RE: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

RE: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Laurent Vivier]

>On Wed, Feb 02, 2011 at 08:28:15PM +0100, Stefan Weil wrote:
>> [PATCH] linux-user: Fix possible realloc memory leak  
>> (http://patchwork.ozlabs.org/patch/79217/)
>
>Looks ok for me.

And this one ?

linux-user: correct core dump format

http://patchwork.ozlabs.org/patch/78464/

Laurent

-- 
--------------------- Laurent@vivier.eu  ---------------------
"Tout ce qui est impossible reste à accomplir"    Jules Verne
"Things are only impossible until they're not" Jean-Luc Picard

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[riku voipio]

On 02/03/2011 12:16 PM, Laurent Vivier wrote:
>> On Wed, Feb 02, 2011 at 08:28:15PM +0100, Stefan Weil wrote:
>>> [PATCH] linux-user: Fix possible realloc memory leak
>>> (http://patchwork.ozlabs.org/patch/79217/)
>> Looks ok for me.

> And this one ?
>
> linux-user: correct core dump format
>
> http://patchwork.ozlabs.org/patch/78464/

Do you need it for 0.14 ?

I intend to include this and other in a pull request for master soon.

Riku

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Anthony Liguori]

On 02/05/2011 01:40 PM, riku voipio wrote:
> On 02/03/2011 12:16 PM, Laurent Vivier wrote:
>>> On Wed, Feb 02, 2011 at 08:28:15PM +0100, Stefan Weil wrote:
>>>> [PATCH] linux-user: Fix possible realloc memory leak
>>>> (http://patchwork.ozlabs.org/patch/79217/)
>>> Looks ok for me.
>
>> And this one ?
>>
>> linux-user: correct core dump format
>>
>> http://patchwork.ozlabs.org/patch/78464/
>
> Do you need it for 0.14 ?
>
> I intend to include this and other in a pull request for master soon.

This looks reasonably important.  Can you just do a pull request for 
master for the patches that might be good for 0.14?

If you can do that in the next couple days, it'll make it into -rc2 no 
problem.

Regards,

Anthony Liguori

> Riku
>

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Riku Voipio]

> >Actually that patch is not ok. The issues you fix on m68k appear
> >on arm/eabi after applying your patch. bswap part appears ok, but
> >the padding is needed atleast on arm.
 
> How do you test this patch ? Do you use native gdb on ARM ? Because, for m68k, cross-compiled gdb does not work (it cannot have sizeof(prstatus_t) for m68k).

I tried both native and cross-gdb, version 7.2. Both fail with your patch applied
and work fine without it.

Before your patch, objdump (CodeSourcery cs2009q3 cross) output looks like:

-snip-
Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 note0         000001bc  00000000  00000000  000000d4  2**0
                  CONTENTS, READONLY
  1 .reg/23968    00000048  00000000  00000000  00000130  2**2
                  CONTENTS
  2 .reg          00000048  00000000  00000000  00000130  2**2
                  CONTENTS
  3 .auxv         00000070  00000000  00000000  00000220  2**2
                  CONTENTS
  4 load1         00000000  00008000  00000000  00001000  2**12
                  ALLOC, READONLY, CODE
-snip-

after:

-snip-
Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 note0         000001bc  00000000  00000000  000000d4  2**0
                  CONTENTS, READONLY
  1 .auxv         00000070  00000000  00000000  00000220  2**2
                  CONTENTS
  2 load1         00000000  00008000  00000000  00001000  2**12
                  ALLOC, READONLY, CODE
                  CONTENTS, ALLOC, LOAD
-snip-

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Laurent Vivier]

>On 9 February 2011 09:20, Laurent Vivier <Laurent@vivier.eu> wrote:
>> It is strange as we have in gdb, if I remember correctly, an explicit check 
>of the size of prstatus:
>>
>> gdb/bfd/elf.c:
>>
>>    if defined (HAVE_PRSTATUS_T)
>>
>>    static bfd_boolean
>>    elfcore_grok_prstatus (bfd *abfd, Elf_Internal_Note *note)
>> ...
>>      if (note->descsz == sizeof (prstatus_t))
>> ...
>>
>> How do you test this patch ? Do you use native gdb on ARM ? Because, for 
>m68k, cross-compiled gdb does not work (it cannot have sizeof(prstatus_t) for 
>m68k).
>
>I dunno about m68k, but if ARM gdb behaves differently natively
>versus the cross-tools version (ie hosted on x86 to target ARM)
>then that sounds like a bug we in Linaro would like to know
>about :-)

What I saw for m68k is it cannot access to the saved registers of the core dump  (i.e. you cannot use cross-compiled gdb to debug a core dump ...)

