Pymol stopped working

[mtiberti (Matteo Tiberti)]

Hello, I'm trying to use PyMOL 1.6.0 on Mac OSX Mavericks, installed through MacPorts, but the same happens with pymol 1.7.1 as well (see ticket #42860). Up to a couple of days ago everything was working perfectly fine. At a certain point I have probably upgraded the installed base (selfupdate + upgrade outdated), and now when I run pymol I get an error (log file attached) and then PyMOL crashes.

I have then tried the following:

• reinstalled PyMOL 1.6.0
• removed PyMOL 1.6.0 and installed 1.7.1
• updated the XQuartz X server to the latest version
• reinstalled the XQuartz X server to an older version
• wiped MacPorts clean and reinstalled PyMOL and its prerequisites from scratch
• installed the gfxCardStatus program and tried both using the integrated gpu and the discrete one
• searched on the internet (of course :) )

I have then asked for assistance in the PyMOL mailing list, and Thomas Holder, a PyMOL developer from Schrodinger, confirmed that he was experiencing the same. It looks like the issue only happens when PyMOL is built in the macports environment, and it's not related to any recent changes to PyMOL. This means that most likely something in macports got broken in the meanwhile.

Thanks for your attention

[mtiberti (Matteo Tiberti)]

crash log

[howarthjw@…]

I would suspect that the MacPorts glew package is the obvious candidate for the offender as the errors are related to shaders which is part of glew and the package has been changed several times in the last two weeks. Can you try reverting back through the earlier revisions glew to confirm which release and revision was last working for you? I don't see any problems with the pymol package that I maintain on fink against the fink 1.10.0 glew so the breakage is specific to the MacPorts build of glew. On fink, we are building glew with clang on 10.6 and later (and with gcc-4.2 on 10.5 and earlier). The build is done as…

make GLEW_DEST=%p GLEW_APPLE_GLX=1 CC=$CC LD=$CC

without any additional patches and the resulting package works fine with fink pymol 1.7.1.1.

[howarthjw@…]

log of failed 'sudo port -d install pymol'

[howarth.at.macports@…]

crash of pymol installed on MacPorts 2.3.0

[howarth.at.macports@…]

Okay, removing the previous /Application/MacPorts python subdirectories, I am able to complete…

sudo port -d install pymol

but he resulting python crashes as…

Process:         Python [3965]
Path:            /opt/local/Library/Frameworks/Python.framework/Versions/2.7/Resources/Python.app/Contents/MacOS/Python
Identifier:      Python
Version:         ??? (???)
Code Type:       X86-64 (Native)
Parent Process:  tcsh [3779]

Date/Time:       2014-05-24 15:34:35.289 -0400
OS Version:      Mac OS X 10.7.5 (11G63)
Report Version:  9

Interval Since Last Report:          5070847 sec
Crashes Since Last Report:           2146
Per-App Crashes Since Last Report:   8
Anonymous UUID:                      C628C6CC-8DD8-4CC4-94C4-B6BD1EF9DE51

Crashed Thread:  0  Dispatch queue: com.apple.main-thread

Exception Type:  EXC_CRASH (SIGABRT)
Exception Codes: 0x0000000000000000, 0x0000000000000000

Application Specific Information:
abort() called
objc[3965]: garbage collection is OFF

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
0   libsystem_kernel.dylib        	0x00007fff96b60ce2 __pthread_kill + 10
1   libsystem_c.dylib             	0x00007fff944de7d2 pthread_kill + 95
2   libsystem_c.dylib             	0x00007fff944cfa7a abort + 143
3   libGL.1.dylib                 	0x00000001085d648a apple_visual_create_pfobj + 935
4   libGL.1.dylib                 	0x00000001085d3ece apple_glx_create_context + 190
5   libGL.1.dylib                 	0x00000001085dcdf3 applegl_create_context + 163
6   libGL.1.dylib                 	0x00000001085d88df CreateContext + 123
7   libGL.1.dylib                 	0x00000001085d9a69 glXCreateNewContext + 40
8   libglut.3.dylib               	0x000000010872025f fghCreateNewContext + 310
9   libglut.3.dylib               	0x000000010871ff58 fgOpenWindow + 896
10  libglut.3.dylib               	0x000000010871eaa3 fgCreateWindow + 356
11  libglut.3.dylib               	0x000000010872045a glutCreateWindow + 118
12  _cmd.so                       	0x0000000107eaad1c main_common + 940
13  _cmd.so                       	0x0000000107eaafcc main_shared + 108
14  _cmd.so                       	0x0000000107e9ff47 CmdRunPyMOL + 87
15  org.python.python             	0x0000000107786b17 PyEval_EvalFrameEx + 9911
16  org.python.python             	0x0000000107784416 PyEval_EvalCodeEx + 1990
….

