Thread overview | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
|
July 15, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #1 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> --- Here's what I get in GDB: Program received signal SIGSEGV, Segmentation fault. __memset_sse2 () at ../sysdeps/x86_64/multiarch/../memset.S:80 80 ../sysdeps/x86_64/multiarch/../memset.S: No such file or directory. (gdb) bt #0 __memset_sse2 () at ../sysdeps/x86_64/multiarch/../memset.S:80 #1 0x0000000000469120 in _d_arraysetlengthT (ti=<optimized out>, newlength=newlength@entry=2226, p=p@entry=0x7fffffffd7c0) at ../../../../src/libphobos/libdruntime/rt/lifetime.d:1617 #2 0x00000000004053ce in D main (args=...) at fuzzalloc.d:25 -- You are receiving this mail because: You are watching all bug changes. |
July 15, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #2 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> --- By the way, here's my version. $ gdc --version gdc (Debian 5.4.0-6) 5.4.0 20160609 Copyright (C) 2015 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. I'm unable to reproduce the issue with a manually compiled gdc+phobos from git ("gdc-5" branch) ; it might depend on build options, or it might already be fixed. (BTW does anybody knows how to tell the git version which was used to build a debian gdc?) -- You are receiving this mail because: You are watching all bug changes. |
July 15, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #3 from Iain Buclaw <ibuclaw@gdcproject.org> --- I can't reproduce it in master, which is using the next version of druntime compared to what is shipped in Debian. Is the segfault consistently in that location? -- You are receiving this mail because: You are watching all bug changes. |
July 15, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #4 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> --- Yes, it seems so. (gdb) run Starting program: /home/ace/projects/fuzzing/crashed/fuzzalloc [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".0 100000 200000 Program received signal SIGSEGV, Segmentation fault. __memset_sse2 () at ../sysdeps/x86_64/multiarch/../memset.S:80 80 ../sysdeps/x86_64/multiarch/../memset.S: No such file or directory. (gdb) bt #0 __memset_sse2 () at ../sysdeps/x86_64/multiarch/../memset.S:80 #1 0x0000000000469120 in _d_arraysetlengthT (ti=<optimized out>, newlength=newlength@entry=8760, p=p@entry=0x7fffffffbd70) at ../../../../src/libphobos/libdruntime/rt/lifetime.d:1617 #2 0x00000000004053ce in D main (args=...) at fuzzalloc.d:25 (gdb) run Starting program: /home/ace/projects/fuzzing/crashed/fuzzalloc [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".0 100000 200000 300000 400000 500000 600000 Program received signal SIGSEGV, Segmentation fault. __memset_sse2 () at ../sysdeps/x86_64/multiarch/../memset.S:80 80 ../sysdeps/x86_64/multiarch/../memset.S: No such file or directory. (gdb) bt #0 __memset_sse2 () at ../sysdeps/x86_64/multiarch/../memset.S:80 #1 0x0000000000469120 in _d_arraysetlengthT (ti=<optimized out>, newlength=newlength@entry=2186, p=p@entry=0x7fffffffd580) at ../../../../src/libphobos/libdruntime/rt/lifetime.d:1617 #2 0x00000000004053ce in D main (args=...) at fuzzalloc.d:25 -- You are receiving this mail because: You are watching all bug changes. |
August 22, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #5 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> --- I was hoping recent phobos updates that made their way in debian-gdc would solve these issues. They don't ; I can still reproduce this crash with recent debian-gdc compilers. It's not deterministic, sometimes it doesn't seem to happen at all ; sometimes it gets killed by OOM killer ... By the way, we might have found a workaround: by rounding all of our allocations to the next multiple of 64, it seems not to happen. $ gdc fuzzalloc.d -o fuzzalloc && ./fuzzalloc 12345 0 100000 200000 300000 400000 500000 [1] 9162 segmentation fault ./fuzzalloc 12345 $ gdc fuzzalloc.d -o fuzzalloc && ./fuzzalloc 12345 0 100000 200000 300000 400000 500000 600000 700000 800000 900000 1000000 1100000 1200000 1300000 1400000 1500000 1600000 1700000 1800000 1900000 2000000 2100000 2200000 2300000 2400000 2500000 2600000 2700000 2800000 2900000 3000000 3100000 [1] 9196 segmentation fault ./fuzzalloc 12345 Real: 24,57s User: 7,86s System: 15,35s Percent: 94% Cmd: ./fuzzalloc 12345 $ gdc --version gdc (Debian 6.1.1-11) 6.1.1 20160802 Copyright (C) 2016 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -- You are receiving this mail because: You are watching all bug changes. |
August 22, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #6 from Iain Buclaw <ibuclaw@gdcproject.org> --- Does this happen on all systems, or just a few? Have you eliminated bad memory - for instance, if memtester fails then you know it isn't druntime. -- You are receiving this mail because: You are watching all bug changes. |
August 23, 2016 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
http://bugzilla.gdcproject.org/show_bug.cgi?id=233 --- Comment #7 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> --- It consistently happens at least on 4 different Debian (stretch) PCs: - two DELL desktop at my office (16Gb of RAM each) - one production HP bladeserver (72Gb of RAM) - my home PC (8Gb of RAM). I think we can eliminate bad memory. We first encountered the problem in the bladeserver, with a real program (our production system). It's a blocking issue to us so we had to find a workaround (round memory allocations to the next multiple of 64). In the meantime, I found a way to reproduce this crash with a reduced test case (the one I posted here). However, the reduced test case isn't deterministic, and I don't get the crash at each run (I would say, 1/3 of the times). -- You are receiving this mail because: You are watching all bug changes. |
July 15, 2017 [Bug 233] crash in memory allocator | ||||
---|---|---|---|---|
| ||||
https://bugzilla.gdcproject.org/show_bug.cgi?id=233 Sebastien Alaiwan <sebastien.alaiwan@gmail.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution|--- |FIXED --- Comment #8 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> --- OK, I can't reproduce it anymore. Seems like the problem was fixed (maybe an issue in core.gc?). I'm closing this. -- You are receiving this mail because: You are watching all bug changes. |
Copyright © 1999-2021 by the D Language Foundation