082f4454 t set_brk ~ $ ~ $ nm -n built-in.a| grep set_brk 00000000000000d4 t set_brk 00000000000000d4 t set_brk ~ $ ~ $ nm -n built-in.a | grep set_brk -B100| egrep "set_brk|\.o:$" fs/binfmt_elf.o: 00000000000000d4 t set_brk fs/compat_binfmt_elf.o: 00000000000000d4 t set_brk ``` These two symbols come from fs/binfmt_elf.o and fs/compat_binfmt_elf.o, and they are just two symbols that happen to share the same name, as is common in the kernel. at the same time, addr2line reports symbols to be generated from the same file. ``` ~ $ llvm-addr2line-14 -fe vmlinux ffff8000082f1d2c ffff8000082f4454 set_brk /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 set_brk /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 ~ $ ~ $ addr2line -fe vmlinux ffff8000082f1d2c ffff8000082f4454 set_brk /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 set_brk /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 ``` looking at the source code: https://elixir.bootlin.com/linux/v6.4/source/fs/compat_binfmt_elf.c#L144 the cause of this behavior, which is unexpected but correct. The rationale is that using source file + line number iproduces better kallsyms table, but it is still do not produce unique names. BR Alessandro > As a final note, please understand that my patch was not intended to undermine > anyone's work. I simply encountered a problem that I wanted to help solve. > Attached to this message is my code, in case anyone wants to replicate it. > I would appreciate being kept in the loop, as I genuinely want to assist in > fixing this problem. > > BR > Alessandro > --- > init/Kconfig | 36 ++++ > scripts/Makefile | 4 + > scripts/kas_alias/Makefile | 4 + > scripts/kas_alias/a2l.c | 268 ++++++++++++++++++++++++++++ > scripts/kas_alias/a2l.h | 32 ++++ > scripts/kas_alias/duplicates_list.c | 70 ++++++++ > scripts/kas_alias/duplicates_list.h | 15 ++ > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++ > scripts/kas_alias/item_list.h | 26 +++ > scripts/kas_alias/kas_alias.c | 217 ++++++++++++++++++++++ > scripts/link-vmlinux.sh | 11 +- > 11 files changed, 910 insertions(+), 3 deletions(-) > create mode 100644 scripts/kas_alias/Makefile > create mode 100644 scripts/kas_alias/a2l.c > create mode 100644 scripts/kas_alias/a2l.h > create mode 100644 scripts/kas_alias/duplicates_list.c > create mode 100644 scripts/kas_alias/duplicates_list.h > create mode 100644 scripts/kas_alias/item_list.c > create mode 100644 scripts/kas_alias/item_list.h > create mode 100644 scripts/kas_alias/kas_alias.c > > diff --git a/init/Kconfig b/init/Kconfig > index f7f65af4ee12..bc69fcd9cbc8 100644 > --- a/init/Kconfig > +++ b/init/Kconfig > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > time constants, and no relocation pass is required at runtime to fix > up the entries based on the runtime load address of the kernel. > > +config KALLSYMS_ALIAS > + bool "Produces alias for duplicated symbols" if EXPERT > + depends on KALLSYMS && (DEBUG_INFO_DWARF4 || DEBUG_INFO_DWARF5) > + help > + It is not uncommon for drivers or modules related to similar > + peripherals to have symbols with the exact same name. > + While this is not a problem for the kernel's binary itself, it > + becomes an issue when attempting to trace or probe specific > + functions using infrastructure like ftrace or kprobe. > + > + This option addresses this challenge by extending the symbol names > + with unique suffixes during the kernel build process. > + The newly created aliases for these duplicated symbols are unique > + names that can be fed to the ftrace sysfs interface. By doing so, it > + enables previously unreachable symbols to be probed. > + > +config CONFIG_KALLSYMS_ALIAS_DATA > + bool "Produces alias also for data" > + depends on KALLSYMS_ALIAS > + help > + Sometimes it can be useful to refer to data. In live patch scenarios, > + you may find yourself needing to use symbols that are shared with > + other functions. Since symbols face the same issue as functions, this > + option allows you to create aliases for data as well. > + > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > + bool "Removes all filter when producing data alias" > + depends on CONFIG_KALLSYMS_ALIAS_DATA > + help > + When selecting data aliases, not all symbols are included in the set > + This is because many symbols are unlikely to be used. If you choose > + to have an alias for all data symbols, be aware that it will > + significantly increase the size. > + > + If unsure, say N. > + > # end of the "standard kernel features (expert users)" menu > > # syscall, maps, verifier > diff --git a/scripts/Makefile b/scripts/Makefile > index 32b6ba722728..65fafe17cfe5 100644 > --- a/scripts/Makefile > +++ b/scripts/Makefile > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > # Let clean descend into subdirs > subdir- += basic dtc gdb kconfig mod > + > +# KALLSyms alias > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > + > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > new file mode 100644 > index 000000000000..e1fde69232b4 > --- /dev/null > +++ b/scripts/kas_alias/Makefile > @@ -0,0 +1,4 @@ > +# SPDX-License-Identifier: GPL-2.0 > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > + > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o a2l.o > diff --git a/scripts/kas_alias/a2l.c b/scripts/kas_alias/a2l.c > new file mode 100644 > index 000000000000..a9692ac30180 > --- /dev/null > +++ b/scripts/kas_alias/a2l.c > @@ -0,0 +1,268 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdio.h> > +#include <stdlib.h> > +#include <string.h> > +#include <unistd.h> > +#include <sys/types.h> > +#include <sys/wait.h> > +#include <string.h> > +#include <stdint.h> > +#include <stdbool.h> > + > +#include "a2l.h" > + > +int addr2line_pid = -1; > +int a2l_in[2]; > +int a2l_out[2]; > +char line[MAX_BUF]; > +char vmlinux_path[MAX_BUF]; > +char addr2line_cmd[MAX_CMD_LEN]; > +FILE *a2l_stdin, *a2l_stdout; > + > +static char *normalize_path(const char *input_path, char *output_path) > +{ > + char *prev_token = NULL; > + char *delimiter = "/"; > + char inbuf[MAX_BUF]; > + char *token; > + char *pos; > + > + memset(inbuf, 0, MAX_BUF); > + *output_path = '\0'; > + strncpy(inbuf, input_path, MAX_BUF); > + if (!input_path || !output_path || strlen(input_path) == 0) > + return NULL; > + > + token = strtok(inbuf, delimiter); > + while (token) { > + if (strcmp(token, "..") == 0 && prev_token) { > + pos = strrchr(output_path, '/'); > + if (pos) > + *pos = '\0'; > + > + } else if (strcmp(token, ".") != 0) { > + strcat(output_path, "/"); > + strcat(output_path, token); > + } > + > + prev_token = token; > + token = strtok(NULL, delimiter); > + } > + > + return output_path; > +} > + > +static void path_of(const char *full_path, char *path) > +{ > + const char *last_slash = strrchr(full_path, '/'); > + size_t path_length; > + char cwd[MAX_BUF]; > + > + if (!last_slash) { > + if (getcwd(cwd, sizeof(cwd))) > + strcpy(path, cwd); > + else > + strcpy(path, "."); > + } else { > + path_length = last_slash - full_path; > + strncpy(path, full_path, path_length); > + path[path_length] = '\0'; > + } > +} > + > +static bool file_exists(const char *file_path) > +{ > + FILE *file; > + > + file = fopen(file_path, "r"); > + if (file) { > + fclose(file); > + return true; > + } > + return false; > +} > + > +int addr2line_init(const char *cmd, const char *vmlinux) > +{ > + if ((!file_exists(cmd)) || (!file_exists(vmlinux))) { > + printf("file not found\n"); > + return 0; > + } > + > + path_of(vmlinux, vmlinux_path); > + if (pipe(a2l_in) == -1) { > + printf("Failed to create pipe\n"); > + return 0; > + } > + > + if (pipe(a2l_out) == -1) { > + printf("Failed to create pipe\n"); > + return 0; > + } > + > + addr2line_pid = fork(); > + if (addr2line_pid == -1) { > + printf("Failed to fork process\n"); > + close(a2l_in[P_READ]); > + close(a2l_in[P_WRITE]); > + close(a2l_out[P_READ]); > + close(a2l_out[P_WRITE]); > + return 0; > + } > + > + if (addr2line_pid == 0) { > + dup2(a2l_in[P_READ], 0); > + dup2(a2l_out[P_WRITE], 1); > + close(a2l_in[P_WRITE]); > + close(a2l_out[P_READ]); > + > + execlp(cmd, cmd, ADDR2LINE_ARGS, vmlinux, NULL); > + > + printf("Failed to execute addr2line command\n"); > + exit(1); > + } else { > + close(a2l_in[P_READ]); > + close(a2l_out[P_WRITE]); > + } > + > + a2l_stdin = fdopen(a2l_in[P_WRITE], "w"); > + if (!a2l_stdin) { > + printf("Failed to open pipe a2l_in\n"); > + return 0; > + } > + > + a2l_stdout = fdopen(a2l_out[P_READ], "r"); > + if (!a2l_stdout) { > + printf("Failed to open pipe a2l_out\n"); > + fclose(a2l_stdin); > + return 0; > + } > + > + return 1; > +} > + > +const char *remove_subdir(const char *home, const char *f_path) > +{ > + int i = 0; > + > + while (*(home + i) == *(f_path + i)) > + i++; > + > + return (strlen(home) != i) ? NULL : f_path + i; > +} > + > +char *addr2line_get_lines(uint64_t address) > +{ > + char buf[MAX_BUF]; > + > + fprintf(a2l_stdin, "%08lx\n", address); > + fflush(a2l_stdin); > + > + if (!fgets(line, sizeof(line), a2l_stdout)) { > + printf("Failed to read lines from addr2line\n"); > + return NULL; > + } > + > + if (!fgets(line, sizeof(line), a2l_stdout)) { > + printf("Failed to read lines from addr2line\n"); > + return NULL; > + } > + > + line[strcspn(line, "\n")] = '\0'; > + strncpy(buf, line, MAX_BUF); > + return normalize_path(buf, line); > +} > + > +int addr2line_cleanup(void) > +{ > + int status; > + > + if (addr2line_pid != -1) { > + kill(addr2line_pid, SIGKILL); > + waitpid(addr2line_pid, &status, 0); > + fclose(a2l_stdin); > + fclose(a2l_stdout); > + addr2line_pid = -1; > + } > + > + return 1; > +} > + > +static char *find_executable(const char *command) > +{ > + char *path_env = getenv("PATH"); > + char *executable_path; > + char *path_copy; > + char *path; > + int n; > + > + if (!path_env) > + return NULL; > + > + path_copy = strdup(path_env); > + if (!path_copy) > + return NULL; > + > + path = strtok(path_copy, ":"); > + while (path) { > + n = snprintf(0, 0, "%s/%s", path, command); > + executable_path = (char *)malloc(n + 1); > + snprintf(executable_path, n + 1, "%s/%s", path, command); > + if (access(executable_path, X_OK) == 0) { > + free(path_copy); > + return executable_path; > + } > + > + path = strtok(NULL, ":"); > + free(executable_path); > + executable_path = NULL; > + } > + > + free(path_copy); > + if (executable_path) > + free(executable_path); > + return NULL; > +} > + > +const char *get_addr2line(int mode) > +{ > + char *buf = ""; > + > + switch (mode) { > + case A2L_CROSS: > + buf = getenv("CROSS_COMPILE"); > + memcpy(addr2line_cmd, buf, strlen(buf)); > + case A2L_DEFAULT: > + memcpy(addr2line_cmd + strlen(buf), ADDR2LINE, strlen(ADDR2LINE)); > + buf = find_executable(addr2line_cmd); > + if (buf) { > + memcpy(addr2line_cmd, buf, strlen(buf)); > + free(buf); > + } > + return addr2line_cmd; > + case A2L_LLVM: > + default: > + return NULL; > + } > +} > + > +char *get_vmlinux(char *input) > +{ > + const char *match_string1 = ".syms"; > + const char *match_string2 = ".tmp_vmlinux.kallsyms"; > + char *result = NULL; > + char *match_pos; > + > + match_pos = strstr(input, match_string1); > + if (!match_pos) > + return NULL; > + > + match_pos = strstr(input, match_string2); > + if (!match_pos) > + return NULL; > + > + result = strdup(input); > + match_pos = strstr(result, match_string1); > + *match_pos = '\0'; > + return result; > +} > diff --git a/scripts/kas_alias/a2l.h b/scripts/kas_alias/a2l.h > new file mode 100644 > index 000000000000..ca6419229dde > --- /dev/null > +++ b/scripts/kas_alias/a2l.h > @@ -0,0 +1,32 @@ > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > +#ifndef A2L_H > +#define A2L_H > +#include <stdint.h> > + > +#define ADDR2LINE "addr2line" > +#define ADDR2LINE_ARGS "-fe" > +//#define VMLINUX "vmlinux" > +#define MAX_BUF 4096 > +#define MAX_CMD_LEN 256 > +#define P_READ 0 > +#define P_WRITE 1 > +#define A2L_DEFAULT 1 > +#define A2L_CROSS 2 > +#define A2L_LLVM 3 > +#define A2L_MAKE_VALUE 2 > + > +extern int addr2line_pid; > +extern int a2l_in[2]; > +extern int a2l_out[2]; > +extern char line[MAX_BUF]; > +extern char vmlinux_path[MAX_BUF]; > +extern char addr2line_cmd[MAX_CMD_LEN]; > + > +int addr2line_init(const char *cmd, const char *vmlinux); > +char *addr2line_get_lines(uint64_t address); > +int addr2line_cleanup(void); > +const char *remove_subdir(const char *home, const char *f_path); > +const char *get_addr2line(int mode); > +char *get_vmlinux(char *input); > + > +#endif > diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c > new file mode 100644 > index 000000000000..e7a3d2917937 > --- /dev/null > +++ b/scripts/kas_alias/duplicates_list.c > @@ -0,0 +1,70 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdint.h> > +#include <stdio.h> > +#include <string.h> > +#include <stdlib.h> > +#include <stdbool.h> > + > +#include "item_list.h" > +#include "duplicates_list.h" > + > +struct duplicate_item *find_duplicates(struct item *list) > +{ > + struct duplicate_item *current_duplicate = NULL; > + struct duplicate_item *duplicates = NULL; > + struct duplicate_item *new_duplicate; > + struct item *current_item = list; > + bool prev_was_duplicate = false; > + struct item *prev_item = NULL; > + > + while (current_item) { > + if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) || > + prev_was_duplicate) { > + if (!duplicates) { > + duplicates = malloc(sizeof(struct duplicate_item)); > + if (!duplicates) > + return NULL; > + > + duplicates->original_item = prev_item; > + duplicates->next = NULL; > + current_duplicate = duplicates; > + } else { > + new_duplicate = malloc(sizeof(struct duplicate_item)); > + if (!new_duplicate) { > + free_duplicates(&duplicates); > + return NULL; > + } > + > + new_duplicate->original_item = prev_item; > + new_duplicate->next = NULL; > + current_duplicate->next = new_duplicate; > + current_duplicate = new_duplicate; > + > + if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) && > + (prev_was_duplicate)) > + prev_was_duplicate = false; > + else > + prev_was_duplicate = true; > + } > + } > + > + prev_item = current_item; > + current_item = current_item->next; > + } > + > + return duplicates; > +} > + > +void free_duplicates(struct duplicate_item **duplicates) > +{ > + struct duplicate_item *duplicates_iterator = *duplicates; > + struct duplicate_item *app; > + > + while (duplicates_iterator) { > + app = duplicates_iterator; > + duplicates_iterator = duplicates_iterator->next; > + free(app); > + } > + > + *duplicates = NULL; > +} > diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h > new file mode 100644 > index 000000000000..76aa73e584bc > --- /dev/null > +++ b/scripts/kas_alias/duplicates_list.h > @@ -0,0 +1,15 @@ > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > +#ifndef DUPLICATES_LIST_H > +#define DUPLICATES_LIST_H > + > +#include "item_list.h" > + > +struct duplicate_item { > + struct item *original_item; > + struct duplicate_item *next; > +}; > + > +struct duplicate_item *find_duplicates(struct item *list); > +void free_duplicates(struct duplicate_item **duplicates); > + > +#endif > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > new file mode 100644 > index 000000000000..48f2e525592a > --- /dev/null > +++ b/scripts/kas_alias/item_list.c > @@ -0,0 +1,230 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdio.h> > +#include <stdlib.h> > +#include <stdint.h> > +#include <string.h> > +#include <stdbool.h> > +#include <assert.h> > +#include "item_list.h" > + > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0) > + > +struct item *list_index[96] = {0}; > + > +void build_index(struct item *list) > +{ > + char current_first_letter = ' '; > + struct item *current = list; > + > + while (current) { > + if (current->symb_name[0] != current_first_letter) { > + current_first_letter = current->symb_name[0]; > + list_index[current_first_letter - 32] = current; > + } > + current = current->next; > + } > +} > + > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr) > +{ > + struct item *new_item; > + struct item *current; > + > + new_item = malloc(sizeof(struct item)); > + if (!new_item) > + return NULL; > + > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > + new_item->addr = addr; > + new_item->stype = stype; > + new_item->next = NULL; > + > + if (!(*list)) { > + *list = new_item; > + } else { > + current = *list; > + while (current->next) > + current = current->next; > + > + current->next = new_item; > + } > + return new_item; > +} > + > +void sort_list(struct item **list, int sort_by) > +{ > + struct item *current = *list; > + struct item *sorted = NULL; > + struct item *next_item; > + struct item *temp; > + > + if (!(*list) || !((*list)->next)) > + return; > + > + while (current) { > + next_item = current->next; > + if (!sorted || > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) || > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > + current->next = sorted; > + sorted = current; > + } else { > + temp = sorted; > + while (temp->next && > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) || > + CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) > + temp = temp->next; > + > + current->next = temp->next; > + temp->next = current; > + } > + current = next_item; > + } > + > + *list = sorted; > +} > + > +struct item *merge(struct item *left, struct item *right, int sort_by) > +{ > + struct item *current = NULL; > + struct item *result = NULL; > + > + if (!left) > + return right; > + if (!right) > + return left; > + > + if (sort_by == BY_NAME) { > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > + result = left; > + left = left->next; > + } else { > + result = right; > + right = right->next; > + } > + } else { > + if (sort_by == BY_ADDRESS) { > + if (left->addr <= right->addr) { > + result = left; > + left = left->next; > + } else { > + result = right; > + right = right->next; > + } > + } > + } > + > + current = result; > + > + while (left && right) { > + if (sort_by == BY_NAME) { > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > + current->next = left; > + left = left->next; > + } else { > + current->next = right; > + right = right->next; > + } > + } else { > + if (sort_by == BY_ADDRESS) { > + if (left->addr <= right->addr) { > + current->next = left; > + left = left->next; > + } else { > + current->next = right; > + right = right->next; > + } > + } > + } > + > + current = current->next; > + } > + > + if (left) { > + current->next = left; > + } else { > + if (right) > + current->next = right; > + } > + > + return result; > +} > + > +struct item *merge_sort(struct item *head, int sort_by) > +{ > + struct item *right; > + struct item *slow; > + struct item *fast; > + struct item *left; > + > + if (!head || !head->next) > + return head; > + > + slow = head; > + fast = head->next; > + > + while (fast && fast->next) { > + slow = slow->next; > + fast = fast->next->next; > + } > + > + left = head; > + right = slow->next; > + slow->next = NULL; > + > + left = merge_sort(left, sort_by); > + right = merge_sort(right, sort_by); > + > + return merge(left, right, sort_by); > +} > + > +void sort_list_m(struct item **head, int sort_by) > +{ > + if (!(*head) || !((*head)->next)) > + return; > + > + *head = merge_sort(*head, sort_by); > +} > + > +int insert_after(struct item *list, const uint64_t search_addr, > + const char *name, uint64_t addr, char stype) > +{ > + struct item *new_item; > + struct item *current; > + int ret = 0; > + > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list; > + while (current) { > + if (current->addr == search_addr) { > + new_item = malloc(sizeof(struct item)); > + if (!new_item) > + return ret; > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > + new_item->addr = addr; > + new_item->stype = stype; > + new_item->next = current->next; > + current->next = new_item; > + ret = 1; > + break; > + } > + current = current->next; > + } > + return ret; > +} > + > +void free_items(struct item **head) > +{ > + struct item *app, *item_iterator = *head; > + > + while (item_iterator) { > + app = item_iterator; > + item_iterator = item_iterator->next; > + free(app); > + } > + *head = NULL; > +} > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > new file mode 100644 > index 000000000000..b4891cb088ee > --- /dev/null > +++ b/scripts/kas_alias/item_list.h > @@ -0,0 +1,26 @@ > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > +#ifndef ITEM_LIST_H > +#define ITEM_LIST_H > +#include <stdint.h> > + > +#define MAX_NAME_SIZE 256 > +#define BY_ADDRESS 1 > +#define BY_NAME 2 > + > +struct item { > + char symb_name[MAX_NAME_SIZE]; > + uint64_t addr; > + char stype; > + struct item *next; > +}; > + > +void build_index(struct item *list); > +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr); > +void sort_list(struct item **list, int sort_by); > +struct item *merge(struct item *left, struct item *right, int sort_by); > +struct item *merge_sort(struct item *head, int sort_by); > +void sort_list_m(struct item **head, int sort_by); > +int insert_after(struct item *list, const uint64_t search_addr, > + const char *name, uint64_t addr, char stype); > +void free_items(struct item **head); > +#endif > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > new file mode 100644 > index 000000000000..532aeb39f851 > --- /dev/null > +++ b/scripts/kas_alias/kas_alias.c > @@ -0,0 +1,217 @@ > +// SPDX-License-Identifier: GPL-2.0-or-later > +#include <stdio.h> > +#include <stdlib.h> > +#include <stdint.h> > +#include <unistd.h> > +#include <string.h> > +#include <stdbool.h> > +#include <stdarg.h> > +#include <regex.h> > + > +#include "item_list.h" > +#include "duplicates_list.h" > +#include "a2l.h" > + > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \ > + (((s)->stype) == 'd') || (((s)->stype) == 'D') || \ > + (((s)->stype) == 'r') || (((s)->stype) == 'R')) > +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > +#else > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > +#endif > +#define FNOMATCH 0 > +#define FMATCH 1 > +#define EREGEX 2 > + > +const char *ignore_regex[] = { > + "^__cfi_.