| Message ID | 1622542057-14632-1-git-send-email-faiyazm@codeaurora.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v8] mm: slub: move sysfs slab alloc/free interfaces to debugfs | expand |
On Tue, Jun 01, 2021 at 03:37:37PM +0530, Faiyaz Mohammed wrote: > alloc_calls and free_calls implementation in sysfs have two issues, > one is PAGE_SIZE limitiation of sysfs and other is it does not adhere > to "one value per file" rule. > > To overcome this issues, move the alloc_calls and free_calls implemeation > to debugfs. > > Debugfs cache will be created if SLAB_STORE_USER flag is set. > > Rename the alloc_calls/free_calls to alloc_traces/free_traces, > to be inline with what it does. > > Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Hi Faiyaz,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on linus/master]
[also build test ERROR on v5.13-rc4]
[cannot apply to hnaz-linux-mm/master next-20210601]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
base: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56
config: i386-tinyconfig (attached as .config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce (this is a W=1 build):
# https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a
# save the attached .config to linux build tree
make W=1 ARCH=i386
If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>
All errors (new ones prefixed by >>):
mm/slab_common.c: In function 'slab_caches_to_rcu_destroy_workfn':
>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror=implicit-function-declaration]
452 | debugfs_slab_release(s);
| ^~~~~~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
vim +/debugfs_slab_release +452 mm/slab_common.c
427
428 static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
429 {
430 LIST_HEAD(to_destroy);
431 struct kmem_cache *s, *s2;
432
433 /*
434 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
435 * @slab_caches_to_rcu_destroy list. The slab pages are freed
436 * through RCU and the associated kmem_cache are dereferenced
437 * while freeing the pages, so the kmem_caches should be freed only
438 * after the pending RCU operations are finished. As rcu_barrier()
439 * is a pretty slow operation, we batch all pending destructions
440 * asynchronously.
441 */
442 mutex_lock(&slab_mutex);
443 list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
444 mutex_unlock(&slab_mutex);
445
446 if (list_empty(&to_destroy))
447 return;
448
449 rcu_barrier();
450
451 list_for_each_entry_safe(s, s2, &to_destroy, list) {
> 452 debugfs_slab_release(s);
453 kfence_shutdown_cache(s);
454 #ifdef SLAB_SUPPORTS_SYSFS
455 sysfs_slab_release(s);
456 #else
457 slab_kmem_cache_release(s);
458 #endif
459 }
460 }
461
---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
On 6/1/21 1:08 PM, kernel test robot wrote: > Hi Faiyaz, > > Thank you for the patch! Yet something to improve: > > [auto build test ERROR on linus/master] > [also build test ERROR on v5.13-rc4] > [cannot apply to hnaz-linux-mm/master next-20210601] > [If your patch is applied to the wrong git tree, kindly drop us a note. > And when submitting patch, we suggest to use '--base' as documented in > https://git-scm.com/docs/git-format-patch] > > url: https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903 > base: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56 > config: i386-tinyconfig (attached as .config) > compiler: gcc-9 (Debian 9.3.0-22) 9.3.0 > reproduce (this is a W=1 build): > # https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a > git remote add linux-review https://github.com/0day-ci/linux > git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903 > git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a > # save the attached .config to linux build tree > make W=1 ARCH=i386 > > If you fix the issue, kindly add following tag as appropriate > Reported-by: kernel test robot <lkp@intel.com> > > All errors (new ones prefixed by >>): > > mm/slab_common.c: In function 'slab_caches_to_rcu_destroy_workfn': >>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror=implicit-function-declaration] > 452 | debugfs_slab_release(s); > | ^~~~~~~~~~~~~~~~~~~~ > cc1: some warnings being treated as errors Oh, right, CONFIG_SLOB. How about moving the ifdef-ed declaration from slub-def.h to mm/slab.h and discarding the slab-def.h empty one. That will take care of SLOB too. > vim +/debugfs_slab_release +452 mm/slab_common.c > > 427 > 428 static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work) > 429 { > 430 LIST_HEAD(to_destroy); > 431 struct kmem_cache *s, *s2; > 432 > 433 /* > 434 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the > 435 * @slab_caches_to_rcu_destroy list. The slab pages are freed > 436 * through RCU and the associated kmem_cache are dereferenced > 437 * while freeing the pages, so the kmem_caches should be freed only > 438 * after the pending RCU operations are finished. As rcu_barrier() > 439 * is a pretty slow operation, we batch all pending destructions > 440 * asynchronously. > 441 */ > 442 mutex_lock(&slab_mutex); > 443 list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy); > 444 mutex_unlock(&slab_mutex); > 445 > 446 if (list_empty(&to_destroy)) > 447 return; > 448 > 449 rcu_barrier(); > 450 > 451 list_for_each_entry_safe(s, s2, &to_destroy, list) { > > 452 debugfs_slab_release(s); > 453 kfence_shutdown_cache(s); > 454 #ifdef SLAB_SUPPORTS_SYSFS > 455 sysfs_slab_release(s); > 456 #else > 457 slab_kmem_cache_release(s); > 458 #endif > 459 } > 460 } > 461 > > --- > 0-DAY CI Kernel Test Service, Intel Corporation > https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org >