What you can try is to generate a core dump with a "kill -SIGILL" on a qemu-arm process (don't forget to do "ulimit -c unlimited" before) and then load the core with your cross-compiled gdb.

Regards,
Laurent
-- 
--------------------- Laurent@vivier.eu  ---------------------
"Tout ce qui est impossible reste à accomplir"    Jules Verne
"Things are only impossible until they're not" Jean-Luc Picard

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Peter Maydell]

On 9 February 2011 09:20, Laurent Vivier <Laurent@vivier.eu> wrote:
> It is strange as we have in gdb, if I remember correctly, an explicit check of the size of prstatus:
>
> gdb/bfd/elf.c:
>
>    if defined (HAVE_PRSTATUS_T)
>
>    static bfd_boolean
>    elfcore_grok_prstatus (bfd *abfd, Elf_Internal_Note *note)
> ...
>      if (note->descsz == sizeof (prstatus_t))
> ...
>
> How do you test this patch ? Do you use native gdb on ARM ? Because, for m68k, cross-compiled gdb does not work (it cannot have sizeof(prstatus_t) for m68k).

I dunno about m68k, but if ARM gdb behaves differently natively
versus the cross-tools version (ie hosted on x86 to target ARM)
then that sounds like a bug we in Linaro would like to know
about :-)

-- PMM

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Laurent Vivier]

>On Sat, Feb 05, 2011 at 10:49:51PM +0100, Laurent Vivier wrote:
>> >On 02/03/2011 12:16 PM, Laurent Vivier wrote:
>> >> And this one ?
>> >>
>> >> linux-user: correct core dump format
>> >>
>> >> http://patchwork.ozlabs.org/patch/78464/
>
>Actually that patch is not ok. The issues you fix on m68k appear
>on arm/eabi after applying your patch. bswap part appears ok, but
>the padding is needed atleast on arm.

It is strange as we have in gdb, if I remember correctly, an explicit check of the size of prstatus:

gdb/bfd/elf.c:

    if defined (HAVE_PRSTATUS_T)

    static bfd_boolean
    elfcore_grok_prstatus (bfd *abfd, Elf_Internal_Note *note)
...
      if (note->descsz == sizeof (prstatus_t))
...

How do you test this patch ? Do you use native gdb on ARM ? Because, for m68k, cross-compiled gdb does not work (it cannot have sizeof(prstatus_t) for m68k).

Regards,
Laurent


-- 
--------------------- Laurent@vivier.eu  ---------------------
"Tout ce qui est impossible reste à accomplir"    Jules Verne
"Things are only impossible until they're not" Jean-Luc Picard

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Riku Voipio]

On Sat, Feb 05, 2011 at 10:49:51PM +0100, Laurent Vivier wrote:
> >On 02/03/2011 12:16 PM, Laurent Vivier wrote:
> >> And this one ?
> >>
> >> linux-user: correct core dump format
> >>
> >> http://patchwork.ozlabs.org/patch/78464/

Actually that patch is not ok. The issues you fix on m68k appear
on arm/eabi after applying your patch. bswap part appears ok, but
the padding is needed atleast on arm.

Riku

Re: [Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Laurent Vivier]

Hi,

I've no need, just put it in your tree where you want...

Regards,
Laurent

riku voipio <riku.voipio@iki.fi> a écrit :

>On 02/03/2011 12:16 PM, Laurent Vivier wrote:
>>> On Wed, Feb 02, 2011 at 08:28:15PM +0100, Stefan Weil wrote:
>>>> [PATCH] linux-user: Fix possible realloc memory leak
>>>> (http://patchwork.ozlabs.org/patch/79217/)
>>> Looks ok for me.
>
>> And this one ?
>>
>> linux-user: correct core dump format
>>
>> http://patchwork.ozlabs.org/patch/78464/
>
>Do you need it for 0.14 ?
>
>I intend to include this and other in a pull request for master soon.
>
>Riku

[Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Riku Voipio]

On Wed, Feb 02, 2011 at 08:28:15PM +0100, Stefan Weil wrote:
> [PATCH] linux-user: Fix possible realloc memory leak  
> (http://patchwork.ozlabs.org/patch/79217/)

Looks ok for me.

Riku

[Qemu-devel] Re: [PING 0.14] Missing patches (mostly fixes)

[Michael S. Tsirkin]

On Wed, Feb 02, 2011 at 08:28:15PM +0100, Stefan Weil wrote:
> [PATCH 1/3] pci: Fix memory leak (http://patchwork.ozlabs.org/patch/79996/)

Looks good. Pls cc me on patches next time to make sure I do not miss a
patch.