Panic during unit tests (coerce_match)

[frewsxcv]

Currently on bd90b93

Running on OSX

cc'ing @kballard and @barosl since I saw both of them also had this issue in #rust-internals on IRC

...
test [run-pass] run-pass/yield2.rs ... ok
test [run-pass] run-pass/zero-size-type-destructors.rs ... ok
test [run-pass] run-pass/yield1.rs ... ok
test [run-pass] run-pass/vector-sort-panic-safe.rs ... ok
test [run-pass] run-pass/yield.rs ... ok

using metrics ratchet: tmp/check-stage2-T-x86_64-apple-darwin-H-x86_64-apple-darwin-rpass-metrics.json
result of ratchet: 0 metrics added, 0 removed, 0 improved, 0 regressed, 0 noise
updated ratchet file

failures:

---- [run-pass] run-pass/coerce-match.rs stdout ----

error: test run failed!
status: signal: 4
command: x86_64-apple-darwin/test/run-pass/coerce-match.stage2-x86_64-apple-darwin
stdout:
------------------------------------------

------------------------------------------
stderr:
------------------------------------------

------------------------------------------

thread '[run-pass] run-pass/coerce-match.rs' panicked at 'explicit panic', /Users/coreyf/Development/rust/src/compiletest/runtest.rs:1487



failures:
    [run-pass] run-pass/coerce-match.rs

test result: FAILED. 1745 passed; 1 failed; 27 ignored; 0 measured

thread '<main>' panicked at 'Some tests failed', /Users/coreyf/Development/rust/src/compiletest/compiletest.rs:267
make: *** [tmp/check-stage2-T-x86_64-apple-darwin-H-x86_64-apple-darwin-rpass.ok] Error 101

[barosl]

I should also note that the test code SIGILLs only when being compiled using an optimization flag. Without the flag, it runs well.

pub fn main() {
    let _: Box<[int]> = if true { box [1i, 2, 3] } else { box [1i] };

    let _: Box<[int]> = match true { true => box [1i, 2, 3], false => box [1i] };

    // Check we don't get over-keen at propagating coercions in the case of casts.
    let x = if true { 42 } else { 42u8 } as u16;
    let x = match true { true => 42, false => 42u8 } as u16;
}

[frewsxcv]

@jroesch is also having this issue. He says that tests stopped passing for him sometime after the night of December 16th

[jroesch]

Yeah both of my open PR's were exhibiting this on Mac OS X Yosemite. You can find them at #20002 and #20067 for reference.

[lilyball]

Here's the backtrace from a crash log:

Exception Type:        EXC_BAD_INSTRUCTION (SIGILL)
Exception Codes:       0x0000000000000001, 0x0000000000000000

Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
0   librustrt-4e7c5e5c.dylib        0x0000000103941f6b stack_overflow::imp::signal_handler::term::h4e1f7a7875fabf2cTLa + 59
1   librustrt-4e7c5e5c.dylib        0x0000000103941ee5 stack_overflow::imp::signal_handler::h49baa5350251116fALa + 101
2   libsystem_platform.dylib        0x00007fff91953f1a _sigtramp + 26
3   librustrt-4e7c5e5c.dylib        0x00000001039664f8 je_extent_tree_ad_remove + 216
4   librustrt-4e7c5e5c.dylib        0x0000000103967224 je_huge_dalloc + 68
5   librustrt-4e7c5e5c.dylib        0x0000000103971c16 je_sdallocx + 534
6   libstd-4e7c5e5c.dylib           0x00000001036ee4f4 thunk::F.Invoke$LT$A$C$$u{20}R$GT$::invoke::h14050218259616233623 + 52
7   libstd-4e7c5e5c.dylib           0x00000001036ee5d4 rt::start::closure.32341 + 164
8   librustrt-4e7c5e5c.dylib        0x00000001039aaafc rust_try_inner + 12
9   librustrt-4e7c5e5c.dylib        0x00000001039aaae6 rust_try + 6
10  librustrt-4e7c5e5c.dylib        0x00000001039458c7 unwind::try::hc34b1ff6cb3d2cbdquc + 71
11  librustrt-4e7c5e5c.dylib        0x000000010394566c task::Task::run::h5eecf441ff21ebbepLb + 124
12  libstd-4e7c5e5c.dylib           0x00000001036ee3af rt::start::h98e85f0930041adcIZx + 511
13  libstd-4e7c5e5c.dylib           0x00000001036ee19d rt::lang_start::hdf0f81f47bc259ddZYx + 109
14  libdyld.dylib                   0x00007fff93b1a5c9 start + 1

