Bug#763755: blender: SIGILL on startup
On 10/10/2014 09:47 AM, Matteo F. Vescovi wrote: > > Then, feel free to provide an updated patch. I'll give it a try. Here. but it's against the .71, and looks pretty much the same as the earlier one. I don't have enough free space to dare the pbuilder thing. I also added that architecture.mk mentioned earlier. Description: as usual TODO: Put a short summary on the line above and replace this paragraph with a longer explanation of this change. Complete the meta-information with other relevant fields (see below for details). To make it easier, the information below has been extracted from the changelog. Adjust it or drop it. . blender (2.71+dfsg1-2) UNRELEASED; urgency=medium . * change rules file to set architecture flags Author: Johann Klammer --- The information above should follow the Patch Tagging Guidelines, please checkout http://dep.debian.net/deps/dep3/ to learn about the format. Here are templates for supplementary fields that you might want to add: Origin: , Bug: Bug-Debian: http://bugs.debian.org/ Bug-Ubuntu: https://launchpad.net/bugs/ Forwarded: Reviewed-By: Last-Update: --- blender-2.71+dfsg1.orig/build_files/cmake/macros.cmake +++ blender-2.71+dfsg1/build_files/cmake/macros.cmake @@ -446,34 +446,42 @@ macro(TEST_SSE_SUPPORT set(CMAKE_REQUIRED_FLAGS "${${_sse_flags}} ${${_sse2_flags}}") - if(NOT DEFINED SUPPORT_SSE_BUILD) - # result cached - check_c_source_runs(" - #include - int main(void) { __m128 v = _mm_setzero_ps(); return 0; }" - SUPPORT_SSE_BUILD) - if(SUPPORT_SSE_BUILD) - message(STATUS "SSE Support: detected.") - else() - message(STATUS "SSE Support: missing.") + if(NOT ( $ENV{DEB_HOST_ARCH_CPU} STREQUAL "i386" ) ) + + if(NOT DEFINED SUPPORT_SSE_BUILD) + # result cached + check_c_source_runs(" + #include + int main(void) { __m128 v = _mm_setzero_ps(); return 0; }" + SUPPORT_SSE_BUILD) + + if(SUPPORT_SSE_BUILD) + message(STATUS "SSE Support: detected.") + else() + message(STATUS "SSE Support: missing.") + endif() endif() - endif() - if(NOT DEFINED SUPPORT_SSE2_BUILD) - # result cached - check_c_source_runs(" - #include - int main(void) { __m128d v = _mm_setzero_pd(); return 0; }" - SUPPORT_SSE2_BUILD) + if(NOT DEFINED SUPPORT_SSE2_BUILD) + # result cached + check_c_source_runs(" + #include + int main(void) { __m128d v = _mm_setzero_pd(); return 0; }" + SUPPORT_SSE2_BUILD) - if(SUPPORT_SSE2_BUILD) - message(STATUS "SSE2 Support: detected.") - else() - message(STATUS "SSE2 Support: missing.") + if(SUPPORT_SSE2_BUILD) + message(STATUS "SSE2 Support: detected.") + else() + message(STATUS "SSE2 Support: missing.") + endif() endif() + else() + message(STATUS "SSE Support: missing.") + message(STATUS "SSE2 Support: missing.") endif() + unset(CMAKE_REQUIRED_FLAGS) endmacro() #!/usr/bin/make -f #they say I should not rely on DEB_HOST_ARCH being correctly set automagically, and do this to be sure... include /usr/share/dpkg/architecture.mk export REPACK_SH=$(CURDIR)/debian/repack.sh INSTDIR = debian/tmp PY3VERS = $(shell py3versions -dv) SRC_VER = $(shell uscan --report | grep blender-.* | sed 's/^.*-//' | sed 's/\.tar.*$$//') get-orig-source: uscan --force-download --repack --compression xz $(REPACK_SH) --upstream-version $(SRC_VER) ../blender_$(SRC_VER).orig.tar.xz override_dh_auto_configure: dh_auto_configure -- \ -DCMAKE_INSTALL_PREFIX=/usr \ -DCMAKE_SKIP_RPATH=ON \ -DCMAKE_VERBOSE_MAKEFILE=ON \ -DFREETYPE_INCLUDE_DIRS="/usr/include/freetype2" \ -DPYTHON_VERSION=$(PY3VERS) \ -DWITH_IMAGE_OPENJPEG=ON \ -DWITH_FFTW3=ON \ -DWITH_JACK=ON \ -DWITH_PLAYER=ON \ -DWITH_INSTALL_PORTABLE
Bug#763755: blender: SIGILL on startup
On 10/10/2014 08:48 AM, Matteo F. Vescovi wrote: > > No, it doesn't. Your patch wasn't applied in 2.71. > I modified it locally... > > > Oh, gotta check this either... but I've no X on my laptop at the moment. > > Thanks for reporting. > Not sure it's a bug.. it's quite possible they don't support that old stuff. ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#763755: blender: SIGILL on startup
