I am going to quote Lee Jones who has been doing some snprintf ->
scnprintf refactorings:
"There is a general misunderstanding amongst engineers that
{v}snprintf() returns the length of the data *actually* encoded into the
destination array. However, as per the C99 standard {v}snprintf()
really returns the length of the data that *would have been* written if
there were enough space for it. This misunderstanding has led to
buffer-overruns in the past. It's generally considered safer to use the
{v}scnprintf() variants in their place (or even sprintf() in simple
cases). So let's do that."
To help prevent new instances of snprintf() from popping up, let's add a
check to checkpatch.pl.
Suggested-by: Finn Thain <[email protected]>
Signed-off-by: Justin Stitt <[email protected]>
---
Changes in v2:
- Had a vim moment and deleted a character before sending the patch.
- Replaced the character :)
- Link to v1: https://lore.kernel.org/r/[email protected]
---
From a discussion here [1].
[1]: https://lore.kernel.org/all/[email protected]/
---
scripts/checkpatch.pl | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
index 9c4c4a61bc83..64025a6e6155 100755
--- a/scripts/checkpatch.pl
+++ b/scripts/checkpatch.pl
@@ -7012,6 +7012,12 @@ sub process {
"Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr);
}
+# snprintf uses that should likely be {v}scnprintf
+ if ($line =~ /\bsnprintf\s*\(\s*/) {
+ WARN("SNPRINTF",
+ "Prefer scnprintf over snprintf\n" . $herecurr);
+ }
+
# ethtool_sprintf uses that should likely be ethtool_puts
if ($line =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\)/) {
if (WARN("PREFER_ETHTOOL_PUTS",
---
base-commit: b401b621758e46812da61fa58a67c3fd8d91de0d
change-id: 20240221-snprintf-checkpatch-a864ed67ebd0
Best regards,
--
Justin Stitt <[email protected]>
On Wed, Feb 21, 2024 at 10:11:59PM +0000, Justin Stitt wrote:
> I am going to quote Lee Jones who has been doing some snprintf ->
> scnprintf refactorings:
>
> "There is a general misunderstanding amongst engineers that
> {v}snprintf() returns the length of the data *actually* encoded into the
> destination array. However, as per the C99 standard {v}snprintf()
> really returns the length of the data that *would have been* written if
> there were enough space for it. This misunderstanding has led to
> buffer-overruns in the past. It's generally considered safer to use the
> {v}scnprintf() variants in their place (or even sprintf() in simple
> cases). So let's do that."
>
> To help prevent new instances of snprintf() from popping up, let's add a
> check to checkpatch.pl.
>
> Suggested-by: Finn Thain <[email protected]>
> Signed-off-by: Justin Stitt <[email protected]>
Yes please! :)
Reviewed-by: Kees Cook <[email protected]>
-Kees
> ---
> Changes in v2:
> - Had a vim moment and deleted a character before sending the patch.
> - Replaced the character :)
> - Link to v1: https://lore.kernel.org/r/[email protected]
> ---
> From a discussion here [1].
>
> [1]: https://lore.kernel.org/all/[email protected]/
> ---
> scripts/checkpatch.pl | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> index 9c4c4a61bc83..64025a6e6155 100755
> --- a/scripts/checkpatch.pl
> +++ b/scripts/checkpatch.pl
> @@ -7012,6 +7012,12 @@ sub process {
> "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr);
> }
>
> +# snprintf uses that should likely be {v}scnprintf
> + if ($line =~ /\bsnprintf\s*\(\s*/) {
> + WARN("SNPRINTF",
> + "Prefer scnprintf over snprintf\n" . $herecurr);
> + }
> +
> # ethtool_sprintf uses that should likely be ethtool_puts
> if ($line =~ /\bethtool_sprintf\s*\(\s*$FuncArg\s*,\s*$FuncArg\s*\)/) {
> if (WARN("PREFER_ETHTOOL_PUTS",
>
> ---
> base-commit: b401b621758e46812da61fa58a67c3fd8d91de0d
> change-id: 20240221-snprintf-checkpatch-a864ed67ebd0
>
> Best regards,
> --
> Justin Stitt <[email protected]>
>
--
Kees Cook
On Wed, 2024-02-21 at 22:11 +0000, Justin Stitt wrote:
> I am going to quote Lee Jones who has been doing some snprintf ->
> scnprintf refactorings:
>
> "There is a general misunderstanding amongst engineers that
> {v}snprintf() returns the length of the data *actually* encoded into the
> destination array. However, as per the C99 standard {v}snprintf()
> really returns the length of the data that *would have been* written if
> there were enough space for it. This misunderstanding has led to
> buffer-overruns in the past. It's generally considered safer to use the
> {v}scnprintf() variants in their place (or even sprintf() in simple
> cases). So let's do that."