[howarth.at.macports@…]

I really doubt this is breakage in pymol itself but some other MacPorts package is at fault.

[howarth.at.macports@…]

Is there a way to ask port to install a particular previous archived version and revision of some package from binary archives? I really don't want to have to do a regression hunt through the supporting packages using manual builds.

[howarth.at.macports@…]

Actually, the recent update of the mesa package from version 7.2 to 8.0.5 seems like the most likely candidate. Few programs exercise the GL/GLX support as much as pymol.

[jeremyhu (Jeremy Huddleston Sequoia)]

I assume you mean from 8.0.4 to 8.0.5 ... It hasn't been at 7.2 since around 2009 ...

[jeremyhu (Jeremy Huddleston Sequoia)]

Along with that abort, a message should be printed to stderr. What is the message?

[mtiberti (Matteo Tiberti)]

Replying to howarth.at.macports@…:

Actually, the recent update of the mesa package from version 7.2 to 8.0.5 seems like the most likely candidate. Few programs exercise the GL/GLX support as much as pymol.

Thanks for your help! Actually pymol installs on MacPorts 2.2.1 which I am using right now (or at least it did yesterday). I have tried installing glew revision 114325 (6 months ago) but nothing changed. I'll try with the mesa package then.

[mtiberti (Matteo Tiberti)]

might be related to #43810

[jeremyhu (Jeremy Huddleston Sequoia)]

I suspect that maybe there's an issue with the 0013-darwin-Fix-test-for-kCGLPFAOpenGLProfile-support-at-.patch ... I wonder if you're getting an error other than kCGLBadAttribute for some reason.

[mtiberti (Matteo Tiberti)]

I can confirm that it works with mesa @8.0.4+py27. However, the command window doesn't show up, but I think it's unrelated

EDIT: installed tcl +threads-corefoundation as in #37062 but it still doesn't work as expected

[howarth.at.macports@…]

I get…

error: invalid pixel format attribute Abort

in the command line when I execute pymol under the current mesa.

[mtiberti (Matteo Tiberti)]

one moment... I will just upgrade outdated and see if I get it as well. I'm on 2.2.1 right now

[howarth.at.macports@…]

FYI, the top command window can take several minutes to open on the first launch. I've never gotten an explanation for that from upstream but assume that it is related to the code probing the video card. As long as you don't logout, the next time pymol is executed the top window opens immediately. Weird bug.

[mtiberti (Matteo Tiberti)]

Replying to howarth.at.macports@…:

FYI, the top command window can take several minutes to open on the first launch. I've never gotten an explanation for that from upstream but assume that it is related to the code probing the video card. As long as you don't logout, the next time pymol is executed the top window opens immediately. Weird bug.

Solved it by installing tk -quartz. I can confirm everything works with mesa@8.0.4+py27

[howarth.at.macports@…]

replacement setup_py.diff for pymol 1.7.1.1

[howarth.at.macports@…]

replacement pymol-clang.diff for pymol 1.7.1.1

[howarth.at.macports@…]

You can also use the replacement setup_py.diff and pymol-clang.diff with the Portfile.diff to update pymol to the latest stable release of 1.7.1.1. None of the more recent svn commits are related to display graphics. This is the same version that we are using in fink without issues.

[howarth.at.macports@…]

Same error…

error: invalid pixel format attribute Abort

occurs under pymol-1.7.1.1.

[jeremyhu (Jeremy Huddleston Sequoia)]

Thanks, FWIW, r120370 has better error reporting in the crash log.

error: invalid pixel format attribute Abort

That indicates that you're getting kCGLBadAttribute. I suspect the kCGLBadAttribute is because kCGLPFAOpenGLProfile is still in the list for some reason.

Someone who is reproducing this will need to provide more information or an update to 0013-darwin-Fix-test-for-kCGLPFAOpenGLProfile-support-at-.patch to address the issue.