*$", // __cfi_ preamble > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > + "^_*TRACE_SYSTEM.*$", > + "^__already_done\\.[0-9]+$", // Call a function once data > + "^___tp_str\\.[0-9]+$", > + "^___done\\.[0-9]+$", > + "^__print_once\\.[0-9]+$", > + "^_rs\\.[0-9]+$", > + "^__compound_literal\\.[0-9]+$", > + "^___once_key\\.[0-9]+$", > + "^__func__\\.[0-9]+$", > + "^__msg\\.[0-9]+$", > + "^CSWTCH\\.[0-9]+$", > + "^__flags\\.[0-9]+$", > + "^__wkey.*$", > + "^__mkey.*$", > + "^__key.*$", > +#endif > + "^__pfx_.*$" // NOP-padding > +}; > + > +int suffix_serial; > + > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > +{ > + va_list args; > + > + va_start(args, fmt); > + if (verbose) > + printf(fmt, args); > + > + va_end(args); > +} > + > +static void create_suffix(const char *name, char *output_suffix) > +{ > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > +} > + > +static void create_file_suffix(const char *name, uint64_t address, char *output_suffix, char *cwd) > +{ > + const char *f_path; > + char *buf; > + int i = 0; > + > + buf = addr2line_get_lines(address); > + f_path = remove_subdir(cwd, buf); > + if (f_path) { > + sprintf(output_suffix, "%s@%s", name, f_path); > + while (*(output_suffix + i) != '\0') { > + switch (*(output_suffix + i)) { > + case '/': > + case ':': > + case '.': > + *(output_suffix + i) = '_'; > + break; > + default: > + } > + i++; > + } > + } else { > + create_suffix(name, output_suffix); > + } > +} > + > +static int filter_symbols(char *symbol, const char **ignore_list, int regex_no) > +{ > + regex_t regex; > + int res, i; > + > + for (i = 0; i < regex_no; i++) { > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > + if (res) > + return -EREGEX; > + > + res = regexec(®ex, symbol, 0, NULL, 0); > + regfree(®ex); > + switch (res) { > + case 0: > + return FMATCH; > + case REG_NOMATCH: > + break; > + default: > + return -EREGEX; > + } > + } > + > + return FNOMATCH; > +} > + > +int main(int argc, char *argv[]) > +{ > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > + struct duplicate_item *duplicate_iterator; > + struct duplicate_item *duplicate; > + struct item *head = {NULL}; > + bool need_2_process = true; > + struct item *last = {NULL}; > + struct item *current; > + int verbose_mode = 0; > + uint64_t address; > + FILE *fp; > + int res; > + > + if (argc < 2 || argc > 3) { > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > + return 1; > + } > + > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > + verbose_mode = 1; > + > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > + > + fp = fopen(argv[1], "r"); > + if (!fp) { > + printf("Can't open input file.\n"); > + return 1; > + } > + > + if (!addr2line_init(get_addr2line(A2L_DEFAULT), get_vmlinux(argv[1]))) > + return 1; > + > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > + if (strstr(sym_name, "@_")) { > + if (verbose_mode && need_2_process) > + printf("Already processed\n"); > + need_2_process = false; > + } > + last = add_item(&last, sym_name, t, address); > + if (!last) { > + printf("Error in allocate memory\n"); > + free_items(&head); > + return 1; > + } > + > + if (!head) > + head = last; > + } > + > + fclose(fp); > + > + if (need_2_process) { > + verbose_msg(verbose_mode, "Sorting nm data\n"); > + sort_list_m(&head, BY_NAME); > + verbose_msg(verbose_mode, "Scanning nm data for duplicates\n"); > + duplicate = find_duplicates(head); > + if (!duplicate) { > + printf("Error in duplicates list\n"); > + return 1; > + } > + > + verbose_msg(verbose_mode, "Applying suffixes\n"); > + build_index(head); > + duplicate_iterator = duplicate; > + while (duplicate_iterator) { > + res = filter_symbols(duplicate_iterator->original_item->symb_name, > + ignore_regex, sizeof(ignore_regex) / > + sizeof(ignore_regex[0])); > + if (res != FMATCH && > + SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { > + if (res < 0) > + return 1; > + > + create_file_suffix(duplicate_iterator->original_item->symb_name, > + duplicate_iterator->original_item->addr, > + new_name, vmlinux_path); > + if (!insert_after(head, duplicate_iterator->original_item->addr, > + new_name, duplicate_iterator->original_item->addr, > + duplicate_iterator->original_item->stype)) > + return 1; > + } > + > + duplicate_iterator = duplicate_iterator->next; > + } > + > + sort_list_m(&head, BY_ADDRESS); > + } > + current = head; > + while (current) { > + printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name); > + current = current->next; > + } > + > + free_items(&head); > + free_duplicates(&duplicate); > + addr2line_cleanup(); > + return 0; > +} > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > index a432b171be82..cacf60b597ce 100755 > --- a/scripts/link-vmlinux.sh > +++ b/scripts/link-vmlinux.sh > @@ -89,8 +89,9 @@ vmlinux_link() > > ldflags="${ldflags} ${wl}--script=${objtree}/${KBUILD_LDS}" > > - # The kallsyms linking does not need debug symbols included. > - if [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > + # The kallsyms linking does not need debug symbols included, unless the KALLSYMS_ALIAS. > + if [ ! is_enabled CONFIG_KALLSYMS_ALIAS ] && \ > + [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > ldflags="${ldflags} ${wl}--strip-debug" > fi > > @@ -161,7 +162,11 @@ kallsyms() > fi > > info KSYMS ${2} > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > + ALIAS=".alias" > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > + fi > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > } > > # Perform one step in kallsyms generation, including temporary linking of > -- > 2.34.1 >
From: Alessandro Carminati <alessandro.carminati@gmail.com> Date: Fri, 21 Jul 2023 14:40:54 +0200 > Hello, > > I apologize for being noisy today. > > In an effort to be collaborative, I would like to share my thoughts on why I > see duplicate symbols in fs/binfmt_elf.c. [...] > ~ $ nm -n built-in.a | grep set_brk -B100| egrep "set_brk|\.o:$" > fs/binfmt_elf.o: > 00000000000000d4 t set_brk > fs/compat_binfmt_elf.o: > 00000000000000d4 t set_brk > ``` > > These two symbols come from fs/binfmt_elf.o and fs/compat_binfmt_elf.o, and > they are just two symbols that happen to share the same name, as is common > in the kernel. > > at the same time, addr2line reports symbols to be generated from the same file. > > ``` > ~ $ llvm-addr2line-14 -fe vmlinux ffff8000082f1d2c ffff8000082f4454 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > ~ $ > ~ $ addr2line -fe vmlinux ffff8000082f1d2c ffff8000082f4454 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > ``` > looking at the source code: > https://elixir.bootlin.com/linux/v6.4/source/fs/compat_binfmt_elf.c#L144 > the cause of this behavior, which is unexpected but correct. Oh man, that's some deep investigation! Let me go back to my old tree I was sending RFC from and check that particular case there... $ grep set_brk test_ksyms ffffffff814b46c0 t fs/binfmt_elf.c:set_brk ffffffff814b7040 t fs/compat_binfmt_elf.c:set_brk Looks a bit better :D > > The rationale is that using source file + line number iproduces better kallsyms > table, but it is still do not produce unique names. > > BR > Alessandro >> As a final note, please understand that my patch was not intended to undermine >> anyone's work. I simply encountered a problem that I wanted to help solve. >> Attached to this message is my code, in case anyone wants to replicate it. >> I would appreciate being kept in the loop, as I genuinely want to assist in >> fixing this problem. Thanks, Olek
Hi. Le vendredi 21 juillet 2023, 14:40:54 CEST Alessandro Carminati a écrit : > Hello, > > I apologize for being noisy today. > > In an effort to be collaborative, I would like to share my thoughts on why I > see duplicate symbols in fs/binfmt_elf.c. > > Il giorno ven 21 lug 2023 alle ore 11:22 Alessandro Carminati > > <alessandro.carminati@gmail.com> ha scritto: > > Hello Steven, Alexander, and Nick, > > > > Since now I realize that a lot of people are working on my very same > > argument, I'm writing you this mail just with the intent of not leaving > > my statements floating in the air, and help the cause by sharing my > > experience, as little it is, to solve this problem. > > > > The addr2line strategy, I tried to implement, allows me to achieve the > > level of accuracy that Luis mentioned in his comment about live-patching > > suggestions. The addr2line-based approach is feasible, however, to make > > this approach operable, I had to ensure that vmlinux contains the > > necessary debug information. > > Let me clarify. > > During the kernel build, I do need the vmlinux to contain debug > > information. The approach I pursued parses the nm output and, when it > > detects a duplicate symbol, it needs to query vmlinux using addr2line to > > resolve the address where the symbol is reported to be. > > The whole process takes place during the building phase. > > I've completed the work to integrate the new alias within the compilation > > phase, and I wanted to share with you how it looks now. > > > > ``` > > ~ # uname -a > > Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 > > GNU/Linux ~ # cat /proc/kallsyms | grep gic_mask_irq > > ffffacf2eb64dae4 t gic_mask_irq > > ffffacf2eb64dae4 t gic_mask_irq@_drivers_irqchip_irq-gic_c_167 > > ffffacf2eb650960 t gic_mask_irq > > ffffacf2eb650960 t gic_mask_irq@_drivers_irqchip_irq-gic-v3_c_404 > > ~ # > > ``` > > > > Despite this appearing to have a sufficient level of accuracy, when I > > tested it, I realized something I really did not expect. Multiple > > instances of the same function exist, and when I say "same function," I > > don't just mean the name, but also the function's body. I apologize for > > stating something that may sound obvious to you, but I found this really > > unexpected. > > > > ``` > > ~ # uname -a > > Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 > > GNU/Linux ~ # cat /proc/kallsyms | grep -E " [ttT] " | cut -d" " > > -f3|sort| uniq -d| grep @_ > > BIT_initDStream@_lib_zstd_common_bitstream_h_259 > > __dev_hold@_include_linux_netdevice_h_4059 > > __dev_put@_include_linux_netdevice_h_4048 > > __flush_tlb_range@_arch_arm64_include_asm_tlbflush_h_291 > > __flush_tlb_range_constprop_0@_arch_arm64_include_asm_tlbflush_h_291 > > __kern_my_cpu_offset@_arch_arm64_include_asm_percpu_h_40 > > __kvm_nvhe_$x@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22 > > __kvm_nvhe___kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34 > > __kvm_nvhe_hyp_page_count@_arch_arm64_kvm_hyp_include_nvhe_memory_h_47 > > __kvm_nvhe_hyp_phys_to_virt@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22 > > __kvm_nvhe_hyp_virt_to_phys@_arch_arm64_kvm_hyp_include_nvhe_memory_h_27 > > __kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34 > > __nodes_weight_constprop_0@_include_linux_nodemask_h_239 > > __pi_$x@_scripts_dtc_libfdt_libfdt_h_145 > > __pi_fdt32_ld@_scripts_dtc_libfdt_libfdt_h_145 > > __preempt_count_dec_and_test@_arch_arm64_include_asm_current_h_19 > > acpi_dev_filter_resource_type_cb@_include_linux_acpi_h_520 > > add_quirk@_drivers_mmc_core_card_h_158 > > bpf_enable_instrumentation@_include_linux_bpf_h_1937 > > btf_id_cmp_func@_include_linux_btf_h_469 > > copy_from_sockptr_offset_constprop_0@_include_linux_sockptr_h_44 > > cpucap_multi_entry_cap_matches@_arch_arm64_include_asm_cpufeature_h_395 > > cpufreq_register_em_with_opp@_include_linux_cpufreq_h_1228 > > cpumask_weight@_include_linux_cpumask_h_691 > > cpumask_weight_constprop_0@_include_linux_cpumask_h_690 > > css_put@_include_linux_cgroup_refcnt_h_77 > > dma_cookie_status@_drivers_dma_dmaengine_h_74 > > dst_discard@_include_net_dst_h_392 > > dst_output@_include_net_dst_h_457 > > elf_core_dump@_fs_binfmt_elf_c_2027 > > fixup_use_fwh_lock@_drivers_mtd_chips_fwh_lock_h_102 > > flush_tlb_mm@_arch_arm64_include_asm_tlbflush_h_250 > > fwh_lock_varsize@_drivers_mtd_chips_fwh_lock_h_81 > > fwh_unlock_varsize@_drivers_mtd_chips_fwh_lock_h_92 > > fwh_xxlock_oneblock@_drivers_mtd_chips_fwh_lock_h_31 > > get_net_ns@_include_net_net_namespace_h_231 > > inline_map_read_isra_0@_include_linux_mtd_map_h_393 > > inline_map_write_part_0@_include_linux_mtd_map_h_426 > > io_run_task_work@_io_uring_io_uring_h_282 > > ipv6_portaddr_hash_isra_0@_include_net_ipv6_h_732 > > irq_find_host@_include_linux_irqdomain_h_322 > > jhash@_include_linux_jhash_h_76 > > jhash_constprop_0@_include_linux_jhash_h_76 > > ktime_get_boottime@_include_linux_timekeeping_h_94 > > ktime_get_real@_include_linux_timekeeping_h_78 > > load_elf_binary@_fs_binfmt_elf_c_824 > > load_elf_phdrs@_fs_binfmt_elf_c_468 > > may_use_simd@_arch_arm64_include_asm_alternative-macros_h_232 > > netif_tx_disable@_include_linux_netdevice_h_4486 > > of_parse_phandle@_include_linux_of_h_946 > > of_parse_phandle_constprop_0@_include_linux_of_h_943 > > padzero@_fs_binfmt_elf_c_142 > > percpu_down_read_constprop_0@_include_linux_percpu-rwsem_h_47 > > percpu_ref_get_many@_include_linux_percpu-refcount_h_199 > > percpu_ref_get_many_constprop_0@_include_linux_percpu-refcount_h_198 > > percpu_ref_put_many_constprop_0@_include_linux_percpu-refcount_h_326 > > percpu_up_read@_include_linux_percpu-rwsem_h_98 > > percpu_up_read_constprop_0@_include_linux_percpu-rwsem_h_97 > > pinconf_generic_dt_node_to_map_all@_include_linux_pinctrl_pinconf-generic_ > > h_223 > > pinconf_generic_dt_node_to_map_group@_include_linux_pinctrl_pinconf-gener > > ic_h_207 > > pinconf_generic_dt_node_to_map_pin@_include_linux_pinctrl_pinconf-generic > > _h_215 > > platform_device_register_resndata_constprop_0@_include_linux_platform_dev > > ice_h_125 > > ptrauth_keys_install_user@_arch_arm64_include_asm_alternative-macros_h_23 > > 2 readl@_arch_arm64_include_asm_io_h_75 > > reqsk_put@_include_net_request_sock_h_133 > > rht_key_get_hash_constprop_0_isra_0@_include_linux_rhashtable_h_133 > > rht_key_get_hash_isra_0@_include_linux_rhashtable_h_125 > > role_show@_include_linux_device_h_763 > > sdhci_and_cqhci_reset@_drivers_mmc_host_sdhci-cqhci_h_16 > > set_active_memcg_part_0@_include_linux_sched_mm_h_410 > > set_brk@_fs_binfmt_elf_c_114 > > set_is_seen@_arch_arm64_include_asm_current_h_19 > > set_lookup@_arch_arm64_include_asm_current_h_19 > > sha1_base_init@_include_crypto_sha1_base_h_22 > > sha224_base_init@_include_crypto_sha256_base_h_23 > > sha256_base_init@_include_crypto_sha256_base_h_31 > > spi_sync_transfer_constprop_0@_include_linux_spi_spi_h_1339 > > spi_write@_include_linux_spi_spi_h_1363 > > tlb_flush@_arch_arm64_include_asm_tlb_h_54 > > trace_xhci_dbg_quirks@_drivers_usb_host_xhci-trace_h_48 > > udp_lib_close@_include_net_udp_h_197 > > udp_lib_hash@_include_net_udp_h_189 > > virtio_net_hdr_to_skb_constprop_0@_include_linux_virtio_net_h_48 > > writenote@_fs_binfmt_elf_c_1479 > > zero_user_segments@_include_linux_highmem_h_271 > > zero_user_segments_constprop_0@_include_linux_highmem_h_268 > > ``` > > > > The previous evidence demonstrates that adding a tag with a file and line > > number does not make a symbol unique. > > > > Here, I inspect the function body to verify if the functions are indeed > > the same. > > > > ``` > > Welcome to Buildroot > > buildroot login: root > > ~ # cat /proc/kallsyms | grep set_brk > > ffffa8a9d1cf1d2c t set_brk > > ffffa8a9d1cf1d2c t set_brk@_fs_binfmt_elf_c_114 > > ffffa8a9d1cf4454 t set_brk > > ffffa8a9d1cf4454 t set_brk@_fs_binfmt_elf_c_114 > > ~ # QEMU: Terminated > > $ cat System.map | grep set_brk > > ffff8000082f1d2c t set_brk > > ffff8000082f4454 t set_brk > > ~ $ r2 vmlinux > > Warning: run r2 with -e bin.cache=true to fix relocations in disassembly > > > > -- Reduce the delta where flag resolving by address is used with > > cfg.delta > > > > [0xffff800008000000]> s 0xffff8000082f1d2c > > [0xffff8000082f1d2c]> aa > > [x] Analyze all flags starting with sym. and entry0 (aa) > > [x] Analyze all functions arguments/locals > > [0xffff8000082f1d2c]> pdf > > > > ; CALL XREFS from sym.load_elf_binary @ 0xffff8000082f352c(x), > > 0xffff8000082f38d4(x), 0xffff8000082f3a68(x)> > > ┌ 112: sym.set_brk (int64_t arg1, int64_t arg2, int64_t arg3); > > │ ; arg int64_t arg1 @ x0 > > │ ; arg int64_t arg2 @ x1 > > │ ; arg int64_t arg3 @ x2 > > │ ; var int64_t var_10h @ sp+0x10 > > │ 0xffff8000082f1d2c 3f2303d5 paciasp ; > > binfmt_elf.c:114 │ 0xffff8000082f1d30 fd7bbea9 stp > > x29, x30, [sp, -0x20]! │ 0xffff8000082f1d34 00fc3f91 > > add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1 │ > > 0xffff8000082f1d38 fd030091 mov x29, sp ; > > binfmt_elf.c:114 │ 0xffff8000082f1d3c 21fc3f91 add > > x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2 │ 0xffff8000082f1d40 > > 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; > > arg1 │ 0xffff8000082f1d44 f30b00f9 str x19, [sp, > > 0x10] ; binfmt_elf.c:116 │ 0xffff8000082f1d48 33cc7492 > > and x19, x1, 0xfffffffffffff000 ; arg2 │ 0xffff8000082f1d4c > > 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1 │ > > ┌─< 0xffff8000082f1d50 63010054 b.lo 0xffff8000082f1d7c │ > > ┌──> 0xffff8000082f1d54 014138d5 mrs x1, sp_el0 ; > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f1d58 22fc41f9 ldr > > x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217 │ ╎│ > > 0xffff8000082f1d5c 00008052 mov w0, 0 ; > > binfmt_elf.c:129 │ ╎│ 0xffff8000082f1d60 539400f9 str > > x19, [x2, 0x128] ; binfmt_elf.c:128 │ ╎│ 0xffff8000082f1d64 > > 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217 │ ╎│ > > 0xffff8000082f1d68 339000f9 str x19, [x1, 0x120] ; > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f1d6c f30b40f9 ldr > > x19, [var_10h] ; binfmt_elf.c:129 ; 5 │ ╎│ 0xffff8000082f1d70 > > fd7bc2a8 ldp x29, x30, [sp], 0x20 │ ╎│ 0xffff8000082f1d74 > > bf2303d5 autiasp > > │ ╎│ 0xffff8000082f1d78 c0035fd6 ret > > │ ╎└─> 0xffff8000082f1d7c 42007e92 and x2, x2, 4 ; > > binfmt_elf.c:123 ; arg3 │ â•Ž 0xffff8000082f1d80 610200cb > > sub x1, x19, x0 ; arg1 │ â•Ž 0xffff8000082f1d84 cae3fc97 > > bl sym.vm_brk_flags │ └──< 0xffff8000082f1d88 60feff34 > > cbz w0, 0xffff8000082f1d54 ; binfmt_elf.c:125 │ > > 0xffff8000082f1d8c f30b40f9 ldr x19, [var_10h] ; > > binfmt_elf.c:130 ; 5 │ 0xffff8000082f1d90 fd7bc2a8 > > ldp x29, x30, [sp], 0x20 │ 0xffff8000082f1d94 bf2303d5 > > autiasp > > â”” 0xffff8000082f1d98 c0035fd6 ret > > [0xffff8000082f1d2c]> s 0xffff8000082f4454 > > [0xffff8000082f4454]> pdf > > > > ; CALL XREFS from sym.load_elf_binary_1 @ > > 0xffff8000082f5cc0(x), 0xffff8000082f5e5c(x), > > 0xffff8000082f6648(x)> > > ┌ 112: sym.set_brk_1 (int64_t arg1, int64_t arg2, int64_t arg3); > > │ ; arg int64_t arg1 @ x0 > > │ ; arg int64_t arg2 @ x1 > > │ ; arg int64_t arg3 @ x2 > > │ ; var int64_t var_10h @ sp+0x10 > > │ 0xffff8000082f4454 3f2303d5 paciasp ; > > binfmt_elf.c:114 │ 0xffff8000082f4458 fd7bbea9 stp > > x29, x30, [sp, -0x20]! │ 0xffff8000082f445c 00fc3f91 > > add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1 │ > > 0xffff8000082f4460 fd030091 mov x29, sp ; > > binfmt_elf.c:114 │ 0xffff8000082f4464 21fc3f91 add > > x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2 │ 0xffff8000082f4468 > > 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; > > arg1 │ 0xffff8000082f446c f30b00f9 str x19, [sp, > > 0x10] ; binfmt_elf.c:116 │ 0xffff8000082f4470 33cc7492 > > and x19, x1, 0xfffffffffffff000 ; arg2 │ 0xffff8000082f4474 > > 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1 │ > > ┌─< 0xffff8000082f4478 63010054 b.lo 0xffff8000082f44a4 │ > > ┌──> 0xffff8000082f447c 014138d5 mrs x1, sp_el0 ; > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f4480 22fc41f9 ldr > > x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217 │ ╎│ > > 0xffff8000082f4484 00008052 mov w0, 0 ; > > binfmt_elf.c:129 │ ╎│ 0xffff8000082f4488 539400f9 str > > x19, [x2, 0x128] ; binfmt_elf.c:128 │ ╎│ 0xffff8000082f448c > > 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217 │ ╎│ > > 0xffff8000082f4490 339000f9 str x19, [x1, 0x120] ; > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f4494 f30b40f9 ldr > > x19, [var_10h] ; binfmt_elf.c:129 ; 5 │ ╎│ 0xffff8000082f4498 > > fd7bc2a8 ldp x29, x30, [sp], 0x20 │ ╎│ 0xffff8000082f449c > > bf2303d5 autiasp > > │ ╎│ 0xffff8000082f44a0 c0035fd6 ret > > │ ╎└─> 0xffff8000082f44a4 42007e92 and x2, x2, 4 ; > > binfmt_elf.c:123 ; arg3 │ â•Ž 0xffff8000082f44a8 610200cb > > sub x1, x19, x0 ; arg1 │ â•Ž 0xffff8000082f44ac 00dafc97 > > bl sym.vm_brk_flags │ └──< 0xffff8000082f44b0 60feff34 > > cbz w0, 0xffff8000082f447c ; binfmt_elf.c:125 │ > > 0xffff8000082f44b4 f30b40f9 ldr x19, [var_10h] ; > > binfmt_elf.c:130 ; 5 │ 0xffff8000082f44b8 fd7bc2a8 > > ldp x29, x30, [sp], 0x20 │ 0xffff8000082f44bc bf2303d5 > > autiasp > > â”” 0xffff8000082f44c0 c0035fd6 ret > > [0xffff8000082f4454]> > > ``` > > > > While I can speculate on a reasonable explanation for why I see symbols > > coming from header files being duplicated – when a header file is > > included in a C file and it produces a function that the compiler does > > not want to inline, it is replicated every time the file is included – I > > have no convincing explanation as to why the fs/binfmt.c functions exist > > in more than one instance. > > > > ``` > > elf_core_dump@_fs_binfmt_elf_c_2027 > > load_elf_binary@_fs_binfmt_elf_c_824 > > load_elf_phdrs@_fs_binfmt_elf_c_468 > > padzero@_fs_binfmt_elf_c_142 > > set_brk@_fs_binfmt_elf_c_114 > > writenote@_fs_binfmt_elf_c_1479 > > ``` > > After examining the contents of built-in.a, I have come up with a new > interpretation of what I am observing. > According to built-in.a, System.map, and vmlinux there are two symbols named > set_brk. > > ``` > ~ $ cat System.map | grep set_brk > ffff8000082f1d2c t set_brk > ffff8000082f4454 t set_brk > ~ $ > ~ $ nm -n vmlinux | grep set_brk > ffff8000082f1d2c t set_brk > ffff8000082f4454 t set_brk > ~ $ > ~ $ nm -n built-in.a| grep set_brk > 00000000000000d4 t set_brk > 00000000000000d4 t set_brk > ~ $ > ~ $ nm -n built-in.a | grep set_brk -B100| egrep "set_brk|\.o:$" > fs/binfmt_elf.o: > 00000000000000d4 t