Add line numbers to windows-gnu backtraces #28004

Diggsey · 2015-08-26T00:52:39Z

Technically this could also be used for windows-msvc targets, as I believe they have both dwarf and pdb debug information, but I haven't enabled it there as it should really use the native windows APIs for that, instead of libbacktrace.

I wasn't exactly sure where I should put "gnu" specific stuff, so tell me if I should structure things differently.

This is still a WIP, and I haven't tested properly to make sure I haven't broken msvc/linux builds yet.

rust-highfive · 2015-08-26T00:52:53Z

r? @alexcrichton

(rust_highfive has picked a reviewer for you, use r? to override)

alexcrichton · 2015-08-26T16:05:41Z

src/libstd/sys/unix/backtrace/printing/mod.rs

@@ -15,5 +15,5 @@ pub use self::imp::*;
 mod imp;

 #[cfg(not(any(target_os = "macos", target_os = "ios")))]
-#[path = "libbacktrace.rs"]
+#[path = "../../../gnu/libbacktrace.rs"]


Can this file instead live inside of src/libstd/sys/common? It can be #[cfg]'d off there and just pub use'd here

alexcrichton · 2015-08-26T16:08:47Z

Nice! I've wondered in the past if we should just start using libbacktrace for everything instead of just symbols/filenames/line numbers, it's got a pretty good track record so far.

Diggsey · 2015-08-27T00:26:07Z

Accidentally closed the PR there for a second! Anyway, I've made the changes you suggested - I'm mostly held back by how long it takes to build everything, so I've still been unable to test properly in all configurations.

alexcrichton · 2015-08-27T16:37:55Z

src/libstd/sys/common/mod.rs

+        all(unix, not(any(target_os = "macos", target_os = "ios"))),
+        all(windows, target_env = "gnu")
+    )
+)]


Typically we don't format this with the newlines here, instead something like:

#[cfg(any(all(unix, not(any(target_os = "macos", target_os = "ios"))), all(windows, target_env = "gnu")))]

alexcrichton · 2015-08-27T16:39:36Z

Thanks! Just one minor nit and r=me after a squashing of the last 6 commits

Diggsey · 2015-08-27T17:07:18Z

All done

alexcrichton · 2015-08-27T18:52:51Z

@bors: r+ 4b2910c99ece3dd042f558c7303d372b78395fc0

alexcrichton · 2015-08-27T18:53:35Z

Er actually, just remembered, there's a test that tests for line numbers and can you be sure that it's enabled on GNU windows now? I believe the test is src/test/run-pass/backtrace-debuginfo.rs

@bors: r-

Diggsey · 2015-08-27T20:25:08Z

I've enabled the tests, and they seem to be passing.

alexcrichton · 2015-08-27T20:39:02Z

@bors: r+ 889fee33d47106bd51b06a9927c939fd8c396061

Thanks!

bors · 2015-08-28T02:46:05Z

⌛ Testing commit 889fee3 with merge 4686e32...

bors · 2015-08-28T02:49:44Z

💔 Test failed - auto-mac-64-opt

aturon · 2015-08-28T05:25:18Z

@bors: r=alexcrichton

bors · 2015-08-28T05:25:18Z

📌 Commit b6c0d66 has been approved by alexcrichton

bors · 2015-08-28T06:06:22Z

⌛ Testing commit b6c0d66 with merge bccd8c6...

bors · 2015-08-28T06:11:09Z

💔 Test failed - auto-win-gnu-64-nopt-t

eddyb · 2015-08-28T14:44:40Z

@bors: r=alexcrichton

alexcrichton · 2015-08-31T21:52:39Z

@bors: r+ 9173b937807b8cfd749340c5ddbc573c84018a3b

bors · 2015-08-31T23:29:14Z

⌛ Testing commit 9173b93 with merge ab23511...

bors · 2015-09-01T00:32:06Z

💔 Test failed - auto-win-gnu-32-opt

Diggsey · 2015-09-03T05:28:27Z

Fixing this is currently blocked by the fact that I'm unable to build rustc to stage2 under i686-pc-windows-gnu, due to the stage1 rustc segfaulting while compiling stage2 libsyntax. It appears to be an llvm bug, and it's proving to be very difficult to reproduce with debug builds of llvm (rustc seems to hang instead of crashing, or it could just be very slow), and Release+Asserts did not help.