I suggest you start by figuring out what error is when we test for the kCGLPFAOpenGLProfile support here:

   /* Request an OpenGL 3.2 profile if one is available and supported */
   attr[numattr++] = kCGLPFAOpenGLProfile;
   attr[numattr++] = kCGLOGLPVersion_3_2_Core;

   /* Test for kCGLPFAOpenGLProfile support at runtime and roll it out if not supported */
   attr[numattr] = 0;
   error = apple_cgl.choose_pixel_format(attr, pfobj, &vsref);
   apple_cgl.destroy_pixel_format(pfobj);
   if (error == kCGLBadAttribute)
      numattr -= 2;

[jeremyhu (Jeremy Huddleston Sequoia)]

Also, I assume everyone seeing this issue is on Snow Leopard.

[jeremyhu (Jeremy Huddleston Sequoia)]

Try running this and tell me what it prints:

$ clang testgl.c -framework OpenGL -o testgl && ./testgl

Where testgl.c is:

#include <stdio.h>
#include <OpenGL/OpenGL.h>

#ifndef kCGLPFAOpenGLProfile
#define kCGLPFAOpenGLProfile 99
#endif

#ifndef kCGLOGLPVersion_3_2_Core
#define kCGLOGLPVersion_3_2_Core 0x3200
#endif

int main() {
    CGLPixelFormatAttribute attr[3];
    CGLPixelFormatObj pfobj;
    GLint vsref = 0;
    CGLError error = 0;

    /* Request an OpenGL 3.2 profile if one is available and supported */
    attr[0] = kCGLPFAOpenGLProfile;
    attr[1] = kCGLOGLPVersion_3_2_Core;
    attr[2] = 0;

    error = CGLChoosePixelFormat(attr, &pfobj, &vsref);
    CGLDestroyPixelFormat(pfobj);

    printf("%ld: %s\n", (long)error, CGLErrorString(error));

    return 0;
}

[howarth.at.macports@…]

The new mesa causes pymol to emit …

CGLChoosePixelFormat error: invalid pixel format attribute Abort

[howarth.at.macports@…]

% clang testgl.c -framework OpenGL -o testgl && ./testgl 0: no error

on late 2006 MacBook Pro with Radeon X1600 graphics

[jeremyhu (Jeremy Huddleston Sequoia)]

Replying to howarth.at.macports@…:

% clang testgl.c -framework OpenGL -o testgl && ./testgl 0: no error

on late 2006 MacBook Pro with Radeon X1600 graphics

What OS version?

[jeremyhu (Jeremy Huddleston Sequoia)]

Ok, at least one case here is with 10.7, so that's odd... 10.7 knows about kCGLPFAOpenGLProfile.

[howarth.at.macports@…]

I'm on 10.7 as well

[howarth.at.macports@…]

This is Xquartz 2.7.6. Should I move to the beta?

[jeremyhu (Jeremy Huddleston Sequoia)]

Replying to howarth.at.macports@…:

This is Xquartz 2.7.6. Should I move to the beta?

This has nothing to do with your X11 server version. This is an issue inside of mesa.

Can you please break on apple_visual_create_pfobj and figure out what attributes are being requested?

[howarth.at.macports@…]

I can get to…

 lldb /opt/local/Library/Frameworks/Python.framework/Versions/2.7/bin/python2.7
Current executable set to '/opt/local/Library/Frameworks/Python.framework/Versions/2.7/bin/python2.7' (x86_64).
(lldb) b apple_visual_create_pfobj                                                                                                      Breakpoint 1: no locations (pending).
WARNING:  Unable to resolve breakpoint to any actual locations.
(lldb) r /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pymol/__init__.py -d _ set stereo_double_pump_mono,quiet=1 -M
Process 15080 launched: '/opt/local/Library/Frameworks/Python.framework/Versions/2.7/bin/python2.7' (x86_64)
Process 15080 stopped
* thread #1: tid = 0x1f03, 0x00007fff5fc01028 dyld`_dyld_start, stop reason = exec
    frame #0: 0x00007fff5fc01028 dyld`_dyld_start
dyld`_dyld_start:
-> 0x7fff5fc01028:  pushq  $0
   0x7fff5fc0102a:  movq   %rsp, %rbp
   0x7fff5fc0102d:  andq   $-16, %rsp
   0x7fff5fc01031:  movq   8(%rbp), %rdi
(lldb) 

but at this point I'm lost as to what lldb commands to use in order to extract the parameters passed to the call.