>
> To help prevent new instances of snprintf() from popping up, let's add a
> check to checkpatch.pl.
>
> Suggested-by: Finn Thain <[email protected]>
> Signed-off-by: Justin Stitt <[email protected]>
> ---
> Changes in v2:
> - Had a vim moment and deleted a character before sending the patch.
> - Replaced the character :)
> - Link to v1: https://lore.kernel.org/r/[email protected]
> ---
> From a discussion here [1].
>
> [1]: https://lore.kernel.org/all/[email protected]/
> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
[]
> @@ -7012,6 +7012,12 @@ sub process {
> "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr);
> }
>
> +# snprintf uses that should likely be {v}scnprintf
> + if ($line =~ /\bsnprintf\s*\(\s*/) {
> + WARN("SNPRINTF",
> + "Prefer scnprintf over snprintf\n" . $herecurr);
There really should be some sort of reference link here
similar to the one above this.
Also, I rather doubt _all_ of these should be changed just
for churn's sake.
Maybe add a test for some return value use like
if (defined($stat) &&
$stat =~ /$Lval\s*=\s*snprintf\s*\(/) {
etc...
Maybe offer to --fix it too.
From: Justin Stitt
> Sent: 21 February 2024 22:12
>
> I am going to quote Lee Jones who has been doing some snprintf ->
> scnprintf refactorings:
>
> "There is a general misunderstanding amongst engineers that
> {v}snprintf() returns the length of the data *actually* encoded into the
> destination array. However, as per the C99 standard {v}snprintf()
> really returns the length of the data that *would have been* written if
> there were enough space for it. This misunderstanding has led to
> buffer-overruns in the past. It's generally considered safer to use the
> {v}scnprintf() variants in their place (or even sprintf() in simple
> cases). So let's do that."
While generally true, there are places that really do want to
detect (and error) overflow.
That isn't possible with scnprintf().
I'm not sure what the solution is though.
Having a function that returns a negative value on overflow is also
likely to get misused.
seq_printf() (or whatever it is called) may let you check,
but it is hardly a cheap wrapper and a bit of a PITA to use.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)
On Wed, 21 Feb 2024, Joe Perches wrote:
> On Wed, 2024-02-21 at 22:11 +0000, Justin Stitt wrote:
> > I am going to quote Lee Jones who has been doing some snprintf ->
> > scnprintf refactorings:
> >
> > "There is a general misunderstanding amongst engineers that
> > {v}snprintf() returns the length of the data *actually* encoded into the
> > destination array. However, as per the C99 standard {v}snprintf()
> > really returns the length of the data that *would have been* written if
> > there were enough space for it. This misunderstanding has led to
> > buffer-overruns in the past. It's generally considered safer to use the
> > {v}scnprintf() variants in their place (or even sprintf() in simple
> > cases). So let's do that."
> >
> > To help prevent new instances of snprintf() from popping up, let's add a
> > check to checkpatch.pl.
> >
> > Suggested-by: Finn Thain <[email protected]>
> > Signed-off-by: Justin Stitt <[email protected]>
> > ---
> > Changes in v2:
> > - Had a vim moment and deleted a character before sending the patch.
> > - Replaced the character :)
> > - Link to v1: https://lore.kernel.org/r/[email protected]
> > ---
> > From a discussion here [1].
> >
> > [1]: https://lore.kernel.org/all/[email protected]/
>
> > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> []
> > @@ -7012,6 +7012,12 @@ sub process {
> > "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr);
> > }
> >
> > +# snprintf uses that should likely be {v}scnprintf
> > + if ($line =~ /\bsnprintf\s*\(\s*/) {
> > + WARN("SNPRINTF",
> > + "Prefer scnprintf over snprintf\n" . $herecurr);
>
> There really should be some sort of reference link here
> similar to the one above this.