On 10/09/2014 03:01 PM, Matteo F. Vescovi wrote: > Hi again! > > On 2014-10-09 at 14:39 (CEST), Matteo F. Vescovi wrote: >> Re-applying it with next upcoming revision. Sorry for the mess. > > FTR, test build on i386 with your patch applied on top of > 2.72 now fails [1]. > > I hope to find a solution soon. > > Cheers. > > > [1] > http://debomatic-i386.debian.net/distribution#unstable/blender/2.72+dfsg1-2/buildlog > perhaps they need a `include /usr/share/dpkg/architecture.mk` in debian/rules now? There was something about that in `man dpkg-architecture`. The .71 in testing seems to build OK with it... The menus look fishy, but that's probably this ancient NVIDIA VANTA here(no shader progs).. ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#763755: blender: SIGILL on startup
On Thu, 2 Oct 2014 14:43:30 +0200 "Matteo F. Vescovi" wrote: > Control: tag -1 + upstream > > Hi! > > On 2014-10-02 at 13:54 (CEST), Daniele Forsi wrote: > > Dear Maintainer, > > > > could Blender be changed to check if it is running on a supported CPU > > and fail with a nicely informative error message? > > At [1] you can see that the former patch that used to check about the > compatibility of the CPU with Blender has been dropped since upstream > developers declared that a better (and elegant) solution was found and > that patch wasn't useful anymore. Hi, Do you have a URL to that statement? Because if they still build their x86 package inside an i386 emulator like before, that thing is not an actual bug upstream and will thus /never/ get fixed. Bye > > > On my PC blender is terminated on startup with: > > Program received signal SIGILL, Illegal instruction. > > Well, probably the upstream approach hasn't been efficient enough to > prevent all the cases. I already got in touch with them and we'll see > how to solve this long-persecuting issue. > > Cheers. > > > [1] > http://anonscm.debian.org/cgit/pkg-multimedia/blender.git/commit/?id=fce9104b648ad2d81fc82362eb9f377c6eb5641d > > -- > Matteo F. Vescovi ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#632080: vlc: VLC cannot open mpeg (TS) files recorded from satellite channels
I have a .ts snippet, which does not display anything here. Works in xine. Was part of a larger file. vlc would just grind the disk, but not display anything. http://members.aon.at/~aklamme4/scratch/hang.ts ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#665732: reopen: vlc: GUI latency/unresponsiveness to clicks on control options
Rémi Denis-Courmont wrote: tags 665732 = moreinfo help unreproducible thanks Hello, Le mardi 26 février 2013 17:33:21, Johann Klammer a écrit : Adding a usleep(1) after the last aout_unlock() in aout_DecPlay() in src/audio_output/dec.c solves the problem for me. Unfortunately that does not really help pinpoint the problem. It is likely a race, but it could be anywhere. Without a stack trace of the freeze or a way to reproduce the problem, there is nothing I, as a VLC developer, can do about this bug report. Could only find those two points where the audio output lock was taken... Changing volume (with ALSA) exhibits a noticeable delay of a few tenth of a second, due to buffering in the ALSA driver. That is neither new and nor fixable. Several seconds of delay is abnormal. it's more than just seconds(perhaps because it's single core box)... Lockup can be broken by doing stuff to shake up the scheduling... switching console<->xorg, disk activity etc.. While a kernel bug cannot be completely ruled out, a bug in libpthread seems more probable. With that in mind, it would be good to know if the bug reproduces with *unoptimized* libpthread builds (on i386, this can be achieved by removing the libc6-i686 package). I am using the i386 versions of everything... Starting the gdb while it's locked up shows that it blocks in ___lll_lock_wait in eglibc's lowlevellock.S... the i486 variant...(there's no i386 because of xchg insn) ...also it's the same for i686(which just includes the 486 version)... And there's a __GI_poll done somewhere inside libasound. ...it sleeps there... Yes, the sched_yield() does the trick, too. ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#665732: reopen: vlc: GUI latency/unresponsiveness to clicks on control options
