Assertion failure in libuv

[staticfloat]

I came across this assertion failure at the tail end of a make testall run. @Keno any ideas? I don't really know what's going on there in uv__io_stop().

        From worker 4:       * repl
        From worker 2:       * test
        From worker 5:       * examples
        From worker 2:       * goto
        From worker 2:       * unicode
        From worker 2:       * parallel
Assertion failed: (loop->watchers[w->fd] == w), function uv__io_stop, file src/unix/core.c, line 764.

signal (6): Abort trap: 6
__pthread_kill at /usr/lib/system/libsystem_kernel.dylib (unknown line)
    SUCCESS

[tkelman]

I get this one fairly often on Ubuntu 14.04. Been happening for a while but looked harmless, though recently there's much more info from it:

        From worker 2:       * parallel
julia: src/unix/core.c:764: uv__io_stop: Assertion `loop->watchers[w->fd] == w' failed.

signal (6): Aborted
gsignal at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
abort at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unkown function (ip: -761442762)
unkown function (ip: -761442590)
uv__io_stop at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
unkown function (ip: -779250618)
uv_close at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
jl_close_uv at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
uvfinalize at socket.jl:306
jlcall_uvfinalize;36274 at  (unknown line)
jl_apply_generic at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
unkown function (ip: -779440258)
unkown function (ip: -779437088)
uv_atexit_hook at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
jl_exit at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
exit at client.jl:37
jlcall_exit;36270 at  (unknown line)
jl_apply_generic at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
jl_f_apply at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
anonymous at multi.jl:873
run_work_thunk at multi.jl:621
run_work_thunk at multi.jl:630
jlcall_run_work_thunk;19844 at  (unknown line)
jl_apply_generic at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
anonymous at task.jl:873
unkown function (ip: -779498256)
unkown function (ip: -779498120)
unkown function (ip: -779498120)
julia_trampoline at /home/tkelman/Julia/julia/usr/bin/../lib/libjulia.so (unknown line)
unkown function (ip: 4199613)
__libc_start_main at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unkown function (ip: 4199667)
unkown function (ip: 0)
    SUCCESS

[vtjnash]

it's unclear which worker threw this exception, but it is known (PR #7275) that the pollfd tests can trigger this assertion (translation: the current implementation of FDWatcher is fundamentally broken in 0.3 / don't use poll_fd)

[staticfloat]

I haven't seen this in a while, I'm going to close this until it happens a bunch again.

[tkelman]

Still happens somewhat often for me on Ubuntu. No idea what's causing it or how to fix it though.

[vtjnash]

I still suspect the poll_fd test. I discussed with Jeff how to fix, it just requires a fair amount of work (time) to implement

[tkelman]

Maybe I was making stuff up this morning, sorry. I am having a hard time reproducing this now. @vtjnash would your most recent libuv bump for #8789 have anything to do with this?

[vtjnash]

I don't think so. From the ordering of the tests, it seems that the pollfd test usually runs near the end, so it's apparently pretty unlike to cause this failure most of the time

[Divisible8737]

I get a similar crash each time I use PyPlot. The figures appear, then when try to manipulate them I get crashes like this:
julia> called Tcl_FindHashEntry on deleted table

signal (6): Abort trap: 6
__pthread_kill at /usr/lib/system/libsystem_kernel.dylib (unknown line)
Abort trap: 6

[tkelman]

Got another one of these on Ubuntu 14.04. Backtrace looks almost identical.

[tkelman]

closed by #12237? I'll keep an eye out for it and reopen if not