[perl #41095] [BUG] Segfault in test.exe during Configuration

Nikolay Ananiev

unread,

Dec 16, 2006, 11:13:33 AM12/16/06

to bugs-bi...@rt.perl.org

# New Ticket Created by "Nikolay Ananiev"
# Please include the string: [perl #41095]
# in the subject line of all future correspondence about this issue.
# <URL: http://rt.perl.org/rt3/Ticket/Display.html?id=41095 >

test.exe or config/auto/jit/test_c.in segfault during configuration on Win32. Note that this happent on
Pentium MMX ( 586 ). The test tests for "fucomip"
cpu instruction, but the 586 generation CPUs don't
have it. It's first introduced in
Pentium PRO (686). The segfault opens a debug
window in win32,
where the user must press OK for the Configure
script to continue. After this the configuration and
compilation are successful, because test.exe returns
false after the segfault and "fucomip" is not used.

jnthn@jnthn.net via RT

unread,

Dec 18, 2006, 7:40:25 PM12/18/06

to perl6-i...@perl.org

On Sat Dec 16 08:13:32 2006, ana...@thegdb.com wrote:
> test.exe or config/auto/jit/test_c.in segfault during configuration on
> Win32. Note that this happent on
> Pentium MMX ( 586 ). The test tests for "fucomip"
> cpu instruction, but the 586 generation CPUs don't
> have it. It's first introduced in
> Pentium PRO (686). The segfault opens a debug
> window in win32,
> where the user must press OK for the Configure
> script to continue. After this the configuration and
> compilation are successful, because test.exe returns
> false after the segfault and "fucomip" is not used.

Thanks for reporting this; kelli++ is looking into using CPUID instead
of just trying to use the opcode and seeing if we crash and burn, as
happens now.

Jonathan

James Keenan via RT

unread,

Apr 13, 2008, 10:36:55 PM4/13/08

to perl6-i...@perl.org

Jonathan: Do you know if we've overcome this problem?

Thank you very much.

kid51

Nikolay Ananiev

unread,

Apr 14, 2008, 2:09:33 AM4/14/08

to parrotbug...@parrotcode.org

I'll be able to test Parrot in the next 48 hours
on my pentium mmx and tell you what the result is.

Nikolay Ananiev

unread,

Apr 16, 2008, 5:57:40 PM4/16/08

to perl6-i...@perl.org

Confirmed. This bug is still there as of r27009

"Nikolay Ananiev" <ana...@thegdb.com> wrote in message
news:007301c89df6$1c9176f0$0700a8c0@ananiev...

Nikolay Ananiev

unread,

Dec 7, 2008, 2:41:10 PM12/7/08

to Will Coleda, parrotbug...@parrotcode.org

Windows 2000 SP4
Visual Studio 2003 SP1

The hardware is:
Pentium MMX (586) with 128mb ram.

The problem with the test file is that it does not work reliably for the
processors that do not support the fcomip cpu instruction (586 and earlier).
I tried to fix this myself by using
CPUID, but I don't have enough knowledge and also I don't know how to handle
the non-x86 CPUs (because they
don't support CPUID).

----- Original Message -----
From: "Will Coleda" <wi...@coleda.com>
To: "Nikolay Ananiev" <ana...@thegdb.com>
Cc: <parrotbug...@parrotcode.org>
Sent: Sunday, December 07, 2008 6:57 PM
Subject: Re: [perl #41095] [BUG] Segfault in test.exe during Configuration

> Thanks - can you also let us know the following:
>
> What version of windows are you using?
>
> Which compiler (including version)?
>
> On Sat, Dec 6, 2008 at 11:47 AM, Nikolay Ananiev <ana...@thegdb.com>
> wrote:
>> Tested with r33568. The bug is still there.
>>
>> ----- Original Message ----- From: "Will Coleda via RT"
>> <parrotbug...@parrotcode.org>
>> To: <ana...@thegdb.com>
>> Sent: Wednesday, December 03, 2008 10:00 PM

>> Subject: [perl #41095] [BUG] Segfault in test.exe during Configuration
>>

>>> Can someone test this with a recent parrot?
>>>
>>> Based on a recent conversation on list, we should probably start getting
>>> OS / Compiler versions as well with bug reports.

>>> --
>>> Will "Coke" Coleda
>>
>
>
>
> --
> Will "Coke" Coleda