On 6/1/2021 5:44 PM, Vlastimil Babka wrote: > On 6/1/21 1:08 PM, kernel test robot wrote: >> Hi Faiyaz, >> >> Thank you for the patch! Yet something to improve: >> >> [auto build test ERROR on linus/master] >> [also build test ERROR on v5.13-rc4] >> [cannot apply to hnaz-linux-mm/master next-20210601] >> [If your patch is applied to the wrong git tree, kindly drop us a note. >> And when submitting patch, we suggest to use '--base' as documented in >> https://git-scm.com/docs/git-format-patch] >> >> url: https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903 >> base: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56 >> config: i386-tinyconfig (attached as .config) >> compiler: gcc-9 (Debian 9.3.0-22) 9.3.0 >> reproduce (this is a W=1 build): >> # https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a >> git remote add linux-review https://github.com/0day-ci/linux >> git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903 >> git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a >> # save the attached .config to linux build tree >> make W=1 ARCH=i386 >> >> If you fix the issue, kindly add following tag as appropriate >> Reported-by: kernel test robot <lkp@intel.com> >> >> All errors (new ones prefixed by >>): >> >> mm/slab_common.c: In function 'slab_caches_to_rcu_destroy_workfn': >>>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror=implicit-function-declaration] >> 452 | debugfs_slab_release(s); >> | ^~~~~~~~~~~~~~~~~~~~ >> cc1: some warnings being treated as errors > > Oh, right, CONFIG_SLOB. > How about moving the ifdef-ed declaration from slub-def.h to mm/slab.h and > discarding the slab-def.h empty one. That will take care of SLOB too. > Yes, doing the same. >> vim +/debugfs_slab_release +452 mm/slab_common.c >> >> 427 >> 428 static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work) >> 429 { >> 430 LIST_HEAD(to_destroy); >> 431 struct kmem_cache *s, *s2; >> 432 >> 433 /* >> 434 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the >> 435 * @slab_caches_to_rcu_destroy list. The slab pages are freed >> 436 * through RCU and the associated kmem_cache are dereferenced >> 437 * while freeing the pages, so the kmem_caches should be freed only >> 438 * after the pending RCU operations are finished. As rcu_barrier() >> 439 * is a pretty slow operation, we batch all pending destructions >> 440 * asynchronously. >> 441 */ >> 442 mutex_lock(&slab_mutex); >> 443 list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy); >> 444 mutex_unlock(&slab_mutex); >> 445 >> 446 if (list_empty(&to_destroy)) >> 447 return; >> 448 >> 449 rcu_barrier(); >> 450 >> 451 list_for_each_entry_safe(s, s2, &to_destroy, list) { >> > 452 debugfs_slab_release(s); >> 453 kfence_shutdown_cache(s); >> 454 #ifdef SLAB_SUPPORTS_SYSFS >> 455 sysfs_slab_release(s); >> 456 #else >> 457 slab_kmem_cache_release(s); >> 458 #endif >> 459 } >> 460 } >> 461 >> >> --- >> 0-DAY CI Kernel Test Service, Intel Corporation >> https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org >> >
On 6/1/21 12:07 PM, Faiyaz Mohammed wrote: > alloc_calls and free_calls implementation in sysfs have two issues, > one is PAGE_SIZE limitiation of sysfs and other is it does not adhere > to "one value per file" rule. > > To overcome this issues, move the alloc_calls and free_calls implemeation > to debugfs. > > Debugfs cache will be created if SLAB_STORE_USER flag is set. > > Rename the alloc_calls/free_calls to alloc_traces/free_traces, > to be inline with what it does. > > Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org> When reading the files I'm getting dmesg errors like this: [ 37.998152] seq_file: buggy .next function slab_debugfs_next did not update position index Don't recall this from previous versions? > diff --git a/mm/slab_common.c b/mm/slab_common.c > index a4a5714..ca39e84 100644 > --- a/mm/slab_common.c > +++ b/mm/slab_common.c > @@ -449,6 +449,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work) > rcu_barrier(); > > list_for_each_entry_safe(s, s2, &to_destroy, list) { > + debugfs_slab_release(s); You should remove this one, you already do the release when queueing the work below. This is not like sysfs where we distinguish _unlink() when queueing and _release when processing the work. > +static struct dentry *slab_debugfs_root; > +struct loc_track t = { 0, 0, NULL }; Ooops, this can't be a global variable, especially when shared without any lock protection - but even with a protection it would be very suboptimal. You'll need to make it a seq_file private, see e.g. here for an example: https://lore.kernel.org/linux-mm/20210521121127.24653-3-glittao@gmail.com/ > static void free_loc_track(struct loc_track *t) > { > if (t->max) > @@ -4825,82 +4840,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s, > add_location(t, s, get_track(s, p, alloc)); > put_map(map); > } > - > -static int list_locations(struct kmem_cache *s, char *buf, > - enum track_item alloc) > -{ > - int len = 0; > - unsigned long i; > - struct loc_track t = { 0, 0, NULL }; > - int node; > - struct kmem_cache_node *n; > - > - if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location), > - GFP_KERNEL)) { > - return sysfs_emit(buf, "Out of memory\n"); > - } > - /* Push