[PATCH v3] arm64/sysreg: refactor deprecated strncpy

[PATCH v3] arm64/sysreg: refactor deprecated strncpy

[Justin Stitt]

strncpy is deprecated [1] and should not be used if the src string is
not NUL-terminated.

When dealing with `cmdline` we are counting the number of characters
until a space then copying these over into `buf`. Let's not use any of
the str*() functions since the src string is not necessarily NUL-terminated.

Prefer `memcpy()` alongside a forced NUL-termination as it more
accurately describes what is going on within this function, i.e: copying
from non NUL-terminated buffer into a NUL-terminated buffer.

Link: https://www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings [1]
Link: https://github.com/KSPP/linux/issues/90
Cc: linux-hardening@vger.kernel.org
Suggested-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Justin Stitt <justinstitt@google.com>
---
Here's a quick rundown on the history of this patch:
1) v1 (changes requested)
2) v2 (applied to arm64 (for-next/misc))
3) v2 reverted (https://lore.kernel.org/all/20230831162227.2307863-1-smostafa@google.com/)
4) v3 (fixes problems with both v1 and v2)

Changes in v3:
- Fix faulty logic and use memcpy over strscpy (thanks Mostafa and Kees)
- Use '\0' instead of 0 to make it abundantly clear that `buf` is a NUL-terminated string
- Link to v2: https://lore.kernel.org/r/20230811-strncpy-arch-arm64-v2-1-ba84eabffadb@google.com

Changes in v2:
- Utilize return value from strscpy and check for truncation (thanks Kees)
- Link to v1: https://lore.kernel.org/r/20230810-strncpy-arch-arm64-v1-1-f67f3685cd64@google.com
---
 arch/arm64/kernel/idreg-override.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/idreg-override.c b/arch/arm64/kernel/idreg-override.c
index 2fe2491b692c..3addc09f8746 100644
--- a/arch/arm64/kernel/idreg-override.c
+++ b/arch/arm64/kernel/idreg-override.c
@@ -263,8 +263,8 @@ static __init void __parse_cmdline(const char *cmdline, bool parse_aliases)
 			return;
 
 		len = min(len, ARRAY_SIZE(buf) - 1);
-		strncpy(buf, cmdline, len);
-		buf[len] = 0;
+		memcpy(buf, cmdline, len);
+		buf[len] = '\0';
 
 		if (strcmp(buf, "--") == 0)
 			return;

---
base-commit: 706a741595047797872e669b3101429ab8d378ef
change-id: 20230810-strncpy-arch-arm64-1f3d328bd9b8

Best regards,
--
Justin Stitt <justinstitt@google.com>

Re: [PATCH v3] arm64/sysreg: refactor deprecated strncpy

[Mostafa Saleh]

Hi Justin,

On Thu, Aug 31, 2023 at 10:55:59PM +0000, Justin Stitt wrote:
> strncpy is deprecated [1] and should not be used if the src string is
> not NUL-terminated.
> 
> When dealing with `cmdline` we are counting the number of characters
> until a space then copying these over into `buf`. Let's not use any of
> the str*() functions since the src string is not necessarily NUL-terminated.
> 
> Prefer `memcpy()` alongside a forced NUL-termination as it more
> accurately describes what is going on within this function, i.e: copying
> from non NUL-terminated buffer into a NUL-terminated buffer.
> 
> Link: https://www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings [1]
> Link: https://github.com/KSPP/linux/issues/90
> Cc: linux-hardening@vger.kernel.org
> Suggested-by: Kees Cook <keescook@chromium.org>
> Signed-off-by: Justin Stitt <justinstitt@google.com>
> ---
> Here's a quick rundown on the history of this patch:
> 1) v1 (changes requested)
> 2) v2 (applied to arm64 (for-next/misc))
> 3) v2 reverted (https://lore.kernel.org/all/20230831162227.2307863-1-smostafa@google.com/)

This is just a patch, it is not reverted yet, also Marc replied with
another proposal to use strscpy instead but with the correct length.

> 4) v3 (fixes problems with both v1 and v2)
> 
> Changes in v3:
> - Fix faulty logic and use memcpy over strscpy (thanks Mostafa and Kees)
> - Use '\0' instead of 0 to make it abundantly clear that `buf` is a NUL-terminated string
> - Link to v2: https://lore.kernel.org/r/20230811-strncpy-arch-arm64-v2-1-ba84eabffadb@google.com
> 
> Changes in v2:
> - Utilize return value from strscpy and check for truncation (thanks Kees)
> - Link to v1: https://lore.kernel.org/r/20230810-strncpy-arch-arm64-v1-1-f67f3685cd64@google.com
> ---
>  arch/arm64/kernel/idreg-override.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm64/kernel/idreg-override.c b/arch/arm64/kernel/idreg-override.c
> index 2fe2491b692c..3addc09f8746 100644
> --- a/arch/arm64/kernel/idreg-override.c
> +++ b/arch/arm64/kernel/idreg-override.c
> @@ -263,8 +263,8 @@ static __init void __parse_cmdline(const char *cmdline, bool parse_aliases)
>  			return;
>  
>  		len = min(len, ARRAY_SIZE(buf) - 1);
> -		strncpy(buf, cmdline, len);
> -		buf[len] = 0;
> +		memcpy(buf, cmdline, len);
> +		buf[len] = '\0';
>  
>  		if (strcmp(buf, "--") == 0)
>  			return;
> 
> ---
> base-commit: 706a741595047797872e669b3101429ab8d378ef
> change-id: 20230810-strncpy-arch-arm64-1f3d328bd9b8

Your patch will not apply on mainline but on the revert, otherwise

Tested-by: Mostafa Saleh <smostafa@google.com>

> Best regards,
> --
> Justin Stitt <justinstitt@google.com>

Thanks,
Mostafa

Re: [PATCH v3] arm64/sysreg: refactor deprecated strncpy

[Will Deacon]

On Fri, Sep 01, 2023 at 12:09:19PM +0000, Mostafa Saleh wrote:
> On Thu, Aug 31, 2023 at 10:55:59PM +0000, Justin Stitt wrote:
> > diff --git a/arch/arm64/kernel/idreg-override.c b/arch/arm64/kernel/idreg-override.c
> > index 2fe2491b692c..3addc09f8746 100644
> > --- a/arch/arm64/kernel/idreg-override.c
> > +++ b/arch/arm64/kernel/idreg-override.c
> > @@ -263,8 +263,8 @@ static __init void __parse_cmdline(const char *cmdline, bool parse_aliases)
> >  			return;
> >  
> >  		len = min(len, ARRAY_SIZE(buf) - 1);
> > -		strncpy(buf, cmdline, len);
> > -		buf[len] = 0;
> > +		memcpy(buf, cmdline, len);
> > +		buf[len] = '\0';
> >  
> >  		if (strcmp(buf, "--") == 0)
> >  			return;
> > 
> > ---
> > base-commit: 706a741595047797872e669b3101429ab8d378ef
> > change-id: 20230810-strncpy-arch-arm64-1f3d328bd9b8
> 
> Your patch will not apply on mainline but on the revert, otherwise
> 
> Tested-by: Mostafa Saleh <smostafa@google.com>

If somebody can post a v4 based against mainline, then I'm happy to pick
it up as a fix.

Thanks.

Will