Compiler panic while compiling libs when /tmp is full

[Kingdread]

Running cargo build on a project with external dependencies, it tries to compile them but fails with

thread 'rustc' panicked at 'called `Result::unwrap()` on an `Err` value: couldn't copy path (couldn't write to file (No space left on device; path=/tmp/rs-16901-0-rsar/rustc-serialize-bef146d997674220.o); from=/home/daniel/Code/digital_composer/target/deps/rustc-serialize-bef146d997674220.o; to=/tmp/rs-16901-0-rsar/rustc-serialize-bef146d997674220.o)', /home/rustbuild/src/rust-buildbot/slave/nightly-linux/build/src/libcore/result.rs:746

The underlying issue is pretty clear (no space left on /tmp), but the error message may sound confusing.

Traceback:

   Compiling rustc-serialize v0.2.1
     Running `rustc /home/daniel/.cargo/registry/src/github.1485827954.workers.dev-1285ae84e5963aae/rustc-serialize-0.2.1/src/lib.rs --crate-name rustc-serialize --crate-type lib -g -C metadata=bef146d997674220 -C extra-filename=-bef146d997674220 --out-dir /home/daniel/Code/digital_composer/target/deps --emit=dep-info,link -L /home/daniel/Code/digital_composer/target/deps -L /home/daniel/Code/digital_composer/target/deps -Awarnings`
error: internal compiler error: unexpected panic
note: the compiler unexpectedly panicked. this is a bug.
note: we would appreciate a bug report: http://doc.rust-lang.org/complement-bugreport.html
note: run with `RUST_BACKTRACE=1` for a backtrace
thread 'rustc' panicked at 'called `Result::unwrap()` on an `Err` value: couldn't copy path (couldn't write to file (No space left on device; path=/tmp/rs-16901-0-rsar/rustc-serialize-bef146d997674220.o); from=/home/daniel/Code/digital_composer/target/deps/rustc-serialize-bef146d997674220.o; to=/tmp/rs-16901-0-rsar/rustc-serialize-bef146d997674220.o)', /home/rustbuild/src/rust-buildbot/slave/nightly-linux/build/src/libcore/result.rs:746

stack backtrace:
   1: 0xb71ed1f0 - sys::backtrace::write::h8c2ed683dda82704ist
   2: 0xb7210e70 - failure::on_fail::hebe4057108e7dc03xLz
   3: 0xb717b7d0 - rt::unwind::begin_unwind_inner::h420b24b59a4d3530ypz
   4: 0xb717c1f0 - rt::unwind::begin_unwind_fmt::h82a51dddeba5bd4bdnz
   5: 0xb7210b80 - rust_begin_unwind
   6: 0xb725fdb0 - panicking::panic_fmt::h54f05751f027ad53eCl
   7: 0xb68119f0 - back::link::link_rlib::h8250a4523f2e5662tHa
   8: 0xb680d4f0 - back::link::link_binary::h88d6b7cda509381a1ua
   9: 0xb7537050 - driver::phase_6_link_output::hc73f9c1c01d961e2cIa
  10: 0xb75142f0 - driver::compile_input::h05aad328a73a6525wba
  11: 0xb766e160 - thunk::F.Invoke<A, R>::invoke::h118516395371660300
  12: 0xb766cf70 - rt::unwind::try::try_fn::h14176178842317427281
  13: 0xb7280cd0 - rust_try_inner
  14: 0xb7280ca0 - rust_try
  15: 0xb766d2b0 - thunk::F.Invoke<A, R>::invoke::h7075950477878468722
  16: 0xb71fe520 - sys::thread::thread_start::hf2d9dfd6d613d3d2dlw
  17: 0xb35460d0 - start_thread
  18: 0xb704b882 - __clone
  19:        0x0 - <unknown>

Rust version:

rustc 0.13.0-nightly (39d740266 2015-01-01 15:51:08 +0000)
binary: rustc
commit-hash: 39d74026663597a8d4ad0ab04e6d117bf9fd6ad4
commit-date: 2015-01-01 15:51:08 +0000
host: i686-unknown-linux-gnu
release: 0.13.0-nightly

[brson]

Seems like this error be handled properly and not just unwrapped.

[XMPPwocky]

This appears to be fixed already as of rustc 1.0.0-dev (89c4e37 2015-01-17 10:58:43 +0000)

[steveklabnik]

@Kingdread can you confirm what @XMPPwocky sees?

[Kingdread]

@steveklabnik The error seems to persist, even with rustc 1.0.0-nightly (d3732a12e 2015-02-06 23:30:17 +0000):

thread 'rustc' panicked at 'called `Result::unwrap()` on an `Err` value: IoError { kind: OtherIoError, desc: "couldn\'t copy path", detail: Some("couldn\'t write to file (No space left on device; path=/tmp/rsar.D7fuw95ZSU5H/rustc-serialize-ec565150ee89f7b5.o); from=\"/home/daniel/Code/digital_composer/target/deps/rustc-serialize-ec565150ee89f7b5.o\"; to=\"/tmp/rsar.D7fuw95ZSU5H/rustc-serialize-ec565150ee89f7b5.o\"") }', /home/rustbuild/src/rust-buildbot/slave/nightly-dist-rustc-linux/build/src/libcore/result.rs:743

[steveklabnik]

Triage: I'm not aware of any changes here, and I don't have /tmp mounted on a separate partition to be able to try this out.

[istankovic]

I think this one can be closed, I've just tried it out with latest nightly
(rustc 1.9.0-nightly (924da295c 2016-04-10)) and I get a nice error:

error: couldn't create a temp dir: No space left on device (os error 28).

rustc seems to honor the TMPDIR environment variable which allowed me
to simply set it to a loop-mounted filesystem that was made full.

[alexcrichton]

Yay!