back cpu slabs */ > - flush_all(s); > - > - for_each_kmem_cache_node(s, node, n) { > - unsigned long flags; > - struct page *page; > - > - if (!atomic_long_read(&n->nr_slabs)) > - continue; > - > - spin_lock_irqsave(&n->list_lock, flags); > - list_for_each_entry(page, &n->partial, slab_list) > - process_slab(&t, s, page, alloc); > - list_for_each_entry(page, &n->full, slab_list) > - process_slab(&t, s, page, alloc); > - spin_unlock_irqrestore(&n->list_lock, flags); > - } > - > - for (i = 0; i < t.count; i++) { > - struct location *l = &t.loc[i]; > - > - len += sysfs_emit_at(buf, len, "%7ld ", l->count); > - > - if (l->addr) > - len += sysfs_emit_at(buf, len, "%pS", (void *)l->addr); > - else > - len += sysfs_emit_at(buf, len, "<not-available>"); > - > - if (l->sum_time != l->min_time) > - len += sysfs_emit_at(buf, len, " age=%ld/%ld/%ld", > - l->min_time, > - (long)div_u64(l->sum_time, > - l->count), > - l->max_time); > - else > - len += sysfs_emit_at(buf, len, " age=%ld", l->min_time); > - > - if (l->min_pid != l->max_pid) > - len += sysfs_emit_at(buf, len, " pid=%ld-%ld", > - l->min_pid, l->max_pid); > - else > - len += sysfs_emit_at(buf, len, " pid=%ld", > - l->min_pid); > - > - if (num_online_cpus() > 1 && > - !cpumask_empty(to_cpumask(l->cpus))) > - len += sysfs_emit_at(buf, len, " cpus=%*pbl", > - cpumask_pr_args(to_cpumask(l->cpus))); > - > - if (nr_online_nodes > 1 && !nodes_empty(l->nodes)) > - len += sysfs_emit_at(buf, len, " nodes=%*pbl", > - nodemask_pr_args(&l->nodes)); > - > - len += sysfs_emit_at(buf, len, "\n"); > - } > - > - free_loc_track(&t); > - if (!t.count) > - len += sysfs_emit_at(buf, len, "No data\n"); > - > - return len; > -} > +#endif /* CONFIG_DEBUG_FS */ > #endif /* CONFIG_SLUB_DEBUG */ > > #ifdef SLUB_RESILIENCY_TEST > @@ -5350,21 +5290,6 @@ static ssize_t validate_store(struct kmem_cache *s, > } > SLAB_ATTR(validate); > > -static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf) > -{ > - if (!(s->flags & SLAB_STORE_USER)) > - return -ENOSYS; > - return list_locations(s, buf, TRACK_ALLOC); > -} > -SLAB_ATTR_RO(alloc_calls); > - > -static ssize_t free_calls_show(struct kmem_cache *s, char *buf) > -{ > - if (!(s->flags & SLAB_STORE_USER)) > - return -ENOSYS; > - return list_locations(s, buf, TRACK_FREE); > -} > -SLAB_ATTR_RO(free_calls); > #endif /* CONFIG_SLUB_DEBUG */ > > #ifdef CONFIG_FAILSLAB > @@ -5528,8 +5453,6 @@ static struct attribute *slab_attrs[] = { > &poison_attr.attr, > &store_user_attr.attr, > &validate_attr.attr, > - &alloc_calls_attr.attr, > - &free_calls_attr.attr, > #endif > #ifdef CONFIG_ZONE_DMA > &cache_dma_attr.attr, > @@ -5818,6 +5741,203 @@ static int __init slab_sysfs_init(void) > __initcall(slab_sysfs_init); > #endif /* CONFIG_SYSFS */ > > +#if defined(CONFIG_SLUB_DEBUG) && defined(CONFIG_DEBUG_FS) > +static int slab_debugfs_show(struct seq_file *seq, void *v) > +{ > + > + struct location *l; > + unsigned int idx = *(unsigned int *)v; > + > + if (idx < t.count) { > + l = &t.loc[idx]; > + > + seq_printf(seq, "%7ld ", l->count); > + > + if (l->addr) > + seq_printf(seq, "%pS", (void *)l->addr); > + else > + seq_puts(seq, "<not-available>"); > + > + if (l->sum_time != l->min_time) { > + seq_printf(seq, " age=%ld/%ld/%ld", > + l->min_time, > + (long)div_u64(l->sum_time, l->count), > + l->max_time); > + } else > + seq_printf(seq, " age=%ld", > + l->min_time); > + > + if (l->min_pid != l->max_pid) > + seq_printf(seq, " pid=%ld-%ld", > + l->min_pid, l->max_pid); > + else > + seq_printf(seq, " pid=%ld", > + l->min_pid); > + > + if (num_online_cpus() > 1 && > + !cpumask_empty(to_cpumask(l->cpus))) > + seq_printf(seq, " cpus=%*pbl", > + cpumask_pr_args(to_cpumask(l->cpus))); > + > + if (nr_online_nodes > 1 && !nodes_empty(l->nodes)) > + seq_printf(seq, " nodes=%*pbl", > + nodemask_pr_args(&l->nodes)); > + > + seq_puts(seq, "\n"); > + } > + > + if (t.count == 0) > + seq_puts(seq, "No data\n"); > + > + return 0; > +} > + > +static void slab_debugfs_stop(struct seq_file *seq, void *v) > +{ > + if (!v && t.max) { > + free_loc_track(&t); > + t.max = 0; > + } > +} > + > +static void *slab_debugfs_next(struct seq_file *seq, void *v, loff_t *ppos) > +{ > + loff_t *spos = v; > + > + if (*ppos < t.count) { > + *spos = *spos + 1; > + *ppos = *spos; > + return spos; > + } > + > + return NULL; > +} > + > +static void *slab_debugfs_start(struct seq_file *seq, loff_t *ppos) > +{ > + struct kmem_cache_node *n; > + struct kmem_cache *s; > + enum track_item alloc; > + int node; > + loff_t *spos = kmalloc(sizeof(loff_t), GFP_KERNEL); > + > + s = seq->file->f_inode->i_private; > + > + if (!spos) > + return NULL; > + > + if (*ppos == 0) { > + > + t.count = 0; > + t.max = 0; > + t.loc = NULL; > + if (strcmp(seq->file->f_path.dentry->d_name.name, "alloc_traces") == 0) > + alloc = TRACK_ALLOC; > + else > + alloc = TRACK_FREE; > + > + if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location), > + GFP_KERNEL)) { > + seq_puts(seq, "Out of memory\n"); > + kfree(spos); > + return ERR_PTR(-ENOMEM); > + } > + /* Push back cpu slabs */ > + flush_all(s); > + > + for_each_kmem_cache_node(s, node, n) { > + unsigned long flags; > + struct page *page; > + > + if (!atomic_long_read(&n->nr_slabs)) > + continue; > + > + spin_lock_irqsave(&n->list_lock, flags); > + list_for_each_entry(page, &n->partial, slab_list) > + process_slab(&t, s, page, alloc); > + list_for_each_entry(page, &n->full, slab_list) > + process_slab(&t, s, page, alloc); > + spin_unlock_irqrestore(&n->list_lock, flags); > + } > + } > + > + if (*ppos < t.count) { > + *spos = *ppos; > + return spos; > + } > + > + kfree(spos); > + return NULL; > +} > + > +static const struct seq_operations slab_debugfs_sops = { > + .start = slab_debugfs_start, > + .next = slab_debugfs_next, > + .stop = slab_debugfs_stop, > + .show = slab_debugfs_show > +}; > +DEFINE_SEQ_ATTRIBUTE(slab_debugfs); > + > +static void debugfs_slab_add(struct kmem_cache *s) > +{ > + const char *name; > + struct dentry *slab_cache_dir; > + int unmergeable = slab_unmergeable(s); > + > + if (unlikely(!slab_debugfs_root)) > + return; > + > + if (!unmergeable && disable_higher_order_debug && > + (slub_debug & DEBUG_METADATA_FLAGS)) > + unmergeable = 1; > + > + if (unmergeable) { > + /* > + * Slabcache can never be merged so we can use the name proper. > + * This is typically the case for debug situations. In that > + * case we can catch duplicate names easily. > + */ > + slab_cache_dir = debugfs_lookup(s->name, slab_debugfs_root); > + debugfs_remove_recursive(slab_cache_dir); > + name = s->name; > + } else { > + /* > + * Create a unique name for the slab as a target > + * for the symlinks. > + */ > + name = create_unique_id(s); > + } > + > + slab_cache_dir = debugfs_create_dir(name, slab_debugfs_root); > + > + debugfs_create_file("alloc_traces", 0400, > + slab_cache_dir, s, &slab_debugfs_fops); > + > + debugfs_create_file("free_traces", 0400, > + slab_cache_dir, s, &slab_debugfs_fops); > +} > + > +void debugfs_slab_release(struct kmem_cache *s) > +{ > + debugfs_remove_recursive(debugfs_lookup(s->name, > + slab_debugfs_root)); > +} > + > +static int __init slab_debugfs_init(void) > +{ > + struct kmem_cache *s; > + > + slab_debugfs_root = debugfs_create_dir("slab", NULL); > + > + list_for_each_entry(s, &slab_caches, list) > + if (s->flags & SLAB_STORE_USER) > + debugfs_slab_add(s); > + > + return 0; > + > +} > +__initcall(slab_debugfs_init); > +#endif > /* > * The /proc/slabinfo ABI > */ >
Hi Faiyaz,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on linus/master]
[also build test ERROR on v5.13-rc4]
[cannot apply to hnaz-linux-mm/master next-20210601]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/0day-ci/linux/commits/Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
base: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git c2131f7e73c9e9365613e323d65c7b9e5b910f56
config: x86_64-randconfig-a001-20210601 (attached as .config)
compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project db26cd30b6dd65e88d786e97a1e453af5cd48966)
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# install x86_64 cross compiling tool for clang build
# apt-get install binutils-x86-64-linux-gnu
# https://github.com/0day-ci/linux/commit/9540acc4691d680b7124d8daa1a2eb98a97ee19a
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Faiyaz-Mohammed/mm-slub-move-sysfs-slab-alloc-free-interfaces-to-debugfs/20210601-180903
git checkout 9540acc4691d680b7124d8daa1a2eb98a97ee19a
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=x86_64
If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>
All errors (new ones prefixed by >>):
>> mm/slab_common.c:452:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror,-Wimplicit-function-declaration]
debugfs_slab_release(s);
^
mm/slab_common.c:473:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror,-Wimplicit-function-declaration]
debugfs_slab_release(s);
^
mm/slab_common.c:481:3: error: implicit declaration of function 'debugfs_slab_release' [-Werror,-Wimplicit-function-declaration]
debugfs_slab_release(s);
^
3 errors generated.
vim +/debugfs_slab_release +452 mm/slab_common.c
427
428 static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work)
429 {
430 LIST_HEAD(to_destroy);
431 struct kmem_cache *s, *s2;
432
433 /*
434 * On destruction, SLAB_TYPESAFE_BY_RCU kmem_caches are put on the
435 * @slab_caches_to_rcu_destroy list. The slab pages are freed
436 * through RCU and the associated kmem_cache are dereferenced
437 * while freeing the pages, so the kmem_caches should be freed only
438 * after the pending RCU operations are finished. As rcu_barrier()
439 * is a pretty slow operation, we batch all pending destructions
440 * asynchronously.
441 */
442 mutex_lock(&slab_mutex);
443 list_splice_init(&slab_caches_to_rcu_destroy, &to_destroy);
444 mutex_unlock(&slab_mutex);
445
446 if (list_empty(&to_destroy))
447 return;
448
449 rcu_barrier();
450
451 list_for_each_entry_safe(s, s2, &to_destroy, list) {
> 452 debugfs_slab_release(s);
453 kfence_shutdown_cache(s);
454 #ifdef SLAB_SUPPORTS_SYSFS
455 sysfs_slab_release(s);
456 #else
457 slab_kmem_cache_release(s);
458 #endif
459 }
460 }
461
---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