[howarth.at.macports@…]

(lldb) c
Process 15099 resuming
1 location added to breakpoint 1
Process 15099 stopped
* thread #1: tid = 0x1f03, 0x00000001006de00f libGL.1.dylib`apple_visual_create_pfobj, stop reason = breakpoint 1.1
    frame #0: 0x00000001006de00f libGL.1.dylib`apple_visual_create_pfobj
libGL.1.dylib`apple_visual_create_pfobj:
-> 0x1006de00f:  pushq  %rbp
   0x1006de010:  movq   %rsp, %rbp
   0x1006de013:  pushq  %r15
   0x1006de015:  pushq  %r14
(lldb) disassemble --frame
libGL.1.dylib`apple_visual_create_pfobj:
-> 0x1006de00f:  pushq  %rbp
   0x1006de010:  movq   %rsp, %rbp
   0x1006de013:  pushq  %r15
   0x1006de015:  pushq  %r14
   0x1006de017:  pushq  %r13
   0x1006de019:  pushq  %r12
   0x1006de01b:  pushq  %rbx
   0x1006de01c:  subq   $280, %rsp
   0x1006de023:  movl   %r8d, %r12d
   0x1006de026:  movq   %rcx, -312(%rbp)
   0x1006de02d:  movq   %rdx, -304(%rbp)
   0x1006de034:  movq   %rsi, %r13
   0x1006de037:  movq   %rdi, %r14
   0x1006de03a:  movq   335839(%rip), %rax        ; (void *)0x00007fff7ddf4f60: __stack_chk_guard
   0x1006de041:  movq   (%rax), %rax
   0x1006de044:  movq   %rax, -48(%rbp)
   0x1006de048:  movl   $0, -292(%rbp)
   0x1006de052:  movl   $99, -288(%rbp)
   0x1006de05c:  movl   $12800, -284(%rbp)
   0x1006de066:  movl   $0, -280(%rbp)
   0x1006de070:  leaq   -288(%rbp), %rdi
   0x1006de077:  leaq   -292(%rbp), %rdx
   0x1006de07e:  leaq   386331(%rip), %rbx        ; apple_cgl
   0x1006de085:  movq   %r14, %rsi
   0x1006de088:  callq  *16(%rbx)
   0x1006de08b:  movl   %eax, %r15d
   0x1006de08e:  movq   %r14, %rdi
   0x1006de091:  callq  *24(%rbx)
   0x1006de094:  cmpl   $10000, %r15d
   0x1006de09b:  setne  %al
   0x1006de09e:  movzbl %al, %ebx
   0x1006de0a1:  addl   %ebx, %ebx
   0x1006de0a3:  testb  %r12b, %r12b
   0x1006de0a6:  je     0x1006de0fd               ; apple_visual_create_pfobj + 238
   0x1006de0a8:  leaq   207700(%rip), %rsi        ; "apple_visual.c"
   0x1006de0af:  leaq   207708(%rip), %rdx        ; "apple_visual_create_pfobj"
   0x1006de0b6:  leaq   207727(%rip), %r8         ; "offscreen rendering enabled.  Using kCGLPFAOffScreen\n"
   0x1006de0bd:  movl   $7, %edi
   0x1006de0c2:  movl   $99, %ecx
   0x1006de0c7:  xorb   %al, %al
   0x1006de0c9:  callq  0x1006dd0d0               ; _apple_glx_log
   0x1006de0ce:  movl   $53, -288(%rbp,%rbx,4)
   0x1006de0d9:  orl    $1, %ebx
   0x1006de0dc:  movl   $8, -288(%rbp,%rbx,4)
   0x1006de0e7:  leal   1(%rbx), %eax
   0x1006de0ea:  movl   $32, -288(%rbp,%rax,4)
   0x1006de0f5:  addl   $2, %ebx
   0x1006de0f8:  jmpq   0x1006de198               ; apple_visual_create_pfobj + 393
   0x1006de0fd:  leaq   207710(%rip), %rdi        ; "LIBGL_ALWAYS_SOFTWARE"
   0x1006de104:  callq  0x10070fa26               ; symbol stub for: getenv
   0x1006de109:  testq  %rax, %rax
   0x1006de10c:  je     0x1006de151               ; apple_visual_create_pfobj + 322
   0x1006de10e:  leaq   207598(%rip), %rsi        ; "apple_visual.c"
   0x1006de115:  leaq   207606(%rip), %rdx        ; "apple_visual_create_pfobj"
