| Message ID | 20230822-strncpy-arch-x86-platform-uv-uv_nmi-v1-1-931f2943de0d@google.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | x86/platform/uv: refactor deprecated strcpy and strncpy | expand |
On Wed, Aug 23, 2023 at 1:32 AM Justin Stitt <justinstitt@google.com> wrote: > > Both `strncpy` and `strcpy` are deprecated for use on NUL-terminated > destination strings [1]. > > A suitable replacement is `strscpy` [2] due to the fact that it > guarantees NUL-termination on its destination buffer argument which is > _not_ the case for `strncpy` or `strcpy`! > > In this case, we can drop both the forced NUL-termination and the `... -1` from: > | strncpy(arg, val, ACTION_LEN - 1); > as `strscpy` implicitly has this behavior. ... > char arg[ACTION_LEN], *p; > > /* (remove possible '\n') */ > - strncpy(arg, val, ACTION_LEN - 1); > - arg[ACTION_LEN - 1] = '\0'; > + strscpy(arg, val, ACTION_LEN); > p = strchr(arg, '\n'); > if (p) > *p = '\0'; https://lore.kernel.org/all/202212091545310085328@zte.com.cn/ ... > + strscpy(uv_nmi_action, arg, strlen(uv_nmi_action)); strlen() on the destination?! ... > - strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > + strscpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); Again, this is weird.
On Wed, Aug 23, 2023 at 4:07 AM Andy Shevchenko <andy.shevchenko@gmail.com> wrote: > > On Wed, Aug 23, 2023 at 1:32 AM Justin Stitt <justinstitt@google.com> wrote: > > > > Both `strncpy` and `strcpy` are deprecated for use on NUL-terminated > > destination strings [1]. > > > > A suitable replacement is `strscpy` [2] due to the fact that it > > guarantees NUL-termination on its destination buffer argument which is > > _not_ the case for `strncpy` or `strcpy`! > > > > In this case, we can drop both the forced NUL-termination and the `... -1` from: > > | strncpy(arg, val, ACTION_LEN - 1); > > as `strscpy` implicitly has this behavior. > > ... > > > char arg[ACTION_LEN], *p; > > > > /* (remove possible '\n') */ > > - strncpy(arg, val, ACTION_LEN - 1); > > - arg[ACTION_LEN - 1] = '\0'; > > + strscpy(arg, val, ACTION_LEN); > > p = strchr(arg, '\n'); > > if (p) > > *p = '\0'; > > https://lore.kernel.org/all/202212091545310085328@zte.com.cn/ > > ... > > > + strscpy(uv_nmi_action, arg, strlen(uv_nmi_action)); > > strlen() on the destination?! > > ... > > > - strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > + strscpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > Again, this is weird. This is a common pattern with `strxcpy` and `sizeof` if you `$ rg "strncpy\(.*sizeof"`. Do you recommend I switch the strlen(dest) to strlen(src)? I only kept as-is because that's what was there originally and I assumed some greater purpose of it. > > -- > With Best Regards, > Andy Shevchenko
On Wed, Aug 23, 2023 at 03:49:34PM -0700, Justin Stitt wrote: > On Wed, Aug 23, 2023 at 4:07 AM Andy Shevchenko > <andy.shevchenko@gmail.com> wrote: > > > > On Wed, Aug 23, 2023 at 1:32 AM Justin Stitt <justinstitt@google.com> wrote: > > > > > > Both `strncpy` and `strcpy` are deprecated for use on NUL-terminated > > > destination strings [1]. > > > > > > A suitable replacement is `strscpy` [2] due to the fact that it > > > guarantees NUL-termination on its destination buffer argument which is > > > _not_ the case for `strncpy` or `strcpy`! > > > > > > In this case, we can drop both the forced NUL-termination and the `... -1` from: > > > | strncpy(arg, val, ACTION_LEN - 1); > > > as `strscpy` implicitly has this behavior. > > > > ... > > > > > char arg[ACTION_LEN], *p; > > > > > > /* (remove possible '\n') */ > > > - strncpy(arg, val, ACTION_LEN - 1); > > > - arg[ACTION_LEN - 1] = '\0'; > > > + strscpy(arg, val, ACTION_LEN); > > > p = strchr(arg, '\n'); > > > if (p) > > > *p = '\0'; > > > > https://lore.kernel.org/all/202212091545310085328@zte.com.cn/ > > > > ... > > > > > + strscpy(uv_nmi_action, arg, strlen(uv_nmi_action)); > > > > strlen() on the destination?! > > > > ... > > > > > - strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > > + strscpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > > > Again, this is weird. > > This is a common pattern with `strxcpy` and `sizeof` if you `$ rg > "strncpy\(.*sizeof"`. Do you recommend I switch the strlen(dest) to > strlen(src)? I only kept as-is because that's what was there > originally and I assumed some greater purpose of it. It's best to avoid any assumptions. If it can't be answered through code inspection, the next best thing would be to ask for clarification. In looking I see uv_nmi_action is a string: arch/x86/platform/uv/uv_nmi.c:193:typedef char action_t[ACTION_LEN]; arch/x86/platform/uv/uv_nmi.c: strcpy(uv_nmi_action, arg); arch/x86/platform/uv/uv_nmi.c:module_param_named(action, uv_nmi_action, action, 0644); arch/x86/platform/uv/uv_nmi.c: return (strncmp(uv_nmi_action, action, strlen(action)) == 0); arch/x86/platform/uv/uv_nmi.c: strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); using strlen() here seems "accidentally safe", as it's overwriting "kdump": if (uv_nmi_action_is("kdump")) { uv_nmi_kdump(cpu, master, regs); /* Unexpected return, revert action to "dump" */ if (master) strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); anyway, a simple "sizeof" should be used AFAICT.
