[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] FPU x86 instructions error

From: Clemens Kolbitsch
Subject: Re: [Qemu-devel] FPU x86 instructions error
Date: Sun, 26 May 2013 17:08:40 -0700

On Sun, May 26, 2013 at 7:54 AM, Andreas Färber <address@hidden> wrote:
> Hi,
> Am 24.05.2013 23:44, schrieb Paolo Bonzini:
>> Il 24/05/2013 23:39, Clemens Kolbitsch ha scritto:
>>> we recently had an issue with running a program using FPU instructions
>>> to obtain the current EIP (basically a weird way of "call 0; pop eax")
>>> that was not working on QEMU (with TCG).
>>> Looking at the problem, we found this patch to be useful/fixing the issue:
>>> https://launchpadlibrarian.net/140457932/patch-qemu-1.5.0-fpip.diff
>>> Looking through the DEVEL archives, I found this patch
>>> http://lists.nongnu.org/archive/html/qemu-devel/2011-06/msg01206.html
>>> that adds the FPU flags to the environment, but is only using them for KVM.
>>> I was wondering - since the above patch is rather old, you have
>>> probably come across it before - if there was a reason for not
>>> including it in QEMU (I checked in git:master and it's not applied).
>>> If there isn't, maybe it'd be worth re-considering :)
>> For the TCG patch, there is no Signed-off-by and using a helper is not
>> necessary.
> Clemens, generally we can't just take another person's patch and apply
> it - that's what we need the Signed-off-by for. Your post is the only
> Google hit for that link and no hits for fpip in my archive - you'll
> need to contact the author to obtain her Sob and properly submit it to
> qemu-devel - or post a patch yourself that is not based on that one.

Andreas, thanks for that info - that makes it much clearer. Since the
patch is rather short, it will be difficult, but we can give it a shot

> http://wiki.qemu.org/Contribute/SubmitAPatch
>> For the KVM patch, it simply fell through the cracks, I believe.
> It didn't:
> http://git.qemu.org/?p=qemu.git;a=commit;h=42cc8fa620cbc73e349e96d84cf46469e828ec34
> (I was about to suggest placing the non-TCG fields into X86CPU. :))

Yes, I think that was a misunderstanding. I was trying to say that for
first one didn't get picked up and the second one is just a
prerequisite -- not that it was missing as well :)


reply via email to

[Prev in Thread] Current Thread [Next in Thread]