* [PATCH] NFS: Prefer strscpy over strlcpy calls
@ 2023-05-12 15:57 Azeem Shaikh
2023-05-16 21:04 ` Kees Cook
2023-05-22 19:39 ` Kees Cook
0 siblings, 2 replies; 4+ messages in thread
From: Azeem Shaikh @ 2023-05-12 15:57 UTC (permalink / raw)
To: Trond Myklebust, Anna Schumaker
Cc: linux-hardening, Azeem Shaikh, linux-nfs, linux-kernel
strlcpy() reads the entire source buffer first.
This read may exceed the destination size limit.
This is both inefficient and can lead to linear read
overflows if a source string is not NUL-terminated [1].
Check for strscpy()'s return value of -E2BIG on truncate for safe
replacement with strlcpy().
This is part of a tree-wide cleanup to remove the strlcpy() function
entirely from the kernel [2].
[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89
Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
---
Note to reviewers: the one case where this patch would modify existing behavior
is when strlen(src)==destlen==0. Current behavior returns 0, with this patch it
would return -1.
Not sure what the implication of this updated behavior would be,
so bringing it to your attention.
fs/nfs/nfsroot.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/nfs/nfsroot.c b/fs/nfs/nfsroot.c
index 620329b7e6ae..7600100ba26f 100644
--- a/fs/nfs/nfsroot.c
+++ b/fs/nfs/nfsroot.c
@@ -164,7 +164,7 @@ __setup("nfsroot=", nfs_root_setup);
static int __init root_nfs_copy(char *dest, const char *src,
const size_t destlen)
{
- if (strlcpy(dest, src, destlen) > destlen)
+ if (strscpy(dest, src, destlen) == -E2BIG)
return -1;
return 0;
}
--
2.40.1.606.ga4b1b128d6-goog
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] NFS: Prefer strscpy over strlcpy calls
2023-05-12 15:57 [PATCH] NFS: Prefer strscpy over strlcpy calls Azeem Shaikh
@ 2023-05-16 21:04 ` Kees Cook
2023-05-22 16:19 ` Azeem Shaikh
2023-05-22 19:39 ` Kees Cook
1 sibling, 1 reply; 4+ messages in thread
From: Kees Cook @ 2023-05-16 21:04 UTC (permalink / raw)
To: Azeem Shaikh
Cc: Trond Myklebust, Anna Schumaker, linux-hardening, linux-nfs,
linux-kernel
On Fri, May 12, 2023 at 03:57:49PM +0000, Azeem Shaikh wrote:
> strlcpy() reads the entire source buffer first.
> This read may exceed the destination size limit.
> This is both inefficient and can lead to linear read
> overflows if a source string is not NUL-terminated [1].
> Check for strscpy()'s return value of -E2BIG on truncate for safe
> replacement with strlcpy().
>
> This is part of a tree-wide cleanup to remove the strlcpy() function
> entirely from the kernel [2].
>
> [1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
> [2] https://github.com/KSPP/linux/issues/89
>
> Signed-off-by: Azeem Shaikh <azeemshaikh38@gmail.com>
> ---
> Note to reviewers: the one case where this patch would modify existing behavior
> is when strlen(src)==destlen==0. Current behavior returns 0, with this patch it
> would return -1.
>
> Not sure what the implication of this updated behavior would be,
> so bringing it to your attention.
I'm not sure either, but I would prefer non-terminated strings produce
an error, which this change does. :)
Reviewed-by: Kees Cook <keescook@chromium.org>
-Kees
>
> fs/nfs/nfsroot.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/nfs/nfsroot.c b/fs/nfs/nfsroot.c
> index 620329b7e6ae..7600100ba26f 100644
> --- a/fs/nfs/nfsroot.c
> +++ b/fs/nfs/nfsroot.c
> @@ -164,7 +164,7 @@ __setup("nfsroot=", nfs_root_setup);
> static int __init root_nfs_copy(char *dest, const char *src,
> const size_t destlen)
> {
> - if (strlcpy(dest, src, destlen) > destlen)
> + if (strscpy(dest, src, destlen) == -E2BIG)
> return -1;
> return 0;
> }
> --
> 2.40.1.606.ga4b1b128d6-goog
>
>
--
Kees Cook
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] NFS: Prefer strscpy over strlcpy calls
2023-05-16 21:04 ` Kees Cook
@ 2023-05-22 16:19 ` Azeem Shaikh
0 siblings, 0 replies; 4+ messages in thread
From: Azeem Shaikh @ 2023-05-22 16:19 UTC (permalink / raw)
To: Kees Cook
Cc: Trond Myklebust, Anna Schumaker, linux-hardening, linux-nfs,
linux-kernel
> > ---
> > Note to reviewers: the one case where this patch would modify existing behavior
> > is when strlen(src)==destlen==0. Current behavior returns 0, with this patch it
> > would return -1.
> >
> > Not sure what the implication of this updated behavior would be,
> > so bringing it to your attention.
>
> I'm not sure either, but I would prefer non-terminated strings produce
> an error, which this change does. :)
>
> Reviewed-by: Kees Cook <keescook@chromium.org>
>
Friendly ping.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] NFS: Prefer strscpy over strlcpy calls
2023-05-12 15:57 [PATCH] NFS: Prefer strscpy over strlcpy calls Azeem Shaikh
2023-05-16 21:04 ` Kees Cook
@ 2023-05-22 19:39 ` Kees Cook
1 sibling, 0 replies; 4+ messages in thread
From: Kees Cook @ 2023-05-22 19:39 UTC (permalink / raw)
To: azeemshaikh38, anna, trond.myklebust
Cc: Kees Cook, linux-nfs, linux-kernel, linux-hardening
On Fri, 12 May 2023 15:57:49 +0000, Azeem Shaikh wrote:
> strlcpy() reads the entire source buffer first.
> This read may exceed the destination size limit.
> This is both inefficient and can lead to linear read
> overflows if a source string is not NUL-terminated [1].
> Check for strscpy()'s return value of -E2BIG on truncate for safe
> replacement with strlcpy().
>
> [...]
Applied to for-next/hardening, thanks!
[1/1] NFS: Prefer strscpy over strlcpy calls
https://git.kernel.org/kees/c/8ca25e00cf81
--
Kees Cook
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2023-05-22 19:40 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-05-12 15:57 [PATCH] NFS: Prefer strscpy over strlcpy calls Azeem Shaikh
2023-05-16 21:04 ` Kees Cook
2023-05-22 16:19 ` Azeem Shaikh
2023-05-22 19:39 ` Kees Cook
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).