Re: [E-devel] Segfault when starting e17 (evas module)
On Tue, 21 Jun 2011 15:28:01 +0200 Svante Signell svante.sign...@telia.com said: On Sat, 2011-06-18 at 08:42 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 16:38:48 +0200 Svante Signell svante.sign...@telia.com said: ... A full backtrace, see http://paste.debian.net/120170/ shows that something goes wrong within the function scale_rgba_in_to_out_clip_sample_internal() #0: evas_scale_sample.c:284 at scale_rgba_in_to_out_clip_sample_internal() Arguments: src_region_h=80, dst_region_h=768 Local variables: src_h = 31439232, dst_h = 9584648 Totally wrong and yes - totally wrong. also there is code to protect against nutty dimensions like this... so i'm wondering if this is stack corruption? Thank you for your help so far. I have now compiled and installed evas 1.0.1 and eina 1.0.1, the problem remains. How to debug a stack corruption? Valgrind is unfortunately not available. well stack corruption happens because of a bug. the only way is either valgrind (even that can only help a bit for tack corruption), OR... you literally trap, and step bit by bit until u get to the known bug point and print stacks along the way until it goes weird. no valgrind makes your life a lot harder. -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com said: On Fri, 2011-06-17 at 12:25 +0900, Carsten Haitzler wrote: On Wed, 15 Jun 2011 11:35:07 +0200 Svante Signell svante.sign...@telia.com said: ... Loks looks the problems are in evas_image_scalecache.c:776 in function evas_common_rgba_image_scalecache_do(). First thought was that since SSE/MMX was enabled, and GNU/Hurd is compiled for i486, it could be the problem, but it was not. I'm running Hurd under qemu-kvm. ... could you try valgrind and get a more detailed print of the line it died at? it makes little sense dying where it did. Thank you for your reply. Unfortunately valgrind is not available for GNU/Hurd. It is on the high-priority list though. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 17 Jun 2011 11:39:29 +0200 Svante Signell svante.sign...@telia.com said: On Fri, 2011-06-17 at 16:35 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com .. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. When compiling eina with gcc-54.6 I get one test failure: 98%: Checks: 88, Failures: 1, Errors: 0 eina_test_main.c:51:F:Main:eina_hamster:0: Failure 'eina_hamster_count() = 0' occured it's a joke. why the hell did someone put that in a test? -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 17 Jun 2011 18:53:54 +0900 Carsten Haitzler (The Rasterman) ras...@rasterman.com wrote: On Fri, 17 Jun 2011 11:39:29 +0200 Svante Signell svante.sign...@telia.com said: On Fri, 2011-06-17 at 16:35 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com .. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. When compiling eina with gcc-54.6 I get one test failure: 98%: Checks: 88, Failures: 1, Errors: 0 eina_test_main.c:51:F:Main:eina_hamster:0: Failure 'eina_hamster_count() = 0' occured it's a joke. why the hell did someone put that in a test? They REALLY like hamsters? -- A big old stinking pile of genius that no one wants coz there are too many silver coated monkeys in the world. signature.asc Description: PGP signature -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 17 Jun 2011 16:38:48 +0200 Svante Signell svante.sign...@telia.com said: On Fri, 2011-06-17 at 16:35 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com said: .. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. I have now compiled all EFL libraries with gcc-4.6, the problem remains. A full backtrace, see http://paste.debian.net/120170/ shows that something goes wrong within the function scale_rgba_in_to_out_clip_sample_internal() #0: evas_scale_sample.c:284 at scale_rgba_in_to_out_clip_sample_internal() Arguments: src_region_h=80, dst_region_h=768 Local variables: src_h = 31439232, dst_h = 9584648 Totally wrong #1: evas_image_scalecache.c:776 at evas_common_rgba_image_scalecache_do() Arguments: src_region_h=800, dst_region_h=768 #2: evas_engine.c:491 at eng_image_draw () Arguments: src_h=800,dst_h=768 Any hints how to check further, or should I upgrade to the 1.0.1 versions of EFL? upgrade -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 2011-06-17 at 16:35 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com .. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. When compiling eina with gcc-54.6 I get one test failure: 98%: Checks: 88, Failures: 1, Errors: 0 eina_test_main.c:51:F:Main:eina_hamster:0: Failure 'eina_hamster_count() = 0' occured What does this test do? -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 2011-06-17 at 16:35 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com said: .. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. I have now compiled all EFL libraries with gcc-4.6, the problem remains. A full backtrace, see http://paste.debian.net/120170/ shows that something goes wrong within the function scale_rgba_in_to_out_clip_sample_internal() #0: evas_scale_sample.c:284 at scale_rgba_in_to_out_clip_sample_internal() Arguments: src_region_h=80, dst_region_h=768 Local variables: src_h = 31439232, dst_h = 9584648 Totally wrong #1: evas_image_scalecache.c:776 at evas_common_rgba_image_scalecache_do() Arguments: src_region_h=800, dst_region_h=768 #2: evas_engine.c:491 at eng_image_draw () Arguments: src_h=800,dst_h=768 Any hints how to check further, or should I upgrade to the 1.0.1 versions of EFL? -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 2011-06-17 at 18:53 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 11:39:29 +0200 Svante Signell svante.sign...@telia.com said: When compiling eina with gcc-4.6 I get one test failure: 98%: Checks: 88, Failures: 1, Errors: 0 eina_test_main.c:51:F:Main:eina_hamster:0: Failure 'eina_hamster_count() = 0' occured it's a joke. why the hell did someone put that in a test? Problem solved: I had LANG set to en_GB.UTF-8. Changing it to C made the test pass. -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 2011-06-17 at 12:25 +0900, Carsten Haitzler wrote: On Wed, 15 Jun 2011 11:35:07 +0200 Svante Signell svante.sign...@telia.com said: ... Loks looks the problems are in evas_image_scalecache.c:776 in function evas_common_rgba_image_scalecache_do(). First thought was that since SSE/MMX was enabled, and GNU/Hurd is compiled for i486, it could be the problem, but it was not. I'm running Hurd under qemu-kvm. ... could you try valgrind and get a more detailed print of the line it died at? it makes little sense dying where it did. Thank you for your reply. Unfortunately valgrind is not available for GNU/Hurd. It is on the high-priority list though. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Thu, 2011-06-16 at 22:51 +0200, Vincent Torri wrote: On Wed, 15 Jun 2011, Svante Signell wrote: Hi, I'm compiling e17 on Debian GNU/Hurd and it compiles now with some recent hurd/libc changes. However when starting it segfaults, seems to be in the evas module. All EFL versions in Debian are 1.0.0. Is there any known issue since then, now most libs are 1.0.1? I have a backtrace in case you are interested, see http://paste.debian.net/119789/ How to best debug e17? for debugging e17: http://trac.enlightenment.org/e/wiki/Debugging Thank you for your link to the debugging page. Attaching to the crashed e17 enables you to get a bt from the crash, but not do single-stepping, etc in gdb. However, I got a tip from Manuel Menal, who had e17 running in GNU/Hurd some time ago. In order to do single stepping in gdb you could do the following: Start X with only an xterm xinit /usr/bin/xterm Then run e17 from that xterm: export E_START=enlightenment_start ; gdb enlightenment This worked for me an I found out that the crash happened at: evas-1.0.0/src/lib/engines/common/evas_scale_sample.c:284 DATA32 *ptr, *dst_ptr, *src_data, *dst_data; ... { 280: dst_ptr = dptr; for (x = 0; x dst_clip_w; x++) dst_clip_w = 36 { ptr = row_ptr[y] + lin_ptr[x]; ptr= (DATA32 *) 0x8306bf8 284: *dst_ptr = *ptr; *ptr= 4284900966 - crash here dst_ptr++; } } dptr += dst_w; *dptr= 4284900966 } } Are these pointer values reasonable? The seem very large to me. -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Fri, 17 Jun 2011 16:38:48 +0200 Svante Signell svante.sign...@telia.com said: On Fri, 2011-06-17 at 16:35 +0900, Carsten Haitzler wrote: On Fri, 17 Jun 2011 08:20:56 +0200 Svante Signell svante.sign...