[Bf-committers] ATI+CoreDuo+Mac10.5 crashes

Damien Plisson damien.plisson at yahoo.fr
Sun Oct 18 18:06:21 CEST 2009


This a known bug (thx devroo for the link) : http://lists.apple.com/archives/mac-opengl//2009/Apr/msg00000.html

Some ATI cards report to have the GL_ARB_texture_non_power_of_two  
extension while it is completely crap...

I propose I write a change to the glew.c init function to change the  
test for GL_ARB_texture_non_power_of_two extension, to return false if  
a mac ati board is present.

Damien

Le 17 oct. 2009 à 14:35, Martin Poirier a écrit :

> Could anyone that has this problem test a one line change?
>
> source/blender/editors/interface/interface_icons.c
>
> replace line 492 with
>
> if (0) {
>
> And see if it changes anything.
>
> Martin
>
> --- On Sat, 10/17/09, Damien Plisson <damien.plisson at yahoo.fr> wrote:
>
>> From: Damien Plisson <damien.plisson at yahoo.fr>
>> Subject: Re: [Bf-committers] ATI+CoreDuo+Mac10.5 crashes
>> To: "bf-blender developers" <bf-committers at blender.org>
>> Received: Saturday, October 17, 2009, 7:57 AM
>> Hi all,
>>
>> I've got confirmed report that the bug that makes these
>> macbook pros
>> w/ ATI crash has been introduced with svn rev 23559.
>>
>> FYI, test showed on same MBP/ATI that rev 23558 works, and
>> rev 23559
>> triggers kernel panic.
>>
>> This severe issue is related to the one reported by devroo
>> (bug #19637
>> "missing letters in gui") of display issue on mac ati
>> systems,
>> introduced with this rev 23559.
>>
>> Damien
>>
>>
>> Le 15 oct. 2009 à 10:51, Damien Plisson a écrit :
>>
>>> Hi Jasper & Galen,
>>>
>>> Yes, there are some hints it may be independent from
>> cocoa, as there
>>> are also reports of systems crashes at blender launch
>> on non-cocoa
>>> builds, for example, this build that is not cocoa
>> (cocoa ghost was not
>>> compiling on 10.4 at that time):
>>> http://www.graphicall.org/builds/builds/showbuild.php?action=show&id=1095
>>>
>>> Another point is that it seems to appear on systems
>> with ATI graphics
>>> AND Core Duo, the early version before the core 2,
>> that is not 64bit
>>> compatible. Is there a compatibility issue w/ ATI
>> Leopard drivers?
>>> From google, it seems there are also issues w/ those
>> systems on other
>>> SW as well.
>>>
>>> Has anyone such a crash issue on a different setup ?
>>>
>>> Anyway, if Blender 2.49 works well on those systems,
>> we should find a
>>> workaround to make 2.5 work.
>>> As Galen stated, any console logs, any traces,
>> anything may help, as I
>>> have no other hints to debug...
>>> FYI: I have no issues running 2.5 cocoa & carbon
>> on my MBP late 2008
>>> (C2D & nvidia), either on 10.5.8 or 10.6.1
>>>
>>> I've been able to make small changes to have ghost
>> cocoa compile on
>>> 10.4 + gcc4.0, so if you prefer, you can test with
>> such a build.
>>>
>>> Damien
>>>
>>> Le 15 oct. 2009 à 02:39, Jasper Mine a écrit :
>>>
>>>> Hello,
>>>>
>>>> With cocoa set to on.  Affirmative.  I
>> had done that.
>>>>
>>>> It is difficult to contain because of the
>> panic.  Is there a tool I
>>>> can use that logs to disk?  Will that be lost
>> on panic?
>>>>
>>>> I should note that the first time I had this
>> happen I was using nan
>>>> make files, which may have avoided your
>> code.  So it might be
>>>> possible
>>>> the panic comes from something/someone else?
>>>>
>>>> My graphics details:
>>>> ATI Radeon X1600
>>>> 256MB
>>>> 1440x900
>>>> 32-bit color
>>>> core image, and quartz extreme supported
>>>> ATIRadeonX1000 1.5.48.6 (6967)
>>>> ATY_Wormy 1.5.48.6 (1.5f37)
>>>>
>>>> The machine is a mac book pro core duo at 2 Ghz
>>>>
>>>> I will again look to see if any of the apple tools
>> will log to a file
>>>> instead of just displaying info in a window, as
>> that will be lost
>>>> after the event.
>>>>
>>>> Jasper
>>>>>
>>>>> On Oct 14, 2009, at 5:22 AM, Damien Plisson
>> wrote:
>>>>>
>>>>>> Hi Jasper,
>>>>>>
>>>>>> Strange... A user level app that triggers
>> a kernel panic...
>>>>>>
>>>>>> Anyway, from your backtrace, it seems the
>> ATI driver is doing
>>>>>> something not clear.
>>>>>>
>>>>>> The only option to change in cmake is to
>> set WITH_COCOA to on to
>>>>>> get
>>>>>> cocoa build.
>>>>>> If you are able (but maybe difficult with
>> a kernel panic) to locate
>>>>>> (using xcode) the blender code line that
>> triggers it, that'll be
>>>>>> useful.
>>>>>>
>>>>>> Damien
>>>>>>
>>>>>> Le 14 oct. 2009 à 01:00, Jasper Mine a
>> écrit :
>>>>>>
>>>>>>> Damien:
>>>>>>>
>>>>>>> I'm still getting the kernel panic
>> using cmake and xcode.  If I
>>>>>>> can
>>>>>>> possibly trace this using the apple
>> tools I can do that.  Would I
>>>>>>> change any of the settings in ccmake
>> gui?  Here is the report I
>>>>>>> receive when I reboot.
>>>>>>>
>>>>>>> Tue Oct 13 18:51:50 2009
>>>>>>> panic(cpu 1 caller 0x001AB0FE): Kernel
>> trap at 0x00989a80, type
>>>>>>> 14=page fault, registers:
>>>>>>> CR0: 0x80010033, CR2: 0xff66a5d2, CR3:
>> 0x01019000, CR4: 0x000006e0
>>>>>>> EAX: 0x02fef000, EBX: 0xff66a5d2, ECX:
>> 0x00000000, EDX: 0x06be3230
>>>>>>> CR2: 0xff66a5d2, EBP: 0x324d77e8, ESI:
>> 0x00000000, EDI: 0x00000008
>>>>>>> EFL: 0x00010246, EIP: 0x00989a80,
>> CS:  0x00000004, DS:  0x0000000c
>>>>>>> Error code: 0x00000000
>>>>>>>
>>>>>>> Backtrace (CPU 1), Frame : Return
>> Address (4 potential args on
>>>>>>> stack)
>>>>>>> 0x324d75b8 : 0x12b4c6 (0x45f91c
>> 0x324d75ec 0x13355c 0x0)
>>>>>>> 0x324d7608 : 0x1ab0fe (0x469a98
>> 0x989a80 0xe 0x469248)
>>>>>>> 0x324d76e8 : 0x1a1713 (0x324d7708
>> 0x2ff0158 0x324d77ac 0x324d7788)
>>>>>>> 0x324d7700 : 0x989a80 (0xe 0x990048
>> 0x6be000c 0xc000c)
>>>>>>> 0x324d77e8 : 0x989e5d (0x2fef000
>> 0x3d65700 0x6be3230 0x0)
>>>>>>> 0x324d7818 : 0x96df57 (0x2fef000
>> 0x3d65700 0x6be3230 0x0)
>>>>>>> 0x324d7858 : 0x9850d0 (0x2fef000
>> 0x3d65700 0x6be3230 0x1596e3f8)
>>>>>>> 0x324d7878 : 0x96b526 (0x2fef000
>> 0x3d65700 0x6be3230 0xf9)
>>>>>>> 0x324d7898 : 0x98511a (0x2fef000
>> 0x6be322c 0x0 0x0)
>>>>>>> 0x324d78b8 : 0x970da7 (0x2fef000
>> 0x6be322c 0x0 0x0)
>>>>>>> 0x324d78e8 : 0x99bf16 (0x6be3000
>> 0x6be322c 0xc8 0x0)
>>>>>>> 0x324d7938 : 0x99c093 (0x6be3000
>> 0x324d7b54 0x0 0x0)
>>>>>>> 0x324d7968 : 0x99d5bc (0x6be3000
>> 0x324d7b54 0x3ace500 0x1)
>>>>>>> 0x324d79d8 : 0x996f07 (0x6be3000
>> 0x324d7b54 0xa6 0xb)
>>>>>>> 0x324d7b88 : 0x972029 (0x6be3000
>> 0x324d7c60 0x324d7bd8 0x12432a)
>>>>>>> 0x324d7c88 : 0x43faec (0x6be3000 0x1
>> 0x324d7cdc 0x324d7cd8)
>>>>>>>     Backtrace
>> continues...
>>>>>>>     Kernel loadable modules
>> in backtrace (with dependencies):
>>>>>>>
>>    com.apple.ATIRadeonX1000(5.4.8)@0x968000->0x9c6fff
>>>>>>>
>> dependency: com.apple.iokit.IONDRVSupport(1.7.3)@0x917000
>>>>>>>
>> dependency: com.apple.iokit.IOPCIFamily(2.6)@0x5ce000
>>>>>>>
>> dependency:
>>>>>>>
>> com.apple.iokit.IOGraphicsFamily(1.7.3)@0x8fa000
>>>>>>>
>>>>>>> BSD process name corresponding to
>> current thread: blender
>>>>>>>
>>>>>>> Mac OS version:
>>>>>>> 9L30
>>>>>>>
>>>>>>> Kernel version:
>>>>>>> Darwin Kernel Version 9.8.0: Wed Jul
>> 15 16:55:01 PDT 2009;
>>>>>>> root:xnu-1228.15.4~1/RELEASE_I386
>>>>>>> System model name: MacBookPro1,1
>> (Mac-F425BEC8
>>>>>>>
>>>>>>> Jasper
>>>>>>>
>>>>>>>
>>>> _______________________________________________
>>>> Bf-committers mailing list
>>>> Bf-committers at blender.org
>>>> http://lists.blender.org/mailman/listinfo/bf-committers
>>>
>>> _______________________________________________
>>> Bf-committers mailing list
>>> Bf-committers at blender.org
>>> http://lists.blender.org/mailman/listinfo/bf-committers
>>
>> _______________________________________________
>> Bf-committers mailing list
>> Bf-committers at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-committers
>>
>
>
>       
> __________________________________________________________________
> Looking for the perfect gift? Give the gift of Flickr!
>
> http://www.flickr.com/gift/
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers



More information about the Bf-committers mailing list