t/pmc/eval.t: freeze/thaw test not passing on OpenBSD/AMD 64

[jkeenan]

I'm reporting this strictly on the basis of examination of Smolder reports on this particular OS/platform combination. I believe the reporter is darbelo.

Test #13 in t/pmc/eval.t is failing steadily as follows:

not ok 13 - eval.freeze+thaw

#   Failed test 'eval.freeze+thaw'
#   at t/pmc/eval.t line 418.
# Exited with error code: 134
# Received:
# written
# hello from foo_1
# hello from foo_1
# src/call/pcc.c:563: failed assertion 'PObj_is_PMC_TEST(sig_pmc)'
# Abort trap (core dumped) 
# 
# Expected:
# written
# hello from foo_1
# hello from foo_1
# 

See, e.g.,  Smolder 24453.

The test appears to be passing on OpenBSD on other chips, and it appears to be passing on AMD64 other than OpenBSD.

[darbelo]

I'm the reporter, and in case anyone needs it here's the backtrace for this failure.

(gdb) bt
#0  0x000000020799433a in kill () from /usr/lib/libc.so.51.0
#1  0x00000002079e0725 in abort () at /usr/src/lib/libc/stdlib/abort.c:68
#2  0x00000002037e1e68 in Parrot_confess (cond=0x203b0fb38 "PObj_is_PMC_TEST(sig_pmc)", 
    file=0x203b0fae0 "src/call/pcc.c", line=563) at src/exceptions.c:605
#3  0x00000002037f99f5 in Parrot_init_arg_indexes_and_sig_pmc (interp=0x2053f3000, ctx=0x20639b700, 
    indexes=0x206d774b8, sig_pmc=0x209e91120, sti=0x7f7ffffc3d00) at src/call/pcc.c:563
#4  0x00000002037fca1b in parrot_pass_args (interp=0x2053f3000, src_ctx=0x20639b700, 
    dest_ctx=0x20639bb00, src_indexes=0x206d774b0, dest_indexes=0x205cff780, 
    param_or_result=PARROT_PASS_RESULTS) at src/call/pcc.c:1887
#5  0x0000000203763fc3 in Parrot_set_returns_pc (cur_opcode=0x206d774b0, interp=0x2053f3000)
    at core.ops:667
#6  0x0000000203844c55 in runops_slow_core (interp=0x2053f3000, pc=0x206d774b0)
    at src/runcore/cores.c:462
#7  0x0000000203843bbd in runops_int (interp=0x2053f3000, offset=0) at src/runcore/main.c:987
#8  0x0000000203801ca5 in runops (interp=0x2053f3000, offs=0) at src/call/ops.c:107
#9  0x0000000203802078 in runops_args (interp=0x2053f3000, sub=0x209e917b0, obj=0x204c75dc0, 
    meth_unused=0x0, sig=0x203b0e4d3 "vP", ap=0x7f7ffffc4070) at src/call/ops.c:256
#10 0x0000000203802355 in Parrot_runops_fromc_args (interp=0x2053f3000, sub=0x209e917b0, 
    sig=0x203b0e4d3 "vP") at src/call/ops.c:325
#11 0x00000002037dda70 in Parrot_runcode (interp=0x2053f3000, argc=1, argv=0x7f7ffffc42c8)
    at src/embed.c:1021
#12 0x00000002039e44f0 in imcc_run_pbc (interp=0x2053f3000, obj_file=0, output_file=0x0, argc=1, 
    argv=0x7f7ffffc42c8) at compilers/imcc/main.c:801
#13 0x00000002039e50c6 in imcc_run (interp=0x2053f3000, 
    sourcefile=0x7f7ffffc47a9 "/home/darbelo/parrot/trunk/t/pmc/eval_13.pir", argc=1, 
    argv=0x7f7ffffc42c8) at compilers/imcc/main.c:1092
#14 0x0000000000400eb5 in main (argc=1, argv=0x7f7ffffc42c8) at src/main.c:60
(gdb)

I should mention too that the machine generating this is not under my direct control, so I can't promise that the smokes will happen regularly. Feel free to ping me if you need more information on this issue.

[rg]

I'm seeing the same failure on FreeBSD/amd64. The test passes when running parrot with -G which I guess indicates a gc bug (maybe one of those you're chasing anyway). Unfortunately, I don't have much spare time recently, so I won't be any help in debugging. Just thought I should mention the -G part.

[darbelo]

I should note that I'm no longer seeing this failure. As far as I can tell some of bug squashing in the past few days fixed or made invisible this bug for OpenBSD/amd64.

[rg]

I'm with darbelo++ on that. Closing ticket.

[jkeenan]

Replying to darbelo:

I should note that I'm no longer seeing this failure. As far as I can tell some of bug squashing in the past few days fixed or made invisible this bug for OpenBSD/amd64.

The Smolder reports do seem to indicate that this bug has cleared up on both OpenBSD/amd64 and FreeBSD/amd64.