On 6/1/2021 7:01 PM, Vlastimil Babka wrote: > On 6/1/21 12:07 PM, Faiyaz Mohammed wrote: >> alloc_calls and free_calls implementation in sysfs have two issues, >> one is PAGE_SIZE limitiation of sysfs and other is it does not adhere >> to "one value per file" rule. >> >> To overcome this issues, move the alloc_calls and free_calls implemeation >> to debugfs. >> >> Debugfs cache will be created if SLAB_STORE_USER flag is set. >> >> Rename the alloc_calls/free_calls to alloc_traces/free_traces, >> to be inline with what it does. >> >> Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org> > > When reading the files I'm getting dmesg errors like this: > > [ 37.998152] seq_file: buggy .next function slab_debugfs_next did not update > position index > > Don't recall this from previous versions? > slab_debugfs_next() is not updating the pos after last read, so it's retuning null. In fs/seq_file.c if pos is not updated in ->next() fn, it's printing the above info but I will address this in next patch version. >> diff --git a/mm/slab_common.c b/mm/slab_common.c >> index a4a5714..ca39e84 100644 >> --- a/mm/slab_common.c >> +++ b/mm/slab_common.c >> @@ -449,6 +449,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work) >> rcu_barrier(); >> >> list_for_each_entry_safe(s, s2, &to_destroy, list) { >> + debugfs_slab_release(s); > > You should remove this one, you already do the release when queueing the work > below. This is not like sysfs where we distinguish _unlink() when queueing and > _release when processing the work. > oh, yes, I will update in next patch version. >> +static struct dentry *slab_debugfs_root; >> +struct loc_track t = { 0, 0, NULL }; > > Ooops, this can't be a global variable, especially when shared without any lock > protection - but even with a protection it would be very suboptimal. You'll need > to make it a seq_file private, see e.g. here for an example: > https://lore.kernel.org/linux-mm/20210521121127.24653-3-glittao@gmail.com/ > I will address this in next patch version. >> static void free_loc_track(struct loc_track *t) >> { >> if (t->max) >> @@ -4825,82 +4840,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s, >> add_location(t, s, get_track(s, p, alloc)); >> put_map(map); >> } >> - >> -static int list_locations(struct kmem_cache *s, char *buf, >> - enum track_item alloc) >> -{ >> - int len = 0; >> - unsigned long i; >> - struct loc_track t = { 0, 0, NULL }; >> - int node; >> - struct kmem_cache_node *n; >> - >> - if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location), >> - GFP_KERNEL)) { >> - return sysfs_emit(buf, "Out of memory\n"); >> - } >> - /* Push back cpu slabs */ >> - flush_all(s); >> - >> - for_each_kmem_cache_node(s, node, n) { >> - unsigned long flags; >> - struct page *page; >> - >> - if (!atomic_long_read(&n->nr_slabs)) >> - continue; >> - >> - spin_lock_irqsave(&n->list_lock, flags); >> - list_for_each_entry(page, &n->partial, slab_list) >> - process_slab(&t, s, page, alloc); >> - list_for_each_entry(page, &n->full, slab_list) >> - process_slab(&t, s, page, alloc); >> - spin_unlock_irqrestore(&n->list_lock, flags); >> - } >> - >> - for (i = 0; i < t.count; i++) { >> - struct location *l = &t.loc[i]; >> - >> - len += sysfs_emit_at(buf, len, "%7ld ", l->count); >> - >> - if (l->addr) >> - len += sysfs_emit_at(buf, len, "%pS", (void *)l->addr); >> - else >> - len += sysfs_emit_at(buf, len, "<not-available>"); >> - >> - if (l->sum_time != l->min_time) >> - len += sysfs_emit_at(buf, len, " age=%ld/%ld/%ld", >> - l->min_time, >> - (long)div_u64(l->sum_time, >> - l->count), >> - l->max_time); >> - else >> - len += sysfs_emit_at(buf, len, " age=%ld", l->min_time); >> - >> - if (l->min_pid != l->max_pid) >> - len += sysfs_emit_at(buf, len, " pid=%ld-%ld", >> - l->min_pid, l->max_pid); >> - else >> - len += sysfs_emit_at(buf, len, " pid=%ld", >> - l->min_pid); >> - >> - if (num_online_cpus() > 1 && >> - !cpumask_empty(to_cpumask(l->cpus))) >> - len += sysfs_emit_at(buf, len, " cpus=%*pbl", >> - cpumask_pr_args(to_cpumask(l->cpus))); >> - >> - if (nr_online_nodes > 1 && !nodes_empty(l->nodes)) >> - len += sysfs_emit_at(buf, len, " nodes=%*pbl", >> - nodemask_pr_args(&l->nodes)); >> - >> - len += sysfs_emit_at(buf, len, "\n"); >> - } >> - >> - free_loc_track(&t); >> - if (!t.count) >> - len += sysfs_emit_at(buf, len, "No data\n"); >> - >> - return len; >> -} >> +#endif /* CONFIG_DEBUG_FS */ >> #endif /* CONFIG_SLUB_DEBUG */ >> >> #ifdef SLUB_RESILIENCY_TEST >> @@ -5350,21 +5290,6 @@ static ssize_t validate_store(struct kmem_cache *s, >> } >> SLAB_ATTR(validate); >> >> -static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf) >> -{ >> - if (!(s->flags & SLAB_STORE_USER)) >> - return -ENOSYS; >> - return list_locations(s, buf, TRACK_ALLOC); >> -} >> -SLAB_ATTR_RO(alloc_calls); >> - >> -static ssize_t free_calls_show(struct kmem_cache *s, char *buf) >> -{ >> - if (!(s->flags & SLAB_STORE_USER)) >> - return -ENOSYS; >> - return list_locations(s, buf, TRACK_FREE); >> -} >> -SLAB_ATTR_RO(free_calls); >> #endif /* CONFIG_SLUB_DEBUG */ >> >> #ifdef CONFIG_FAILSLAB >> @@ -5528,8 +5453,6 @@ static struct attribute *slab_attrs[] = { >> &poison_attr.attr, >> &store_user_attr.attr, >> &validate_attr.attr, >> - &alloc_calls_attr.attr, >> - &free_calls_attr.attr, >> #endif >> #ifdef CONFIG_ZONE_DMA >> &cache_dma_attr.attr, >> @@ -5818,6 +5741,203 @@ static int __init slab_sysfs_init(void) >> __initcall(slab_sysfs_init); >> #endif /* CONFIG_SYSFS */ >> >> +#if defined(CONFIG_SLUB_DEBUG) && defined(CONFIG_DEBUG_FS) >> +static int slab_debugfs_show(struct seq_file *seq, void *v) >> +{ >> + >> + struct location *l; >> + unsigned int idx = *(unsigned int *)v; >> + >> + if (idx < t.count) { >> + l = &t.loc[idx]; >> + >> + seq_printf(seq, "%7ld ", l->count); >> + >> + if (l->addr) >> + seq_printf(seq, "%pS", (void *)l->addr); >> + else >> + seq_puts(seq, "<not-available>"); >> + >> + if (l->sum_time != l->min_time) { >> + seq_printf(seq, " age=%ld/%ld/%ld", >> + l->min_time, >> + (long)div_u64(l->sum_time, l->count), >> + l->max_time); >> + } else >> + seq_printf(seq, " age=%ld", >> + l->min_time); >> + >> + if (l->min_pid != l->max_pid) >> + seq_printf(seq, " pid=%ld-%ld", >> + l->min_pid, l->max_pid); >> + else >> + seq_printf(seq, " pid=%ld", >> + l->min_pid); >> + >> + if (num_online_cpus() > 1 && >> + !cpumask_empty(to_cpumask(l->cpus))) >> + seq_printf(seq, " cpus=%*pbl", >> + cpumask_pr_args(to_cpumask(l->cpus))); >> + >> + if (nr_online_nodes > 1 && !nodes_empty(l->nodes)) >> + seq_printf(seq, " nodes=%*pbl", >> + nodemask_pr_args(&l->nodes)); >> + >> + seq_puts(seq, "\n"); >> + } >> + >> + if (t.count == 0) >> + seq_puts(seq, "No data\n"); >> + >> + return 0; >> +} >> + >> +static void slab_debugfs_stop(struct seq_file *seq, void *v) >> +{ >> + if (!v && t.max) { >> + free_loc_track(&t); >> + t.max = 0; >> + } >> +} >> + >> +static void *slab_debugfs_next(struct seq_file *seq, void *v, loff_t *ppos) >> +{ >> + loff_t *spos = v; >> + >> + if (*ppos < t.count) { >> + *spos = *spos + 1; >> + *ppos = *spos; >> + return spos; >> + } >> + >> + return NULL; >> +} >> + >> +static void *slab_debugfs_start(struct seq_file *seq, loff_t *ppos) >> +{ >> + struct kmem_cache_node *n; >> + struct kmem_cache *s; >> + enum track_item alloc; >> + int node; >> + loff_t *spos = kmalloc(sizeof(loff_t), GFP_KERNEL); >> + >> + s = seq->file->f_inode->i_private; >> + >> + if (!spos) >> + return NULL; >> + >> + if (*ppos == 0) { >> + >> + t.count = 0; >> + t.max = 0; >> + t.loc = NULL; >> + if (strcmp(seq->file->f_path.dentry->d_name.name, "alloc_traces") == 0) >> + alloc = TRACK_ALLOC; >> + else >> + alloc = TRACK_FREE; >> + >> + if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location), >> + GFP_KERNEL)) { >> + seq_puts(seq, "Out of memory\n"); >> + kfree(spos); >> + return ERR_PTR(-ENOMEM); >> + } >> + /* Push back cpu slabs */ >> + flush_all(s); >> + >> + for_each_kmem_cache_node(s, node, n) { >> + unsigned long flags; >> + struct page *page; >> + >> + if (!atomic_long_read(&n->nr_slabs)) >> + continue; >> + >> + spin_lock_irqsave(&n->list_lock, flags); >> + list_for_each_entry(page, &n->partial, slab_list) >> + process_slab(&t, s, page, alloc); >> + list_for_each_entry(page, &n->full, slab_list) >> + process_slab(&t, s, page, alloc); >> + spin_unlock_irqrestore(&n->list_lock, flags); >> + } >> + } >> + >> + if (*ppos < t.count) { >> + *spos = *ppos; >> + return spos; >> + } >> + >> + kfree(spos); >> + return NULL; >> +} >> + >> +static const struct seq_operations slab_debugfs_sops = { >> + .start = slab_debugfs_start, >> + .next = slab_debugfs_next, >> + .stop = slab_debugfs_stop, >> + .show = slab_debugfs_show >> +}; >> +DEFINE_SEQ_ATTRIBUTE(slab_debugfs); >> + >> +static void debugfs_slab_add(struct kmem_cache *s) >> +{ >> + const char *name; >> + struct dentry *slab_cache_dir; >> + int unmergeable = slab_unmergeable(s); >> + >> + if (unlikely(!slab_debugfs_root)) >> + return; >> + >> + if (!unmergeable && disable_higher_order_debug && >> + (slub_debug & DEBUG_METADATA_FLAGS)) >> + unmergeable = 1; >> + >> + if (unmergeable) { >> + /* >> + * Slabcache can never be merged so we can use the name proper. >> + * This is typically the case for debug situations. In that >> + * case we can catch duplicate names easily. >> + */ >> + slab_cache_dir = debugfs_lookup(s->name, slab_debugfs_root); >> + debugfs_remove_recursive(slab_cache_dir); >> + name = s->name; >> + } else { >> + /* >> + * Create a unique name for the slab as a target >> + * for the symlinks. >> + */ >> + name = create_unique_id(s); >> + } >> + >> + slab_cache_dir = debugfs_create_dir(name, slab_debugfs_root); >> + >> + debugfs_create_file("alloc_traces", 0400, >> + slab_cache_dir, s, &slab_debugfs_fops); >> + >> + debugfs_create_file("free_traces", 0400, >> + slab_cache_dir, s, &slab_debugfs_fops); >> +} >> + >> +void debugfs_slab_release(struct kmem_cache *s) >> +{ >> + debugfs_remove_recursive(debugfs_lookup(s->name, >> + slab_debugfs_root)); >> +} >> + >> +static int __init slab_debugfs_init(void) >> +{ >> + struct kmem_cache *s; >> + >> + slab_debugfs_root = debugfs_create_dir("slab", NULL); >> + >> + list_for_each_entry(s, &slab_caches, list) >> + if (s->flags & SLAB_STORE_USER) >> + debugfs_slab_add(s); >> + >> + return 0; >> + >> +} >> +__initcall(slab_debugfs_init); >> +#endif >> /* >> * The /proc/slabinfo ABI >> */ >> >
diff --git a/include/linux/slab_def.h b/include/linux/slab_def.h index 3aa5e1e..02ca311 100644 --- a/include/linux/slab_def.h +++ b/include/linux/slab_def.h @@ -120,4 +120,6 @@ static inline int objs_per_slab_page(const struct kmem_cache *cache, return cache->num; } +static inline void debugfs_slab_release(struct kmem_cache *s) { } + #endif /* _LINUX_SLAB_DEF_H */ diff --git a/include/linux/slub_def.h b/include/linux/slub_def.h index dcde82a..90d659a 100644 --- a/include/linux/slub_def.h +++ b/include/linux/slub_def.h @@ -159,6 +159,14 @@ static inline void sysfs_slab_release(struct kmem_cache *s) } #endif +#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_SLUB_DEBUG) +void debugfs_slab_release(struct kmem_cache *); +#else +static inline void debugfs_slab_release(struct kmem_cache *s) +{ +} +#endif + void object_err(struct kmem_cache *s, struct page *page, u8 *object, char *reason); diff --git a/mm/slab_common.c b/mm/slab_common.c index a4a5714..ca39e84 100644 --- a/mm/slab_common.c +++ b/mm/slab_common.c @@ -449,6 +449,7 @@ static void slab_caches_to_rcu_destroy_workfn(struct