The Release backtrace only showed that it might be happening inside llvm's tail-call elimination pass.

alexcrichton · 2015-09-03T16:35:05Z

@Diggsey yeah I've seen that happen unfortunately with gcc 5.2.0, although I'm not sure how to downgrade... Last I checked (a few weeks ago I think) it also happened with the HEAD LLVM, and unfortunately I haven't tracked it down much farther than that :(

Diggsey · 2015-09-03T23:35:14Z

Ok, I've just disabled the two backtrace tests on 32-bit windows-gnu again, since the problem seems to be unrelated to this PR (#28218).

bors · 2015-09-04T00:06:25Z

☔ The latest upstream changes (presumably #28200) made this pull request unmergeable. Please resolve the merge conflicts.

This includes the changes from commits: - cd8f317 - fcb30a0

Fix formatting Remove unused imports Refactor Fix msvc build Fix line lengths Formatting Enable backtrace tests Fix using directive on mac pwd info Work-around buildbot PWD bug, and fix libbacktrace configuration Use alternative to `env -u` which is not supported on bitrig Disable tests on 32-bit windows gnu

Diggsey · 2015-09-04T00:33:05Z

Rebased

eddyb · 2015-09-04T01:48:11Z

@bors r=alexcrichton

bors · 2015-09-04T01:48:12Z

📌 Commit d4fc3ec has been approved by alexcrichton

bors · 2015-09-04T04:17:45Z

⌛ Testing commit d4fc3ec with merge 074c4b6...

bors · 2015-09-04T05:15:21Z

💔 Test failed - auto-linux-32-nopt-t

Diggsey · 2015-09-04T05:47:48Z

Wat???

alexcrichton · 2015-09-04T06:09:36Z

@bors: retry

On Thu, Sep 3, 2015 at 10:47 PM, Diggory Blake [email protected]
wrote:

Wat???

—
Reply to this email directly or view it on GitHub
#28004 (comment).

Technically this could also be used for `windows-msvc` targets, as I believe they have *both* dwarf and pdb debug information, but I haven't enabled it there as it should really use the native windows APIs for that, instead of libbacktrace. I wasn't exactly sure where I should put "gnu" specific stuff, so tell me if I should structure things differently. This is still a WIP, and I haven't tested properly to make sure I haven't broken msvc/linux builds yet.

bors · 2015-09-04T07:08:18Z

⌛ Testing commit d4fc3ec with merge 9d1f820...

bors · 2015-09-04T08:50:55Z

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-x-android-t, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-win-gnu-32-nopt-t, auto-win-gnu-32-opt, auto-win-gnu-64-nopt-t, auto-win-gnu-64-opt, auto-win-msvc-32-opt, auto-win-msvc-64-opt

retep998 · 2015-09-04T08:53:45Z

🎊

jonas-schievink · 2015-09-05T19:15:08Z

Hmm... I think this caused the backtrace.rs and backtrace-debuginfo.rs tests to fail for me (on 64-Bit Windows 10, with up-to-date MSYS2):

---- [run-pass] run-pass/backtrace.rs stdout ----

error: test run failed!
status: exit code: 101
command: PATH="x86_64-pc-windows-gnu/stage1/bin/rustlib/x86_64-pc-windows-gnu/lib;C:\Users\Jonas\Documents\rust\build\x86_64-pc-windows-gnu\stage1\bin;C:\msys64\mingw64\bin;C:\msys64\usr\local\bin;C:\msys64\usr\bin;C:\msys64\usr\bin;C:\WINDOWS\system32;C:\WINDOWS;C:\WINDOWS\System32\Wbem;C:\WINDOWS\System32\WindowsPowerShell\v1.0;C:\Program Files\Rust nightly 1.4\bin;C:\Users\Jonas\AppData\Local\GitHub\PortableGit_c2ba306e536fdf878271f7fe636a147ff37326ad\bin;C:\Users\Jonas\AppData\Local\atom\bin;C:\Users\Jonas\bin;C:\msys64\usr\bin\site_perl;C:\msys64\usr\bin\vendor_perl;C:\msys64\usr\bin\core_perl" x86_64-pc-windows-gnu/test/run-pass\backtrace.stage1-x86_64-pc-windows-gnu.exe
stdout:
------------------------------------------