On Wed, Aug 23, 2023 at 04:00:13PM -0700, Kees Cook wrote: > On Wed, Aug 23, 2023 at 03:49:34PM -0700, Justin Stitt wrote: > > On Wed, Aug 23, 2023 at 4:07 AM Andy Shevchenko > > <andy.shevchenko@gmail.com> wrote: > > > > > > On Wed, Aug 23, 2023 at 1:32 AM Justin Stitt <justinstitt@google.com> wrote: > > > > > > > > Both `strncpy` and `strcpy` are deprecated for use on NUL-terminated > > > > destination strings [1]. > > > > > > > > A suitable replacement is `strscpy` [2] due to the fact that it > > > > guarantees NUL-termination on its destination buffer argument which is > > > > _not_ the case for `strncpy` or `strcpy`! > > > > > > > > In this case, we can drop both the forced NUL-termination and the `... -1` from: > > > > | strncpy(arg, val, ACTION_LEN - 1); > > > > as `strscpy` implicitly has this behavior. > > > > > > ... > > > > > > > char arg[ACTION_LEN], *p; > > > > > > > > /* (remove possible '\n') */ > > > > - strncpy(arg, val, ACTION_LEN - 1); > > > > - arg[ACTION_LEN - 1] = '\0'; > > > > + strscpy(arg, val, ACTION_LEN); > > > > p = strchr(arg, '\n'); > > > > if (p) > > > > *p = '\0'; > > > > > > https://lore.kernel.org/all/202212091545310085328@zte.com.cn/ > > > > > > ... > > > > > > > + strscpy(uv_nmi_action, arg, strlen(uv_nmi_action)); > > > > > > strlen() on the destination?! The original code for the above (strcpy()), copies strlen(arg) assuming null termination, so strlen(uv_nmi_action) is not correct for this case. You probably want to use sizeof of the destination. > > > > > > ... > > > > > > > - strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > > > + strscpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > > > > > Again, this is weird. > > > > This is a common pattern with `strxcpy` and `sizeof` if you `$ rg > > "strncpy\(.*sizeof"`. Do you recommend I switch the strlen(dest) to > > strlen(src)? I only kept as-is because that's what was there > > originally and I assumed some greater purpose of it. > > It's best to avoid any assumptions. If it can't be answered through code > inspection, the next best thing would be to ask for clarification. In > looking I see uv_nmi_action is a string: > > arch/x86/platform/uv/uv_nmi.c:193:typedef char action_t[ACTION_LEN]; > > arch/x86/platform/uv/uv_nmi.c: strcpy(uv_nmi_action, arg); > arch/x86/platform/uv/uv_nmi.c:module_param_named(action, uv_nmi_action, action, 0644); > arch/x86/platform/uv/uv_nmi.c: return (strncmp(uv_nmi_action, action, strlen(action)) == 0); > arch/x86/platform/uv/uv_nmi.c: strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > using strlen() here seems "accidentally safe", as it's overwriting > "kdump": > > if (uv_nmi_action_is("kdump")) { > uv_nmi_kdump(cpu, master, regs); > > /* Unexpected return, revert action to "dump" */ > if (master) > strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); > > anyway, a simple "sizeof" should be used AFAICT. > I agree.
diff --git a/arch/x86/platform/uv/uv_nmi.c b/arch/x86/platform/uv/uv_nmi.c index a60af0230e27..45d784143a13 100644 --- a/arch/x86/platform/uv/uv_nmi.c +++ b/arch/x86/platform/uv/uv_nmi.c @@ -205,8 +205,7 @@ static int param_set_action(const char *val, const struct kernel_param *kp) char arg[ACTION_LEN], *p; /* (remove possible '\n') */ - strncpy(arg, val, ACTION_LEN - 1); - arg[ACTION_LEN - 1] = '\0'; + strscpy(arg, val, ACTION_LEN); p = strchr(arg, '\n'); if (p) *p = '\0'; @@ -216,7 +215,7 @@ static int param_set_action(const char *val, const struct kernel_param *kp) break; if (i < n) { - strcpy(uv_nmi_action, arg); + strscpy(uv_nmi_action, arg, strlen(uv_nmi_action)); pr_info("UV: New NMI action:%s\n", uv_nmi_action); return 0; } @@ -959,7 +958,7 @@ static int uv_handle_nmi(unsigned int reason, struct pt_regs *regs) /* Unexpected return, revert action to "dump" */ if (master) - strncpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); + strscpy(uv_nmi_action, "dump", strlen(uv_nmi_action)); } /* Pause as all CPU's enter the NMI handler */
Both `strncpy` and `strcpy` are deprecated for use on NUL-terminated destination strings [1]. A suitable replacement is `strscpy` [2] due to the fact that it guarantees NUL-termination on its destination buffer argument which is _not_ the case for `strncpy` or `strcpy`! In this case, we can drop both the forced NUL-termination and the `... -1` from: | strncpy(arg, val, ACTION_LEN - 1); as `strscpy` implicitly has this behavior. Link: www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings[1] Link: https://manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html [2] Link: https://github.com/KSPP/linux/issues/90 Cc: linux-hardening@vger.kernel.org Signed-off-by: Justin Stitt <justinstitt@google.com> --- Note: build-tested only --- arch/x86/platform/uv/uv_nmi.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) --- base-commit: 706a741595047797872e669b3101429ab8d378ef change-id: 20230822-strncpy-arch-x86-platform-uv-uv_nmi-474e5295c2c1 Best regards, -- Justin Stitt <justinstitt@google.com>