work_struct *work) rcu_barrier(); list_for_each_entry_safe(s, s2, &to_destroy, list) { + debugfs_slab_release(s); kfence_shutdown_cache(s); #ifdef SLAB_SUPPORTS_SYSFS sysfs_slab_release(s); @@ -469,6 +470,7 @@ static int shutdown_cache(struct kmem_cache *s) list_del(&s->list); if (s->flags & SLAB_TYPESAFE_BY_RCU) { + debugfs_slab_release(s); #ifdef SLAB_SUPPORTS_SYSFS sysfs_slab_unlink(s); #endif @@ -476,6 +478,7 @@ static int shutdown_cache(struct kmem_cache *s) schedule_work(&slab_caches_to_rcu_destroy_work); } else { kfence_shutdown_cache(s); + debugfs_slab_release(s); #ifdef SLAB_SUPPORTS_SYSFS sysfs_slab_unlink(s); sysfs_slab_release(s); diff --git a/mm/slub.c b/mm/slub.c index 3f96e09..f7735c1 100644 --- a/mm/slub.c +++ b/mm/slub.c @@ -36,6 +36,7 @@ #include <linux/memcontrol.h> #include <linux/random.h> +#include <linux/debugfs.h> #include <trace/events/kmem.h> #include "internal.h" @@ -225,6 +226,12 @@ static inline int sysfs_slab_alias(struct kmem_cache *s, const char *p) { return 0; } #endif +#if defined(CONFIG_DEBUG_FS) && defined(CONFIG_SLUB_DEBUG) +static void debugfs_slab_add(struct kmem_cache *); +#else +static inline void debugfs_slab_add(struct kmem_cache *s) { } +#endif + static inline void stat(const struct kmem_cache *s, enum stat_item si) { #ifdef CONFIG_SLUB_STATS @@ -4546,6 +4553,9 @@ int __kmem_cache_create(struct kmem_cache *s, slab_flags_t flags) if (err) __kmem_cache_release(s); + if (s->flags & SLAB_STORE_USER) + debugfs_slab_add(s); + return err; } @@ -4686,6 +4696,8 @@ static long validate_slab_cache(struct kmem_cache *s) return count; } + +#ifdef CONFIG_DEBUG_FS /* * Generate lists of code addresses where slabcache objects are allocated * and freed. @@ -4709,6 +4721,9 @@ struct loc_track { struct location *loc; }; +static struct dentry *slab_debugfs_root; +struct loc_track t = { 0, 0, NULL }; + static void free_loc_track(struct loc_track *t) { if (t->max) @@ -4825,82 +4840,7 @@ static void process_slab(struct loc_track *t, struct kmem_cache *s, add_location(t, s, get_track(s, p, alloc)); put_map(map); } - -static int list_locations(struct kmem_cache *s, char *buf, - enum track_item alloc) -{ - int len = 0; - unsigned long i; - struct loc_track t = { 0, 0, NULL }; - int node; - struct kmem_cache_node *n; - - if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location), - GFP_KERNEL)) { - return sysfs_emit(buf, "Out of memory\n"); - } - /* Push back cpu slabs */ - flush_all(s); - - for_each_kmem_cache_node(s, node, n) { - unsigned long flags; - struct page *page; - - if (!atomic_long_read(&n->nr_slabs)) - continue; - - spin_lock_irqsave(&n->list_lock, flags); - list_for_each_entry(page, &n->partial, slab_list) - process_slab(&t, s, page, alloc); - list_for_each_entry(page, &n->full, slab_list) - process_slab(&t, s, page, alloc); - spin_unlock_irqrestore(&n->list_lock, flags); - } - - for (i = 0; i < t.count; i++) { - struct location *l = &t.loc[i]; - - len += sysfs_emit_at(buf, len, "%7ld ", l->count); - - if (l->addr) - len += sysfs_emit_at(buf, len, "%pS", (void *)l->addr); - else - len += sysfs_emit_at(buf, len, "<not-available>"); - - if (l->sum_time != l->min_time) - len += sysfs_emit_at(buf, len, " age=%ld/%ld/%ld", - l->min_time, - (long)div_u64(l->sum_time, - l->count), - l->max_time); - else - len += sysfs_emit_at(buf, len, " age=%ld", l->min_time); - - if (l->min_pid != l->max_pid) - len += sysfs_emit_at(buf, len, " pid=%ld-%ld", - l->min_pid, l->max_pid); - else - len += sysfs_emit_at(buf, len, " pid=%ld", - l->min_pid); - - if (num_online_cpus() > 1 && - !cpumask_empty(to_cpumask(l->cpus))) - len += sysfs_emit_at(buf, len, " cpus=%*pbl", - cpumask_pr_args(to_cpumask(l->cpus))); - - if (nr_online_nodes > 1 && !nodes_empty(l->nodes)) - len += sysfs_emit_at(buf, len, " nodes=%*pbl", - nodemask_pr_args(&l->nodes)); - - len += sysfs_emit_at(buf, len, "\n"); - } - - free_loc_track(&t); - if (!t.count) - len += sysfs_emit_at(buf, len, "No data\n"); - - return len; -} +#endif /* CONFIG_DEBUG_FS */ #endif /* CONFIG_SLUB_DEBUG */ #ifdef SLUB_RESILIENCY_TEST @@ -5350,21 +5290,6 @@ static ssize_t validate_store(struct kmem_cache *s, } SLAB_ATTR(validate); -static ssize_t alloc_calls_show(struct kmem_cache *s, char *buf) -{ - if (!(s->flags & SLAB_STORE_USER)) - return -ENOSYS; - return list_locations(s, buf, TRACK_ALLOC); -} -SLAB_ATTR_RO(alloc_calls); - -static ssize_t free_calls_show(struct kmem_cache *s, char *buf) -{ - if (!(s->flags & SLAB_STORE_USER)) - return -ENOSYS; - return list_locations(s, buf, TRACK_FREE); -} -SLAB_ATTR_RO(free_calls); #endif /* CONFIG_SLUB_DEBUG */ #ifdef CONFIG_FAILSLAB @@ -5528,8 +5453,6 @@ static struct attribute *slab_attrs[] = { &poison_attr.attr, &store_user_attr.attr, &validate_attr.attr, - &alloc_calls_attr.attr, - &free_calls_attr.attr, #endif #ifdef CONFIG_ZONE_DMA &cache_dma_attr.attr, @@ -5818,6 +5741,203 @@ static int __init slab_sysfs_init(void) __initcall(slab_sysfs_init); #endif /* CONFIG_SYSFS */ +#if defined(CONFIG_SLUB_DEBUG) && defined(CONFIG_DEBUG_FS) +static int slab_debugfs_show(struct seq_file *seq, void *v) +{ + + struct location *l; + unsigned int idx = *(unsigned int *)v; + + if (idx < t.count) { + l = &t.loc[idx]; + + seq_printf(seq, "%7ld ", l->count); + + if (l->addr) + seq_printf(seq, "%pS", (void *)l->addr); + else + seq_puts(seq, "<not-available>"); + + if (l->sum_time != l->min_time) { + seq_printf(seq, " age=%ld/%ld/%ld", + l->min_time, + (long)div_u64(l->sum_time, l->count), + l->max_time); + } else + seq_printf(seq, " age=%ld", + l->min_time); + + if (l->min_pid != l->max_pid) + seq_printf(seq, " pid=%ld-%ld", + l->min_pid, l->max_pid); + else + seq_printf(seq, " pid=%ld", + l->min_pid); + + if (num_online_cpus() > 1 && + !cpumask_empty(to_cpumask(l->cpus))) + seq_printf(seq, " cpus=%*pbl", + cpumask_pr_args(to_cpumask(l->cpus))); + + if (nr_online_nodes > 1 && !nodes_empty(l->nodes)) + seq_printf(seq, " nodes=%*pbl", + nodemask_pr_args(&l->nodes)); + + seq_puts(seq, "\n"); + } + + if (t.count == 0) + seq_puts(seq, "No data\n"); + + return 0; +} + +static void slab_debugfs_stop(struct seq_file *seq, void *v) +{ + if (!v && t.max) { + free_loc_track(&t); + t.max = 0; + } +} + +static void *slab_debugfs_next(struct seq_file *seq, void *v, loff_t *ppos) +{ + loff_t *spos = v; + + if (*ppos < t.count) { + *spos = *spos + 1; + *ppos = *spos; + return spos; + } + + return NULL; +} + +static void *slab_debugfs_start(struct seq_file *seq, loff_t *ppos) +{ + struct kmem_cache_node *n; + struct kmem_cache *s; + enum track_item alloc; + int node; + loff_t *spos = kmalloc(sizeof(loff_t), GFP_KERNEL); + + s = seq->file->f_inode->i_private; + + if (!spos) + return NULL; + + if (*ppos == 0) { + + t.count = 0; + t.max = 0; + t.loc = NULL; + if (strcmp(seq->file->f_path.dentry->d_name.name, "alloc_traces") == 0) + alloc = TRACK_ALLOC; + else + alloc = TRACK_FREE; + + if (!alloc_loc_track(&t, PAGE_SIZE / sizeof(struct location), + GFP_KERNEL)) { + seq_puts(seq, "Out of memory\n"); + kfree(spos); + return ERR_PTR(-ENOMEM); + } + /* Push back cpu slabs */ + flush_all(s); + + for_each_kmem_cache_node(s, node, n) { + unsigned long flags; + struct page *page; + + if (!atomic_long_read(&n->nr_slabs)) + continue; + + spin_lock_irqsave(&n->list_lock, flags); + list_for_each_entry(page, &n->partial, slab_list) + process_slab(&t, s, page, alloc); + list_for_each_entry(page, &n->full, slab_list) + process_slab(&t, s, page, alloc); + spin_unlock_irqrestore(&n->list_lock, flags); + } + } + + if (*ppos < t.count) { + *spos = *ppos; + return spos; + } + + kfree(spos); + return NULL; +} + +static const struct seq_operations slab_debugfs_sops = { + .start = slab_debugfs_start, + .next = slab_debugfs_next, + .stop = slab_debugfs_stop, + .show = slab_debugfs_show +}; +DEFINE_SEQ_ATTRIBUTE(slab_debugfs); + +static void debugfs_slab_add(struct kmem_cache *s) +{ + const char *name; + struct dentry *slab_cache_dir; + int unmergeable = slab_unmergeable(s); + + if (unlikely(!slab_debugfs_root)) + return; + + if (!unmergeable && disable_higher_order_debug && + (slub_debug & DEBUG_METADATA_FLAGS)) + unmergeable = 1; + + if (unmergeable) { + /* + * Slabcache can never be merged so we can use the name proper. + * This is typically the case for debug situations. In that + * case we can catch duplicate names easily. + */ + slab_cache_dir = debugfs_lookup(s->name, slab_debugfs_root); + debugfs_remove_recursive(slab_cache_dir); + name = s->name; + } else { + /* + * Create a unique name for the slab as a target + * for the symlinks. + */ + name = create_unique_id(s); + } + + slab_cache_dir = debugfs_create_dir(name, slab_debugfs_root); + + debugfs_create_file("alloc_traces", 0400, + slab_cache_dir, s, &slab_debugfs_fops); + + debugfs_create_file("free_traces", 0400, + slab_cache_dir, s, &slab_debugfs_fops); +} + +void debugfs_slab_release(struct kmem_cache *s) +{ + debugfs_remove_recursive(debugfs_lookup(s->name, + slab_debugfs_root)); +} + +static int __init slab_debugfs_init(void) +{ + struct kmem_cache *s; + + slab_debugfs_root = debugfs_create_dir("slab", NULL); + + list_for_each_entry(s, &slab_caches, list) + if (s->flags & SLAB_STORE_USER) + debugfs_slab_add(s); + + return 0; + +} +__initcall(slab_debugfs_init); +#endif /* * The /proc/slabinfo ABI */
alloc_calls and free_calls implementation in sysfs have two issues, one is PAGE_SIZE limitiation of sysfs and other is it does not adhere to "one value per file" rule. To overcome this issues, move the alloc_calls and free_calls implemeation to debugfs. Debugfs cache will be created if SLAB_STORE_USER flag is set. Rename the alloc_calls/free_calls to alloc_traces/free_traces, to be inline with what it does. Signed-off-by: Faiyaz Mohammed <faiyazm@codeaurora.org> --- changes in v8: - Debugfs cache file will not be created if SLAB_STORE_USER flag is not set. - Make sure the debugfs slab release happen before slab kmem cahce release. changes in V7: - https://lore.kernel.org/linux-mm/1621928285-751-1-git-send-email-faiyazm@codeaurora.org/ changes in v6: - https://lore.kernel.org/linux-mm/1621341949-26762-1-git-send-email-faiyazm@codeaurora.org/ changes in v5: - https://lore.kernel.org/linux-mm/1620296523-21922-1-git-send-email-faiyazm@codeaurora.org/ changes in v4: - https://lore.kernel.org/linux-mm/1618583239-18124-1-git-send-email-faiyazm@codeaurora.org/ changes in v3: - https://lore.kernel.org/linux-mm/1617712064-12264-1-git-send-email-faiyazm@codeaurora.org/ changes in v2: - https://lore.kernel.org/linux-mm/3ac1d3e6-6207-96ad-16a1-0f5139d8b2b5@codeaurora.org/ changes in v1 - https://lore.kernel.org/linux-mm/1610443287-23933-1-git-send-email-faiyazm@codeaurora.org/ include/linux/slab_def.h | 2 + include/linux/slub_def.h | 8 ++ mm/slab_common.c | 3 + mm/slub.c | 306 +++++++++++++++++++++++++++++++++-------------- 4 files changed, 226 insertions(+), 93 deletions(-)