   0x1006de11c:  leaq   207701(%rip), %r8         ; "Software rendering requested.  Using kCGLRendererGenericFloatID.\n"
   0x1006de123:  movl   $7, %edi
   0x1006de128:  movl   $107, %ecx
   0x1006de12d:  xorb   %al, %al
   0x1006de12f:  callq  0x1006dd0d0               ; _apple_glx_log
   0x1006de134:  movl   $70, -288(%rbp,%rbx,4)
   0x1006de13f:  orl    $1, %ebx
   0x1006de142:  movl   $132096, -288(%rbp,%rbx,4)
   0x1006de14d:  incl   %ebx
   0x1006de14f:  jmp    0x1006de198               ; apple_visual_create_pfobj + 393
   0x1006de151:  leaq   207714(%rip), %rdi        ; "LIBGL_ALLOW_SOFTWARE"
   0x1006de158:  callq  0x10070fa26               ; symbol stub for: getenv
   0x1006de15d:  testq  %rax, %rax
   0x1006de160:  je     0x1006de18a               ; apple_visual_create_pfobj + 379
   0x1006de162:  leaq   207514(%rip), %rsi        ; "apple_visual.c"
   0x1006de169:  leaq   207522(%rip), %rdx        ; "apple_visual_create_pfobj"
   0x1006de170:  leaq   207704(%rip), %r8         ; "Software rendering is not being excluded.  Not using kCGLPFAAccelerated.\n"
   0x1006de177:  movl   $7, %edi
   0x1006de17c:  movl   $113, %ecx
   0x1006de181:  xorb   %al, %al
   0x1006de183:  callq  0x1006dd0d0               ; _apple_glx_log
   0x1006de188:  jmp    0x1006de198               ; apple_visual_create_pfobj + 393
   0x1006de18a:  movl   $73, -288(%rbp,%rbx,4)
   0x1006de195:  orl    $1, %ebx
   0x1006de198:  movslq %ebx, %rcx
   0x1006de19b:  movl   $74, -288(%rbp,%rcx,4)
   0x1006de1a6:  incl   %ecx
   0x1006de1a8:  xorb   %al, %al
   0x1006de1aa:  cmpl   $0, 16(%r13)
   0x1006de1af:  je     0x1006de1c8               ; apple_visual_create_pfobj + 441
   0x1006de1b1:  movslq %ecx, %rcx
   0x1006de1b4:  movl   $6, -288(%rbp,%rcx,4)
   0x1006de1bf:  addl   $2, %ebx
   0x1006de1c2:  movb   $1, %dl
   0x1006de1c4:  movl   %ebx, %ecx
   0x1006de1c6:  jmp    0x1006de1ca               ; apple_visual_create_pfobj + 443
   0x1006de1c8:  xorb   %dl, %dl
   0x1006de1ca:  movq   -312(%rbp), %rsi
   0x1006de1d1:  movb   %dl, (%rsi)
   0x1006de1d3:  testb  %r12b, %r12b
   0x1006de1d6:  jne    0x1006de1f1               ; apple_visual_create_pfobj + 482
   0x1006de1d8:  cmpl   $0, 12(%r13)
   0x1006de1dd:  je     0x1006de1f1               ; apple_visual_create_pfobj + 482
   0x1006de1df:  movslq %ecx, %rcx
   0x1006de1e2:  movl   $5, -288(%rbp,%rcx,4)
   0x1006de1ed:  incl   %ecx
   0x1006de1ef:  movb   $1, %al
   0x1006de1f1:  movq   -304(%rbp), %rdx
   0x1006de1f8:  movb   %al, (%rdx)
   0x1006de1fa:  movslq %ecx, %rdx
   0x1006de1fd:  movl   $8, -288(%rbp,%rdx,4)
   0x1006de208:  movl   24(%r13), %eax
   0x1006de20c:  addl   20(%r13), %eax
   0x1006de210:  addl   28(%r13), %eax
   0x1006de214:  leal   1(%rdx), %esi
   0x1006de217:  movslq %esi, %rsi
   0x1006de21a:  movl   %eax, -288(%rbp,%rsi,4)
   0x1006de221:  leal   2(%rdx), %eax
   0x1006de224:  movslq %eax, %rax
   0x1006de227:  movl   $11, -288(%rbp,%rax,4)
   0x1006de232:  leal   4(%rdx), %eax
   0x1006de235:  addl   $3, %edx
   0x1006de238:  movslq %edx, %rdx
   0x1006de23b:  movl   32(%r13), %esi