set_brk > fs/compat_binfmt_elf.o: > 00000000000000d4 t set_brk > ``` > > These two symbols come from fs/binfmt_elf.o and fs/compat_binfmt_elf.o, and > they are just two symbols that happen to share the same name, as is common > in the kernel. > > at the same time, addr2line reports symbols to be generated from the same > file. > > ``` > ~ $ llvm-addr2line-14 -fe vmlinux ffff8000082f1d2c ffff8000082f4454 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > ~ $ > ~ $ addr2line -fe vmlinux ffff8000082f1d2c ffff8000082f4454 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > set_brk > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > ``` > looking at the source code: > https://elixir.bootlin.com/linux/v6.4/source/fs/compat_binfmt_elf.c#L144 > the cause of this behavior, which is unexpected but correct. Thank you for the investigation, this an interesting reading! > The rationale is that using source file + line number iproduces better > kallsyms table, but it is still do not produce unique names. When I first read your cover letter, I also thought it would be cool to have the filename rather than a serial ID. So, your latest modification is really good as it goes into this direction. Regarding having non unique names, I am not sure if this is a problem. In my case, which is using kprobe to attach BPF code to trace some events, your contribution would make me on the way to know which address I should use because I could choose regarding the filename. I am curious about a potential v3 of this patch! > BR > Alessandro > > > As a final note, please understand that my patch was not intended to > > undermine anyone's work. I simply encountered a problem that I wanted to > > help solve. Attached to this message is my code, in case anyone wants to > > replicate it. I would appreciate being kept in the loop, as I genuinely > > want to assist in fixing this problem. > > > > BR > > Alessandro > > --- > > > > init/Kconfig | 36 ++++ > > scripts/Makefile | 4 + > > scripts/kas_alias/Makefile | 4 + > > scripts/kas_alias/a2l.c | 268 ++++++++++++++++++++++++++++ > > scripts/kas_alias/a2l.h | 32 ++++ > > scripts/kas_alias/duplicates_list.c | 70 ++++++++ > > scripts/kas_alias/duplicates_list.h | 15 ++ > > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++ > > scripts/kas_alias/item_list.h | 26 +++ > > scripts/kas_alias/kas_alias.c | 217 ++++++++++++++++++++++ > > scripts/link-vmlinux.sh | 11 +- > > 11 files changed, 910 insertions(+), 3 deletions(-) > > create mode 100644 scripts/kas_alias/Makefile > > create mode 100644 scripts/kas_alias/a2l.c > > create mode 100644 scripts/kas_alias/a2l.h > > create mode 100644 scripts/kas_alias/duplicates_list.c > > create mode 100644 scripts/kas_alias/duplicates_list.h > > create mode 100644 scripts/kas_alias/item_list.c > > create mode 100644 scripts/kas_alias/item_list.h > > create mode 100644 scripts/kas_alias/kas_alias.c > > > > diff --git a/init/Kconfig b/init/Kconfig > > index f7f65af4ee12..bc69fcd9cbc8 100644 > > --- a/init/Kconfig > > +++ b/init/Kconfig > > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > > > > time constants, and no relocation pass is required at runtime to > > fix > > up the entries based on the runtime load address of the kernel. > > > > +config KALLSYMS_ALIAS > > + bool "Produces alias for duplicated symbols" if EXPERT > > + depends on KALLSYMS && (DEBUG_INFO_DWARF4 || DEBUG_INFO_DWARF5) > > + help > > + It is not uncommon for drivers or modules related to similar > > + peripherals to have symbols with the exact same name. > > + While this is not a problem for the kernel's binary itself, it > > + becomes an issue when attempting to trace or probe specific > > + functions using infrastructure like ftrace or kprobe. > > + > > + This option addresses this challenge by extending the symbol > > names + with unique suffixes during the kernel build process. > > + The newly created aliases for these duplicated symbols are > > unique > > + names that can be fed to the ftrace sysfs interface. By doing > > so, it + enables previously unreachable symbols to be probed. > > + > > +config CONFIG_KALLSYMS_ALIAS_DATA > > + bool "Produces alias also for data" > > + depends on KALLSYMS_ALIAS > > + help > > + Sometimes it can be useful to refer to data. In live patch > > scenarios, + you may find yourself needing to use symbols that > > are shared with + other functions. Since symbols face the same > > issue as functions, this + option allows you to create aliases > > for data as well. > > + > > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > > + bool "Removes all filter when producing data alias" > > + depends on CONFIG_KALLSYMS_ALIAS_DATA > > + help > > + When selecting data aliases, not all symbols are included in the > > set + This is because many symbols are unlikely to be used. If > > you choose + to have an alias for all data symbols, be aware that > > it will + significantly increase the size. > > + > > + If unsure, say N. > > + > > > > # end of the "standard kernel features (expert users)" menu > > > > # syscall, maps, verifier > > > > diff --git a/scripts/Makefile b/scripts/Makefile > > index 32b6ba722728..65fafe17cfe5 100644 > > --- a/scripts/Makefile > > +++ b/scripts/Makefile > > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > > > # Let clean descend into subdirs > > subdir- += basic dtc gdb kconfig mod > > > > + > > +# KALLSyms alias > > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > + > > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > > new file mode 100644 > > index 000000000000..e1fde69232b4 > > --- /dev/null > > +++ b/scripts/kas_alias/Makefile > > @@ -0,0 +1,4 @@ > > +# SPDX-License-Identifier: GPL-2.0 > > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > + > > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o a2l.o > > diff --git a/scripts/kas_alias/a2l.c b/scripts/kas_alias/a2l.c > > new file mode 100644 > > index 000000000000..a9692ac30180 > > --- /dev/null > > +++ b/scripts/kas_alias/a2l.c > > @@ -0,0 +1,268 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdio.h> > > +#include <stdlib.h> > > +#include <string.h> > > +#include <unistd.h> > > +#include <sys/types.h> > > +#include <sys/wait.h> > > +#include <string.h> > > +#include <stdint.h> > > +#include <stdbool.h> > > + > > +#include "a2l.h" > > + > > +int addr2line_pid = -1; > > +int a2l_in[2]; > > +int a2l_out[2]; > > +char line[MAX_BUF]; > > +char vmlinux_path[MAX_BUF]; > > +char addr2line_cmd[MAX_CMD_LEN]; > > +FILE *a2l_stdin, *a2l_stdout; > > + > > +static char *normalize_path(const char *input_path, char *output_path) > > +{ > > + char *prev_token = NULL; > > + char *delimiter = "/"; > > + char inbuf[MAX_BUF]; > > + char *token; > > + char *pos; > > + > > + memset(inbuf, 0, MAX_BUF); > > + *output_path = '\0'; > > + strncpy(inbuf, input_path, MAX_BUF); > > + if (!input_path || !output_path || strlen(input_path) == 0) > > + return NULL; > > + > > + token = strtok(inbuf, delimiter); > > + while (token) { > > + if (strcmp(token, "..") == 0 && prev_token) { > > + pos = strrchr(output_path, '/'); > > + if (pos) > > + *pos = '\0'; > > + > > + } else if (strcmp(token, ".") != 0) { > > + strcat(output_path, "/"); > > + strcat(output_path, token); > > + } > > + > > + prev_token = token; > > + token = strtok(NULL, delimiter); > > + } > > + > > + return output_path; > > +} > > + > > +static void path_of(const char *full_path, char *path) > > +{ > > + const char *last_slash = strrchr(full_path, '/'); > > + size_t path_length; > > + char cwd[MAX_BUF]; > > + > > + if (!last_slash) { > > + if (getcwd(cwd, sizeof(cwd))) > > + strcpy(path, cwd); > > + else > > + strcpy(path, "."); > > + } else { > > + path_length = last_slash - full_path; > > + strncpy(path, full_path, path_length); > > + path[path_length] = '\0'; > > + } > > +} > > + > > +static bool file_exists(const char *file_path) > > +{ > > + FILE *file; > > + > > + file = fopen(file_path, "r"); > > + if (file) { > > + fclose(file); > > + return true; > > + } > > + return false; > > +} > > + > > +int addr2line_init(const char *cmd, const char *vmlinux) > > +{ > > + if ((!file_exists(cmd)) || (!file_exists(vmlinux))) { > > + printf("file not found\n"); > > + return 0; > > + } > > + > > + path_of(vmlinux, vmlinux_path); > > + if (pipe(a2l_in) == -1) { > > + printf("Failed to create pipe\n"); > > + return 0; > > + } > > + > > + if (pipe(a2l_out) == -1) { > > + printf("Failed to create pipe\n"); > > + return 0; > > + } > > + > > + addr2line_pid = fork(); > > + if (addr2line_pid == -1) { > > + printf("Failed to fork process\n"); > > + close(a2l_in[P_READ]); > > + close(a2l_in[P_WRITE]); > > + close(a2l_out[P_READ]); > > + close(a2l_out[P_WRITE]); > > + return 0; > > + } > > + > > + if (addr2line_pid == 0) { > > + dup2(a2l_in[P_READ], 0); > > + dup2(a2l_out[P_WRITE], 1); > > + close(a2l_in[P_WRITE]); > > + close(a2l_out[P_READ]); > > + > > + execlp(cmd, cmd, ADDR2LINE_ARGS, vmlinux, NULL); > > + > > + printf("Failed to execute addr2line command\n"); > > + exit(1); > > + } else { > > + close(a2l_in[P_READ]); > > + close(a2l_out[P_WRITE]); > > + } > > + > > + a2l_stdin = fdopen(a2l_in[P_WRITE], "w"); > > + if (!a2l_stdin) { > > + printf("Failed to open pipe a2l_in\n"); > > + return 0; > > + } > > + > > + a2l_stdout = fdopen(a2l_out[P_READ], "r"); > > + if (!a2l_stdout) { > > + printf("Failed to open pipe a2l_out\n"); > > + fclose(a2l_stdin); > > + return 0; > > + } > > + > > + return 1; > > +} > > + > > +const char *remove_subdir(const char *home, const char *f_path) > > +{ > > + int i = 0; > > + > > + while (*(home + i) == *(f_path + i)) > > + i++; > > + > > + return (strlen(home) != i) ? NULL : f_path + i; > > +} > > + > > +char *addr2line_get_lines(uint64_t address) > > +{ > > + char buf[MAX_BUF]; > > + > > + fprintf(a2l_stdin, "%08lx\n", address); > > + fflush(a2l_stdin); > > + > > + if (!fgets(line, sizeof(line), a2l_stdout)) { > > + printf("Failed to read lines from addr2line\n"); > > + return NULL; > > + } > > + > > + if (!fgets(line, sizeof(line), a2l_stdout)) { > > + printf("Failed to read lines from addr2line\n"); > > + return NULL; > > + } > > + > > + line[strcspn(line, "\n")] = '\0'; > > + strncpy(buf, line, MAX_BUF); > > + return normalize_path(buf, line); > > +} > > + > > +int addr2line_cleanup(void) > > +{ > > + int status; > > + > > + if (addr2line_pid != -1) { > > + kill(addr2line_pid, SIGKILL); > > + waitpid(addr2line_pid, &status, 0); > > + fclose(a2l_stdin); > > + fclose(a2l_stdout); > > + addr2line_pid = -1; > > + } > > + > > + return 1; > > +} > > + > > +static char *find_executable(const char *command) > > +{ > > + char *path_env = getenv("PATH"); > > + char *executable_path; > > + char *path_copy; > > + char *path; > > + int n; > > + > > + if (!path_env) > > + return NULL; > > + > > + path_copy = strdup(path_env); > > + if (!path_copy) > > + return NULL; > > + > > + path = strtok(path_copy, ":"); > > + while (path) { > > + n = snprintf(0, 0, "%s/%s", path, command); > > + executable_path = (char *)malloc(n + 1); > > + snprintf(executable_path, n + 1, "%s/%s", path, command); > > + if (access(executable_path, X_OK) == 0) { > > + free(path_copy); > > + return executable_path; > > + } > > + > > + path = strtok(NULL, ":"); > > + free(executable_path); > > + executable_path = NULL; > > + } > > + > > + free(path_copy); > > + if (executable_path) > > + free(executable_path); > > + return NULL; > > +} > > + > > +const char *get_addr2line(int mode) > > +{ > > + char *buf = ""; > > + > > + switch (mode) { > > + case A2L_CROSS: > > + buf = getenv("CROSS_COMPILE"); > > + memcpy(addr2line_cmd, buf, strlen(buf)); > > + case A2L_DEFAULT: > > + memcpy(addr2line_cmd + strlen(buf), ADDR2LINE, > > strlen(ADDR2LINE)); + buf = find_executable(addr2line_cmd); > > + if (buf) { > > + memcpy(addr2line_cmd, buf, strlen(buf)); > > + free(buf); > > + } > > + return addr2line_cmd; > > + case A2L_LLVM: > > + default: > > + return NULL; > > + } > > +} > > + > > +char *get_vmlinux(char *input) > > +{ > > + const char *match_string1 = ".syms"; > > + const char *match_string2 = ".tmp_vmlinux.kallsyms"; > > + char *result = NULL; > > + char *match_pos; > > + > > + match_pos = strstr(input, match_string1); > > + if (!match_pos) > > + return NULL; > > + > > + match_pos = strstr(input, match_string2); > > + if (!match_pos) > > + return NULL; > > + > > + result = strdup(input); > > + match_pos = strstr(result, match_string1); > > + *match_pos = '\0'; > > + return result; > > +} > > diff --git a/scripts/kas_alias/a2l.h b/scripts/kas_alias/a2l.h > > new file mode 100644 > > index 000000000000..ca6419229dde > > --- /dev/null > > +++ b/scripts/kas_alias/a2l.h > > @@ -0,0 +1,32 @@ > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > +#ifndef A2L_H > > +#define A2L_H > > +#include <stdint.h> > > + > > +#define ADDR2LINE "addr2line" > > +#define ADDR2LINE_ARGS "-fe" > > +//#define VMLINUX "vmlinux" > > +#define MAX_BUF 4096 > > +#define MAX_CMD_LEN 256 > > +#define P_READ 0 > > +#define P_WRITE 1 > > +#define A2L_DEFAULT 1 > > +#define A2L_CROSS 2 > > +#define A2L_LLVM 3 > > +#define A2L_MAKE_VALUE 2 > > + > > +extern int addr2line_pid; > > +extern int a2l_in[2]; > > +extern int a2l_out[2]; > > +extern char line[MAX_BUF]; > > +extern char vmlinux_path[MAX_BUF]; > > +extern char addr2line_cmd[MAX_CMD_LEN]; > > + > > +int addr2line_init(const char *cmd, const char *vmlinux); > > +char *addr2line_get_lines(uint64_t address); > > +int addr2line_cleanup(void); > > +const char *remove_subdir(const char *home, const char *f_path); > > +const char *get_addr2line(int mode); > > +char *get_vmlinux(char *input); > > + > > +#endif > > diff --git a/scripts/kas_alias/duplicates_list.c > > b/scripts/kas_alias/duplicates_list.c new file mode 100644 > > index 000000000000..e7a3d2917937 > > --- /dev/null > > +++ b/scripts/kas_alias/duplicates_list.c > > @@ -0,0 +1,70 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdint.h> > > +#include <stdio.h> > > +#include <string.h> > > +#include <stdlib.h> > > +#include <stdbool.h> > > + > > +#include "item_list.h" > > +#include "duplicates_list.h" > > + > > +struct duplicate_item *find_duplicates(struct item *list) > > +{ > > + struct duplicate_item *current_duplicate = NULL; > > + struct duplicate_item *duplicates = NULL; > > + struct duplicate_item *new_duplicate; > > + struct item *current_item = list; > > + bool prev_was_duplicate = false; > > + struct item *prev_item = NULL; > > + > > + while (current_item) { > > + if ((prev_item && (strcmp(current_item->symb_name, > > prev_item->symb_name) == 0)) || + prev_was_duplicate) { > > + if (!duplicates) { > > + duplicates = malloc(sizeof(struct > > duplicate_item)); + if (!duplicates) > > + return NULL; > > + > > + duplicates->original_item = prev_item; > > + duplicates->next = NULL; > > + current_duplicate = duplicates; > > + } else { > > + new_duplicate = malloc(sizeof(struct > > duplicate_item)); + if (!new_duplicate) { > > + free_duplicates(&duplicates); > > + return NULL; > > + } > > + > > + new_duplicate->original_item = prev_item; > > + new_duplicate->next = NULL; > > + current_duplicate->next = new_duplicate; > > + current_duplicate = new_duplicate; > > + > > + if ((strcmp(current_item->symb_name, > > prev_item->symb_name) != 0) && + > > (prev_was_duplicate)) > > + prev_was_duplicate = false; > > + else > > + prev_was_duplicate = true; > > + } > > + } > > + > > + prev_item = current_item; > > + current_item = current_item->next; > > + } > > + > > + return duplicates; > > +} > > + > > +void free_duplicates(struct duplicate_item **duplicates) > > +{ > > + struct duplicate_item *duplicates_iterator = *duplicates; > > + struct duplicate_item *app; > > + > > + while (duplicates_iterator) { > > + app = duplicates_iterator; > > + duplicates_iterator = duplicates_iterator->next; > > + free(app); > > + } > > + > > + *duplicates = NULL; > > +} > > diff --git a/scripts/kas_alias/duplicates_list.h > > b/scripts/kas_alias/duplicates_list.h new file mode 100644 > > index 000000000000..76aa73e584bc > > --- /dev/null > > +++ b/scripts/kas_alias/duplicates_list.h > > @@ -0,0 +1,15 @@ > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > +#ifndef DUPLICATES_LIST_H > > +#define DUPLICATES_LIST_H > > + > > +#include "item_list.h" > > + > > +struct duplicate_item { > > + struct item *original_item; > > + struct duplicate_item *next; > > +}; > > + > > +struct duplicate_item *find_duplicates(struct item *list); > > +void free_duplicates(struct duplicate_item **duplicates); > > + > > +#endif > > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > > new file mode 100644 > > index 000000000000..48f2e525592a > > --- /dev/null > > +++ b/scripts/kas_alias/item_list.c > > @@ -0,0 +1,230 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdio.h> > > +#include <stdlib.h> > > +#include <stdint.h> > > +#include <string.h> > > +#include <stdbool.h> > > +#include <assert.h> > > +#include "item_list.h" > > + > > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, > > (temp)->symb_name) op 0) + > > +struct item *list_index[96] = {0}; > > + > > +void build_index(struct item *list) > > +{ > > + char current_first_letter = ' '; > > + struct item *current = list; > > + > > + while (current) { > > + if (current->symb_name[0] != current_first_letter) { > > + current_first_letter = current->symb_name[0]; > > + list_index[current_first_letter - 32] = current; > > + } > > + current = current->next; > > + } > > +} > > + > > +struct item *add_item(struct item **list, const char *name, char stype, > > uint64_t addr) +{ > > + struct item *new_item; > > + struct item *current; > > + > > + new_item = malloc(sizeof(struct item)); > > + if (!new_item) > > + return NULL; > > + > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > + new_item->addr = addr; > > + new_item->stype = stype; > > + new_item->next = NULL; > > + > > + if (!(*list)) { > > + *list = new_item; > > + } else { > > + current = *list; > > + while (current->next) > > + current = current->next; > > + > > + current->next = new_item; > > + } > > + return new_item; > > +} > > + > > +void sort_list(struct item **list, int sort_by) > > +{ > > + struct item *current = *list; > > + struct item *sorted = NULL; > > + struct item *next_item; > > + struct item *temp; > > + > > + if (!(*list) || !((*list)->next)) > > + return; > > + > > + while (current) { > > + next_item = current->next; > > + if (!sorted || > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) > > || > > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > > + current->next = sorted; > > + sorted = current; > > + } else { > > + temp = sorted; > > + while (temp->next && > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, > > temp->next, >=) || + > > CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) + > > temp = temp->next; > > + > > + current->next = temp->next; > > + temp->next = current; > > + } > > + current = next_item; > > + } > > + > > + *list = sorted; > > +} > > + > > +struct item *merge(struct item *left, struct item *right, int sort_by) > > +{ > > + struct item *current = NULL; > > + struct item *result = NULL; > > + > > + if (!left) > > + return right; > > + if (!right) > > + return left; > > + > > + if (sort_by == BY_NAME) { > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > + result = left; > > + left = left->next; > > + } else { > > + result = right; > > + right = right->next; > > + } > > + } else { > > + if (sort_by == BY_ADDRESS) { > > + if (left->addr <= right->addr) { > > + result = left; > > + left = left->next; > > + } else { > > + result = right; > > + right = right->next; > > + } > > + } > > + } > > + > > + current = result; > > + > > + while (left && right) { > > + if (sort_by == BY_NAME) { > > + if (strcmp(left->symb_name, right->symb_name) <= > > 0) { + current->next = left; > > + left = left->next; > > + } else { > > + current->next = right; > > + right = right->next; > > + } > > + } else { > > + if (sort_by == BY_ADDRESS) { > > + if (left->addr <= right->addr) { > > + current->next = left; > > + left = left->next; > > + } else { > > + current->next = right; > > + right = right->next; > > + } > > + } > > + } > > + > > + current = current->next; > > + } > > + > > + if (left) { > > + current->next = left; > > + } else { > > + if (right) > > + current->next = right; > > + } > > + > > + return result; > > +} > > + > > +struct item *merge_sort(struct item *head, int sort_by) > > +{ > > + struct item *right; > > + struct item *slow; > > + struct item *fast; > > + struct item *left; > > + > > + if (!head || !head->next) > > + return head; > > + > > + slow = head; > > + fast = head->next; > > + > > + while (fast && fast->next) { > > + slow = slow->next; > > + fast = fast->next->next; > > + } > > + > > + left = head; > > + right = slow->next; > > + slow->next = NULL; > > + > > + left = merge_sort(left, sort_by); > > + right = merge_sort(right, sort_by); > > + > > + return merge(left, right, sort_by); > > +} > > + > > +void sort_list_m(struct item **head, int sort_by) > > +{ > > + if (!(*head) || !