Memory leak with 1.0.0-beta.2 on Mac

[abhi-bit]

Sample code used:

fn create_box() {
    let _function_box = Box::new(3i32);
}

fn main() {
    let _box_int = Box::new(5i32);

    {
        let _short_lived_box = Box::new(4i32);
    }

    for _ in 0u32..1_000_000 {
        create_box();
    }
}

Valgrind output from Linux:

==72139== Memcheck, a memory error detector
==72139== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==72139== Using Valgrind-3.10.1 and LibVEX; rerun with -h for copyright info
==72139== Command: ./raii
==72139==
==72139==
==72139== HEAP SUMMARY:
==72139==     in use at exit: 0 bytes in 0 blocks
==72139==   total heap usage: 1,000,018 allocs, 1,000,018 frees, 8,001,360 bytes allocated
==72139==
==72139== All heap blocks were freed -- no leaks are possible
==72139==
==72139== For counts of detected and suppressed errors, rerun with: -v
==72139== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

Valgrind output from Mac(Yosemite):

==2290== Memcheck, a memory error detector
==2290== Copyright (C) 2002-2013, and GNU GPL'd, by Julian Seward et al.
==2290== Using Valgrind-3.11.0.SVN and LibVEX; rerun with -h for copyright info
==2290== Command: ./raii
==2290==
--2290-- ./raii:
--2290-- dSYM directory is missing; consider using --dsymutil=yes
==2290==
==2290== HEAP SUMMARY:
==2290==     in use at exit: 35,270 bytes in 428 blocks
==2290==   total heap usage: 1,000,517 allocs, 1,000,089 frees, 8,041,742 bytes
allocated
==2290==
==2290== LEAK SUMMARY:
==2290==    definitely lost: 16 bytes in 1 blocks
==2290==    indirectly lost: 0 bytes in 0 blocks
==2290==      possibly lost: 13,098 bytes in 118 blocks
==2290==    still reachable: 22,156 bytes in 309 blocks
==2290==         suppressed: 0 bytes in 0 blocks
==2290== Rerun with --leak-check=full to see details of leaked memory
==2290==
==2290== For counts of detected and suppressed errors, rerun with: -v
==2290== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

Please let me know if you aren't able to replicate this issue, I could pass valgrind output with --leak-check=full --show-reachable=yes

Rust Version info:

rustc 1.0.0-beta.2 (e9080ec39 2015-04-16) (built 2015-04-16)
binary: rustc
commit-hash: e9080ec39d9a44eb0773d648b348575b19f655e5
commit-date: 2015-04-16
build-date: 2015-04-16
host: x86_64-apple-darwin
release: 1.0.0-beta.2

[alexcrichton]

Can you pass the extra flags to valgrind? It's possible that this could just be a bunch of false positives

[sfackler]

I don't think this has anything to do with Rust. A noop C program (int main() {}) has the same leakage: https://gist.github.com/sfackler/f2399049857ecd5ee5c6

[abhi-bit]

valgrind dump with extra flags: https://gist.github.com/abhi-bit/94e1f4dd11a227aae24f

@sfackler maybe right, I suspect.

[alexcrichton]

@abhi-bit yeah it looks like @sfackler is right, so closing this. Thanks for the report though!

[CaseyLeask]

I'm getting a memory leak on OS X with an empty main rust program, but it doesn't seem to be as serious as this one.

But I'm not getting any memory leaks with the C example.

valgrind dump with extra flags for both examples: https://gist.github.com/CaseyLeask/c8d61ec45c2e146d5c1f

[alexcrichton]

@CaseyLeask that may be a dupe of #19776 as pthread destructors do not run on process exit