   0x1006de23f:  movl   %esi, -288(%rbp,%rdx,4)
   0x1006de246:  movl   64(%r13), %edx
   0x1006de24a:  addl   60(%r13), %edx
   0x1006de24e:  addl   68(%r13), %edx
   0x1006de252:  testl  %edx, %edx
   0x1006de254:  jle    0x1006de27a               ; apple_visual_create_pfobj + 619
   0x1006de256:  movslq %eax, %rax
   0x1006de259:  movl   $14, -288(%rbp,%rax,4)
   0x1006de264:  addl   72(%r13), %edx
   0x1006de268:  leal   5(%rcx), %eax
   0x1006de26b:  movslq %eax, %rax
   0x1006de26e:  movl   %edx, -288(%rbp,%rax,4)
   0x1006de275:  addl   $6, %ecx
   0x1006de278:  movl   %ecx, %eax
   0x1006de27a:  movl   76(%r13), %ecx
   0x1006de27e:  testl  %ecx, %ecx
   0x1006de280:  jle    0x1006de2a0               ; apple_visual_create_pfobj + 657
   0x1006de282:  movslq %eax, %rax
   0x1006de285:  movl   $12, -288(%rbp,%rax,4)
   0x1006de290:  leal   1(%rax), %edx
   0x1006de293:  movslq %edx, %rdx
   0x1006de296:  movl   %ecx, -288(%rbp,%rdx,4)
   0x1006de29d:  addl   $2, %eax
   0x1006de2a0:  movl   80(%r13), %ecx
   0x1006de2a4:  testl  %ecx, %ecx
   0x1006de2a6:  jle    0x1006de2c6               ; apple_visual_create_pfobj + 695
   0x1006de2a8:  movslq %eax, %rax
   0x1006de2ab:  movl   $13, -288(%rbp,%rax,4)
   0x1006de2b6:  leal   1(%rax), %edx
   0x1006de2b9:  movslq %edx, %rdx
   0x1006de2bc:  movl   %ecx, -288(%rbp,%rdx,4)
   0x1006de2c3:  addl   $2, %eax
   0x1006de2c6:  movl   132(%r13), %ecx
   0x1006de2cd:  testl  %ecx, %ecx
   0x1006de2cf:  jle    0x1006de325               ; apple_visual_create_pfobj + 790
   0x1006de2d1:  movslq %eax, %rax
   0x1006de2d4:  movl   $59, -288(%rbp,%rax,4)
   0x1006de2df:  leal   1(%rax), %edx
   0x1006de2e2:  movslq %edx, %rdx
   0x1006de2e5:  movl   $55, -288(%rbp,%rdx,4)
   0x1006de2f0:  leal   2(%rax), %edx
   0x1006de2f3:  movslq %edx, %rdx
   0x1006de2f6:  movl   %ecx, -288(%rbp,%rdx,4)
   0x1006de2fd:  leal   3(%rax), %ecx
   0x1006de300:  movslq %ecx, %rcx
   0x1006de303:  movl   $56, -288(%rbp,%rcx,4)
   0x1006de30e:  leal   4(%rax), %ecx
   0x1006de311:  movslq %ecx, %rcx
   0x1006de314:  movl   136(%r13), %edx
   0x1006de31b:  movl   %edx, -288(%rbp,%rcx,4)
   0x1006de322:  addl   $5, %eax
   0x1006de325:  leaq   385652(%rip), %rbx        ; apple_cgl
   0x1006de32c:  movslq %eax, %rax
   0x1006de32f:  movl   $0, -288(%rbp,%rax,4)
   0x1006de33a:  incl   %eax
   0x1006de33c:  cmpl   $60, %eax
   0x1006de33f:  jge    0x1006de385               ; apple_visual_create_pfobj + 886
   0x1006de341:  leaq   -288(%rbp), %rdi
   0x1006de348:  leaq   -292(%rbp), %rdx
   0x1006de34f:  movq   %r14, %rsi
   0x1006de352:  callq  *16(%rbx)
   0x1006de355:  testl  %eax, %eax
   0x1006de357:  jne    0x1006de3a4               ; apple_visual_create_pfobj + 917
   0x1006de359:  cmpq   $0, (%r14)
   0x1006de35d:  je     0x1006de3d2               ; apple_visual_create_pfobj + 963
   0x1006de35f:  movq   335034(%rip), %rax        ; (void *)0x00007fff7ddf4f60: __stack_chk_guard
   0x1006de366:  movq   (%rax), %rax
   0x1006de369:  cmpq   -48(%rbp), %rax
   0x1006de36d:  jne    0x1006de40d               ; apple_visual_create_pfobj + 1022
