[patch, libfortran] Fix PR 80365, undefined memcpy

[patch, libfortran] Fix PR 80365, undefined memcpy

[Thomas Koenig]

[-- Attachment #1: Type: text/plain, Size: 449 bytes --]

Hello world,

the attached test cases fixes a problematic invocation of
memcpy where src=NULL and n=0 by simply doing nothing
in the surrounding function if there is nothing to be done.

No test case, because this only shows up with a sanitized
libgfortran or under a debugger.

OK for trunk?

Regards

	Thomas

2017-07-23  Thomas Koenig  <tkoenig@gcc.gnu.org>

         * io/unix.c (buf_write):  Return early if there is
         nothing to write.

[-- Attachment #2: p1.diff --]
[-- Type: text/x-patch, Size: 349 bytes --]

Index: io/unix.c
===================================================================
--- io/unix.c	(Revision 249936)
+++ io/unix.c	(Arbeitskopie)
@@ -582,6 +582,9 @@
 static ssize_t
 buf_write (unix_stream *s, const void *buf, ssize_t nbyte)
 {
+  if (nbyte == 0)
+    return 0;
+
   if (s->ndirty == 0)
     s->buffer_offset = s->logical_offset;
 

Re: [patch, libfortran] Fix PR 80365, undefined memcpy

[Janne Blomqvist]

On Sun, Jul 23, 2017 at 10:23 PM, Thomas Koenig <tkoenig@netcologne.de> wrote:
> Hello world,
>
> the attached test cases fixes a problematic invocation of
> memcpy where src=NULL and n=0 by simply doing nothing
> in the surrounding function if there is nothing to be done.
>
> No test case, because this only shows up with a sanitized
> libgfortran or under a debugger.
>
> OK for trunk?

Good catch, Ok.


-- 
Janne Blomqvist

Re: [patch, libfortran] Fix PR 80365, undefined memcpy

[Thomas Koenig]

Hi Janne,

> On Sun, Jul 23, 2017 at 10:23 PM, Thomas Koenig <tkoenig@netcologne.de> wrote:
>> Hello world,
>>
>> the attached test cases fixes a problematic invocation of
>> memcpy where src=NULL and n=0 by simply doing nothing
>> in the surrounding function if there is nothing to be done.
>>
>> No test case, because this only shows up with a sanitized
>> libgfortran or under a debugger.
>>
>> OK for trunk?
> 
> Good catch, Ok.


Committed, thanks for the review.

What is the opinion on backporting?  We should avoid undefined
behavior in the library, I think, so I am leanding towards
doing it.

Regards

	Thomas

Re: [patch, libfortran] Fix PR 80365, undefined memcpy

[Janne Blomqvist]

On Mon, Jul 24, 2017 at 2:12 PM, Thomas Koenig <tkoenig@netcologne.de> wrote:
> Hi Janne,
>
>> On Sun, Jul 23, 2017 at 10:23 PM, Thomas Koenig <tkoenig@netcologne.de>
>> wrote:
>>>
>>> Hello world,
>>>
>>> the attached test cases fixes a problematic invocation of
>>> memcpy where src=NULL and n=0 by simply doing nothing
>>> in the surrounding function if there is nothing to be done.
>>>
>>> No test case, because this only shows up with a sanitized
>>> libgfortran or under a debugger.
>>>
>>> OK for trunk?
>>
>>
>> Good catch, Ok.
>
>
>
> Committed, thanks for the review.
>
> What is the opinion on backporting?  We should avoid undefined
> behavior in the library, I think, so I am leanding towards
> doing it.
>
> Regards
>
>         Thomas
>

Sure, go ahead. I can't see how the patch would cause any bugs (knock
knock...), so it seems quite safe.

BTW, there might be a similar issue in buf_read()?


-- 
Janne Blomqvist