Adding a usleep(1) after the last aout_unlock() in aout_DecPlay() in src/audio_output/dec.c solves the problem for me. It seems that the GUI thread fails to get woken up after the mutex here is unlocked Not sure if this is a kernel bug or just quality of implementation... ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#687735: Testcase for demuxer/psi bug
Fabian Greffrath wrote: Please note that the bug got tagged "upstream" and "moreinfo", not "wontfix". Sorry, Should have been more precise. I was ranting about #665732(not submitted by me, but affects me too) in the second paragraph. ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#687735: Testcase for demuxer/psi bug
There: http://members.aon.at/~aklamme4/scratch/test.ts File was generated using rtpdump, Switching from TELE5 to ProSieben(on different mux). The .ts file crashes vlc at the point of transition. I find it hard to believe that you could not reproduce as it always, reliably happens here when switching between muliplexes, regardless of multiplex used. Unrelated: I have also experienced the delay when changing volume, and it is most severe, not just a small delay but up to several minutes. I am most extremely pissed that you have tagged it wontfix. I have found myself stopping the stream just to change volume and restarting it afterwards. Otherwise the whole UI freezes. I am using Alsa output. May I ask, WHAT THE FUCK ARE YOU SMOKING? ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#687735: VLC crashes when playing back TS-over-RTP and transport stream changes
Package: vlc Version: 2.0.3-2 Hello, I recently upgraded vlc from 1.1.13-1+b1 to 2.0.3-2. I am using it to play back rtp:// streams forwarded by this software: https://github.com/klammerj/dvb-vulture AFAIK The stream contains a valid PCR... When the stream composition changes (by tuning to different service..) the vlc segfaults. It did not happen with previously installed version. The vlc log and a gdb backtrace are attached. regards, JK [0x9b8b908] main libvlc debug: translation test: code is "C" [0x9b8b908] main libvlc debug: CPU has capabilities MMX 3DNow! MMXEXT SSE FPU [0x9b8b908] main libvlc debug: looking for memcpy module: 4 candidates [0x9b8b908] main libvlc debug: using memcpy module "memcpymmxext" [0x9d0ddf8] main input debug: Creating an input for 'Media Library' [0x9d0ddf8] main input debug: Input is a meta file: disabling unneeded options [0x9d0ddf8] main input debug: using timeshift granularity of 50 MiB, in path '/tmp' [0x9d0ddf8] main input debug: `file/xspf-open:///home/klammerj/.local/share/vlc/ml.xspf' gives access `file' demux `xspf-open' path `/home/klammerj/.local/share/vlc/ml.xspf' [0x9d0ddf8] main input debug: creating demux: access='file' demux='xspf-open' location='/home/klammerj/.local/share/vlc/ml.xspf' file='/home/klammerj/.local/share/vlc/ml.xspf' [0x9bad378] main demux debug: looking for access_demux module: 3 candidates [0x9bad378] main demux debug: no access_demux module matching "file" could be loaded [0x9bad378] main demux debug: TIMER module_need() : 6.822 ms - Total 6.822 ms / 1 intvls (Avg 6.822 ms) [0x9d0ddf8] main input debug: creating access 'file' location='/home/klammerj/.local/share/vlc/ml.xspf', path='/home/klammerj/.local/share/vlc/ml.xspf' [0x9b9d510] main access debug: looking for access module: 2 candidates [0x9b9d510] filesystem access debug: opening file `/home/klammerj/.local/share/vlc/ml.xspf' [0x9b9d510] main access debug: using access module "filesystem" [0x9b9d510] main access debug: TIMER module_need() : 2.632 ms - Total 2.632 ms / 1 intvls (Avg 2.632 ms) [0x9b9e468] main stream debug: Using stream method for AStream* [0x9b9e468] main stream debug: starting pre-buffering [0x9b9e468] main stream debug: received first data after 0 ms [0x9b9e468] main stream debug: pre-buffering done 296 bytes in 0s - 9967 KiB/s [0x9b9dce0] main stream debug: looking for stream_filter module: 7 candidates [0x9b9dce0] main stream debug: no stream_filter module matching "any" could be loaded [0x9b9dce0] main stream debug: TIMER module_need() : 6.646 ms - Total 6.646 ms / 1 intvls (Avg 6.646 ms) [0x9b9dce0] main stream debug: looking for stream_filter module: 1 candidate [0x9b9dce0] main stream debug: using stream_filter module "stream_filter_record" [0x9b9dce0] main stream debug: TIMER module_need() : 1.141 ms - Total 1.141 ms / 1 intvls (Avg 1.141 ms) [0x9d0ddf8] main input debug: creating demux: access='file' demux='xspf-open' location='/home/klammerj/.local/share/vlc/ml.xspf' file='/home/klammerj/.local/share/vlc/ml.xspf' [0x9b9c680] main demux debug: looking for demux module: 1 candidate [0x9b9c680] playlist demux debug: using XSPF playlist reader [0x9b9c680] main demux debug: using demux module "playlist" [0x9b9c680] main demux debug: TIMER module_need() : 1.697 ms - Total 1.697 ms / 1 intvls (Avg 1.697 ms) [0x9b9e528] main demux meta debug: looking for meta reader module: 2 candidates [0x9b9e528] lua demux meta debug: Trying Lua scripts in /home/klammerj/.local/share/vlc/lua/meta/reader [0x9b9e528] lua demux meta debug: Trying Lua scripts in /usr/lib/vlc/lua/meta/reader [0x9b9e528] lua demux meta debug: Trying Lua playlist script /usr/lib/vlc/lua/meta/reader/filename.luac [0x9b9e528] lua demux meta debug: Trying Lua scripts in /usr/share/vlc/lua/meta/reader [0x9b9e528] main demux meta debug: no meta reader module matching "any" could be loaded [0x9b9e528] main demux meta debug: TIMER module_need() : 11.190 ms - Total 11.190 ms / 1 intvls (Avg 11.190 ms) [0x9d0ddf8] main input debug: `file/xspf-open:///home/klammerj/.local/share/vlc/ml.xspf' successfully opened [0x9cf4368] main xml reader debug: looking for xml reader module: 1 candidate [0x9cf4368] main xml reader debug: using xml reader module "xml" [0x9cf4368] main xml reader debug: TIMER module_need() : 5.141 ms - Total 5.141 ms / 1 intvls (Avg 5.141 ms) [0x9b9c680] playlist demux debug: parsed 0 tracks successfully [0x9d0ddf8] main input debug: EOF reached [0x9b9c680] main demux debug: removing module "playlist" [0x9b9dce0] main stream debug: removing module "stream_filter_record" [0x9b9d510] main access debug: removing module "filesystem" [0x9d0ddf8] main input debug: TIMER input launching for 'Media Library' : 32.272 ms - Total 32.272 ms / 1 intvls (Avg 32.272 ms) [0x9c15778] main playlist debug: adding services_discovery sap... [0x9b9dc00] main services discovery debug: looking for services_discovery module: 1 candidate [0x9b9dc00] main services discovery debug: using s
Bug#599680: conditional SSE/2 patch
This one is a bit better, as it leaves SSE/2 enabled on targets that support it. Description: Conditionalize SSE and SSE2 support This makes sure it builds compatible binaries also on machines which have sse/sse2. Fixes the SIGILL bug. Checks for DEB_HOST_ARCH_CPU!