>
> Also, I rather doubt _all_ of these should be changed just
> for churn's sake.
This is for new implementations only.
Kees is planning on changing all of the current instances kernel-wide.
> Maybe add a test for some return value use like
>
> if (defined($stat) &&
> $stat =~ /$Lval\s*=\s*snprintf\s*\(/) {
> etc...
>
> Maybe offer to --fix it too.
>
--
Lee Jones [李琼斯]
On Thu, 22 Feb 2024, David Laight wrote:
> From: Justin Stitt
> > Sent: 21 February 2024 22:12
> >
> > I am going to quote Lee Jones who has been doing some snprintf ->
> > scnprintf refactorings:
> >
> > "There is a general misunderstanding amongst engineers that
> > {v}snprintf() returns the length of the data *actually* encoded into the
> > destination array. However, as per the C99 standard {v}snprintf()
> > really returns the length of the data that *would have been* written if
> > there were enough space for it. This misunderstanding has led to
> > buffer-overruns in the past. It's generally considered safer to use the
> > {v}scnprintf() variants in their place (or even sprintf() in simple
> > cases). So let's do that."
>
> While generally true, there are places that really do want to
> detect (and error) overflow.
> That isn't possible with scnprintf().
>
> I'm not sure what the solution is though.
> Having a function that returns a negative value on overflow is also
> likely to get misused.
> seq_printf() (or whatever it is called) may let you check,
> but it is hardly a cheap wrapper and a bit of a PITA to use.
I agree.
spprinf() was my favorite solution, but it seems that the lib string
people don't like to accept new functionality, even if it's a clear
improvement over the currently available solutions.
[0] https://lore.kernel.org/all/[email protected]/
--
Lee Jones [李琼斯]
On Fri, 2024-02-23 at 10:38 +0000, Lee Jones wrote:
> On Wed, 21 Feb 2024, Joe Perches wrote:
>
> > On Wed, 2024-02-21 at 22:11 +0000, Justin Stitt wrote:
> > > I am going to quote Lee Jones who has been doing some snprintf ->
> > > scnprintf refactorings:
> > >
> > > "There is a general misunderstanding amongst engineers that
> > > {v}snprintf() returns the length of the data *actually* encoded into the
> > > destination array. However, as per the C99 standard {v}snprintf()
> > > really returns the length of the data that *would have been* written if
> > > there were enough space for it. This misunderstanding has led to
> > > buffer-overruns in the past. It's generally considered safer to use the
> > > {v}scnprintf() variants in their place (or even sprintf() in simple
> > > cases). So let's do that."
> > >
> > > To help prevent new instances of snprintf() from popping up, let's add a
> > > check to checkpatch.pl.
> > >
> > > Suggested-by: Finn Thain <[email protected]>
> > > Signed-off-by: Justin Stitt <[email protected]>
> > > ---
> > > Changes in v2:
> > > - Had a vim moment and deleted a character before sending the patch.
> > > - Replaced the character :)
> > > - Link to v1: https://lore.kernel.org/r/[email protected]
> > > ---
> > > From a discussion here [1].
> > >
> > > [1]: https://lore.kernel.org/all/[email protected]/
> >
> > > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> > []
> > > @@ -7012,6 +7012,12 @@ sub process {
> > > "Prefer strscpy, strscpy_pad, or __nonstring over strncpy - see: https://github.com/KSPP/linux/issues/90\n" . $herecurr);
> > > }
> > >
> > > +# snprintf uses that should likely be {v}scnprintf
> > > + if ($line =~ /\bsnprintf\s*\(\s*/) {
> > > + WARN("SNPRINTF",
> > > + "Prefer scnprintf over snprintf\n" . $herecurr);
> >
> > There really should be some sort of reference link here
> > similar to the one above this.
> >
> > Also, I rather doubt _all_ of these should be changed just
> > for churn's sake.
>
> This is for new implementations only.
>
> Kees is planning on changing all of the current instances kernel-wide.
I saw that. I also saw pushback.
Not just my own.
Creating a cocci script is easy.
Getting Linus and others to run it isn't.