   0x1006de373:  addq   $280, %rsp
   0x1006de37a:  popq   %rbx
   0x1006de37b:  popq   %r12
   0x1006de37d:  popq   %r13
   0x1006de37f:  popq   %r14
   0x1006de381:  popq   %r15
   0x1006de383:  popq   %rbp
   0x1006de384:  ret    
   0x1006de385:  leaq   206982(%rip), %rdi        ; "apple_visual_create_pfobj"
   0x1006de38c:  leaq   206960(%rip), %rsi        ; "apple_visual.c"
   0x1006de393:  leaq   207231(%rip), %rcx        ; "numattr < MAX_ATTR"
   0x1006de39a:  movl   $173, %edx
   0x1006de39f:  callq  0x10070f99c               ; symbol stub for: __assert_rtn
   0x1006de3a4:  movl   %eax, %edi
   0x1006de3a6:  callq  *80(%rbx)
   0x1006de3a9:  leaq   385760(%rip), %rbx        ; __crashreporter_info_buff__
   0x1006de3b0:  leaq   207221(%rip), %r8         ; "CGLChoosePixelFormat error: %s\n"
   0x1006de3b7:  movq   %rbx, %rdi
   0x1006de3ba:  movl   $4096, %esi
   0x1006de3bf:  xorl   %edx, %edx
   0x1006de3c1:  movl   $4096, %ecx
   0x1006de3c6:  movq   %rax, %r9
   0x1006de3c9:  xorb   %al, %al
   0x1006de3cb:  callq  0x10070f9a8               ; symbol stub for: __snprintf_chk
   0x1006de3d0:  jmp    0x1006de3f6               ; apple_visual_create_pfobj + 999
   0x1006de3d2:  leaq   385719(%rip), %rbx        ; __crashreporter_info_buff__
   0x1006de3d9:  leaq   207212(%rip), %r8         ; "No matching pixelformats found, perhaps try using LIBGL_ALLOW_SOFTWARE\n"
   0x1006de3e0:  movq   %rbx, %rdi
   0x1006de3e3:  movl   $4096, %esi
   0x1006de3e8:  xorl   %edx, %edx
   0x1006de3ea:  movl   $4096, %ecx
   0x1006de3ef:  xorb   %al, %al
   0x1006de3f1:  callq  0x10070f9a8               ; symbol stub for: __snprintf_chk
   0x1006de3f6:  movq   334891(%rip), %rax        ; (void *)0x00007fff7ddef630: __stderrp
   0x1006de3fd:  movq   (%rax), %rsi
   0x1006de400:  movq   %rbx, %rdi
   0x1006de403:  callq  0x10070fa14               ; symbol stub for: fputs
   0x1006de408:  callq  0x10070f9ba               ; symbol stub for: abort
   0x1006de40d:  callq  0x10070f9b4               ; symbol stub for: __stack_chk_fail
(lldb) 

[jeremyhu (Jeremy Huddleston Sequoia)]

You're missing debug symbols. You need to build with debug symbols.

In any event, r120374 reverts the change that is causing this while we further triage.

[howarth.at.macports@…]

Okay. I rebuild pymol with…

reinplace "s|\"-O3\",|\"-O3\", \"-g\",|g" ${worksrcpath}/setup.py

added to the post-patch for -g in the compiles but I get the same debug results as above.

[howarth.at.macports@…]

Portfile diff for pymol 1.7.1.1

[howarth.at.macports@…]

Can some please check in the changes in the attached Portfile.diff for pymol and the replacement setup_py.diff and pymol-clang.diff files for me? These changes will update the pymol package to the latest stable release and compile it with -g for further debugging of this problem with the reverted mesa patches. Thanks in advance.

[mf2k (Frank Schima)]

It was applied in r120407.

[mtiberti (Matteo Tiberti)]

I have installed the new version now.. everything works fine for me, with the new mesa package.