=i386 before tests are attempted. Will try to use SSE/SSE2 on all other architectures. Generated Code will still need MMX support to run. Bug-Debian: http://bugs.debian.org/599680 --- blender-2.63a.orig/build_files/cmake/macros.cmake +++ blender-2.63a/build_files/cmake/macros.cmake @@ -358,35 +358,40 @@ macro(TEST_SSE_SUPPORT endif() set(CMAKE_REQUIRED_FLAGS "${${_sse_flags}} ${${_sse2_flags}}") - - if(NOT DEFINED SUPPORT_SSE_BUILD) - # result cached - check_c_source_runs(" - #include - int main(void) { __m128 v = _mm_setzero_ps(); return 0; }" - SUPPORT_SSE_BUILD) - - if(SUPPORT_SSE_BUILD) - message(STATUS "SSE Support: detected.") - else() - message(STATUS "SSE Support: missing.") + + if(NOT ( $ENV{DEB_HOST_ARCH_CPU} STREQUAL "i386" ) ) + + if(NOT DEFINED SUPPORT_SSE_BUILD) + # result cached + check_c_source_runs(" +#include +int main(void) { __m128 v = _mm_setzero_ps(); return 0; }" + SUPPORT_SSE_BUILD) + + if(SUPPORT_SSE_BUILD) +message(STATUS "SSE Support: detected.") + else() +message(STATUS "SSE Support: missing.") + endif() endif() - endif() - - if(NOT DEFINED SUPPORT_SSE2_BUILD) - # result cached - check_c_source_runs(" + + if(NOT DEFINED SUPPORT_SSE2_BUILD) + # result cached + check_c_source_runs(" #include - int main(void) { __m128d v = _mm_setzero_pd(); return 0; }" - SUPPORT_SSE2_BUILD) +int main(void) { __m128d v = _mm_setzero_pd(); return 0; }" + SUPPORT_SSE2_BUILD) - if(SUPPORT_SSE2_BUILD) - message(STATUS "SSE2 Support: detected.") - else() - message(STATUS "SSE2 Support: missing.") + if(SUPPORT_SSE2_BUILD) +message(STATUS "SSE2 Support: detected.") + else() +message(STATUS "SSE2 Support: missing.") + endif() endif() + else() + message(STATUS "SSE Support: missing.") + message(STATUS "SSE2 Support: missing.") endif() - unset(CMAKE_REQUIRED_FLAGS) endmacro() ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#599680: blender: crashes with: illegal hardware instruction
Arrgh, Just tried Version 2.63a-1 .deb and it _still_ fails with SIGILL. The blender bugzilla/wiki[1][2][3] mentions that their binary build is done on a virtual machine and build flags get set automatically... Obviously this will not work for debianized builds, as (sub)architectures vary, but build should be to match lowest common subset. I' ve appended a patch that modifies one of the build scripts, to disable SSE/SSE2 detection. I do not know for sure if it works, as I do not have the appropriate hardware, but if you apply it and it gets build, it should remove the '-msse' '-msse2' and '-D__SSE__' '-D__SSE2__' args to the compiler... This would be visible in the buildd logs... Thank you. [1]http://projects.blender.org/tracker/?group_id=9&atid=498&func=detail&aid=23257 [2]http://projects.blender.org/tracker/?group_id=9&atid=498&func=detail&aid=27167 [3]http://wiki.blender.org/index.php/User:Nazg-gul/LinuxBuildbotEnvironment Description: Removes SSE and SSE2 support This makes sure it builds compatible binaries also on machines which have sse/sse2. Fixes the SIGILL bug. . --- blender-2.63a.orig/build_files/cmake/macros.cmake +++ blender-2.63a/build_files/cmake/macros.cmake @@ -359,33 +359,9 @@ macro(TEST_SSE_SUPPORT set(CMAKE_REQUIRED_FLAGS "${${_sse_flags}} ${${_sse2_flags}}") - if(NOT DEFINED SUPPORT_SSE_BUILD) - # result cached - check_c_source_runs(" - #include - int main(void) { __m128 v = _mm_setzero_ps(); return 0; }" - SUPPORT_SSE_BUILD) - - if(SUPPORT_SSE_BUILD) - message(STATUS "SSE Support: detected.") - else() - message(STATUS "SSE Support: missing.") - endif() - endif() + message(STATUS "SSE Support: unused.") - if(NOT DEFINED SUPPORT_SSE2_BUILD) - # result cached - check_c_source_runs(" - #include - int main(void) { __m128d v = _mm_setzero_pd(); return 0; }" - SUPPORT_SSE2_BUILD) - - if(SUPPORT_SSE2_BUILD) - message(STATUS "SSE2 Support: detected.") - else() - message(STATUS "SSE2 Support: missing.") - endif() - endif() + message(STATUS "SSE2 Support: unused.") unset(CMAKE_REQUIRED_FLAGS) endmacro() ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers
Bug#599680: blender: crashes with: illegal hardware instruction
Just tried the 2.62-1 version .deb and it still fails with SIGILL The official build from the blender homepage works well, though. ___ pkg-multimedia-maintainers mailing list pkg-multimedia-maintainers@lists.alioth.debian.org http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-multimedia-maintainers