LLDB gives the following stack trace:

* thread #1: tid = 0xc417fe, 0x00000001000c3418 libstd-4e7c5e5c.dylib`je_extent_tree_ad_remove(rbtree=0x000000010039cce0, node=0x0000000000000000) + 216 at extent.c:38, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x18)
    frame #0: 0x00000001000c3418 libstd-4e7c5e5c.dylib`je_extent_tree_ad_remove(rbtree=0x000000010039cce0, node=0x0000000000000000) + 216 at extent.c:38
  * frame #1: 0x00000001000c4144 libstd-4e7c5e5c.dylib`je_huge_dalloc(ptr=<unavailable>) + 68 at huge.c:273
    frame #2: 0x00000001000ceb36 libstd-4e7c5e5c.dylib`je_sdallocx [inlined] je_isdalloct(ptr=<unavailable>) + 534 at jemalloc_internal.h:786
    frame #3: 0x00000001000ceb2b libstd-4e7c5e5c.dylib`je_sdallocx [inlined] je_isqalloc(ptr=<unavailable>) at jemalloc_internal.h:813
    frame #4: 0x00000001000ceb2b libstd-4e7c5e5c.dylib`je_sdallocx [inlined] isfree(ptr=<unavailable>) + 215 at jemalloc.c:1257
    frame #5: 0x00000001000cea54 libstd-4e7c5e5c.dylib`je_sdallocx(ptr=<unavailable>, size=<unavailable>, flags=<unavailable>) + 308 at jemalloc.c:1896
    frame #6: 0x00000001001053c9 libstd-4e7c5e5c.dylib`rust_try_inner + 9
    frame #7: 0x00000001001053b6 libstd-4e7c5e5c.dylib`rust_try + 6
    frame #8: 0x000000010009d43d libstd-4e7c5e5c.dylib`rt::lang_start::hd39f943feeb73f34NCz + 653
    frame #9: 0x00007fff93b1a5c9 libdyld.dylib`start + 1

frame #0 points at a macro to generate red-black tree functions, though as you can see node is NULL, so perhaps that's the cause.

[lilyball]

I'm going to attempt a bisect, though it'll take a while.

[lilyball]

According to the bisect, the bad commit is 46eb724.

[lilyball]

Which was introduced in #19769 /cc @nick29581

[lilyball]

As it turns out, #19769 is what introduced the run-pass/coerce-match.rs test in the first place.

[eddyb]

@kballard can you check whether #20083 affects this issue in any way?

[lilyball]

@eddyb Sadly, run-pass/coerce-match.rs still fails on make check after applying #20083.

[crhino]

I was running make check CFG_VALGRIND="/usr/bin/valgrind" and hit an error on this unit test as well, looks to be related. I'm on 3.13.0-43-generic GNU/Linux. Everything passes fine when not running under Valgrind.

This is running with the merge of #20083., commit d10642e.

failures:

---- [run-pass] run-pass/coerce-match.rs stdout ----

error: test run failed!
status: signal: 11
command: /usr/bin/valgrind x86_64-unknown-linux-gnu/test/run-pass/coerce-match.stage2-x86_64-unknown-linux-gnu
stdout:
------------------------------------------

------------------------------------------
stderr:
------------------------------------------
==22243== Memcheck, a memory error detector
==22243== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==22243== Using Valgrind-3.10.0.SVN and LibVEX; rerun with -h for copyright info
==22243== Command: x86_64-unknown-linux-gnu/test/run-pass/coerce-match.stage2-x86_64-unknown-linux-gnu
==22243== 
==22243== Warning: client switching stacks?  SP change: 0xffeffee60 --> 0xffedf63d8
==22243==          to suppress, use: --max-stackframe=2132616 or greater
==22243== Syscall param read(buf) points to unaddressable byte(s)
==22243==    at 0x40194F7: read (syscall-template.S:81)
==22243==    by 0x4005E0C: open_verify.constprop.6 (dl-load.c:2099)
==22243==    by 0x40061CE: open_path (dl-load.c:2216)
==22243==    by 0x4008F19: _dl_map_object (dl-load.c:2447)
==22243==    by 0x400D601: openaux (dl-deps.c:63)
==22243==    by 0x400FFF3: _dl_catch_error (dl-error.c:187)
==22243==    by 0x400DD04: _dl_map_object_deps (dl-deps.c:254)
==22243==    by 0x400315C: dl_main (rtld.c:1742)
==22243==    by 0x4017564: _dl_sysdep_start (dl-sysdep.c:249)
==22243==    by 0x4004CF7: _dl_start (rtld.c:332)
==22243==    by 0x40012D7: ??? (in /lib/x86_64-linux-gnu/ld-2.19.so)
==22243==  Address 0xffedf63f0 is on thread 1's stack
==22243== 
==22243== Warning: client switching stacks?  SP change: 0xffedf63e0 --> 0xffeffee88
==22243==          to suppress, use: --max-stackframe=2132648 or greater
==22243== Invalid free() / delete / delete[] / realloc()
==22243==    at 0x4F779D8: je_valgrind_freelike_block (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4FC1D28: rust_try_inner (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4FC1D15: rust_try (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5DBC9: rt::lang_start::he0ccf12406aa1099d1z (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x5489EC4: (below main) (libc-start.c:287)
==22243==  Address 0x6c26000 is 0 bytes inside a block of size 32 free'd
==22243==    at 0x4F779D8: je_valgrind_freelike_block (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5D5B2: rt::args::imp::put::hbd0ed9ed6213e26c8Rz (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5D024: rt::args::init::hcc36baa9a67ccf9dlPz (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5DB52: rt::lang_start::he0ccf12406aa1099d1z (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x5489EC4: (below main) (libc-start.c:287)
==22243== 
==22243== Invalid read of size 8
==22243==    at 0x4F92559: je_arena_dalloc_small (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F76FF1: je_sdallocx (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5D1A2: rt::args::cleanup::hec9e413559adae2cDPz (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5DC9C: rt::lang_start::he0ccf12406aa1099d1z (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x5489EC4: (below main) (libc-start.c:287)
==22243==  Address 0x6be92b8 is not stack'd, malloc'd or (recently) free'd
==22243== 
==22243== Invalid read of size 4
==22243==    at 0x5A3C414: pthread_mutex_lock (pthread_mutex_lock.c:66)
==22243==    by 0x4F92565: je_arena_dalloc_small (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F76FF1: je_sdallocx (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5D1A2: rt::args::cleanup::hec9e413559adae2cDPz (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x4F5DC9C: rt::lang_start::he0ccf12406aa1099d1z (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==22243==    by 0x5489EC4: (below main) (libc-start.c:287)
==22243==  Address 0x10 is not stack'd, malloc'd or (recently) free'd
==22243== 
==21446== 
==21446== Process terminating with default action of signal 11 (SIGSEGV)
==21446==    at 0x549EBB9: raise (raise.c:56)
==21446==    by 0x4F4A704: sys::stack_overflow::imp::signal_handler::h7e9a49daa22611feMkw (in /home/chris/workspace/rust/x86_64-unknown-linux-gnu/stage2/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-4e7c5e5c.so)
==21446==    by 0x549EC2F: ??? (in /lib/x86_64-linux-gnu/libc-2.19.so)
==21446==    by 0x5A3C413: pthread_mutex_lock (pthread_mutex_lock.c:63)
==22243== 
==22243== HEAP SUMMARY:
==22243==     in use at exit: 264 bytes in 6 blocks
==22243==   total heap usage: 18 allocs, 13 frees, 1,672 bytes allocated
==22243== 
==22243== LEAK SUMMARY:
==22243==    definitely lost: 0 bytes in 0 blocks
==22243==    indirectly lost: 0 bytes in 0 blocks
==22243==      possibly lost: 0 bytes in 0 blocks
==22243==    still reachable: 264 bytes in 6 blocks
==22243==         suppressed: 0 bytes in 0 blocks
==22243== Rerun with --leak-check=full to see details of leaked memory
==22243== 
==22243== For counts of detected and suppressed errors, rerun with: -v
==22243== ERROR SUMMARY: 4 errors from 4 contexts (suppressed: 2 from 2)

------------------------------------------

thread '[run-pass] run-pass/coerce-match.rs' panicked at 'explicit panic', /home/chris/workspace/rust/src/compiletest/runtest.rs:1487



failures:
    [run-pass] run-pass/coerce-match.rs

test result: FAILED. 1756 passed; 1 failed; 27 ignored; 0 measured

[pnkfelix]

i also see this on my mac