Join GitHub today
GitHub is home to over 31 million developers working together to host and review code, manage projects, and build software together.
Sign up1.14.0 powerpc64le test failures: smoke_dtor, test_typed_arena_drop_small_count #39015
Comments
brson
added
A-testsuite
O-PowerPC
I-wrong
and removed
A-testsuite
labels
Jan 12, 2017
This comment has been minimized.
This comment has been minimized.
racardoso
commented
Feb 9, 2017
|
Those tests aren't failing on master but they're failing on branch 1.15.0 as well
|
This comment has been minimized.
This comment has been minimized.
racardoso
commented
Feb 13, 2017
•
|
The test
But on release mode it fails:
Seems to me that some optimization is making this test fail. |
This comment has been minimized.
This comment has been minimized.
|
I tried a build without the optimization.
output of asm-out-read-uninit as example:
|
This comment has been minimized.
This comment has been minimized.
|
The asm tests are just missing |
This comment has been minimized.
This comment has been minimized.
racardoso
commented
Feb 20, 2017
|
@er-1 Doing a quick looking on the output you sent, the code is warning a dead code and (maybe) the warning output is not expected since the optimizations should do a dead code elimination:
'smoke_dtor' and |
This comment has been minimized.
This comment has been minimized.
|
I've added ignore flags in the related asm test files. I'll let you know how the build goes. |
This comment has been minimized.
This comment has been minimized.
|
As said I added the ignore flags in the related asm test files.
And then I'm able to build and test rust on ppc64el :) |
This comment has been minimized.
This comment has been minimized.
amboar
commented
Feb 22, 2017
|
@er-1 I've been meaning to send a pull req to mark those asm tests as ignored on powerpc64* for a while. Thanks for that. Separately, I've been investigating the failure of Here libarena was extracted from 1.15.0 and hacked up a bit to remove the already passing test cases. It shouldn't matter what version it comes from though as it's the optimiser that's failing, and it's still failing with recent nightly I haven't yet looked into the details of |
er-1
added a commit
to er-1/rust
that referenced
this issue
Feb 22, 2017
frewsxcv
added a commit
to frewsxcv/rust
that referenced
this issue
Feb 23, 2017
eddyb
added a commit
to eddyb/rust
that referenced
this issue
Feb 25, 2017
eddyb
added a commit
to eddyb/rust
that referenced
this issue
Feb 25, 2017
This comment has been minimized.
This comment has been minimized.
amboar
commented
Feb 27, 2017
|
Following up on my comment above, it seems the |
This comment has been minimized.
This comment has been minimized.
amboar
commented
Feb 27, 2017
|
I began testing against upstream LLVM with the reduced optimisation pipeline. The problem has been resolved in upstream, specifically by commit llvm-mirror/llvm@21c68fa I have created an LLVM tree which backports relevant patches to the rust branch: amboar/llvm@ed2c001, and I didn't put much effort into the backport so it's a complete hatchet-job on @hfinkle's patches. I'll try to clean it up in the next day or so and submit a pull request. |
This comment has been minimized.
This comment has been minimized.
amboar
commented
Feb 28, 2017
•
er-1
added a commit
to er-1/rust
that referenced
this issue
Mar 1, 2017
er-1
referenced this issue
Mar 1, 2017
Merged
Add a reference to the dl library to the Makefile of the test issue-2… #40173
GuillaumeGomez
added a commit
to GuillaumeGomez/rust
that referenced
this issue
Mar 2, 2017
arielb1
added
I-nominated
T-compiler
labels
Mar 23, 2017
This comment has been minimized.
This comment has been minimized.
|
cc me |
arielb1
added
P-high
and removed
I-nominated
labels
Mar 30, 2017
This comment has been minimized.
This comment has been minimized.
|
So I said last meeting I had a powerpc virtual machine I could test whether LLVM 4.0 really fixes this. Sadly, turns out this machine is not 64-bit powerpc nor it is the low endian one, sadly. |
brson
added
the
A-LLVM
label
Apr 11, 2017
This comment has been minimized.
This comment has been minimized.
|
triage: P-medium Downgrading to P-medium. Blocked on LLVM 4.0 upgrade (#37609). |
rust-highfive
added
P-medium
and removed
P-high
labels
Apr 13, 2017
This comment has been minimized.
This comment has been minimized.
|
llvm 4.0 upgrade has just landed, so this might be fixed/fixable now |
This comment has been minimized.
This comment has been minimized.
amboar
commented
Apr 25, 2017
|
@shahn thanks for the heads-up, I'll try to test this tomorrow |
This was referenced Jun 7, 2017
This comment has been minimized.
This comment has been minimized.
|
How did the test go? |
This comment has been minimized.
This comment has been minimized.
amboar
commented
Jun 19, 2017
•
This comment has been minimized.
This comment has been minimized.
|
Thanks for testing! Closing. |
infinity0 commentedJan 12, 2017
Build machine: https://db.debian.org/machines.cgi?host=ppc64el-osuosl-01
Build log: https://buildd.debian.org/status/fetch.php?pkg=rustc&arch=ppc64el&ver=1.14.0%2Bdfsg1-3&stamp=1484057235
Raw build log: https://buildd.debian.org/status/fetch.php?pkg=rustc&arch=ppc64el&ver=1.14.0%2Bdfsg1-3&stamp=1484057235&raw=1
This is after applying #38650 and #38675. The LLVM diff between Debian vs Rust is here.
Note that, because of these failures, the whole-compiler tests (e.g. run-pass etc) have not been run, but they probably also have failures.
Relevant links:
https://github.com/rust-lang/rust/blob/1.14.0/src/librustc_back/target/powerpc64le_unknown_linux_gnu.rs
https://github.com/rust-lang/rust/blob/1.14.0/mk/cfg/powerpc64le-unknown-linux-gnu.mk