((*head)->next)) > > + return; > > + > > + *head = merge_sort(*head, sort_by); > > +} > > + > > +int insert_after(struct item *list, const uint64_t search_addr, > > + const char *name, uint64_t addr, char stype) > > +{ > > + struct item *new_item; > > + struct item *current; > > + int ret = 0; > > + > > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : > > list; + while (current) { > > + if (current->addr == search_addr) { > > + new_item = malloc(sizeof(struct item)); > > + if (!new_item) > > + return ret; > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > + new_item->addr = addr; > > + new_item->stype = stype; > > + new_item->next = current->next; > > + current->next = new_item; > > + ret = 1; > > + break; > > + } > > + current = current->next; > > + } > > + return ret; > > +} > > + > > +void free_items(struct item **head) > > +{ > > + struct item *app, *item_iterator = *head; > > + > > + while (item_iterator) { > > + app = item_iterator; > > + item_iterator = item_iterator->next; > > + free(app); > > + } > > + *head = NULL; > > +} > > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > > new file mode 100644 > > index 000000000000..b4891cb088ee > > --- /dev/null > > +++ b/scripts/kas_alias/item_list.h > > @@ -0,0 +1,26 @@ > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > +#ifndef ITEM_LIST_H > > +#define ITEM_LIST_H > > +#include <stdint.h> > > + > > +#define MAX_NAME_SIZE 256 > > +#define BY_ADDRESS 1 > > +#define BY_NAME 2 > > + > > +struct item { > > + char symb_name[MAX_NAME_SIZE]; > > + uint64_t addr; > > + char stype; > > + struct item *next; > > +}; > > + > > +void build_index(struct item *list); > > +struct item *add_item(struct item **list, const char *name, char stype, > > uint64_t addr); +void sort_list(struct item **list, int sort_by); > > +struct item *merge(struct item *left, struct item *right, int sort_by); > > +struct item *merge_sort(struct item *head, int sort_by); > > +void sort_list_m(struct item **head, int sort_by); > > +int insert_after(struct item *list, const uint64_t search_addr, > > + const char *name, uint64_t addr, char stype); > > +void free_items(struct item **head); > > +#endif > > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > > new file mode 100644 > > index 000000000000..532aeb39f851 > > --- /dev/null > > +++ b/scripts/kas_alias/kas_alias.c > > @@ -0,0 +1,217 @@ > > +// SPDX-License-Identifier: GPL-2.0-or-later > > +#include <stdio.h> > > +#include <stdlib.h> > > +#include <stdint.h> > > +#include <unistd.h> > > +#include <string.h> > > +#include <stdbool.h> > > +#include <stdarg.h> > > +#include <regex.h> > > + > > +#include "item_list.h" > > +#include "duplicates_list.h" > > +#include "a2l.h" > > + > > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') > > || \ + (((s)->stype) == 'd') || (((s)->stype) == > > 'D') || \ + (((s)->stype) == 'r') || > > (((s)->stype) == 'R')) +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > > +#else > > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > > +#endif > > +#define FNOMATCH 0 > > +#define FMATCH 1 > > +#define EREGEX 2 > > + > > +const char *ignore_regex[] = { > > + "^__cfi_.*$", // __cfi_ preamble > > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > > + "^_*TRACE_SYSTEM.*$", > > + "^__already_done\\.[0-9]+$", // Call a function once > > data + "^___tp_str\\.[0-9]+$", > > + "^___done\\.[0-9]+$", > > + "^__print_once\\.[0-9]+$", > > + "^_rs\\.[0-9]+$", > > + "^__compound_literal\\.[0-9]+$", > > + "^___once_key\\.[0-9]+$", > > + "^__func__\\.[0-9]+$", > > + "^__msg\\.[0-9]+$", > > + "^CSWTCH\\.[0-9]+$", > > + "^__flags\\.[0-9]+$", > > + "^__wkey.*$", > > + "^__mkey.*$", > > + "^__key.*$", > > +#endif > > + "^__pfx_.*$" // NOP-padding > > +}; > > + > > +int suffix_serial; > > + > > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > > +{ > > + va_list args; > > + > > + va_start(args, fmt); > > + if (verbose) > > + printf(fmt, args); > > + > > + va_end(args); > > +} > > + > > +static void create_suffix(const char *name, char *output_suffix) > > +{ > > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > > +} > > + > > +static void create_file_suffix(const char *name, uint64_t address, char > > *output_suffix, char *cwd) +{ > > + const char *f_path; > > + char *buf; > > + int i = 0; > > + > > + buf = addr2line_get_lines(address); > > + f_path = remove_subdir(cwd, buf); > > + if (f_path) { > > + sprintf(output_suffix, "%s@%s", name, f_path); > > + while (*(output_suffix + i) != '\0') { > > + switch (*(output_suffix + i)) { > > + case '/': > > + case ':': > > + case '.': > > + *(output_suffix + i) = '_'; > > + break; > > + default: > > + } > > + i++; > > + } > > + } else { > > + create_suffix(name, output_suffix); > > + } > > +} > > + > > +static int filter_symbols(char *symbol, const char **ignore_list, int > > regex_no) +{ > > + regex_t regex; > > + int res, i; > > + > > + for (i = 0; i < regex_no; i++) { > > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > > + if (res) > > + return -EREGEX; > > + > > + res = regexec(®ex, symbol, 0, NULL, 0); > > + regfree(®ex); > > + switch (res) { > > + case 0: > > + return FMATCH; > > + case REG_NOMATCH: > > + break; > > + default: > > + return -EREGEX; > > + } > > + } > > + > > + return FNOMATCH; > > +} > > + > > +int main(int argc, char *argv[]) > > +{ > > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > > + struct duplicate_item *duplicate_iterator; > > + struct duplicate_item *duplicate; > > + struct item *head = {NULL}; > > + bool need_2_process = true; > > + struct item *last = {NULL}; > > + struct item *current; > > + int verbose_mode = 0; > > + uint64_t address; > > + FILE *fp; > > + int res; > > + > > + if (argc < 2 || argc > 3) { > > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > > + return 1; > > + } > > + > > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > > + verbose_mode = 1; > > + > > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > > + > > + fp = fopen(argv[1], "r"); > > + if (!fp) { > > + printf("Can't open input file.\n"); > > + return 1; > > + } > > + > > + if (!addr2line_init(get_addr2line(A2L_DEFAULT), > > get_vmlinux(argv[1]))) + return 1; > > + > > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > > + if (strstr(sym_name, "@_")) { > > + if (verbose_mode && need_2_process) > > + printf("Already processed\n"); > > + need_2_process = false; > > + } > > + last = add_item(&last, sym_name, t, address); > > + if (!last) { > > + printf("Error in allocate memory\n"); > > + free_items(&head); > > + return 1; > > + } > > + > > + if (!head) > > + head = last; > > + } > > + > > + fclose(fp); > > + > > + if (need_2_process) { > > + verbose_msg(verbose_mode, "Sorting nm data\n"); > > + sort_list_m(&head, BY_NAME); > > + verbose_msg(verbose_mode, "Scanning nm data for > > duplicates\n"); + duplicate = find_duplicates(head); > > + if (!duplicate) { > > + printf("Error in duplicates list\n"); > > + return 1; > > + } > > + > > + verbose_msg(verbose_mode, "Applying suffixes\n"); > > + build_index(head); > > + duplicate_iterator = duplicate; > > + while (duplicate_iterator) { > > + res = > > filter_symbols(duplicate_iterator->original_item->symb_name, + > > ignore_regex, sizeof(ignore_regex) / + > > sizeof(ignore_regex[0])); + > > if (res != FMATCH && > > + > > SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { + > > if (res < 0) > > + return 1; > > + > > + > > create_file_suffix(duplicate_iterator->original_item->symb_name, + > > > > duplicate_iterator->original_item->addr, + > > new_name, vmlinux_path); + > > if (!insert_after(head, duplicate_iterator->original_item->addr, + > > new_name, > > duplicate_iterator->original_item->addr, + > > duplicate_iterator->original_item->stype)) + > > return 1; > > + } > > + > > + duplicate_iterator = duplicate_iterator->next; > > + } > > + > > + sort_list_m(&head, BY_ADDRESS); > > + } > > + current = head; > > + while (current) { > > + printf("%08lx %c %s\n", current->addr, current->stype, > > current->symb_name); + current = current->next; > > + } > > + > > + free_items(&head); > > + free_duplicates(&duplicate); > > + addr2line_cleanup(); > > + return 0; > > +} > > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > > index a432b171be82..cacf60b597ce 100755 > > --- a/scripts/link-vmlinux.sh > > +++ b/scripts/link-vmlinux.sh > > @@ -89,8 +89,9 @@ vmlinux_link() > > > > ldflags="${ldflags} ${wl}--script=${objtree}/${KBUILD_LDS}" > > > > - # The kallsyms linking does not need debug symbols included. > > - if [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > > + # The kallsyms linking does not need debug symbols included, > > unless the KALLSYMS_ALIAS. + if [ ! is_enabled > > CONFIG_KALLSYMS_ALIAS ] && \ > > + [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > > > > ldflags="${ldflags} ${wl}--strip-debug" > > > > fi > > > > @@ -161,7 +162,11 @@ kallsyms() > > > > fi > > > > info KSYMS ${2} > > > > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > > + ALIAS=".alias" > > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > > + fi > > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > > > > } > > > > # Perform one step in kallsyms generation, including temporary linking of > > > > -- > > 2.34.1 Best regards.
Hello Francis, Thanks for your input! Il giorno gio 24 ago 2023 alle ore 17:35 Francis Laniel <flaniel@linux.microsoft.com> ha scritto: > > Hi. > > Le vendredi 21 juillet 2023, 14:40:54 CEST Alessandro Carminati a écrit : > > Hello, > > > > I apologize for being noisy today. > > > > In an effort to be collaborative, I would like to share my thoughts on why I > > see duplicate symbols in fs/binfmt_elf.c. > > > > Il giorno ven 21 lug 2023 alle ore 11:22 Alessandro Carminati > > > > <alessandro.carminati@gmail.com> ha scritto: > > > Hello Steven, Alexander, and Nick, > > > > > > Since now I realize that a lot of people are working on my very same > > > argument, I'm writing you this mail just with the intent of not leaving > > > my statements floating in the air, and help the cause by sharing my > > > experience, as little it is, to solve this problem. > > > > > > The addr2line strategy, I tried to implement, allows me to achieve the > > > level of accuracy that Luis mentioned in his comment about live-patching > > > suggestions. The addr2line-based approach is feasible, however, to make > > > this approach operable, I had to ensure that vmlinux contains the > > > necessary debug information. > > > Let me clarify. > > > During the kernel build, I do need the vmlinux to contain debug > > > information. The approach I pursued parses the nm output and, when it > > > detects a duplicate symbol, it needs to query vmlinux using addr2line to > > > resolve the address where the symbol is reported to be. > > > The whole process takes place during the building phase. > > > I've completed the work to integrate the new alias within the compilation > > > phase, and I wanted to share with you how it looks now. > > > > > > ``` > > > ~ # uname -a > > > Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 > > > GNU/Linux ~ # cat /proc/kallsyms | grep gic_mask_irq > > > ffffacf2eb64dae4 t gic_mask_irq > > > ffffacf2eb64dae4 t gic_mask_irq@_drivers_irqchip_irq-gic_c_167 > > > ffffacf2eb650960 t gic_mask_irq > > > ffffacf2eb650960 t gic_mask_irq@_drivers_irqchip_irq-gic-v3_c_404 > > > ~ # > > > ``` > > > > > > Despite this appearing to have a sufficient level of accuracy, when I > > > tested it, I realized something I really did not expect. Multiple > > > instances of the same function exist, and when I say "same function," I > > > don't just mean the name, but also the function's body. I apologize for > > > stating something that may sound obvious to you, but I found this really > > > unexpected. > > > > > > ``` > > > ~ # uname -a > > > Linux (none) 6.4.0 #1 SMP PREEMPT Thu Jul 20 10:05:36 UTC 2023 aarch64 > > > GNU/Linux ~ # cat /proc/kallsyms | grep -E " [ttT] " | cut -d" " > > > -f3|sort| uniq -d| grep @_ > > > BIT_initDStream@_lib_zstd_common_bitstream_h_259 > > > __dev_hold@_include_linux_netdevice_h_4059 > > > __dev_put@_include_linux_netdevice_h_4048 > > > __flush_tlb_range@_arch_arm64_include_asm_tlbflush_h_291 > > > __flush_tlb_range_constprop_0@_arch_arm64_include_asm_tlbflush_h_291 > > > __kern_my_cpu_offset@_arch_arm64_include_asm_percpu_h_40 > > > __kvm_nvhe_$x@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22 > > > __kvm_nvhe___kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34 > > > __kvm_nvhe_hyp_page_count@_arch_arm64_kvm_hyp_include_nvhe_memory_h_47 > > > __kvm_nvhe_hyp_phys_to_virt@_arch_arm64_kvm_hyp_include_nvhe_memory_h_22 > > > __kvm_nvhe_hyp_virt_to_phys@_arch_arm64_kvm_hyp_include_nvhe_memory_h_27 > > > __kvm_skip_instr@_arch_arm64_kvm_hyp_include_hyp_adjust_pc_h_34 > > > __nodes_weight_constprop_0@_include_linux_nodemask_h_239 > > > __pi_$x@_scripts_dtc_libfdt_libfdt_h_145 > > > __pi_fdt32_ld@_scripts_dtc_libfdt_libfdt_h_145 > > > __preempt_count_dec_and_test@_arch_arm64_include_asm_current_h_19 > > > acpi_dev_filter_resource_type_cb@_include_linux_acpi_h_520 > > > add_quirk@_drivers_mmc_core_card_h_158 > > > bpf_enable_instrumentation@_include_linux_bpf_h_1937 > > > btf_id_cmp_func@_include_linux_btf_h_469 > > > copy_from_sockptr_offset_constprop_0@_include_linux_sockptr_h_44 > > > cpucap_multi_entry_cap_matches@_arch_arm64_include_asm_cpufeature_h_395 > > > cpufreq_register_em_with_opp@_include_linux_cpufreq_h_1228 > > > cpumask_weight@_include_linux_cpumask_h_691 > > > cpumask_weight_constprop_0@_include_linux_cpumask_h_690 > > > css_put@_include_linux_cgroup_refcnt_h_77 > > > dma_cookie_status@_drivers_dma_dmaengine_h_74 > > > dst_discard@_include_net_dst_h_392 > > > dst_output@_include_net_dst_h_457 > > > elf_core_dump@_fs_binfmt_elf_c_2027 > > > fixup_use_fwh_lock@_drivers_mtd_chips_fwh_lock_h_102 > > > flush_tlb_mm@_arch_arm64_include_asm_tlbflush_h_250 > > > fwh_lock_varsize@_drivers_mtd_chips_fwh_lock_h_81 > > > fwh_unlock_varsize@_drivers_mtd_chips_fwh_lock_h_92 > > > fwh_xxlock_oneblock@_drivers_mtd_chips_fwh_lock_h_31 > > > get_net_ns@_include_net_net_namespace_h_231 > > > inline_map_read_isra_0@_include_linux_mtd_map_h_393 > > > inline_map_write_part_0@_include_linux_mtd_map_h_426 > > > io_run_task_work@_io_uring_io_uring_h_282 > > > ipv6_portaddr_hash_isra_0@_include_net_ipv6_h_732 > > > irq_find_host@_include_linux_irqdomain_h_322 > > > jhash@_include_linux_jhash_h_76 > > > jhash_constprop_0@_include_linux_jhash_h_76 > > > ktime_get_boottime@_include_linux_timekeeping_h_94 > > > ktime_get_real@_include_linux_timekeeping_h_78 > > > load_elf_binary@_fs_binfmt_elf_c_824 > > > load_elf_phdrs@_fs_binfmt_elf_c_468 > > > may_use_simd@_arch_arm64_include_asm_alternative-macros_h_232 > > > netif_tx_disable@_include_linux_netdevice_h_4486 > > > of_parse_phandle@_include_linux_of_h_946 > > > of_parse_phandle_constprop_0@_include_linux_of_h_943 > > > padzero@_fs_binfmt_elf_c_142 > > > percpu_down_read_constprop_0@_include_linux_percpu-rwsem_h_47 > > > percpu_ref_get_many@_include_linux_percpu-refcount_h_199 > > > percpu_ref_get_many_constprop_0@_include_linux_percpu-refcount_h_198 > > > percpu_ref_put_many_constprop_0@_include_linux_percpu-refcount_h_326 > > > percpu_up_read@_include_linux_percpu-rwsem_h_98 > > > percpu_up_read_constprop_0@_include_linux_percpu-rwsem_h_97 > > > pinconf_generic_dt_node_to_map_all@_include_linux_pinctrl_pinconf-generic_ > > > h_223 > > > pinconf_generic_dt_node_to_map_group@_include_linux_pinctrl_pinconf-gener > > > ic_h_207 > > > pinconf_generic_dt_node_to_map_pin@_include_linux_pinctrl_pinconf-generic > > > _h_215 > > > platform_device_register_resndata_constprop_0@_include_linux_platform_dev > > > ice_h_125 > > > ptrauth_keys_install_user@_arch_arm64_include_asm_alternative-macros_h_23 > > > 2 readl@_arch_arm64_include_asm_io_h_75 > > > reqsk_put@_include_net_request_sock_h_133 > > > rht_key_get_hash_constprop_0_isra_0@_include_linux_rhashtable_h_133 > > > rht_key_get_hash_isra_0@_include_linux_rhashtable_h_125 > > > role_show@_include_linux_device_h_763 > > > sdhci_and_cqhci_reset@_drivers_mmc_host_sdhci-cqhci_h_16 > > > set_active_memcg_part_0@_include_linux_sched_mm_h_410 > > > set_brk@_fs_binfmt_elf_c_114 > > > set_is_seen@_arch_arm64_include_asm_current_h_19 > > > set_lookup@_arch_arm64_include_asm_current_h_19 > > > sha1_base_init@_include_crypto_sha1_base_h_22 > > > sha224_base_init@_include_crypto_sha256_base_h_23 > > > sha256_base_init@_include_crypto_sha256_base_h_31 > > > spi_sync_transfer_constprop_0@_include_linux_spi_spi_h_1339 > > > spi_write@_include_linux_spi_spi_h_1363 > > > tlb_flush@_arch_arm64_include_asm_tlb_h_54 > > > trace_xhci_dbg_quirks@_drivers_usb_host_xhci-trace_h_48 > > > udp_lib_close@_include_net_udp_h_197 > > > udp_lib_hash@_include_net_udp_h_189 > > > virtio_net_hdr_to_skb_constprop_0@_include_linux_virtio_net_h_48 > > > writenote@_fs_binfmt_elf_c_1479 > > > zero_user_segments@_include_linux_highmem_h_271 > > > zero_user_segments_constprop_0@_include_linux_highmem_h_268 > > > ``` > > > > > > The previous evidence demonstrates that adding a tag with a file and line > > > number does not make a symbol unique. > > > > > > Here, I inspect the function body to verify if the functions are indeed > > > the same. > > > > > > ``` > > > Welcome to Buildroot > > > buildroot login: root > > > ~ # cat /proc/kallsyms | grep set_brk > > > ffffa8a9d1cf1d2c t set_brk > > > ffffa8a9d1cf1d2c t set_brk@_fs_binfmt_elf_c_114 > > > ffffa8a9d1cf4454 t set_brk > > > ffffa8a9d1cf4454 t set_brk@_fs_binfmt_elf_c_114 > > > ~ # QEMU: Terminated > > > $ cat System.map | grep set_brk > > > ffff8000082f1d2c t set_brk > > > ffff8000082f4454 t set_brk > > > ~ $ r2 vmlinux > > > Warning: run r2 with -e bin.cache=true to fix relocations in disassembly > > > > > > -- Reduce the delta where flag resolving by address is used with > > > cfg.delta > > > > > > [0xffff800008000000]> s 0xffff8000082f1d2c > > > [0xffff8000082f1d2c]> aa > > > [x] Analyze all flags starting with sym. and entry0 (aa) > > > [x] Analyze all functions arguments/locals > > > [0xffff8000082f1d2c]> pdf > > > > > > ; CALL XREFS from sym.load_elf_binary @ 0xffff8000082f352c(x), > > > 0xffff8000082f38d4(x), 0xffff8000082f3a68(x)> > > > ┌ 112: sym.set_brk (int64_t arg1, int64_t arg2, int64_t arg3); > > > │ ; arg int64_t arg1 @ x0 > > > │ ; arg int64_t arg2 @ x1 > > > │ ; arg int64_t arg3 @ x2 > > > │ ; var int64_t var_10h @ sp+0x10 > > > │ 0xffff8000082f1d2c 3f2303d5 paciasp ; > > > binfmt_elf.c:114 │ 0xffff8000082f1d30 fd7bbea9 stp > > > x29, x30, [sp, -0x20]! │ 0xffff8000082f1d34 00fc3f91 > > > add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1 │ > > > 0xffff8000082f1d38 fd030091 mov x29, sp ; > > > binfmt_elf.c:114 │ 0xffff8000082f1d3c 21fc3f91 add > > > x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2 │ 0xffff8000082f1d40 > > > 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; > > > arg1 │ 0xffff8000082f1d44 f30b00f9 str x19, [sp, > > > 0x10] ; binfmt_elf.c:116 │ 0xffff8000082f1d48 33cc7492 > > > and x19, x1, 0xfffffffffffff000 ; arg2 │ 0xffff8000082f1d4c > > > 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1 │ > > > ┌─< 0xffff8000082f1d50 63010054 b.lo 0xffff8000082f1d7c │ > > > ┌──> 0xffff8000082f1d54 014138d5 mrs x1, sp_el0 ; > > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f1d58 22fc41f9 ldr > > > x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217 │ ╎│ > > > 0xffff8000082f1d5c 00008052 mov w0, 0 ; > > > binfmt_elf.c:129 │ ╎│ 0xffff8000082f1d60 539400f9 str > > > x19, [x2, 0x128] ; binfmt_elf.c:128 │ ╎│ 0xffff8000082f1d64 > > > 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217 │ ╎│ > > > 0xffff8000082f1d68 339000f9 str x19, [x1, 0x120] ; > > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f1d6c f30b40f9 ldr > > > x19, [var_10h] ; binfmt_elf.c:129 ; 5 │ ╎│ 0xffff8000082f1d70 > > > fd7bc2a8 ldp x29, x30, [sp], 0x20 │ ╎│ 0xffff8000082f1d74 > > > bf2303d5 autiasp > > > │ ╎│ 0xffff8000082f1d78 c0035fd6 ret > > > │ ╎└─> 0xffff8000082f1d7c 42007e92 and x2, x2, 4 ; > > > binfmt_elf.c:123 ; arg3 │ â•Ž 0xffff8000082f1d80 610200cb > > > sub x1, x19, x0 ; arg1 │ â•Ž 0xffff8000082f1d84 cae3fc97 > > > bl sym.vm_brk_flags │ └──< 0xffff8000082f1d88 60feff34 > > > cbz w0, 0xffff8000082f1d54 ; binfmt_elf.c:125 │ > > > 0xffff8000082f1d8c f30b40f9 ldr x19, [var_10h] ; > > > binfmt_elf.c:130 ; 5 │ 0xffff8000082f1d90 fd7bc2a8 > > > ldp x29, x30, [sp], 0x20 │ 0xffff8000082f1d94 bf2303d5 > > > autiasp > > > â”” 0xffff8000082f1d98 c0035fd6 ret > > > [0xffff8000082f1d2c]> s 0xffff8000082f4454 > > > [0xffff8000082f4454]> pdf > > > > > > ; CALL XREFS from sym.load_elf_binary_1 @ > > > 0xffff8000082f5cc0(x), 0xffff8000082f5e5c(x), > > > 0xffff8000082f6648(x)> > > > ┌ 112: sym.set_brk_1 (int64_t arg1, int64_t arg2, int64_t arg3); > > > │ ; arg int64_t arg1 @ x0 > > > │ ; arg int64_t arg2 @ x1 > > > │ ; arg int64_t arg3 @ x2 > > > │ ; var int64_t var_10h @ sp+0x10 > > > │ 0xffff8000082f4454 3f2303d5 paciasp ; > > > binfmt_elf.c:114 │ 0xffff8000082f4458 fd7bbea9 stp > > > x29, x30, [sp, -0x20]! │ 0xffff8000082f445c 00fc3f91 > > > add x0, x0, 0xfff ; binfmt_elf.c:115 ; arg1 │ > > > 0xffff8000082f4460 fd030091 mov x29, sp ; > > > binfmt_elf.c:114 │ 0xffff8000082f4464 21fc3f91 add > > > x1, x1, 0xfff ; binfmt_elf.c:116 ; arg2 │ 0xffff8000082f4468 > > > 00cc7492 and x0, x0, 0xfffffffffffff000 ; binfmt_elf.c:114 ; > > > arg1 │ 0xffff8000082f446c f30b00f9 str x19, [sp, > > > 0x10] ; binfmt_elf.c:116 │ 0xffff8000082f4470 33cc7492 > > > and x19, x1, 0xfffffffffffff000 ; arg2 │ 0xffff8000082f4474 > > > 1f0013eb cmp x0, x19 ; binfmt_elf.c:117 ; arg1 │ > > > ┌─< 0xffff8000082f4478 63010054 b.lo 0xffff8000082f44a4 │ > > > ┌──> 0xffff8000082f447c 014138d5 mrs x1, sp_el0 ; > > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f4480 22fc41f9 ldr > > > x2, [x1, 0x3f8] ; current.h:21 ; 0xd9 ; 217 │ ╎│ > > > 0xffff8000082f4484 00008052 mov w0, 0 ; > > > binfmt_elf.c:129 │ ╎│ 0xffff8000082f4488 539400f9 str > > > x19, [x2, 0x128] ; binfmt_elf.c:128 │ ╎│ 0xffff8000082f448c > > > 21fc41f9 ldr x1, [x1, 0x3f8] ; current.h:17 ; 0xd9 ; 217 │ ╎│ > > > 0xffff8000082f4490 339000f9 str x19, [x1, 0x120] ; > > > binfmt_elf.c:128 │ ╎│ 0xffff8000082f4494 f30b40f9 ldr > > > x19, [var_10h] ; binfmt_elf.c:129 ; 5 │ ╎│ 0xffff8000082f4498 > > > fd7bc2a8 ldp x29, x30, [sp], 0x20 │ ╎│ 0xffff8000082f449c > > > bf2303d5 autiasp > > > │ ╎│ 0xffff8000082f44a0 c0035fd6 ret > > > │ ╎└─> 0xffff8000082f44a4 42007e92 and x2, x2, 4 ; > > > binfmt_elf.c:123 ; arg3 │ â•Ž 0xffff8000082f44a8 610200cb > > > sub x1, x19, x0 ; arg1 │ â•Ž 0xffff8000082f44ac 00dafc97 > > > bl sym.vm_brk_flags │ └──< 0xffff8000082f44b0 60feff34 > > > cbz w0, 0xffff8000082f447c ; binfmt_elf.c:125 │ > > > 0xffff8000082f44b4 f30b40f9 ldr x19, [var_10h] ; > > > binfmt_elf.c:130 ; 5 │ 0xffff8000082f44b8 fd7bc2a8 > > > ldp x29, x30, [sp], 0x20 │ 0xffff8000082f44bc bf2303d5 > > > autiasp > > > â”” 0xffff8000082f44c0 c0035fd6 ret > > > [0xffff8000082f4454]> > > > ``` > > > > > > While I can speculate on a reasonable explanation for why I see symbols > > > coming from header files being duplicated – when a header file is > > > included in a C file and it produces a function that the compiler does > > > not want to inline, it is replicated every time the file is included – I > > > have no convincing explanation as to why the fs/binfmt.c functions exist > > > in more than one instance. > > > > > > ``` > > > elf_core_dump@_fs_binfmt_elf_c_2027 > > > load_elf_binary@_fs_binfmt_elf_c_824 > > > load_elf_phdrs@_fs_binfmt_elf_c_468 > > > padzero@_fs_binfmt_elf_c_142 > > > set_brk@_fs_binfmt_elf_c_114 > > > writenote@_fs_binfmt_elf_c_1479 > > > ``` > > > > After examining the contents of built-in.a, I have come up with a new > > interpretation of what I am observing. > > According to built-in.a, System.map, and vmlinux there are two symbols named > > set_brk. > > > > ``` > > ~ $ cat System.map | grep set_brk > > ffff8000082f1d2c t set_brk > > ffff8000082f4454 t set_brk > > ~ $ > > ~ $ nm -n vmlinux | grep set_brk > > ffff8000082f1d2c t set_brk > > ffff8000082f4454 t set_brk > > ~ $ > > ~ $ nm -n built-in.a| grep set_brk > > 00000000000000d4 t set_brk > > 00000000000000d4 t set_brk > > ~ $ > > ~ $ nm -n built-in.a | grep set_brk -B100| egrep "set_brk|\.o:$" > > fs/binfmt_elf.o: > > 00000000000000d4 t set_brk > > fs/compat_binfmt_elf.o: > > 00000000000000d4 t set_brk > > ``` > > > > These two symbols come from fs/binfmt_elf.o and fs/compat_binfmt_elf.o, and > > they are just two symbols that happen to share the same name, as is common > > in the kernel. > > > > at the same time, addr2line reports symbols to be generated from the same > > file. > > > > ``` > > ~ $ llvm-addr2line-14 -fe vmlinux ffff8000082f1d2c ffff8000082f4454 > > set_brk > > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > > set_brk > > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > > ~ $ > > ~ $ addr2line -fe vmlinux ffff8000082f1d2c ffff8000082f4454 > > set_brk > > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > > set_brk > > /home/alessandro/src/lka64/linux-6.4/fs/binfmt_elf.c:114 > > ``` > > looking at the source code: > > https://elixir.bootlin.com/linux/v6.4/source/fs/compat_binfmt_elf.c#L144 > > the cause of this behavior, which is unexpected but correct. > > Thank you for the investigation, this an interesting reading! > > > The rationale is that using source file + line number iproduces better > > kallsyms table, but it is still do not produce unique names. > > When I first read your cover letter, I also thought it would be cool to have > the filename rather than a serial ID. > So, your latest modification is really good as it goes into this direction. > Regarding having non unique names, I am not sure if this is a problem. > In my case, which is using kprobe to attach BPF code to trace some events, > your contribution would make me on the way to know which address I should use > because I could choose regarding the filename. This whole thing got trickier than I expected when I started. At first, I was just grappling with different functions having the same name. But during my investigation, it hit me that I've got to handle multiple copies of the same function too. So, those addresses labeled with the same name? They're not just different functions having the same name - they're a mix of functions that happen to share a name and functions that are the same but come from different compile units. Imagine this: one compile unit defines a function foo1() that uses a bar() from a shared header. Another compile unit defines foo2(), also relying on bar(). The snag is, these compile units end up being linked together, carrying their own copies of bar(). So, if I wanna keep an eye on bar(), which copy should I watch? The answer depends on whose bar() I want to watch at. I could go for probing them all, sure, but that's kind of a suboptimal solution that waters down what I'm trying to achieve. > > I am curious about a potential v3 of this patch! The way things are being done right now, even though it's nice that it can track down these copies in the kernel image, it doesn't quite hit the bullseye of what I was going for. But hey, I'm totally open to putting out v3 with the limitations I just talked about, and discuss if it can be improved. > > > BR > > Alessandro > > > > > As a final note, please understand that my patch was not intended to > > > undermine anyone's work. I simply encountered a problem that I wanted to > > > help solve. Attached to this message is my code, in case anyone wants to > > > replicate it. I would appreciate being kept in the loop, as I genuinely > > > want to assist in fixing this problem. > > > > > > BR > > > Alessandro > > > --- > > > > > > init/Kconfig | 36 ++++ > > > scripts/Makefile | 4 + > > > scripts/kas_alias/Makefile | 4 + > > > scripts/kas_alias/a2l.c | 268 ++++++++++++++++++++++++++++ > > > scripts/kas_alias/a2l.h | 32 ++++ > > > scripts/kas_alias/duplicates_list.c | 70 ++++++++ > > > scripts/kas_alias/duplicates_list.h | 15 ++ > > > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++ > > > scripts/kas_alias/item_list.h | 26 +++ > > > scripts/kas_alias/kas_alias.c | 217 ++++++++++++++++++++++ > > > scripts/link-vmlinux.sh | 11 +- > > > 11 files changed, 910 insertions(+), 3 deletions(-) > > > create mode 100644 scripts/kas_alias/Makefile > > > create mode 100644 scripts/kas_alias/a2l.c > > > create mode 100644 scripts/kas_alias/a2l.h > > > create mode 100644 scripts/kas_alias/duplicates_list.c > > > create mode 100644 scripts/kas_alias/duplicates_list.h > > > create mode 100644 scripts/kas_alias/item_list.c > > > create mode 100644 scripts/kas_alias/item_list.h > > > create mode 100644 scripts/kas_alias/kas_alias.c > > > > > > diff --git a/init/Kconfig b/init/Kconfig > > > index f7f65af4ee12..bc69fcd9cbc8 100644 > > > --- a/init/Kconfig > > > +++ b/init/Kconfig > > > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > > > > > > time constants, and no relocation pass is required at runtime to > > > fix > > > up the entries based on the runtime load address of the kernel. > > > > > > +config KALLSYMS_ALIAS > > > + bool "Produces alias for duplicated symbols" if EXPERT > > > + depends on KALLSYMS && (DEBUG_INFO_DWARF4 || DEBUG_INFO_DWARF5) > > > + help > > > + It is not uncommon for drivers or modules related to similar > > > + peripherals to have symbols with the exact same name. > > > + While this is not a problem for the kernel's binary itself, it > > > + becomes an issue when attempting to trace or probe specific > > > + functions using infrastructure like ftrace or kprobe. > > > + > > > + This option addresses this challenge by extending the symbol > > > names + with unique suffixes during the kernel build process. > > > + The newly created aliases for these duplicated symbols are > > > unique > > > + names that can be fed to the ftrace sysfs interface. By doing > > > so, it + enables previously unreachable symbols to be probed. > > > + > > > +config CONFIG_KALLSYMS_ALIAS_DATA > > > + bool "Produces alias also for data" > > > + depends on KALLSYMS_ALIAS > > > + help > > > + Sometimes it can be useful to refer to data. In live patch > > > scenarios, + you may find yourself needing to use symbols that > > > are shared with + other functions. Since symbols face the same > > > issue as functions, this + option allows you to create aliases > > > for data as well. > > > + > > > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > + bool "Removes all filter when producing data alias" > > > + depends on CONFIG_KALLSYMS_ALIAS_DATA > > > + help > > > + When selecting data aliases, not all symbols are included in the > > > set + This is because many symbols are unlikely to be used. If > > > you choose + to have an alias for all data symbols, be aware that > > > it will + significantly increase the size. > > > + > > > + If unsure, say N. > > > + > > > > > > # end of the "standard kernel features (expert users)" menu > > > > > > # syscall, maps, verifier > > > > > > diff --git a/scripts/Makefile b/scripts/Makefile > > > index 32b6ba722728..65fafe17cfe5 100644 > > > --- a/scripts/Makefile > > > +++ b/scripts/Makefile > > > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > > > > > # Let clean descend into subdirs > > > subdir- += basic dtc gdb kconfig mod > > > > > > + > > > +# KALLSyms alias > > > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > + > > > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > > > new file mode 100644 > > > index 000000000000..e1fde69232b4 > > > --- /dev/null > > > +++ b/scripts/kas_alias/Makefile > > > @@ -0,0 +1,4 @@ > > > +# SPDX-License-Identifier: GPL-2.0 > > > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > + > > > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o a2l.o > > > diff --git a/scripts/kas_alias/a2l.c b/scripts/kas_alias/a2l.c > > > new file mode 100644 > > > index 000000000000..a9692ac30180 > > > --- /dev/null > > > +++ b/scripts/kas_alias/a2l.c > > > @@ -0,0 +1,268 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <string.h> > > > +#include <unistd.h> > > > +#include <sys/types.h> > > > +#include <sys/wait.h> > > > +#include <string.h> > > > +#include <stdint.h> > > > +#include <stdbool.h> > > > + > > > +#include "a2l.h" > > > + > > > +int addr2line_pid = -1; > > > +int a2l_in[2]; > > > +int a2l_out[2]; > > > +char line[MAX_BUF]; > > > +char vmlinux_path[MAX_BUF]; > > > +char addr2line_cmd[MAX_CMD_LEN]; > > > +FILE *a2l_stdin, *a2l_stdout; > > > + > > > +static char *normalize_path(const char *input_path, char *output_path) > > > +{ > > > + char *prev_token = NULL; > > > + char *delimiter = "/"; > > > + char inbuf[MAX_BUF]; > > > + char *token; > > > + char *pos; > > > + > > > + memset(inbuf, 0, MAX_BUF); > > > + *output_path = '\0'; > > > + strncpy(inbuf, input_path, MAX_BUF); > > > + if (!input_path || !output_path || strlen(input_path) == 0) > > > + return NULL; > > > + > > > + token = strtok(inbuf, delimiter); > > > + while (token) { > > > + if (strcmp(token, "..") == 0 && prev_token) { > > > + pos = strrchr(output_path, '/'); > > > + if (pos) > > > + *pos = '\0'; > > > + > > > + } else if (strcmp(token, ".") != 0) { > > > + strcat(output_path, "/"); > > > + strcat(output_path, token); > > > + } > > > + > > > + prev_token = token; > > > + token = strtok(NULL, delimiter); > > > + } > > > + > > > + return output_path; > > > +} > > > + > > > +static void path_of(const char *full_path, char *path) > > > +{ > > > + const char *last_slash = strrchr(full_path, '/'); > > > + size_t path_length; > > > + char cwd[MAX_BUF]; > > > + > > > + if (!last_slash) { > > > + if (getcwd(cwd, sizeof(cwd))) > > > + strcpy(path, cwd); > > > + else > > > + strcpy(path, "."); > > > + } else { > > > + path_length = last_slash - full_path; > > > + strncpy(path, full_path, path_length); > > > + path[path_length] = '\0'; > > > + } > > > +} > > > + > > > +static bool file_exists(const char *file_path) > > > +{ > > > + FILE *file; > > > + > > > + file = fopen(file_path, "r"); > > > + if (file) { > > > + fclose(file); > > > + return true; > > > + } > > > + return false; > > > +} > > > + > > > +int addr2line_init(const char *cmd, const char *vmlinux) > > > +{ > > > + if ((!file_exists(cmd)) || (!file_exists(vmlinux))) { > > > + printf("file not found\n"); > > > + return 0; > > > + } > > > + > > > + path_of(vmlinux, vmlinux_path); > > > + if (pipe(a2l_in) == -1) { > > > + printf("Failed to create pipe\n"); > > > + return 0; > > > + } > > > + > > > + if (pipe(a2l_out) == -1) { > > > + printf("Failed to create pipe\n"); > > > + return 0; > > > + } > > > + > > > + addr2line_pid = fork(); > > > + if (addr2line_pid == -1) { > > > + printf("Failed to fork process\n"); > > > + close(a2l_in[P_READ]); > > > + close(a2l_in[P_WRITE]); > > > + close(a2l_out[P_READ]); > > > + close(a2l_out[P_WRITE]); > > > + return 0; > > > + } > > > + > > > + if (addr2line_pid == 0) { > > > + dup2(a2l_in[P_READ], 0); > > > + dup2(a2l_out[P_WRITE], 1); > > > + close(a2l_in[P_WRITE]); > > > + close(a2l_out[P_READ]); > > > + > > > + execlp(cmd, cmd, ADDR2LINE_ARGS, vmlinux, NULL); > > > + > > > + printf("Failed to execute addr2line command\n"); > > > + exit(1); > > > + } else { > > > + close(a2l_in[P_READ]); > > > + close(a2l_out[P_WRITE]); > > > + } > > > + > > > + a2l_stdin = fdopen(a2l_in[P_WRITE], "w"); > > > + if (!a2l_stdin) { > > > + printf("Failed to open pipe a2l_in\n"); > > > + return 0; > > > + } > > > + > > > + a2l_stdout = fdopen(a2l_out[P_READ], "r"); > > > + if (!a2l_stdout) { > > > + printf("Failed to open pipe a2l_out\n"); > > > + fclose(a2l_stdin); > > > + return 0; > > > + } > > > + > > > + return 1; > > > +} > > > + > > > +const char *remove_subdir(const char *home, const char *f_path) > > > +{ > > > + int i = 0; > > > + > > > + while (*(home + i) == *(f_path + i)) > > > + i++; > > > + > > > + return (strlen(home) != i) ? NULL : f_path + i; > > > +} > > > + > > > +char *addr2line_get_lines(uint64_t address) > > > +{ > > > + char buf[MAX_BUF]; > > > + > > > + fprintf(a2l_stdin, "%08lx\n", address); > > > + fflush(a2l_stdin); > > > + > > > + if (!fgets(line, sizeof(line), a2l_stdout)) { > > > + printf("Failed to read lines from addr2line\n"); > > > + return NULL; > > > + } > > > + > > > + if (!fgets(line, sizeof(line), a2l_stdout)) { > > > + printf("Failed to read lines from addr2line\n"); > > > + return NULL; > > > + } > > > + > > > + line[strcspn(line, "\n")] = '\0'; > > > + strncpy(buf, line, MAX_BUF); > > > + return normalize_path(buf, line); > > > +} > > > + > > > +int addr2line_cleanup(void) > > > +{ > > > + int status; > > > + > > > + if (addr2line_pid != -1) { > > > + kill(addr2line_pid, SIGKILL); > > > + waitpid(addr2line_pid, &status, 0); > > > + fclose(a2l_stdin); > > > + fclose(a2l_stdout); > > > + addr2line_pid = -1; > > > + } > > > + > > > + return 1; > > > +} > > > + > > > +static char *find_executable(const char *command) > > > +{ > > > + char *path_env = getenv("PATH"); > > > + char *executable_path; > > > + char *path_copy; > > > + char *path; > > > + int n; > > > + > > > + if (!path_env) > > > + return NULL; > > > + > > > + path_copy = strdup(path_env); > > > + if (!path_copy) > > > + return NULL; > > > + > > > + path = strtok(path_copy, ":"); > > > + while (path) { > > > + n = snprintf(0, 0, "%s/%s", path, command); > > > + executable_path = (char *)malloc(n + 1); > > > + snprintf(executable_path, n + 1, "%s/%s", path, command); > > > + if (access(executable_path, X_OK) == 0) { > > > + free(path_copy); > > > + return executable_path; > > > + } > > > + > > > + path = strtok(NULL, ":"); > > > + free(executable_path); > > > + executable_path = NULL; > > > + } > > > + > > > + free(path_copy); > > > + if (executable_path) > > > + free(executable_path); > > > + return NULL; > > > +} > > > + > > > +const char *get_addr2line(int mode) > > > +{ > > > + char *buf = ""; > > > + > > > + switch (mode) { > > > + case A2L_CROSS: > > > + buf = getenv("CROSS_COMPILE"); > > > + memcpy(addr2line_cmd, buf, strlen(buf)); > > > + case A2L_DEFAULT: > > > + memcpy(addr2line_cmd + strlen(buf), ADDR2LINE, > > > strlen(ADDR2LINE)); + buf = find_executable(addr2line_cmd); > > > + if (buf) { > > > + memcpy(addr2line_cmd, buf, strlen(buf)); > > > + free(buf); > > > + } > > > + return addr2line_cmd; > > > + case A2L_LLVM: > > > + default: > > > + return NULL; > > > + } > > > +} > > > + > > > +char *get_vmlinux(char *input) > > > +{ > > > + const char *match_string1 = ".syms"; > > > + const char *match_string2 = ".tmp_vmlinux.kallsyms"; > > > + char *result = NULL; > > > + char *match_pos; > > > + > > > + match_pos = strstr(input, match_string1); > > > + if (!match_pos) > > > + return NULL; > > > + > > > + match_pos = strstr(input, match_string2); > > > + if (!match_pos) > > > + return NULL; > > > + > > > + result = strdup(input); > > > + match_pos = strstr(result, match_string1); > > > + *match_pos = '\0'; > > > + return result; > > > +} > > > diff --git a/scripts/kas_alias/a2l.h b/scripts/kas_alias/a2l.h > > > new file mode 100644 > > > index 000000000000..ca6419229dde > > > --- /dev/null > > > +++ b/scripts/kas_alias/a2l.h > > > @@ -0,0 +1,32 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef A2L_H > > > +#define A2L_H > > > +#include <stdint.h> > > > + > > > +#define ADDR2LINE "addr2line" > > > +#define ADDR2LINE_ARGS "-fe" > > > +//#define VMLINUX "vmlinux" > > > +#define MAX_BUF 4096 > > > +#define MAX_CMD_LEN 256 > > > +#define P_READ 0 > > > +#define P_WRITE 1 > > > +#define A2L_DEFAULT 1 > > > +#define A2L_CROSS 2 > > > +#define A2L_LLVM 3 > > > +#define A2L_MAKE_VALUE 2 > > > + > > > +extern int addr2line_pid; > > > +extern int a2l_in[2]; > > > +extern int a2l_out[2]; > > > +extern char line[MAX_BUF]; > > > +extern char vmlinux_path[MAX_BUF]; > > > +extern char addr2line_cmd[MAX_CMD_LEN]; > > > + > > > +int addr2line_init(const char *cmd, const char *vmlinux); > > > +char *addr2line_get_lines(uint64_t address); > > > +int addr2line_cleanup(void); > > > +const char *remove_subdir(const char *home, const char *f_path); > > > +const char *get_addr2line(int mode); > > > +char *get_vmlinux(char *input); > > > + > > > +#endif > > > diff --git a/scripts/kas_alias/duplicates_list.c > > > b/scripts/kas_alias/duplicates_list.c new file mode 100644 > > > index 000000000000..e7a3d2917937 > > > --- /dev/null > > > +++ b/scripts/kas_alias/duplicates_list.c > > > @@ -0,0 +1,70 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdint.h> > > > +#include <stdio.h> > > > +#include <string.h> > > > +#include <stdlib.h> > > > +#include <stdbool.h> > > > + > > > +#include "item_list.h" > > > +#include "duplicates_list.h" > > > + > > > +struct duplicate_item *find_duplicates(struct item *list) > > > +{ > > > + struct duplicate_item *current_duplicate = NULL; > > > + struct duplicate_item *duplicates = NULL; > > > + struct duplicate_item *new_duplicate; > > > + struct item *current_item = list; > > > + bool prev_was_duplicate = false; > > > + struct item *prev_item = NULL; > > > + > > > + while (current_item) { > > > + if ((prev_item && (strcmp(current_item->symb_name, > > > prev_item->symb_name) == 0)) || + prev_was_duplicate) { > > > + if (!duplicates) { > > > + duplicates = malloc(sizeof(struct > > > duplicate_item)); + if (!duplicates) > > > + return NULL; > > > + > > > + duplicates->original_item = prev_item; > > > + duplicates->next = NULL; > > > + current_duplicate = duplicates; > > > + } else { > > > + new_duplicate = malloc(sizeof(struct > > > duplicate_item)); + if (!new_duplicate) { > > > + free_duplicates(&duplicates); > > > + return NULL; > > > + } > > > + > > > + new_duplicate->original_item = prev_item; > > > + new_duplicate->next = NULL; > > > + current_duplicate->next = new_duplicate; > > > + current_duplicate = new_duplicate; > > > + > > > + if ((strcmp(current_item->symb_name, > > > prev_item->symb_name) != 0) && + > > > (prev_was_duplicate)) > > > + prev_was_duplicate = false; > > > + else > > > + prev_was_duplicate = true; > > > + } > > > + } > > > + > > > + prev_item = current_item; > > > + current_item = current_item->next; > > > + } > > > + > > > + return duplicates; > > > +} > > > + > > > +void free_duplicates(struct duplicate_item **duplicates) > > > +{ > > > + struct duplicate_item *duplicates_iterator = *duplicates; > > > + struct duplicate_item *app; > > > + > > > + while (duplicates_iterator) { > > > + app = duplicates_iterator; > > > + duplicates_iterator = duplicates_iterator->next; > > > + free(app); > > > + } > > > + > > > + *duplicates = NULL; > > > +} > > > diff --git a/scripts/kas_alias/duplicates_list.h > > > b/scripts/kas_alias/duplicates_list.h new file mode 100644 > > > index 000000000000..76aa73e584bc > > > --- /dev/null > > > +++ b/scripts/kas_alias/duplicates_list.h > > > @@ -0,0 +1,15 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef DUPLICATES_LIST_H > > > +#define DUPLICATES_LIST_H > > > + > > > +#include "item_list.h" > > > + > > > +struct duplicate_item { > > > + struct item *original_item; > > > + struct duplicate_item *next; > > > +}; > > > + > > > +struct duplicate_item *find_duplicates(struct item *list); > > > +void free_duplicates(struct duplicate_item **duplicates); > > > + > > > +#endif > > > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > > > new file mode 100644 > > > index 000000000000..48f2e525592a > > > --- /dev/null > > > +++ b/scripts/kas_alias/item_list.c > > > @@ -0,0 +1,230 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <stdint.h> > > > +#include <string.h> > > > +#include <stdbool.h> > > > +#include <assert.h> > > > +#include "item_list.h" > > > + > > > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > > > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > > > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > > > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, > > > (temp)->symb_name) op 0) + > > > +struct item *list_index[96] = {0}; > > > + > > > +void build_index(struct item *list) > > > +{ > > > + char current_first_letter = ' '; > > > + struct item *current = list; > > > + > > > + while (current) { > > > + if (current->symb_name[0] != current_first_letter) { > > > + current_first_letter = current->symb_name[0]; > > > + list_index[current_first_letter - 32] = current; > > > + } > > > + current = current->next; > > > + } > > > +} > > > + > > > +struct item *add_item(struct item **list, const char *name, char stype, > > > uint64_t addr) +{ > > > + struct item *new_item; > > > + struct item *current; > > > + > > > + new_item = malloc(sizeof(struct item)); > > > + if (!new_item) > > > + return NULL; > > > + > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > + new_item->addr = addr; > > > + new_item->stype = stype; > > > + new_item->next = NULL; > > > + > > > + if (!