Makes mathicgb FTBFS and mathicgb autopkgtest fail
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tbb (Ubuntu) |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
Running mathicgb tests with valgrind reveals crash in tbb 4.4~20160526-
g++ testMain.o Range.o gtestInclude.o gb-test.o ideals.o poly-test.o src/test/ideals.hpp SparseMatrix.o QuadMatrixBuilder.o F4MatrixBuilder.o F4MatrixReducer.o mathicgb.o PrimeField.o MonoMonoid.o Scanner.o MathicIO.o -lmathicgb -lmathic -lmemtailor -lpthread -ltbb -o testMain
valgrind -- ./testMain
==28427== Memcheck, a memory error detector
==28427== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==28427== Using Valgrind-3.12.0 and LibVEX; rerun with -h for copyright info
==28427== Command: ./testMain
==28427==
[==========] Running 237 tests from 27 test cases.
[----------] Global test environment set-up.
[----------] 9 tests from Range
[ RUN ] Range.Simple
[ OK ] Range.Simple (33 ms)
[ RUN ] Range.rangeToVector
[ OK ] Range.rangeToVector (6 ms)
[ RUN ] Range.zip
[ OK ] Range.zip (9 ms)
[ RUN ] Range.intRange
[ OK ] Range.intRange (23 ms)
[ RUN ] Range.indexRange
[ OK ] Range.indexRange (16 ms)
[ RUN ] Range.oppositeP
[ OK ] Range.oppositeP
[ RUN ] Range.adjPairRange
[ OK ] Range.adjPairRange (12 ms)
[ RUN ] Range.flatten
[ OK ] Range.flatten (13 ms)
[ RUN ] Range.flattenRe
[ OK ] Range.flattenRe
[----------] 9 tests from Range (172 ms total)
[----------] 1 test from IO
[ RUN ] IO.ideal
[ OK ] IO.ideal (34 ms)
[----------] 1 test from IO (34 ms total)
[----------] 11 tests from GB
[ RUN ] GB.small
==28427== Invalid read of size 8
==28427== at 0x57920BA: tbb::internal:
==28427== by 0x579218A: tbb::internal:
==28427== by 0x5793CE7: custom_scheduler (custom_
==28427== by 0x5793CE7: tbb::internal:
==28427== by 0x5792691: allocate_scheduler (scheduler.cpp:42)
==28427== by 0x5792691: tbb::internal:
==28427== by 0x578D0F3: tbb::internal:
==28427== by 0x578D1AB: tbb::task_
==28427== by 0x17DC5F: tbb::task_
==28427== by 0x17A0D0: testGB(
==28427== by 0x17ADED: GB_small_
==28427== by 0x16B0BC: void testing:
==28427== by 0x165368: void testing:
==28427== by 0x148EA1: testing:
==28427== Address 0xfffffffffffffff7 is not stack'd, malloc'd or (recently) free'd
==28427==
==28427==
==28427== Process terminating with default action of signal 11 (SIGSEGV)
==28427== Access not within mapped region at address 0xFFFFFFFFFFFFFFF7
==28427== at 0x57920BA: tbb::internal:
==28427== by 0x579218A: tbb::internal:
==28427== by 0x5793CE7: custom_scheduler (custom_
==28427== by 0x5793CE7: tbb::internal:
==28427== by 0x5792691: allocate_scheduler (scheduler.cpp:42)
==28427== by 0x5792691: tbb::internal:
==28427== by 0x578D0F3: tbb::internal:
==28427== by 0x578D1AB: tbb::task_
==28427== by 0x17DC5F: tbb::task_
==28427== by 0x17A0D0: testGB(
==28427== by 0x17ADED: GB_small_
==28427== by 0x16B0BC: void testing:
==28427== by 0x165368: void testing:
==28427== by 0x148EA1: testing:
==28427== If you believe this happened as a result of a stack
==28427== overflow in your program's main thread (unlikely but
==28427== possible), you can try to increase the size of the
==28427== main thread stack using the --main-stacksize= flag.
==28427== The main thread stack size used in this run was 8388608.
==28427==
==28427== HEAP SUMMARY:
==28427== in use at exit: 115,732 bytes in 1,206 blocks
==28427== total heap usage: 3,666 allocs, 2,460 frees, 533,309 bytes allocated
==28427==
==28427== LEAK SUMMARY:
==28427== definitely lost: 176 bytes in 1 blocks
==28427== indirectly lost: 480 bytes in 3 blocks
==28427== possibly lost: 0 bytes in 0 blocks
==28427== still reachable: 115,076 bytes in 1,202 blocks
==28427== of which reachable via heuristic:
==28427== newarray : 792 bytes in 3 blocks
==28427== suppressed: 0 bytes in 0 blocks
==28427== Rerun with --leak-check=full to see details of leaked memory
==28427==
==28427== For counts of detected and suppressed errors, rerun with: -v
==28427== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 0 from 0)
debian/
Debian unstable has an older tbb which does not crash Debian's mathicgb.
Changed in tbb (Ubuntu): | |
status: | New → Fix Released |
And the attached patch fixes the mathicgb tests and probably many other reverse dependencies.