I finally managed to install ubuntu on my new laptop and am using bumblebee to access my 525m nvidia card. I can successfully build my own blender again now (me does happy dance) & with gpu support but if I run blender under bumblebee I get an instant crash the moment I do anything in the new tile compositor:
Dell inspiron 17r n7110
2nd Generation Intel® Core™ i5-2450M processor (2.50 GHz with Turbo Boost 2.0 up to 3.10 GHz)
6GB Dual Channel DDR3 1333MHz
vid cards:
Intel® HD Graphics 3000
NVIDIA® GeForce® GT 525M (128-bit) 2GB
default blend file (cube, BI render)
./blender-> works but no acceleration
optirun ./blender-> 525m shows up for cuda so card is active. If I go directly to compositor from the splash screen and either delete render node or try to add any new node ->instant crash. If I go to debug menu and go to old compositor (debug 200) first, no crash. This is not just with my build, it also happens with this http://www.graphicall.org/834 build. 2.63a works great of course.
I now have compiled a debug build and here is the backtrace:
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffd5ea2700 (LWP 8345)]
0x00007ffff4679dba in _IO_vfprintf_internal (s=<optimized out>,
format=<optimized out>, ap=<optimized out>) at vfprintf.c:1624
1624 vfprintf.c: No such file or directory.
(gdb) bt #0 0x00007ffff4679dba in _IO_vfprintf_internal (s=<optimized out>,
format=<optimized out>, ap=<optimized out>) at vfprintf.c:1624 #1 0x00007ffff46826a9 in __printf (format=<optimized out>) at printf.c:35 #2 0x0000000001de08ae in WorkScheduler::initialize ()
at source/blender/compositor/intern/COM_WorkScheduler.cpp:238 #3 0x0000000001dd6197 in COM_execute (editingtree=0x6e99128, rendering=0)
at source/blender/compositor/intern/COM_compositor.cpp:41 #4 0x0000000001d527c0 in ntreeCompositExecTree (ntree=0x6e99128,
rd=0x5b0b8a0, rendering=0, do_preview=1)
at source/blender/nodes/composite/node_composite_tree.c:686 #5 0x00000000016d3830 in compo_startjob (cjv=0x6f1e8d8, stop=0x6e8979c,
do_update=0x6e8979a, progress=0x6e897a0)
at source/blender/editors/space_node/node_edit.c:188 #6 0x0000000001406988 in do_job_thread (job_v=0x6e89728)
at source/blender/windowmanager/intern/wm_jobs.c:259 #7 0x000000000208656b in tslot_thread_start (tslot_p=0x74271c8)
at source/blender/blenlib/intern/threads.c:217 #8 0x00007ffff3f1ce9a in start_thread (arg=0x7fffd5ea2700)
at pthread_create.c:308 #9 0x00007ffff47224bd in clone ()
at …/sysdeps/unix/sysv/linux/x86_64/clone.S:112
—Type <return> to continue, or q <return> to quit— #10 0x0000000000000000 in ?? ()
No, I haven’t filed a bug report but I want to make sure it’s not just something I have screwed up first (check my sig)
edit: also instantly crashes on loading 2.63a files that use the compositor
Report often they don’t care if its is your fault, they will just close the bug. Plus tile is undergoing a lot of change so getting all the issues with it reported as quickly as possible is helpful
Got no response on #blendercoders (apparently main devs are smart enough to avoid optimus machines) so searched tracker… #31562 seems to be related so I just added my 2 cents worth there
edit: as are most people here since no one else seems to have this prob (or they just stick with windows)
I was using r47055 but just rebuilt to r47081 -> still crashes-> same backtrace, Yeah, guess that means it’s prob related to my cheap 525m card. I guess I need to check my drivers next although I did assume that I had the latest since this is a new clean install of ubuntu (less than a week)
Thanks for taking the time to check it out and getting back to me
Thinking it may be something wrong with my build and the build on graphicall that I tried, I tried todays build from the buildbot and got the same crash with the same readout with gdb that I got with my non-debug build. I am surprised that the buildbot doesn’t offer debug builds…
parprm, 47180 should have fixed the crash. It’s strange that you still get it though I smell some misconfiguration on your system. bumblebee is, unfortunately, a bit unstable recently and developers do not respond to bug reports :/. In any case you should try a clean build of trunk and comment on that bug report if that doesn’t work.
Posted on bug report and now here so any linux/bumblebee people who build blender can be forewarned:
47180 probably fixed it here too… last night I noted that I couldn’t understand why my tomato build worked but trunk still crashed (both r47198) but when I tried again with r47246 and it crashed yet again(same bt), I suddenly realized that I hadn’t bothered to make a tomato debug build. rebuilt trunk without debug and all is good… no crash and CLERROR[-1001]: Power Ballad.
guess I won’t be bother making debug builds on my laptop anymore.
Thanks go to Jeroen Bakker & Monique Dewanchand for fixing this so quickly… ya gotta love the blender Devs, where else can you get a bug fixed in just a few days?
BTW: what does CLERROR[-1001]: Power Ballad signify?
(or they just stick with windows)