(*list)) { > > > + *list = new_item; > > > + } else { > > > + current = *list; > > > + while (current->next) > > > + current = current->next; > > > + > > > + current->next = new_item; > > > + } > > > + return new_item; > > > +} > > > + > > > +void sort_list(struct item **list, int sort_by) > > > +{ > > > + struct item *current = *list; > > > + struct item *sorted = NULL; > > > + struct item *next_item; > > > + struct item *temp; > > > + > > > + if (!(*list) || !((*list)->next)) > > > + return; > > > + > > > + while (current) { > > > + next_item = current->next; > > > + if (!sorted || > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) > > > || > > > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > > > + current->next = sorted; > > > + sorted = current; > > > + } else { > > > + temp = sorted; > > > + while (temp->next && > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, > > > temp->next, >=) || + > > > CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) + > > > temp = temp->next; > > > + > > > + current->next = temp->next; > > > + temp->next = current; > > > + } > > > + current = next_item; > > > + } > > > + > > > + *list = sorted; > > > +} > > > + > > > +struct item *merge(struct item *left, struct item *right, int sort_by) > > > +{ > > > + struct item *current = NULL; > > > + struct item *result = NULL; > > > + > > > + if (!left) > > > + return right; > > > + if (!right) > > > + return left; > > > + > > > + if (sort_by == BY_NAME) { > > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > > + result = left; > > > + left = left->next; > > > + } else { > > > + result = right; > > > + right = right->next; > > > + } > > > + } else { > > > + if (sort_by == BY_ADDRESS) { > > > + if (left->addr <= right->addr) { > > > + result = left; > > > + left = left->next; > > > + } else { > > > + result = right; > > > + right = right->next; > > > + } > > > + } > > > + } > > > + > > > + current = result; > > > + > > > + while (left && right) { > > > + if (sort_by == BY_NAME) { > > > + if (strcmp(left->symb_name, right->symb_name) <= > > > 0) { + current->next = left; > > > + left = left->next; > > > + } else { > > > + current->next = right; > > > + right = right->next; > > > + } > > > + } else { > > > + if (sort_by == BY_ADDRESS) { > > > + if (left->addr <= right->addr) { > > > + current->next = left; > > > + left = left->next; > > > + } else { > > > + current->next = right; > > > + right = right->next; > > > + } > > > + } > > > + } > > > + > > > + current = current->next; > > > + } > > > + > > > + if (left) { > > > + current->next = left; > > > + } else { > > > + if (right) > > > + current->next = right; > > > + } > > > + > > > + return result; > > > +} > > > + > > > +struct item *merge_sort(struct item *head, int sort_by) > > > +{ > > > + struct item *right; > > > + struct item *slow; > > > + struct item *fast; > > > + struct item *left; > > > + > > > + if (!head || !head->next) > > > + return head; > > > + > > > + slow = head; > > > + fast = head->next; > > > + > > > + while (fast && fast->next) { > > > + slow = slow->next; > > > + fast = fast->next->next; > > > + } > > > + > > > + left = head; > > > + right = slow->next; > > > + slow->next = NULL; > > > + > > > + left = merge_sort(left, sort_by); > > > + right = merge_sort(right, sort_by); > > > + > > > + return merge(left, right, sort_by); > > > +} > > > + > > > +void sort_list_m(struct item **head, int sort_by) > > > +{ > > > + if (!(*head) || !((*head)->next)) > > > + return; > > > + > > > + *head = merge_sort(*head, sort_by); > > > +} > > > + > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > + const char *name, uint64_t addr, char stype) > > > +{ > > > + struct item *new_item; > > > + struct item *current; > > > + int ret = 0; > > > + > > > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : > > > list; + while (current) { > > > + if (current->addr == search_addr) { > > > + new_item = malloc(sizeof(struct item)); > > > + if (!new_item) > > > + return ret; > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > + new_item->addr = addr; > > > + new_item->stype = stype; > > > + new_item->next = current->next; > > > + current->next = new_item; > > > + ret = 1; > > > + break; > > > + } > > > + current = current->next; > > > + } > > > + return ret; > > > +} > > > + > > > +void free_items(struct item **head) > > > +{ > > > + struct item *app, *item_iterator = *head; > > > + > > > + while (item_iterator) { > > > + app = item_iterator; > > > + item_iterator = item_iterator->next; > > > + free(app); > > > + } > > > + *head = NULL; > > > +} > > > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > > > new file mode 100644 > > > index 000000000000..b4891cb088ee > > > --- /dev/null > > > +++ b/scripts/kas_alias/item_list.h > > > @@ -0,0 +1,26 @@ > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > +#ifndef ITEM_LIST_H > > > +#define ITEM_LIST_H > > > +#include <stdint.h> > > > + > > > +#define MAX_NAME_SIZE 256 > > > +#define BY_ADDRESS 1 > > > +#define BY_NAME 2 > > > + > > > +struct item { > > > + char symb_name[MAX_NAME_SIZE]; > > > + uint64_t addr; > > > + char stype; > > > + struct item *next; > > > +}; > > > + > > > +void build_index(struct item *list); > > > +struct item *add_item(struct item **list, const char *name, char stype, > > > uint64_t addr); +void sort_list(struct item **list, int sort_by); > > > +struct item *merge(struct item *left, struct item *right, int sort_by); > > > +struct item *merge_sort(struct item *head, int sort_by); > > > +void sort_list_m(struct item **head, int sort_by); > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > + const char *name, uint64_t addr, char stype); > > > +void free_items(struct item **head); > > > +#endif > > > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > > > new file mode 100644 > > > index 000000000000..532aeb39f851 > > > --- /dev/null > > > +++ b/scripts/kas_alias/kas_alias.c > > > @@ -0,0 +1,217 @@ > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > +#include <stdio.h> > > > +#include <stdlib.h> > > > +#include <stdint.h> > > > +#include <unistd.h> > > > +#include <string.h> > > > +#include <stdbool.h> > > > +#include <stdarg.h> > > > +#include <regex.h> > > > + > > > +#include "item_list.h" > > > +#include "duplicates_list.h" > > > +#include "a2l.h" > > > + > > > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > > > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') > > > || \ + (((s)->stype) == 'd') || (((s)->stype) == > > > 'D') || \ + (((s)->stype) == 'r') || > > > (((s)->stype) == 'R')) +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > > > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > > > +#else > > > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > > > +#endif > > > +#define FNOMATCH 0 > > > +#define FMATCH 1 > > > +#define EREGEX 2 > > > + > > > +const char *ignore_regex[] = { > > > + "^__cfi_.*$", // __cfi_ preamble > > > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > + "^_*TRACE_SYSTEM.*$", > > > + "^__already_done\\.[0-9]+$", // Call a function once > > > data + "^___tp_str\\.[0-9]+$", > > > + "^___done\\.[0-9]+$", > > > + "^__print_once\\.[0-9]+$", > > > + "^_rs\\.[0-9]+$", > > > + "^__compound_literal\\.[0-9]+$", > > > + "^___once_key\\.[0-9]+$", > > > + "^__func__\\.[0-9]+$", > > > + "^__msg\\.[0-9]+$", > > > + "^CSWTCH\\.[0-9]+$", > > > + "^__flags\\.[0-9]+$", > > > + "^__wkey.*$", > > > + "^__mkey.*$", > > > + "^__key.*$", > > > +#endif > > > + "^__pfx_.*$" // NOP-padding > > > +}; > > > + > > > +int suffix_serial; > > > + > > > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > > > +{ > > > + va_list args; > > > + > > > + va_start(args, fmt); > > > + if (verbose) > > > + printf(fmt, args); > > > + > > > + va_end(args); > > > +} > > > + > > > +static void create_suffix(const char *name, char *output_suffix) > > > +{ > > > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > > > +} > > > + > > > +static void create_file_suffix(const char *name, uint64_t address, char > > > *output_suffix, char *cwd) +{ > > > + const char *f_path; > > > + char *buf; > > > + int i = 0; > > > + > > > + buf = addr2line_get_lines(address); > > > + f_path = remove_subdir(cwd, buf); > > > + if (f_path) { > > > + sprintf(output_suffix, "%s@%s", name, f_path); > > > + while (*(output_suffix + i) != '\0') { > > > + switch (*(output_suffix + i)) { > > > + case '/': > > > + case ':': > > > + case '.': > > > + *(output_suffix + i) = '_'; > > > + break; > > > + default: > > > + } > > > + i++; > > > + } > > > + } else { > > > + create_suffix(name, output_suffix); > > > + } > > > +} > > > + > > > +static int filter_symbols(char *symbol, const char **ignore_list, int > > > regex_no) +{ > > > + regex_t regex; > > > + int res, i; > > > + > > > + for (i = 0; i < regex_no; i++) { > > > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > > > + if (res) > > > + return -EREGEX; > > > + > > > + res = regexec(®ex, symbol, 0, NULL, 0); > > > + regfree(®ex); > > > + switch (res) { > > > + case 0: > > > + return FMATCH; > > > + case REG_NOMATCH: > > > + break; > > > + default: > > > + return -EREGEX; > > > + } > > > + } > > > + > > > + return FNOMATCH; > > > +} > > > + > > > +int main(int argc, char *argv[]) > > > +{ > > > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > > > + struct duplicate_item *duplicate_iterator; > > > + struct duplicate_item *duplicate; > > > + struct item *head = {NULL}; > > > + bool need_2_process = true; > > > + struct item *last = {NULL}; > > > + struct item *current; > > > + int verbose_mode = 0; > > > + uint64_t address; > > > + FILE *fp; > > > + int res; > > > + > > > + if (argc < 2 || argc > 3) { > > > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > > > + return 1; > > > + } > > > + > > > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > > > + verbose_mode = 1; > > > + > > > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > > > + > > > + fp = fopen(argv[1], "r"); > > > + if (!fp) { > > > + printf("Can't open input file.\n"); > > > + return 1; > > > + } > > > + > > > + if (!addr2line_init(get_addr2line(A2L_DEFAULT), > > > get_vmlinux(argv[1]))) + return 1; > > > + > > > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > > > + if (strstr(sym_name, "@_")) { > > > + if (verbose_mode && need_2_process) > > > + printf("Already processed\n"); > > > + need_2_process = false; > > > + } > > > + last = add_item(&last, sym_name, t, address); > > > + if (!last) { > > > + printf("Error in allocate memory\n"); > > > + free_items(&head); > > > + return 1; > > > + } > > > + > > > + if (!head) > > > + head = last; > > > + } > > > + > > > + fclose(fp); > > > + > > > + if (need_2_process) { > > > + verbose_msg(verbose_mode, "Sorting nm data\n"); > > > + sort_list_m(&head, BY_NAME); > > > + verbose_msg(verbose_mode, "Scanning nm data for > > > duplicates\n"); + duplicate = find_duplicates(head); > > > + if (!duplicate) { > > > + printf("Error in duplicates list\n"); > > > + return 1; > > > + } > > > + > > > + verbose_msg(verbose_mode, "Applying suffixes\n"); > > > + build_index(head); > > > + duplicate_iterator = duplicate; > > > + while (duplicate_iterator) { > > > + res = > > > filter_symbols(duplicate_iterator->original_item->symb_name, + > > > ignore_regex, sizeof(ignore_regex) / + > > > sizeof(ignore_regex[0])); + > > > if (res != FMATCH && > > > + > > > SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { + > > > if (res < 0) > > > + return 1; > > > + > > > + > > > create_file_suffix(duplicate_iterator->original_item->symb_name, + > > > > > > duplicate_iterator->original_item->addr, + > > > new_name, vmlinux_path); + > > > if (!insert_after(head, duplicate_iterator->original_item->addr, + > > > new_name, > > > duplicate_iterator->original_item->addr, + > > > duplicate_iterator->original_item->stype)) + > > > return 1; > > > + } > > > + > > > + duplicate_iterator = duplicate_iterator->next; > > > + } > > > + > > > + sort_list_m(&head, BY_ADDRESS); > > > + } > > > + current = head; > > > + while (current) { > > > + printf("%08lx %c %s\n", current->addr, current->stype, > > > current->symb_name); + current = current->next; > > > + } > > > + > > > + free_items(&head); > > > + free_duplicates(&duplicate); > > > + addr2line_cleanup(); > > > + return 0; > > > +} > > > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > > > index a432b171be82..cacf60b597ce 100755 > > > --- a/scripts/link-vmlinux.sh > > > +++ b/scripts/link-vmlinux.sh > > > @@ -89,8 +89,9 @@ vmlinux_link() > > > > > > ldflags="${ldflags} ${wl}--script=${objtree}/${KBUILD_LDS}" > > > > > > - # The kallsyms linking does not need debug symbols included. > > > - if [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > > > + # The kallsyms linking does not need debug symbols included, > > > unless the KALLSYMS_ALIAS. + if [ ! is_enabled > > > CONFIG_KALLSYMS_ALIAS ] && \ > > > + [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > > > > > > ldflags="${ldflags} ${wl}--strip-debug" > > > > > > fi > > > > > > @@ -161,7 +162,11 @@ kallsyms() > > > > > > fi > > > > > > info KSYMS ${2} > > > > > > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > > > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > > > + ALIAS=".alias" > > > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > > > + fi > > > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > > > > > > } > > > > > > # Perform one step in kallsyms generation, including temporary linking of > > > > > > -- > > > 2.34.1 > > Best regards. > > BR Alessandro
Hi Alessandro, On Fri, 25 Aug 2023 12:15:09 +0200 Alessandro Carminati <alessandro.carminati@gmail.com> wrote: > > Thank you for the investigation, this an interesting reading! > > > > > The rationale is that using source file + line number iproduces better > > > kallsyms table, but it is still do not produce unique names. > > > > When I first read your cover letter, I also thought it would be cool to have > > the filename rather than a serial ID. > > So, your latest modification is really good as it goes into this direction. > > Regarding having non unique names, I am not sure if this is a problem. > > In my case, which is using kprobe to attach BPF code to trace some events, > > your contribution would make me on the way to know which address I should use > > because I could choose regarding the filename. > > This whole thing got trickier than I expected when I started. At first, I was > just grappling with different functions having the same name. > But during my investigation, it hit me that I've got to handle multiple copies > of the same function too. > So, those addresses labeled with the same name? > They're not just different functions having the same name - they're a mix of > functions that happen to share a name and functions that are the same but come > from different compile units. > > Imagine this: > one compile unit defines a function foo1() that uses a bar() from a > shared header. > Another compile unit defines foo2(), also relying on bar(). > The snag is, these compile units end up being linked together, carrying their > own copies of bar(). Ah, there are 1:n mapping issue along with n:1 issue. In that case, both symbols should be probed. ('perf probe' adds probes for all those instances, usually happens on probing inlined function) > So, if I wanna keep an eye on bar(), which copy should I watch? > The answer depends on whose bar() I want to watch at. > I could go for probing them all, sure, but that's kind of a suboptimal > solution that waters down what I'm trying to achieve. I think we can keep the same symbols in the kallsyms in that case, because those may have the same function prototype. (.constprop suffix has different type but we can keep .constprope suffix to the symbols in that case) If your feature ensures that the same-name symbols has the same type, we can probe all same-name symbols with the same BPF programs or probe-event. Again, I think the important point is the same name symbols has the same prototype. > > I am curious about a potential v3 of this patch! > > The way things are being done right now, even though it's nice that it can > track down these copies in the kernel image, it doesn't quite hit the > bullseye of what I was going for. But hey, I'm totally open to putting out > v3 with the limitations I just talked about, and discuss if it can be improved. Thank you for your work! > > > > > > BR > > > Alessandro > > > > > > > As a final note, please understand that my patch was not intended to > > > > undermine anyone's work. I simply encountered a problem that I wanted to > > > > help solve. Attached to this message is my code, in case anyone wants to > > > > replicate it. I would appreciate being kept in the loop, as I genuinely > > > > want to assist in fixing this problem. > > > > > > > > BR > > > > Alessandro > > > > --- > > > > > > > > init/Kconfig | 36 ++++ > > > > scripts/Makefile | 4 + > > > > scripts/kas_alias/Makefile | 4 + > > > > scripts/kas_alias/a2l.c | 268 ++++++++++++++++++++++++++++ > > > > scripts/kas_alias/a2l.h | 32 ++++ > > > > scripts/kas_alias/duplicates_list.c | 70 ++++++++ > > > > scripts/kas_alias/duplicates_list.h | 15 ++ > > > > scripts/kas_alias/item_list.c | 230 ++++++++++++++++++++++++ > > > > scripts/kas_alias/item_list.h | 26 +++ > > > > scripts/kas_alias/kas_alias.c | 217 ++++++++++++++++++++++ > > > > scripts/link-vmlinux.sh | 11 +- > > > > 11 files changed, 910 insertions(+), 3 deletions(-) > > > > create mode 100644 scripts/kas_alias/Makefile > > > > create mode 100644 scripts/kas_alias/a2l.c > > > > create mode 100644 scripts/kas_alias/a2l.h > > > > create mode 100644 scripts/kas_alias/duplicates_list.c > > > > create mode 100644 scripts/kas_alias/duplicates_list.h > > > > create mode 100644 scripts/kas_alias/item_list.c > > > > create mode 100644 scripts/kas_alias/item_list.h > > > > create mode 100644 scripts/kas_alias/kas_alias.c > > > > > > > > diff --git a/init/Kconfig b/init/Kconfig > > > > index f7f65af4ee12..bc69fcd9cbc8 100644 > > > > --- a/init/Kconfig > > > > +++ b/init/Kconfig > > > > @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE > > > > > > > > time constants, and no relocation pass is required at runtime to > > > > fix > > > > up the entries based on the runtime load address of the kernel. > > > > > > > > +config KALLSYMS_ALIAS > > > > + bool "Produces alias for duplicated symbols" if EXPERT > > > > + depends on KALLSYMS && (DEBUG_INFO_DWARF4 || DEBUG_INFO_DWARF5) > > > > + help > > > > + It is not uncommon for drivers or modules related to similar > > > > + peripherals to have symbols with the exact same name. > > > > + While this is not a problem for the kernel's binary itself, it > > > > + becomes an issue when attempting to trace or probe specific > > > > + functions using infrastructure like ftrace or kprobe. > > > > + > > > > + This option addresses this challenge by extending the symbol > > > > names + with unique suffixes during the kernel build process. > > > > + The newly created aliases for these duplicated symbols are > > > > unique > > > > + names that can be fed to the ftrace sysfs interface. By doing > > > > so, it + enables previously unreachable symbols to be probed. > > > > + > > > > +config CONFIG_KALLSYMS_ALIAS_DATA > > > > + bool "Produces alias also for data" > > > > + depends on KALLSYMS_ALIAS > > > > + help > > > > + Sometimes it can be useful to refer to data. In live patch > > > > scenarios, + you may find yourself needing to use symbols that > > > > are shared with + other functions. Since symbols face the same > > > > issue as functions, this + option allows you to create aliases > > > > for data as well. > > > > + > > > > +config CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > > + bool "Removes all filter when producing data alias" > > > > + depends on CONFIG_KALLSYMS_ALIAS_DATA > > > > + help > > > > + When selecting data aliases, not all symbols are included in the > > > > set + This is because many symbols are unlikely to be used. If > > > > you choose + to have an alias for all data symbols, be aware that > > > > it will + significantly increase the size. > > > > + > > > > + If unsure, say N. > > > > + > > > > > > > > # end of the "standard kernel features (expert users)" menu > > > > > > > > # syscall, maps, verifier > > > > > > > > diff --git a/scripts/Makefile b/scripts/Makefile > > > > index 32b6ba722728..65fafe17cfe5 100644 > > > > --- a/scripts/Makefile > > > > +++ b/scripts/Makefile > > > > @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux > > > > > > > > # Let clean descend into subdirs > > > > subdir- += basic dtc gdb kconfig mod > > > > > > > > + > > > > +# KALLSyms alias > > > > +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > > + > > > > diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile > > > > new file mode 100644 > > > > index 000000000000..e1fde69232b4 > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/Makefile > > > > @@ -0,0 +1,4 @@ > > > > +# SPDX-License-Identifier: GPL-2.0 > > > > +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias > > > > + > > > > +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o a2l.o > > > > diff --git a/scripts/kas_alias/a2l.c b/scripts/kas_alias/a2l.c > > > > new file mode 100644 > > > > index 000000000000..a9692ac30180 > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/a2l.c > > > > @@ -0,0 +1,268 @@ > > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > > +#include <stdio.h> > > > > +#include <stdlib.h> > > > > +#include <string.h> > > > > +#include <unistd.h> > > > > +#include <sys/types.h> > > > > +#include <sys/wait.h> > > > > +#include <string.h> > > > > +#include <stdint.h> > > > > +#include <stdbool.h> > > > > + > > > > +#include "a2l.h" > > > > + > > > > +int addr2line_pid = -1; > > > > +int a2l_in[2]; > > > > +int a2l_out[2]; > > > > +char line[MAX_BUF]; > > > > +char vmlinux_path[MAX_BUF]; > > > > +char addr2line_cmd[MAX_CMD_LEN]; > > > > +FILE *a2l_stdin, *a2l_stdout; > > > > + > > > > +static char *normalize_path(const char *input_path, char *output_path) > > > > +{ > > > > + char *prev_token = NULL; > > > > + char *delimiter = "/"; > > > > + char inbuf[MAX_BUF]; > > > > + char *token; > > > > + char *pos; > > > > + > > > > + memset(inbuf, 0, MAX_BUF); > > > > + *output_path = '\0'; > > > > + strncpy(inbuf, input_path, MAX_BUF); > > > > + if (!input_path || !output_path || strlen(input_path) == 0) > > > > + return NULL; > > > > + > > > > + token = strtok(inbuf, delimiter); > > > > + while (token) { > > > > + if (strcmp(token, "..") == 0 && prev_token) { > > > > + pos = strrchr(output_path, '/'); > > > > + if (pos) > > > > + *pos = '\0'; > > > > + > > > > + } else if (strcmp(token, ".") != 0) { > > > > + strcat(output_path, "/"); > > > > + strcat(output_path, token); > > > > + } > > > > + > > > > + prev_token = token; > > > > + token = strtok(NULL, delimiter); > > > > + } > > > > + > > > > + return output_path; > > > > +} > > > > + > > > > +static void path_of(const char *full_path, char *path) > > > > +{ > > > > + const char *last_slash = strrchr(full_path, '/'); > > > > + size_t path_length; > > > > + char cwd[MAX_BUF]; > > > > + > > > > + if (!last_slash) { > > > > + if (getcwd(cwd, sizeof(cwd))) > > > > + strcpy(path, cwd); > > > > + else > > > > + strcpy(path, "."); > > > > + } else { > > > > + path_length = last_slash - full_path; > > > > + strncpy(path, full_path, path_length); > > > > + path[path_length] = '\0'; > > > > + } > > > > +} > > > > + > > > > +static bool file_exists(const char *file_path) > > > > +{ > > > > + FILE *file; > > > > + > > > > + file = fopen(file_path, "r"); > > > > + if (file) { > > > > + fclose(file); > > > > + return true; > > > > + } > > > > + return false; > > > > +} > > > > + > > > > +int addr2line_init(const char *cmd, const char *vmlinux) > > > > +{ > > > > + if ((!file_exists(cmd)) || (!file_exists(vmlinux))) { > > > > + printf("file not found\n"); > > > > + return 0; > > > > + } > > > > + > > > > + path_of(vmlinux, vmlinux_path); > > > > + if (pipe(a2l_in) == -1) { > > > > + printf("Failed to create pipe\n"); > > > > + return 0; > > > > + } > > > > + > > > > + if (pipe(a2l_out) == -1) { > > > > + printf("Failed to create pipe\n"); > > > > + return 0; > > > > + } > > > > + > > > > + addr2line_pid = fork(); > > > > + if (addr2line_pid == -1) { > > > > + printf("Failed to fork process\n"); > > > > + close(a2l_in[P_READ]); > > > > + close(a2l_in[P_WRITE]); > > > > + close(a2l_out[P_READ]); > > > > + close(a2l_out[P_WRITE]); > > > > + return 0; > > > > + } > > > > + > > > > + if (addr2line_pid == 0) { > > > > + dup2(a2l_in[P_READ], 0); > > > > + dup2(a2l_out[P_WRITE], 1); > > > > + close(a2l_in[P_WRITE]); > > > > + close(a2l_out[P_READ]); > > > > + > > > > + execlp(cmd, cmd, ADDR2LINE_ARGS, vmlinux, NULL); > > > > + > > > > + printf("Failed to execute addr2line command\n"); > > > > + exit(1); > > > > + } else { > > > > + close(a2l_in[P_READ]); > > > > + close(a2l_out[P_WRITE]); > > > > + } > > > > + > > > > + a2l_stdin = fdopen(a2l_in[P_WRITE], "w"); > > > > + if (!a2l_stdin) { > > > > + printf("Failed to open pipe a2l_in\n"); > > > > + return 0; > > > > + } > > > > + > > > > + a2l_stdout = fdopen(a2l_out[P_READ], "r"); > > > > + if (!a2l_stdout) { > > > > + printf("Failed to open pipe a2l_out\n"); > > > > + fclose(a2l_stdin); > > > > + return 0; > > > > + } > > > > + > > > > + return 1; > > > > +} > > > > + > > > > +const char *remove_subdir(const char *home, const char *f_path) > > > > +{ > > > > + int i = 0; > > > > + > > > > + while (*(home + i) == *(f_path + i)) > > > > + i++; > > > > + > > > > + return (strlen(home) != i) ? NULL : f_path + i; > > > > +} > > > > + > > > > +char *addr2line_get_lines(uint64_t address) > > > > +{ > > > > + char buf[MAX_BUF]; > > > > + > > > > + fprintf(a2l_stdin, "%08lx\n", address); > > > > + fflush(a2l_stdin); > > > > + > > > > + if (!fgets(line, sizeof(line), a2l_stdout)) { > > > > + printf("Failed to read lines from addr2line\n"); > > > > + return NULL; > > > > + } > > > > + > > > > + if (!fgets(line, sizeof(line), a2l_stdout)) { > > > > + printf("Failed to read lines from addr2line\n"); > > > > + return NULL; > > > > + } > > > > + > > > > + line[strcspn(line, "\n")] = '\0'; > > > > + strncpy(buf, line, MAX_BUF); > > > > + return normalize_path(buf, line); > > > > +} > > > > + > > > > +int addr2line_cleanup(void) > > > > +{ > > > > + int status; > > > > + > > > > + if (addr2line_pid != -1) { > > > > + kill(addr2line_pid, SIGKILL); > > > > + waitpid(addr2line_pid, &status, 0); > > > > + fclose(a2l_stdin); > > > > + fclose(a2l_stdout); > > > > + addr2line_pid = -1; > > > > + } > > > > + > > > > + return 1; > > > > +} > > > > + > > > > +static char *find_executable(const char *command) > > > > +{ > > > > + char *path_env = getenv("PATH"); > > > > + char *executable_path; > > > > + char *path_copy; > > > > + char *path; > > > > + int n; > > > > + > > > > + if (!path_env) > > > > + return NULL; > > > > + > > > > + path_copy = strdup(path_env); > > > > + if (!path_copy) > > > > + return NULL; > > > > + > > > > + path = strtok(path_copy, ":"); > > > > + while (path) { > > > > + n = snprintf(0, 0, "%s/%s", path, command); > > > > + executable_path = (char *)malloc(n + 1); > > > > + snprintf(executable_path, n + 1, "%s/%s", path, command); > > > > + if (access(executable_path, X_OK) == 0) { > > > > + free(path_copy); > > > > + return executable_path; > > > > + } > > > > + > > > > + path = strtok(NULL, ":"); > > > > + free(executable_path); > > > > + executable_path = NULL; > > > > + } > > > > + > > > > + free(path_copy); > > > > + if (executable_path) > > > > + free(executable_path); > > > > + return NULL; > > > > +} > > > > + > > > > +const char *get_addr2line(int mode) > > > > +{ > > > > + char *buf = ""; > > > > + > > > > + switch (mode) { > > > > + case A2L_CROSS: > > > > + buf = getenv("CROSS_COMPILE"); > > > > + memcpy(addr2line_cmd, buf, strlen(buf)); > > > > + case A2L_DEFAULT: > > > > + memcpy(addr2line_cmd + strlen(buf), ADDR2LINE, > > > > strlen(ADDR2LINE)); + buf = find_executable(addr2line_cmd); > > > > + if (buf) { > > > > + memcpy(addr2line_cmd, buf, strlen(buf)); > > > > + free(buf); > > > > + } > > > > + return addr2line_cmd; > > > > + case A2L_LLVM: > > > > + default: > > > > + return NULL; > > > > + } > > > > +} > > > > + > > > > +char *get_vmlinux(char *input) > > > > +{ > > > > + const char *match_string1 = ".syms"; > > > > + const char *match_string2 = ".tmp_vmlinux.kallsyms"; > > > > + char *result = NULL; > > > > + char *match_pos; > > > > + > > > > + match_pos = strstr(input, match_string1); > > > > + if (!match_pos) > > > > + return NULL; > > > > + > > > > + match_pos = strstr(input, match_string2); > > > > + if (!match_pos) > > > > + return NULL; > > > > + > > > > + result = strdup(input); > > > > + match_pos = strstr(result, match_string1); > > > > + *match_pos = '\0'; > > > > + return result; > > > > +} > > > > diff --git a/scripts/kas_alias/a2l.h b/scripts/kas_alias/a2l.h > > > > new file mode 100644 > > > > index 000000000000..ca6419229dde > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/a2l.h > > > > @@ -0,0 +1,32 @@ > > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > > +#ifndef A2L_H > > > > +#define A2L_H > > > > +#include <stdint.h> > > > > + > > > > +#define ADDR2LINE "addr2line" > > > > +#define ADDR2LINE_ARGS "-fe" > > > > +//#define VMLINUX "vmlinux" > > > > +#define MAX_BUF 4096 > > > > +#define MAX_CMD_LEN 256 > > > > +#define P_READ 0 > > > > +#define P_WRITE 1 > > > > +#define A2L_DEFAULT 1 > > > > +#define A2L_CROSS 2 > > > > +#define A2L_LLVM 3 > > > > +#define A2L_MAKE_VALUE 2 > > > > + > > > > +extern int addr2line_pid; > > > > +extern int a2l_in[2]; > > > > +extern int a2l_out[2]; > > > > +extern char line[MAX_BUF]; > > > > +extern char vmlinux_path[MAX_BUF]; > > > > +extern char addr2line_cmd[MAX_CMD_LEN]; > > > > + > > > > +int addr2line_init(const char *cmd, const char *vmlinux); > > > > +char *addr2line_get_lines(uint64_t address); > > > > +int addr2line_cleanup(void); > > > > +const char *remove_subdir(const char *home, const char *f_path); > > > > +const char *get_addr2line(int mode); > > > > +char *get_vmlinux(char *input); > > > > + > > > > +#endif > > > > diff --git a/scripts/kas_alias/duplicates_list.c > > > > b/scripts/kas_alias/duplicates_list.c new file mode 100644 > > > > index 000000000000..e7a3d2917937 > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/duplicates_list.c > > > > @@ -0,0 +1,70 @@ > > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > > +#include <stdint.h> > > > > +#include <stdio.h> > > > > +#include <string.h> > > > > +#include <stdlib.h> > > > > +#include <stdbool.h> > > > > + > > > > +#include "item_list.h" > > > > +#include "duplicates_list.h" > > > > + > > > > +struct duplicate_item *find_duplicates(struct item *list) > > > > +{ > > > > + struct duplicate_item *current_duplicate = NULL; > > > > + struct duplicate_item *duplicates = NULL; > > > > + struct duplicate_item *new_duplicate; > > > > + struct item *current_item = list; > > > > + bool prev_was_duplicate = false; > > > > + struct item *prev_item = NULL; > > > > + > > > > + while (current_item) { > > > > + if ((prev_item && (strcmp(current_item->symb_name, > > > > prev_item->symb_name) == 0)) || + prev_was_duplicate) { > > > > + if (!duplicates) { > > > > + duplicates = malloc(sizeof(struct > > > > duplicate_item)); + if (!duplicates) > > > > + return NULL; > > > > + > > > > + duplicates->original_item = prev_item; > > > > + duplicates->next = NULL; > > > > + current_duplicate = duplicates; > > > > + } else { > > > > + new_duplicate = malloc(sizeof(struct > > > > duplicate_item)); + if (!new_duplicate) { > > > > + free_duplicates(&duplicates); > > > > + return NULL; > > > > + } > > > > + > > > > + new_duplicate->original_item = prev_item; > > > > + new_duplicate->next = NULL; > > > > + current_duplicate->next = new_duplicate; > > > > + current_duplicate = new_duplicate; > > > > + > > > > + if ((strcmp(current_item->symb_name, > > > > prev_item->symb_name) != 0) && + > > > > (prev_was_duplicate)) > > > > + prev_was_duplicate = false; > > > > + else > > > > + prev_was_duplicate = true; > > > > + } > > > > + } > > > > + > > > > + prev_item = current_item; > > > > + current_item = current_item->next; > > > > + } > > > > + > > > > + return duplicates; > > > > +} > > > > + > > > > +void free_duplicates(struct duplicate_item **duplicates) > > > > +{ > > > > + struct duplicate_item *duplicates_iterator = *duplicates; > > > > + struct duplicate_item *app; > > > > + > > > > + while (duplicates_iterator) { > > > > + app = duplicates_iterator; > > > > + duplicates_iterator = duplicates_iterator->next; > > > > + free(app); > > > > + } > > > > + > > > > + *duplicates = NULL; > > > > +} > > > > diff --git a/scripts/kas_alias/duplicates_list.h > > > > b/scripts/kas_alias/duplicates_list.h new file mode 100644 > > > > index 000000000000..76aa73e584bc > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/duplicates_list.h > > > > @@ -0,0 +1,15 @@ > > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > > +#ifndef DUPLICATES_LIST_H > > > > +#define DUPLICATES_LIST_H > > > > + > > > > +#include "item_list.h" > > > > + > > > > +struct duplicate_item { > > > > + struct item *original_item; > > > > + struct duplicate_item *next; > > > > +}; > > > > + > > > > +struct duplicate_item *find_duplicates(struct item *list); > > > > +void free_duplicates(struct duplicate_item **duplicates); > > > > + > > > > +#endif > > > > diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c > > > > new file mode 100644 > > > > index 000000000000..48f2e525592a > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/item_list.c > > > > @@ -0,0 +1,230 @@ > > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > > +#include <stdio.h> > > > > +#include <stdlib.h> > > > > +#include <stdint.h> > > > > +#include <string.h> > > > > +#include <stdbool.h> > > > > +#include <assert.h> > > > > +#include "item_list.h" > > > > + > > > > +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ > > > > + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) > > > > +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ > > > > + ((sort_by) == BY_NAME && strcmp((current)->symb_name, > > > > (temp)->symb_name) op 0) + > > > > +struct item *list_index[96] = {0}; > > > > + > > > > +void build_index(struct item *list) > > > > +{ > > > > + char current_first_letter = ' '; > > > > + struct item *current = list; > > > > + > > > > + while (current) { > > > > + if (current->symb_name[0] != current_first_letter) { > > > > + current_first_letter = current->symb_name[0]; > > > > + list_index[current_first_letter - 32] = current; > > > > + } > > > > + current = current->next; > > > > + } > > > > +} > > > > + > > > > +struct item *add_item(struct item **list, const char *name, char stype, > > > > uint64_t addr) +{ > > > > + struct item *new_item; > > > > + struct item *current; > > > > + > > > > + new_item = malloc(sizeof(struct item)); > > > > + if (!new_item) > > > > + return NULL; > > > > + > > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > > + new_item->addr = addr; > > > > + new_item->stype = stype; > > > > + new_item->next = NULL; > > > > + > > > > + if (!(*list)) { > > > > + *list = new_item; > > > > + } else { > > > > + current = *list; > > > > + while (current->next) > > > > + current = current->next; > > > > + > > > > + current->next = new_item; > > > > + } > > > > + return new_item; > > > > +} > > > > + > > > > +void sort_list(struct item **list, int sort_by) > > > > +{ > > > > + struct item *current = *list; > > > > + struct item *sorted = NULL; > > > > + struct item *next_item; > > > > + struct item *temp; > > > > + > > > > + if (!(*list) || !((*list)->next)) > > > > + return; > > > > + > > > > + while (current) { > > > > + next_item = current->next; > > > > + if (!sorted || > > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) > > > > || > > > > + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { > > > > + current->next = sorted; > > > > + sorted = current; > > > > + } else { > > > > + temp = sorted; > > > > + while (temp->next && > > > > + (CHECK_ORDER_BY_ADDRESS(sort_by, current, > > > > temp->next, >=) || + > > > > CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) + > > > > temp = temp->next; > > > > + > > > > + current->next = temp->next; > > > > + temp->next = current; > > > > + } > > > > + current = next_item; > > > > + } > > > > + > > > > + *list = sorted; > > > > +} > > > > + > > > > +struct item *merge(struct item *left, struct item *right, int sort_by) > > > > +{ > > > > + struct item *current = NULL; > > > > + struct item *result = NULL; > > > > + > > > > + if (!left) > > > > + return right; > > > > + if (!right) > > > > + return left; > > > > + > > > > + if (sort_by == BY_NAME) { > > > > + if (strcmp(left->symb_name, right->symb_name) <= 0) { > > > > + result = left; > > > > + left = left->next; > > > > + } else { > > > > + result = right; > > > > + right = right->next; > > > > + } > > > > + } else { > > > > + if (sort_by == BY_ADDRESS) { > > > > + if (left->addr <= right->addr) { > > > > + result = left; > > > > + left = left->next; > > > > + } else { > > > > + result = right; > > > > + right = right->next; > > > > + } > > > > + } > > > > + } > > > > + > > > > + current = result; > > > > + > > > > + while (left && right) { > > > > + if (sort_by == BY_NAME) { > > > > + if (strcmp(left->symb_name, right->symb_name) <= > > > > 0) { + current->next = left; > > > > + left = left->next; > > > > + } else { > > > > + current->next = right; > > > > + right = right->next; > > > > + } > > > > + } else { > > > > + if (sort_by == BY_ADDRESS) { > > > > + if (left->addr <= right->addr) { > > > > + current->next = left; > > > > + left = left->next; > > > > + } else { > > > > + current->next = right; > > > > + right = right->next; > > > > + } > > > > + } > > > > + } > > > > + > > > > + current = current->next; > > > > + } > > > > + > > > > + if (left) { > > > > + current->next = left; > > > > + } else { > > > > + if (right) > > > > + current->next = right; > > > > + } > > > > + > > > > + return result; > > > > +} > > > > + > > > > +struct item *merge_sort(struct item *head, int sort_by) > > > > +{ > > > > + struct item *right; > > > > + struct item *slow; > > > > + struct item *fast; > > > > + struct item *left; > > > > + > > > > + if (!head || !head->next) > > > > + return head; > > > > + > > > > + slow = head; > > > > + fast = head->next; > > > > + > > > > + while (fast && fast->next) { > > > > + slow = slow->next; > > > > + fast = fast->next->next; > > > > + } > > > > + > > > > + left = head; > > > > + right = slow->next; > > > > + slow->next = NULL; > > > > + > > > > + left = merge_sort(left, sort_by); > > > > + right = merge_sort(right, sort_by); > > > > + > > > > + return merge(left, right, sort_by); > > > > +} > > > > + > > > > +void sort_list_m(struct item **head, int sort_by) > > > > +{ > > > > + if (!(*head) || !((*head)->next)) > > > > + return; > > > > + > > > > + *head = merge_sort(*head, sort_by); > > > > +} > > > > + > > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > > + const char *name, uint64_t addr, char stype) > > > > +{ > > > > + struct item *new_item; > > > > + struct item *current; > > > > + int ret = 0; > > > > + > > > > + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : > > > > list; + while (current) { > > > > + if (current->addr == search_addr) { > > > > + new_item = malloc(sizeof(struct item)); > > > > + if (!new_item) > > > > + return ret; > > > > + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); > > > > + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; > > > > + new_item->addr = addr; > > > > + new_item->stype = stype; > > > > + new_item->next = current->next; > > > > + current->next = new_item; > > > > + ret = 1; > > > > + break; > > > > + } > > > > + current = current->next; > > > > + } > > > > + return ret; > > > > +} > > > > + > > > > +void free_items(struct item **head) > > > > +{ > > > > + struct item *app, *item_iterator = *head; > > > > + > > > > + while (item_iterator) { > > > > + app = item_iterator; > > > > + item_iterator = item_iterator->next; > > > > + free(app); > > > > + } > > > > + *head = NULL; > > > > +} > > > > diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h > > > > new file mode 100644 > > > > index 000000000000..b4891cb088ee > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/item_list.h > > > > @@ -0,0 +1,26 @@ > > > > +/* SPDX-License-Identifier: GPL-2.0-or-later */ > > > > +#ifndef ITEM_LIST_H > > > > +#define ITEM_LIST_H > > > > +#include <stdint.h> > > > > + > > > > +#define MAX_NAME_SIZE 256 > > > > +#define BY_ADDRESS 1 > > > > +#define BY_NAME 2 > > > > + > > > > +struct item { > > > > + char symb_name[MAX_NAME_SIZE]; > > > > + uint64_t addr; > > > > + char stype; > > > > + struct item *next; > > > > +}; > > > > + > > > > +void build_index(struct item *list); > > > > +struct item *add_item(struct item **list, const char *name, char stype, > > > > uint64_t addr); +void sort_list(struct item **list, int sort_by); > > > > +struct item *merge(struct item *left, struct item *right, int sort_by); > > > > +struct item *merge_sort(struct item *head, int sort_by); > > > > +void sort_list_m(struct item **head, int sort_by); > > > > +int insert_after(struct item *list, const uint64_t search_addr, > > > > + const char *name, uint64_t addr, char stype); > > > > +void free_items(struct item **head); > > > > +#endif > > > > diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c > > > > new file mode 100644 > > > > index 000000000000..532aeb39f851 > > > > --- /dev/null > > > > +++ b/scripts/kas_alias/kas_alias.c > > > > @@ -0,0 +1,217 @@ > > > > +// SPDX-License-Identifier: GPL-2.0-or-later > > > > +#include <stdio.h> > > > > +#include <stdlib.h> > > > > +#include <stdint.h> > > > > +#include <unistd.h> > > > > +#include <string.h> > > > > +#include <stdbool.h> > > > > +#include <stdarg.h> > > > > +#include <regex.h> > > > > + > > > > +#include "item_list.h" > > > > +#include "duplicates_list.h" > > > > +#include "a2l.h" > > > > + > > > > +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) > > > > +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') > > > > || \ + (((s)->stype) == 'd') || (((s)->stype) == > > > > 'D') || \ + (((s)->stype) == 'r') || > > > > (((s)->stype) == 'R')) +#ifdef CONFIG_KALLSYMS_ALIAS_DATA > > > > +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) > > > > +#else > > > > +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) > > > > +#endif > > > > +#define FNOMATCH 0 > > > > +#define FMATCH 1 > > > > +#define EREGEX 2 > > > > + > > > > +const char *ignore_regex[] = { > > > > + "^__cfi_.*$", // __cfi_ preamble > > > > +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL > > > > + "^_*TRACE_SYSTEM.*$", > > > > + "^__already_done\\.[0-9]+$", // Call a function once > > > > data + "^___tp_str\\.[0-9]+$", > > > > + "^___done\\.[0-9]+$", > > > > + "^__print_once\\.[0-9]+$", > > > > + "^_rs\\.[0-9]+$", > > > > + "^__compound_literal\\.[0-9]+$", > > > > + "^___once_key\\.[0-9]+$", > > > > + "^__func__\\.[0-9]+$", > > > > + "^__msg\\.[0-9]+$", > > > > + "^CSWTCH\\.[0-9]+$", > > > > + "^__flags\\.[0-9]+$", > > > > + "^__wkey.*$", > > > > + "^__mkey.*$", > > > > + "^__key.*$", > > > > +#endif > > > > + "^__pfx_.