Tab-Expansion of Base in REPL causes SIGABRT

[staticfloat]

$ gdb ./julia
...
julia> Base.<TAB>
<Begins displaying properly>
<Right around Base.@sprintf, things start to go crazy>

Program received signal SIGABRT, Aborted.
0x00007fff8978a212 in __pthread_kill ()
(gdb) bt
#0  0x00007fff8978a212 in __pthread_kill ()
#1  0x00007fff8b030af4 in pthread_kill ()
#2  0x00007fff8b074dce in abort ()
#3  0x00007fff8b0508a5 in szone_error ()
#4  0x00007fff8b056819 in small_malloc_from_free_list ()
#5  0x00007fff8b056406 in szone_malloc_should_clear ()
#6  0x00007fff8b048153 in malloc_zone_malloc ()
#7  0x00007fff8b048ba7 in malloc ()
#8  0x00000001008c4031 in llvm::MallocSlabAllocator::Allocate ()
#9  0x00000001008c3c20 in llvm::BumpPtrAllocator::Allocate ()
#10 0x000000010053c616 in llvm::ScopedHashTable<llvm::Value*, std::pair<llvm::Value*, unsigned int>, llvm::DenseMapInfo<llvm::Value*>, llvm::RecyclingAllocator<llvm::BumpPtrAllocator, llvm::ScopedHashTableVal<llvm::Value*, std::pair<llvm::Value*, unsigned int> >, 40ul, 8ul> >::insertIntoScope ()
#11 0x00000001005393a6 in (anonymous namespace)::EarlyCSE::runOnFunction ()
#12 0x0000000100894670 in llvm::FPPassManager::runOnFunction ()
#13 0x0000000100893f95 in llvm::FunctionPassManagerImpl::run ()
#14 0x0000000100893ea5 in llvm::FunctionPassManager::run ()
#15 0x0000000100062cc2 in to_function ()
#16 0x0000000100062bcd in jl_compile ()
#17 0x00000001000606eb in jl_trampoline ()
#18 0x0000000101e26a03 in ?? ()
#19 0x0000000101e2692a in ?? ()
#20 0x000000010005b173 in jl_apply_generic ()
#21 0x000000010009b5a8 in jl_callback_call ()
#22 0x000000010009c04d in jl_asynccb ()
#23 0x00000001000bdee8 in uv__async_io (loop=0x100c4cb58, w=0x100c4ccd8, events=1) at src/unix/async.c:142
#24 0x00000001000d7b45 in uv__io_poll (loop=0x100c4cb58, timeout=10000) at src/unix/kqueue.c:221
#25 0x00000001000be778 in uv__run (loop=0x100c4cb58) at src/unix/core.c:271
#26 0x00000001000be705 in uv_run (loop=0x100c4cb58) at src/unix/core.c:279
#27 0x0000000101e0d891 in ?? ()
#28 0x0000000101e0d527 in ?? ()
#29 0x000000010005b173 in jl_apply_generic ()
#30 0x0000000101e0d486 in ?? ()
#31 0x0000000100094ebf in start_task ()
#32 0x0000000100093d2a in julia_trampoline ()
#33 0x00000001000018a9 in main ()

[staticfloat]

Interestingly, building a Debug version and running that inside of gdb doesn't crash, although the text layout does still go crazy.

[JeffBezanson]

I don't get a crash, but it definitely behaves oddly. The Display all 1860 possibilities? message appears, but before I can answer it starts printing all the completions, leaving the terminal in a strange state. This only seems to happen in cases with enough completions to print that message. Surely there is something unhealthy about telling readline to read from /dev/null. @vtjnash

[johnmyleswhite]

Thanks for looking into this.

[staticfloat]

It appears that the segfault only happens when running Julia inside of tmux. Running julia inside tmux results in output that looks like this: http://i.imgur.com/t58hFus.png

This is after typing "Base."

[JeffBezanson]

@vtjnash Unfortunately I don't see a simple way to fix this within the current framework. We have readline'd ourselves into a corner.

[vtjnash]

it certainly seems that way. i'm seeing if we can feed readline with a pipe (created by libuv) instead of rl_stuff_char.

[vtjnash]

seems like the best thing to do is comment out the /dev/null stdin redirection for the v0.1 release, until we can replace readline

[staticfloat]

This has been fixed for me. Can anyone else reproduce this error? Otherwise I will close.

[railskipl]

This i what i am getting,,

Program received signal SIGABRT, Aborted.
0x96a63a6a in __pthread_kill ()
(gdb) bt
#0  0x96a63a6a in __pthread_kill ()
#1  0x9590bacf in pthread_kill ()
#2  0x0391357b in abort ()
#3  0x03825f7b in abort_message ()
#4  0x03823a25 in default_terminate ()
#5  0x006170c1 in _objc_terminate ()
#6  0x03775652 in std::terminate ()
#7  0x00617039 in objc_terminate ()
#8  0x038a4027 in _dispatch_client_callout ()
#9  0x038947d5 in _dispatch_main_queue_callback_4CF ()
#10 0x02448af5 in __CFRunLoopRun ()
#11 0x02447f44 in CFRunLoopRunSpecific ()
#12 0x02447e1b in CFRunLoopRunInMode ()
#13 0x0403c7e3 in GSEventRunModal ()
#14 0x0403c668 in GSEventRun ()
#15 0x0098465c in UIApplicationMain ()
#16 0x000a5187 in main ()

[pao: formatting]

[pao]

Our poor early issues. @railskipl, just a heads up, be sure you code-quote (single backtick inline, triple backtick for block) console output and Julia code, since the GitHub autolinker may otherwise drop cross-references in unrelated issues (as it did here with the #nn from the backtrace) or ping random people whose usernames happen to match the names of Julia macros (due to Julia's @macro syntax). Thanks!

[ihnorton]

+1 to close. Can't replicate either.

[pao]

Going to go ahead and close this. @railskipl, if you can reproduce on an up-to-date build of Julia, please open a new issue for it.