@telia.com said: .. Can different compilers have anything to do with the problems? eet-1.4.0 and eina-1.0.0 have been built with gcc-4.4 by the buildd while e17-0.16.999.55225, ecore-1.0.0, edbus-1.0.0, edje-1.0.0, efreet-1.0.0, evas-1.0.0 all have been built by me with gcc-4.6. I think all other depending libraries also have been built with gcc-4.4. used 4.4 and using 4.5 on linux with no problems. so i think not. I have now compiled all EFL libraries with gcc-4.6, the problem remains. A full backtrace, see http://paste.debian.net/120170/ shows that something goes wrong within the function scale_rgba_in_to_out_clip_sample_internal() #0: evas_scale_sample.c:284 at scale_rgba_in_to_out_clip_sample_internal() Arguments: src_region_h=80, dst_region_h=768 Local variables: src_h = 31439232, dst_h = 9584648 Totally wrong and yes - totally wrong. also there is code to protect against nutty dimensions like this... so i'm wondering if this is stack corruption? #1: evas_image_scalecache.c:776 at evas_common_rgba_image_scalecache_do() Arguments: src_region_h=800, dst_region_h=768 #2: evas_engine.c:491 at eng_image_draw () Arguments: src_h=800,dst_h=768 Any hints how to check further, or should I upgrade to the 1.0.1 versions of EFL? -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
[E-devel] Segfault when starting e17 (evas module)
Hi, I'm compiling e17 on Debian GNU/Hurd and it compiles now with some recent hurd/libc changes. However when starting it segfaults, seems to be in the evas module. All EFL versions in Debian are 1.0.0. Is there any known issue since then, now most libs are 1.0.1? I have a backtrace in case you are interested, see http://paste.debian.net/119789/ How to best debug e17? Loks looks the problems are in evas_image_scalecache.c:776 in function evas_common_rgba_image_scalecache_do(). First thought was that since SSE/MMX was enabled, and GNU/Hurd is compiled for i486, it could be the problem, but it was not. I'm running Hurd under qemu-kvm. I tried IRC but got no reply, and had to leave. Are you keeping IRC logs somewhere? Thanks! -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Wed, 15 Jun 2011, Svante Signell wrote: Hi, I'm compiling e17 on Debian GNU/Hurd and it compiles now with some recent hurd/libc changes. However when starting it segfaults, seems to be in the evas module. All EFL versions in Debian are 1.0.0. Is there any known issue since then, now most libs are 1.0.1? I have a backtrace in case you are interested, see http://paste.debian.net/119789/ How to best debug e17? for debugging e17: http://trac.enlightenment.org/e/wiki/Debugging Vincent Loks looks the problems are in evas_image_scalecache.c:776 in function evas_common_rgba_image_scalecache_do(). First thought was that since SSE/MMX was enabled, and GNU/Hurd is compiled for i486, it could be the problem, but it was not. I'm running Hurd under qemu-kvm. I tried IRC but got no reply, and had to leave. Are you keeping IRC logs somewhere? -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
Re: [E-devel] Segfault when starting e17 (evas module)
On Wed, 15 Jun 2011 11:35:07 +0200 Svante Signell svante.sign...@telia.com said: Hi, I'm compiling e17 on Debian GNU/Hurd and it compiles now with some recent hurd/libc changes. However when starting it segfaults, seems to be in the evas module. All EFL versions in Debian are 1.0.0. Is there any known issue since then, now most libs are 1.0.1? I have a backtrace in case you are interested, see http://paste.debian.net/119789/ How to best debug e17? Loks looks the problems are in evas_image_scalecache.c:776 in function evas_common_rgba_image_scalecache_do(). First thought was that since SSE/MMX was enabled, and GNU/Hurd is compiled for i486, it could be the problem, but it was not. I'm running Hurd under qemu-kvm. I tried IRC but got no reply, and had to leave. Are you keeping IRC logs somewhere? could you try valgrind and get a more detailed print of the line it died at? it makes little sense dying where it did. -- - Codito, ergo sum - I code, therefore I am -- The Rasterman (Carsten Haitzler)ras...@rasterman.com -- EditLive Enterprise is the world's most technically advanced content authoring tool. Experience the power of Track Changes, Inline Image Editing and ensure content is compliant with Accessibility Checking. http://p.sf.net/sfu/ephox-dev2dev ___ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel