Re: [PATCH] tools lib traceevent: Fix buffer overflow in arg_eval

Re: [PATCH] tools lib traceevent: Fix buffer overflow in arg_eval

[Steven Rostedt]

On Wed, 27 Feb 2019 17:55:32 -0800
Tony Jones <tonyj@suse.de> wrote:

> Fix buffer overflow observed when running perf test.
> 
> The overflow is when trying to evaluate "1ULL << (64 - 1)" which
> is resulting in -9223372036854775808 which overflows the 20 character
> buffer.
> 
> If is possible this bug has been reported before but I still don't
> see any fix checked in:
> 
> See: https://www.spinics.net/lists/linux-perf-users/msg07714.html
> 
> Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
> Cc: linux-perf-users@vger.kernel.org
> Cc: Steven Rostedt <rostedt@goodmis.org>
> Signed-off-by: Tony Jones <tonyj@suse.de>

Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>

I have to say I've let this slide and it is not the first time a patch
went out with this fix. But this one has the correct fix because we
should use a buffer with a multiple of 4. Anyway, Tony I believe was
the first to report this anyway.

For reference we have:

I first heard about Tony's complaint on a post to linux-perf-users on Jan 18.

But then we had after that:

 Michael Sartain reported it on 1/24 (and fixed by Tzvetomir)
 https://lore.kernel.org/linux-trace-devel/20190125102014.19600-1-tstoyanov@vmware.com/
 
It was later fixed again by Mathias Krause
 https://lore.kernel.org/linux-trace-devel/20190223122404.21137-1-minipli@googlemail.com/

But since Tony was first to report it, and we discussed that it should
be 24 bytes, I would say this is the patch to take.

Again, sorry for not getting this acknowledged earlier and everyone doing
the same thing multiple times. :-/

Arnaldo, please take this patch. But also add:

Reported-by: Michael Sartain <mikesart@fastmail.com>
Reported-by: Mathias Krause <minipli@googlemail.com>

Thanks,

-- Steve


> ---
>  tools/lib/traceevent/event-parse.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/tools/lib/traceevent/event-parse.c
> b/tools/lib/traceevent/event-parse.c index abd4fa5d3088..87494c7c619d
> 100644 --- a/tools/lib/traceevent/event-parse.c
> +++ b/tools/lib/traceevent/event-parse.c
> @@ -2457,7 +2457,7 @@ static int arg_num_eval(struct tep_print_arg
> *arg, long long *val) static char *arg_eval (struct tep_print_arg
> *arg) {
>  	long long val;
> -	static char buf[20];
> +	static char buf[24];
>  
>  	switch (arg->type) {
>  	case TEP_PRINT_ATOM:

Re: [PATCH] tools lib traceevent: Fix buffer overflow in arg_eval

[Arnaldo Carvalho de Melo]

Em Wed, Feb 27, 2019 at 10:12:50PM -0500, Steven Rostedt escreveu:
> On Wed, 27 Feb 2019 17:55:32 -0800
> Tony Jones <tonyj@suse.de> wrote:
> 
> > Fix buffer overflow observed when running perf test.
> > 
> > The overflow is when trying to evaluate "1ULL << (64 - 1)" which
> > is resulting in -9223372036854775808 which overflows the 20 character
> > buffer.
> > 
> > If is possible this bug has been reported before but I still don't
> > see any fix checked in:
> > 
> > See: https://www.spinics.net/lists/linux-perf-users/msg07714.html
> > 
> > Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
> > Cc: linux-perf-users@vger.kernel.org
> > Cc: Steven Rostedt <rostedt@goodmis.org>
> > Signed-off-by: Tony Jones <tonyj@suse.de>
> 
> Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
> 
> I have to say I've let this slide and it is not the first time a patch
> went out with this fix. But this one has the correct fix because we
> should use a buffer with a multiple of 4. Anyway, Tony I believe was
> the first to report this anyway.
> 
> For reference we have:
> 
> I first heard about Tony's complaint on a post to linux-perf-users on Jan 18.
> 
> But then we had after that:
> 
>  Michael Sartain reported it on 1/24 (and fixed by Tzvetomir)
>  https://lore.kernel.org/linux-trace-devel/20190125102014.19600-1-tstoyanov@vmware.com/
>  
> It was later fixed again by Mathias Krause
>  https://lore.kernel.org/linux-trace-devel/20190223122404.21137-1-minipli@googlemail.com/
> 
> But since Tony was first to report it, and we discussed that it should
> be 24 bytes, I would say this is the patch to take.
> 
> Again, sorry for not getting this acknowledged earlier and everyone doing
> the same thing multiple times. :-/
> 
> Arnaldo, please take this patch. But also add:
> 
> Reported-by: Michael Sartain <mikesart@fastmail.com>
> Reported-by: Mathias Krause <minipli@googlemail.com>

Will do.

Thanks for the credit research,

- Arnaldo

Re: [PATCH] tools lib traceevent: Fix buffer overflow in arg_eval

[Arnaldo Carvalho de Melo]

Em Thu, Feb 28, 2019 at 04:01:23PM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Wed, Feb 27, 2019 at 10:12:50PM -0500, Steven Rostedt escreveu:
> > On Wed, 27 Feb 2019 17:55:32 -0800
> > Tony Jones <tonyj@suse.de> wrote:
> > 
> > > Fix buffer overflow observed when running perf test.
> > > 
> > > The overflow is when trying to evaluate "1ULL << (64 - 1)" which
> > > is resulting in -9223372036854775808 which overflows the 20 character
> > > buffer.
> > > 
> > > If is possible this bug has been reported before but I still don't
> > > see any fix checked in:
> > > 
> > > See: https://www.spinics.net/lists/linux-perf-users/msg07714.html
> > > 
> > > Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
> > > Cc: linux-perf-users@vger.kernel.org
> > > Cc: Steven Rostedt <rostedt@goodmis.org>
> > > Signed-off-by: Tony Jones <tonyj@suse.de>
> > 
> > Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
> > 
> > I have to say I've let this slide and it is not the first time a patch
> > went out with this fix. But this one has the correct fix because we
> > should use a buffer with a multiple of 4. Anyway, Tony I believe was
> > the first to report this anyway.
> > 
> > For reference we have:
> > 
> > I first heard about Tony's complaint on a post to linux-perf-users on Jan 18.
> > 
> > But then we had after that:
> > 
> >  Michael Sartain reported it on 1/24 (and fixed by Tzvetomir)
> >  https://lore.kernel.org/linux-trace-devel/20190125102014.19600-1-tstoyanov@vmware.com/
> >  
> > It was later fixed again by Mathias Krause
> >  https://lore.kernel.org/linux-trace-devel/20190223122404.21137-1-minipli@googlemail.com/
> > 
> > But since Tony was first to report it, and we discussed that it should
> > be 24 bytes, I would say this is the patch to take.
> > 
> > Again, sorry for not getting this acknowledged earlier and everyone doing
> > the same thing multiple times. :-/
> > 
> > Arnaldo, please take this patch. But also add:
> > 
> > Reported-by: Michael Sartain <mikesart@fastmail.com>
> > Reported-by: Mathias Krause <minipli@googlemail.com>
> 
> Will do.
> 
> Thanks for the credit research,

Some more research, an approximation, adding:

Cc: Frederic Weisbecker <fweisbec@gmail.com>
Fixes: f7d82350e597 ("tools/events: Add files to create libtraceevent.a")

That is the 'git blame' entry for when that line was added with sizeof
== 20

- Arnaldo

[PATCH] tools lib traceevent: Fix buffer overflow in arg_eval

[Tzvetomir Stoyanov (VMware)]

From: Tony Jones <tonyj@suse.de>

Fix buffer overflow observed when running perf test.

The overflow is when trying to evaluate "1ULL << (64 - 1)" which is
resulting in -9223372036854775808 which overflows the 20 character
buffer.

If is possible this bug has been reported before but I still don't see
any fix checked in:

See: https://www.spinics.net/lists/linux-perf-users/msg07714.html

Reported-by: Michael Sartain <mikesart@fastmail.com>
Reported-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: Tony Jones <tonyj@suse.de>
Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Fixes: f7d82350e597 ("tools/events: Add files to create libtraceevent.a")
Link: http://lkml.kernel.org/r/20190228015532.8941-1-tonyj@suse.de
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
[
  Backport from kernel into the trace-cmd repo.
]

 lib/traceevent/event-parse.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/traceevent/event-parse.c b/lib/traceevent/event-parse.c
index 7cb2ae4..b36b536 100644
--- a/lib/traceevent/event-parse.c
+++ b/lib/traceevent/event-parse.c
@@ -2456,7 +2456,7 @@ static int arg_num_eval(struct tep_print_arg *arg, long long *val)
 static char *arg_eval (struct tep_print_arg *arg)
 {
 	long long val;
-	static char buf[20];
+	static char buf[24];
 
 	switch (arg->type) {
 	case TEP_PRINT_ATOM:
-- 
2.21.0