*$" // NOP-padding > > > > +}; > > > > + > > > > +int suffix_serial; > > > > + > > > > +static inline void verbose_msg(bool verbose, const char *fmt, ...) > > > > +{ > > > > + va_list args; > > > > + > > > > + va_start(args, fmt); > > > > + if (verbose) > > > > + printf(fmt, args); > > > > + > > > > + va_end(args); > > > > +} > > > > + > > > > +static void create_suffix(const char *name, char *output_suffix) > > > > +{ > > > > + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); > > > > +} > > > > + > > > > +static void create_file_suffix(const char *name, uint64_t address, char > > > > *output_suffix, char *cwd) +{ > > > > + const char *f_path; > > > > + char *buf; > > > > + int i = 0; > > > > + > > > > + buf = addr2line_get_lines(address); > > > > + f_path = remove_subdir(cwd, buf); > > > > + if (f_path) { > > > > + sprintf(output_suffix, "%s@%s", name, f_path); > > > > + while (*(output_suffix + i) != '\0') { > > > > + switch (*(output_suffix + i)) { > > > > + case '/': > > > > + case ':': > > > > + case '.': > > > > + *(output_suffix + i) = '_'; > > > > + break; > > > > + default: > > > > + } > > > > + i++; > > > > + } > > > > + } else { > > > > + create_suffix(name, output_suffix); > > > > + } > > > > +} > > > > + > > > > +static int filter_symbols(char *symbol, const char **ignore_list, int > > > > regex_no) +{ > > > > + regex_t regex; > > > > + int res, i; > > > > + > > > > + for (i = 0; i < regex_no; i++) { > > > > + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); > > > > + if (res) > > > > + return -EREGEX; > > > > + > > > > + res = regexec(®ex, symbol, 0, NULL, 0); > > > > + regfree(®ex); > > > > + switch (res) { > > > > + case 0: > > > > + return FMATCH; > > > > + case REG_NOMATCH: > > > > + break; > > > > + default: > > > > + return -EREGEX; > > > > + } > > > > + } > > > > + > > > > + return FNOMATCH; > > > > +} > > > > + > > > > +int main(int argc, char *argv[]) > > > > +{ > > > > + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; > > > > + struct duplicate_item *duplicate_iterator; > > > > + struct duplicate_item *duplicate; > > > > + struct item *head = {NULL}; > > > > + bool need_2_process = true; > > > > + struct item *last = {NULL}; > > > > + struct item *current; > > > > + int verbose_mode = 0; > > > > + uint64_t address; > > > > + FILE *fp; > > > > + int res; > > > > + > > > > + if (argc < 2 || argc > 3) { > > > > + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); > > > > + return 1; > > > > + } > > > > + > > > > + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) > > > > + verbose_mode = 1; > > > > + > > > > + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); > > > > + > > > > + fp = fopen(argv[1], "r"); > > > > + if (!fp) { > > > > + printf("Can't open input file.\n"); > > > > + return 1; > > > > + } > > > > + > > > > + if (!addr2line_init(get_addr2line(A2L_DEFAULT), > > > > get_vmlinux(argv[1]))) + return 1; > > > > + > > > > + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { > > > > + if (strstr(sym_name, "@_")) { > > > > + if (verbose_mode && need_2_process) > > > > + printf("Already processed\n"); > > > > + need_2_process = false; > > > > + } > > > > + last = add_item(&last, sym_name, t, address); > > > > + if (!last) { > > > > + printf("Error in allocate memory\n"); > > > > + free_items(&head); > > > > + return 1; > > > > + } > > > > + > > > > + if (!head) > > > > + head = last; > > > > + } > > > > + > > > > + fclose(fp); > > > > + > > > > + if (need_2_process) { > > > > + verbose_msg(verbose_mode, "Sorting nm data\n"); > > > > + sort_list_m(&head, BY_NAME); > > > > + verbose_msg(verbose_mode, "Scanning nm data for > > > > duplicates\n"); + duplicate = find_duplicates(head); > > > > + if (!duplicate) { > > > > + printf("Error in duplicates list\n"); > > > > + return 1; > > > > + } > > > > + > > > > + verbose_msg(verbose_mode, "Applying suffixes\n"); > > > > + build_index(head); > > > > + duplicate_iterator = duplicate; > > > > + while (duplicate_iterator) { > > > > + res = > > > > filter_symbols(duplicate_iterator->original_item->symb_name, + > > > > ignore_regex, sizeof(ignore_regex) / + > > > > sizeof(ignore_regex[0])); + > > > > if (res != FMATCH && > > > > + > > > > SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { + > > > > if (res < 0) > > > > + return 1; > > > > + > > > > + > > > > create_file_suffix(duplicate_iterator->original_item->symb_name, + > > > > > > > > duplicate_iterator->original_item->addr, + > > > > new_name, vmlinux_path); + > > > > if (!insert_after(head, duplicate_iterator->original_item->addr, + > > > > new_name, > > > > duplicate_iterator->original_item->addr, + > > > > duplicate_iterator->original_item->stype)) + > > > > return 1; > > > > + } > > > > + > > > > + duplicate_iterator = duplicate_iterator->next; > > > > + } > > > > + > > > > + sort_list_m(&head, BY_ADDRESS); > > > > + } > > > > + current = head; > > > > + while (current) { > > > > + printf("%08lx %c %s\n", current->addr, current->stype, > > > > current->symb_name); + current = current->next; > > > > + } > > > > + > > > > + free_items(&head); > > > > + free_duplicates(&duplicate); > > > > + addr2line_cleanup(); > > > > + return 0; > > > > +} > > > > diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh > > > > index a432b171be82..cacf60b597ce 100755 > > > > --- a/scripts/link-vmlinux.sh > > > > +++ b/scripts/link-vmlinux.sh > > > > @@ -89,8 +89,9 @@ vmlinux_link() > > > > > > > > ldflags="${ldflags} ${wl}--script=${objtree}/${KBUILD_LDS}" > > > > > > > > - # The kallsyms linking does not need debug symbols included. > > > > - if [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > > > > + # The kallsyms linking does not need debug symbols included, > > > > unless the KALLSYMS_ALIAS. + if [ ! is_enabled > > > > CONFIG_KALLSYMS_ALIAS ] && \ > > > > + [ "$output" != "${output#.tmp_vmlinux.kallsyms}" ] ; then > > > > > > > > ldflags="${ldflags} ${wl}--strip-debug" > > > > > > > > fi > > > > > > > > @@ -161,7 +162,11 @@ kallsyms() > > > > > > > > fi > > > > > > > > info KSYMS ${2} > > > > > > > > - scripts/kallsyms ${kallsymopt} ${1} > ${2} > > > > + if is_enabled CONFIG_KALLSYMS_ALIAS; then > > > > + ALIAS=".alias" > > > > + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} > > > > + fi > > > > + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} > > > > > > > > } > > > > > > > > # Perform one step in kallsyms generation, including temporary linking of > > > > > > > > -- > > > > 2.34.1 > > > > Best regards. > > > > > BR > Alessandro
diff --git a/init/Kconfig b/init/Kconfig index f7f65af4ee12..5ae7c613cdc4 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -1737,6 +1737,42 @@ config KALLSYMS_BASE_RELATIVE time constants, and no relocation pass is required at runtime to fix up the entries based on the runtime load address of the kernel. +config KALLSYMS_ALIAS + bool "Produces alias for duplicated symbols" if EXPERT + depends on KALLSYMS + help + It is not uncommon for drivers or modules related to similar + peripherals to have symbols with the exact same name. + While this is not a problem for the kernel's binary itself, it + becomes an issue when attempting to trace or probe specific + functions using infrastructure like ftrace or kprobe. + + This option addresses this challenge by extending the symbol names + with unique suffixes during the kernel build process. + The newly created aliases for these duplicated symbols are unique + names that can be fed to the ftrace sysfs interface. By doing so, it + enables previously unreachable symbols to be probed. + +config CONFIG_KALLSYMS_ALIAS_DATA + bool "Produces alias also for data" + depends on KALLSYMS_ALIAS + help + Sometimes it can be useful to refer to data. In live patch scenarios, + you may find yourself needing to use symbols that are shared with + other functions. Since symbols face the same issue as functions, this + option allows you to create aliases for data as well. + +config CONFIG_KALLSYMS_ALIAS_DATA_ALL + bool "Removes all filter when producing data alias" + depends on CONFIG_KALLSYMS_ALIAS_DATA + help + When selecting data aliases, not all symbols are included in the set + This is because many symbols are unlikely to be used. If you choose + to have an alias for all data symbols, be aware that it will + significantly increase the size. + + If unsure, say N. + # end of the "standard kernel features (expert users)" menu # syscall, maps, verifier diff --git a/scripts/Makefile b/scripts/Makefile index 32b6ba722728..65fafe17cfe5 100644 --- a/scripts/Makefile +++ b/scripts/Makefile @@ -49,3 +49,7 @@ subdir-$(CONFIG_SECURITY_SELINUX) += selinux # Let clean descend into subdirs subdir- += basic dtc gdb kconfig mod + +# KALLSyms alias +subdir-$(CONFIG_KALLSYMS_ALIAS) += kas_alias + diff --git a/scripts/kas_alias/Makefile b/scripts/kas_alias/Makefile new file mode 100644 index 000000000000..523fa3441013 --- /dev/null +++ b/scripts/kas_alias/Makefile @@ -0,0 +1,4 @@ +# SPDX-License-Identifier: GPL-2.0 +hostprogs-always-$(CONFIG_KALLSYMS_ALIAS) += kas_alias + +kas_alias-objs := duplicates_list.o item_list.o kas_alias.o diff --git a/scripts/kas_alias/duplicates_list.c b/scripts/kas_alias/duplicates_list.c new file mode 100644 index 000000000000..e7a3d2917937 --- /dev/null +++ b/scripts/kas_alias/duplicates_list.c @@ -0,0 +1,70 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +#include <stdint.h> +#include <stdio.h> +#include <string.h> +#include <stdlib.h> +#include <stdbool.h> + +#include "item_list.h" +#include "duplicates_list.h" + +struct duplicate_item *find_duplicates(struct item *list) +{ + struct duplicate_item *current_duplicate = NULL; + struct duplicate_item *duplicates = NULL; + struct duplicate_item *new_duplicate; + struct item *current_item = list; + bool prev_was_duplicate = false; + struct item *prev_item = NULL; + + while (current_item) { + if ((prev_item && (strcmp(current_item->symb_name, prev_item->symb_name) == 0)) || + prev_was_duplicate) { + if (!duplicates) { + duplicates = malloc(sizeof(struct duplicate_item)); + if (!duplicates) + return NULL; + + duplicates->original_item = prev_item; + duplicates->next = NULL; + current_duplicate = duplicates; + } else { + new_duplicate = malloc(sizeof(struct duplicate_item)); + if (!new_duplicate) { + free_duplicates(&duplicates); + return NULL; + } + + new_duplicate->original_item = prev_item; + new_duplicate->next = NULL; + current_duplicate->next = new_duplicate; + current_duplicate = new_duplicate; + + if ((strcmp(current_item->symb_name, prev_item->symb_name) != 0) && + (prev_was_duplicate)) + prev_was_duplicate = false; + else + prev_was_duplicate = true; + } + } + + prev_item = current_item; + current_item = current_item->next; + } + + return duplicates; +} + +void free_duplicates(struct duplicate_item **duplicates) +{ + struct duplicate_item *duplicates_iterator = *duplicates; + struct duplicate_item *app; + + while (duplicates_iterator) { + app = duplicates_iterator; + duplicates_iterator = duplicates_iterator->next; + free(app); + } + + *duplicates = NULL; +} diff --git a/scripts/kas_alias/duplicates_list.h b/scripts/kas_alias/duplicates_list.h new file mode 100644 index 000000000000..76aa73e584bc --- /dev/null +++ b/scripts/kas_alias/duplicates_list.h @@ -0,0 +1,15 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +#ifndef DUPLICATES_LIST_H +#define DUPLICATES_LIST_H + +#include "item_list.h" + +struct duplicate_item { + struct item *original_item; + struct duplicate_item *next; +}; + +struct duplicate_item *find_duplicates(struct item *list); +void free_duplicates(struct duplicate_item **duplicates); + +#endif diff --git a/scripts/kas_alias/item_list.c b/scripts/kas_alias/item_list.c new file mode 100644 index 000000000000..7c9d5aecca9a --- /dev/null +++ b/scripts/kas_alias/item_list.c @@ -0,0 +1,230 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +#include <stdio.h> +#include <stdlib.h> +#include <stdint.h> +#include <string.h> +#include <stdbool.h> +#include <assert.h> +#include "item_list.h" + +#define CHECK_ORDER_BY_ADDRESS(sort_by, current, temp, op) \ + ((sort_by) == BY_ADDRESS && (current)->addr op (temp)->addr) +#define CHECK_ORDER_BY_NAME(sort_by, current, temp, op) \ + ((sort_by) == BY_NAME && strcmp((current)->symb_name, (temp)->symb_name) op 0) + +struct item *list_index[96] = {0}; + +void build_index(struct item *list) +{ + char current_first_letter = ' '; + struct item *current = list; + + while (current) { + if (current->symb_name[0] != current_first_letter) { + current_first_letter = current->symb_name[0]; + list_index[current_first_letter - 32] = current; + } + current = current->next; + } +} + +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr) +{ + struct item *new_item; + struct item *current; + + new_item = malloc(sizeof(struct item)); + if (!new_item) + return NULL; + + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; + new_item->addr = addr; + new_item->stype = stype; + new_item->next = NULL; + + if (!(*list)) { + *list = new_item; + } else { + current = *list; + while (current->next) + current = current->next; + + current->next = new_item; + } + return new_item; +} + +void sort_list(struct item **list, int sort_by) +{ + struct item *current = *list; + struct item *sorted = NULL; + struct item *next_item; + struct item *temp; + + if (!(*list) || !((*list)->next)) + return; + + while (current) { + next_item = current->next; + if (!sorted || + (CHECK_ORDER_BY_ADDRESS(sort_by, current, sorted, <) || + CHECK_ORDER_BY_NAME(sort_by, current, sorted, >=))) { + current->next = sorted; + sorted = current; + } else { + temp = sorted; + while (temp->next && + (CHECK_ORDER_BY_ADDRESS(sort_by, current, temp->next, >=) || + CHECK_ORDER_BY_NAME(sort_by, current, temp->next, >=))) + temp = temp->next; + + current->next = temp->next; + temp->next = current; + } + current = next_item; + } + + *list = sorted; +} + +struct item *merge(struct item *left, struct item *right, int sort_by) +{ + struct item *current = NULL; + struct item *result = NULL; + + if (!left) + return right; + if (!right) + return left; + + if (sort_by == BY_NAME) { + if (strcmp(left->symb_name, right->symb_name) <= 0) { + result = left; + left = left->next; + } else { + result = right; + right = right->next; + } + } else { + if (sort_by == BY_ADDRESS) { + if (left->addr <= right->addr) { + result = left; + left = left->next; + } else { + result = right; + right = right->next; + } + } + } + + current = result; + + while (left && right) { + if (sort_by == BY_NAME) { + if (strcmp(left->symb_name, right->symb_name) <= 0) { + current->next = left; + left = left->next; + } else { + current->next = right; + right = right->next; + } + } else { + if (sort_by == BY_ADDRESS) { + if (left->addr <= right->addr) { + current->next = left; + left = left->next; + } else { + current->next = right; + right = right->next; + } + } + } + + current = current->next; + } + + if (left) { + current->next = left; + } else { + if (right) + current->next = right; + } + + return result; +} + +struct item *merge_sort(struct item *head, int sort_by) +{ + struct item *right; + struct item *slow; + struct item *fast; + struct item *left; + + if (!head || !head->next) + return head; + + slow = head; + fast = head->next; + + while (fast && fast->next) { + slow = slow->next; + fast = fast->next->next; + } + + left = head; + right = slow->next; + slow->next = NULL; + + left = merge_sort(left, sort_by); + right = merge_sort(right, sort_by); + + return merge(left, right, sort_by); +} + +void sort_list_m(struct item **head, int sort_by) +{ + if (!(*head) || !((*head)->next)) + return; + + *head = merge_sort(*head, sort_by); +} + +int insert_after(struct item *list, const uint64_t search_addr, + const char *name, uint64_t addr, char stype) +{ + struct item *new_item; + struct item *current; + int ret = 0; + + current = (list_index[name[0] - 32]) ? list_index[name[0] - 32] : list; + while (current) { + if (current->addr == search_addr) { + new_item = malloc(sizeof(struct item)); + if (!new_item) + return ret; + strncpy(new_item->symb_name, name, MAX_NAME_SIZE); + new_item->symb_name[MAX_NAME_SIZE - 1] = '\0'; + new_item->addr = addr; + new_item->stype = stype; + new_item->next = current->next; + current->next = new_item; + ret = 1; + break; + } + current = current->next; + } + return ret; +} + +void free_items(struct item **head) +{ + struct item *app, *item_iterator = *head; + + while (item_iterator) { + app = item_iterator; + item_iterator = item_iterator->next; + free(app); + } + *head = NULL; +} diff --git a/scripts/kas_alias/item_list.h b/scripts/kas_alias/item_list.h new file mode 100644 index 000000000000..b4891cb088ee --- /dev/null +++ b/scripts/kas_alias/item_list.h @@ -0,0 +1,26 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +#ifndef ITEM_LIST_H +#define ITEM_LIST_H +#include <stdint.h> + +#define MAX_NAME_SIZE 256 +#define BY_ADDRESS 1 +#define BY_NAME 2 + +struct item { + char symb_name[MAX_NAME_SIZE]; + uint64_t addr; + char stype; + struct item *next; +}; + +void build_index(struct item *list); +struct item *add_item(struct item **list, const char *name, char stype, uint64_t addr); +void sort_list(struct item **list, int sort_by); +struct item *merge(struct item *left, struct item *right, int sort_by); +struct item *merge_sort(struct item *head, int sort_by); +void sort_list_m(struct item **head, int sort_by); +int insert_after(struct item *list, const uint64_t search_addr, + const char *name, uint64_t addr, char stype); +void free_items(struct item **head); +#endif diff --git a/scripts/kas_alias/kas_alias.c b/scripts/kas_alias/kas_alias.c new file mode 100644 index 000000000000..eb873b368023 --- /dev/null +++ b/scripts/kas_alias/kas_alias.c @@ -0,0 +1,186 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +#include <stdio.h> +#include <stdlib.h> +#include <stdint.h> +#include <string.h> +#include <stdbool.h> +#include <stdarg.h> +#include <regex.h> + +#include "item_list.h" +#include "duplicates_list.h" + +#define SYMB_IS_TEXT(s) ((((s)->stype) == 't') || (((s)->stype) == 'T')) +#define SYMB_IS_DATA(s) ((((s)->stype) == 'b') || (((s)->stype) == 'B') || \ + (((s)->stype) == 'd') || (((s)->stype) == 'D') || \ + (((s)->stype) == 'r') || (((s)->stype) == 'R')) +#ifdef CONFIG_KALLSYMS_ALIAS_DATA +#define SYMB_NEEDS_ALIAS(s) (SYMB_IS_TEXT(s) || SYMB_IS_DATA(s)) +#else +#define SYMB_NEEDS_ALIAS(s) SYMB_IS_TEXT(s) +#endif +#define FNOMATCH 0 +#define FMATCH 1 +#define EREGEX 2 + +const char *ignore_regex[] = { + "^__cfi_.*$", // __cfi_ preamble +#ifndef CONFIG_KALLSYMS_ALIAS_DATA_ALL + "^_*TRACE_SYSTEM.*$", + "^__already_done\\.[0-9]+$", // Call a function once data + "^___tp_str\\.[0-9]+$", + "^___done\\.[0-9]+$", + "^__print_once\\.[0-9]+$", + "^_rs\\.[0-9]+$", + "^__compound_literal\\.[0-9]+$", + "^___once_key\\.[0-9]+$", + "^__func__\\.[0-9]+$", + "^__msg\\.[0-9]+$", + "^CSWTCH\\.[0-9]+$", + "^__flags\\.[0-9]+$", + "^__wkey.*$", + "^__mkey.*$", + "^__key.*$", +#endif + "^__pfx_.*$" // NOP-padding +}; + +int suffix_serial; + +static inline void verbose_msg(bool verbose, const char *fmt, ...) +{ + va_list args; + + va_start(args, fmt); + if (verbose) + printf(fmt, args); + + va_end(args); +} + +static void create_suffix(const char *name, char *output_suffix) +{ + sprintf(output_suffix, "%s__alias__%d", name, suffix_serial++); +} + +static int filter_symbols(char *symbol, const char **ignore_list, int regex_no) +{ + regex_t regex; + int res, i; + + for (i = 0; i < regex_no; i++) { + res = regcomp(®ex, ignore_list[i], REG_EXTENDED); + if (res) + return -EREGEX; + + res = regexec(®ex, symbol, 0, NULL, 0); + regfree(®ex); + switch (res) { + case 0: + return FMATCH; + case REG_NOMATCH: + break; + default: + return -EREGEX; + } + + } + + return FNOMATCH; +} + +int main(int argc, char *argv[]) +{ + char t, sym_name[MAX_NAME_SIZE], new_name[MAX_NAME_SIZE + 15]; + struct duplicate_item *duplicate_iterator; + struct duplicate_item *duplicate; + struct item *head = {NULL}; + bool need_2_process = true; + struct item *last = {NULL}; + struct item *current; + int verbose_mode = 0; + uint64_t address; + FILE *fp; + int res; + + if (argc < 2 || argc > 3) { + printf("Usage: %s <nmfile> [-verbose]\n", argv[0]); + return 1; + } + + if (argc == 3 && strcmp(argv[2], "-verbose") == 0) + verbose_mode = 1; + + verbose_msg(verbose_mode, "Scanning nm data(%s)\n", argv[1]); + + fp = fopen(argv[1], "r"); + if (!fp) { + printf("Can't open input file.\n"); + return 1; + } + + while (fscanf(fp, "%lx %c %99s\n", &address, &t, sym_name) == 3) { + if (strstr(sym_name, "__alias__1") != NULL) { + if (verbose_mode && need_2_process) + printf("Already processed\n"); + need_2_process = false; + } + last = add_item(&last, sym_name, t, address); + if (!last) { + printf("Error in allocate memory\n"); + free_items(&head); + return 1; + } + + if (!head) + head = last; + } + + fclose(fp); + + if (need_2_process) { + verbose_msg(verbose_mode, "Sorting nm data\n"); + sort_list_m(&head, BY_NAME); + verbose_msg(verbose_mode, "Scanning nm data for duplicates\n"); + duplicate = find_duplicates(head); + if (!duplicate) { + printf("Error in duplicates list\n"); + return 1; + } + + verbose_msg(verbose_mode, "Applying suffixes\n"); + build_index(head); + duplicate_iterator = duplicate; + while (duplicate_iterator) { + res = filter_symbols(duplicate_iterator->original_item->symb_name, + ignore_regex, sizeof(ignore_regex) / + sizeof(ignore_regex[0])); + if ((res != FMATCH) && + SYMB_NEEDS_ALIAS(duplicate_iterator->original_item)) { + if (res < 0) + return 1; + + create_suffix(duplicate_iterator->original_item->symb_name, + new_name); + if (!insert_after(head, duplicate_iterator->original_item->addr, + new_name, duplicate_iterator->original_item->addr, + duplicate_iterator->original_item->stype)) + return 1; + } + + duplicate_iterator = duplicate_iterator->next; + } + + sort_list_m(&head, BY_ADDRESS); + } + current = head; + while (current) { + printf("%08lx %c %s\n", current->addr, current->stype, current->symb_name); + current = current->next; + } + + free_items(&head); + free_duplicates(&duplicate); + + return 0; +} diff --git a/scripts/link-vmlinux.sh b/scripts/link-vmlinux.sh index a432b171be82..4e3dea3ac2df 100755 --- a/scripts/link-vmlinux.sh +++ b/scripts/link-vmlinux.sh @@ -161,7 +161,11 @@ kallsyms() fi info KSYMS ${2} - scripts/kallsyms ${kallsymopt} ${1} > ${2} + if is_enabled CONFIG_KALLSYMS_ALIAS; then + ALIAS=".alias" + scripts/kas_alias/kas_alias ${1} >${1}${ALIAS} + fi + scripts/kallsyms ${kallsymopt} ${1}${ALIAS} > ${2} } # Perform one step in kallsyms generation, including temporary linking of