------------------------------------------
stderr:
------------------------------------------
thread '<main>' panicked at 'bad output: thread '<main>' panicked at 'explicit panic', C:/Users/Jonas/Documents/rust/src/test/run-pass/backtrace.rs:24
stack backtrace:
   0:         0x678fe84e - <unknown>
   1:         0x67907780 - <unknown>
   2:         0x678c56ee - <unknown>
   3:           0x4017b3 - <unknown>
   4:           0x401731 - <unknown>
   5:           0x4028e6 - <unknown>
   6:         0x67908568 - <unknown>
   7:         0x678f07d3 - <unknown>
   8:         0x67908477 - <unknown>
   9:           0x4013ec - <unknown>
  10:           0x40152a - <unknown>
  11:     0x7ffe43c22d91 - <unknown>
', C:/Users/Jonas/Documents/rust/src/test/run-pass/backtrace.rs:55

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

thread '[run-pass] run-pass/backtrace.rs' panicked at 'explicit panic', C:/Users/Jonas/Documents/rust/src/compiletest\runtest.rs:1501


---- [run-pass] run-pass/backtrace-debuginfo.rs stdout ----

error: test run failed!
status: exit code: 101
command: PATH="x86_64-pc-windows-gnu/stage1/bin/rustlib/x86_64-pc-windows-gnu/lib;C:\Users\Jonas\Documents\rust\build\x86_64-pc-windows-gnu\stage1\bin;C:\msys64\mingw64\bin;C:\msys64\usr\local\bin;C:\msys64\usr\bin;C:\msys64\usr\bin;C:\WINDOWS\system32;C:\WINDOWS;C:\WINDOWS\System32\Wbem;C:\WINDOWS\System32\WindowsPowerShell\v1.0;C:\Program Files\Rust nightly 1.4\bin;C:\Users\Jonas\AppData\Local\GitHub\PortableGit_c2ba306e536fdf878271f7fe636a147ff37326ad\bin;C:\Users\Jonas\AppData\Local\atom\bin;C:\Users\Jonas\bin;C:\msys64\usr\bin\site_perl;C:\msys64\usr\bin\vendor_perl;C:\msys64\usr\bin\core_perl" x86_64-pc-windows-gnu/test/run-pass\backtrace-debuginfo.stage1-x86_64-pc-windows-gnu.exe
stdout:
------------------------------------------

------------------------------------------
stderr:
------------------------------------------
thread '<main>' panicked at 'trace does not match position list: test case 0
thread '<main>' panicked at 'explicit panic', C:/Users/Jonas/Documents/rust/src/test/run-pass/backtrace-debuginfo.rs:84
stack backtrace:
   0:         0x678fe84e - <unknown>
   1:         0x67907780 - <unknown>
   2:         0x678c56ee - <unknown>
   3:           0x401a89 - <unknown>
   4:           0x401979 - <unknown>
   5:           0x401c75 - <unknown>
   6:           0x403e3e - <unknown>
   7:         0x67908568 - <unknown>
   8:         0x678f07d3 - <unknown>
   9:         0x67908477 - <unknown>
  10:           0x4013ec - <unknown>
  11:           0x40152a - <unknown>
  12:     0x7ffe43c22d91 - <unknown>

---
backtrace-debuginfo.rs:125
backtrace-debuginfo.rs:173
', C:/Users/Jonas/Documents/rust/src/test/run-pass/backtrace-debuginfo.rs:139

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

thread '[run-pass] run-pass/backtrace-debuginfo.rs' panicked at 'explicit panic', C:/Users/Jonas/Documents/rust/src/compiletest\runtest.rs:1501

More specifically, ac12c7f works, but d4fc3ec doesn't.

Diggsey · 2015-09-05T19:35:41Z

@jonas-schievink Neither of those tests were enabled on windows before this PR, so it's a safe bet that this caused it.

It looks like for some reason libbacktrace is unable to access the dwarf debug info on the executable. Most likely causes are eiither there's no debug info being generated in the first place, or there's some new security restriction on windows 10 that's preventing it.

rust-highfive assigned alexcrichton Aug 26, 2015

alexcrichton reviewed Aug 26, 2015
View reviewed changes

Diggsey closed this Aug 27, 2015

Diggsey reopened this Aug 27, 2015

alexcrichton reviewed Aug 27, 2015
View reviewed changes

Diggsey force-pushed the win-backtrace branch from 359298c to 4b2910c Compare August 27, 2015 17:06

Diggsey force-pushed the win-backtrace branch from 4b2910c to 889fee3 Compare August 27, 2015 20:23

Diggsey force-pushed the win-backtrace branch from 889fee3 to b6c0d66 Compare August 28, 2015 02:58

Diggsey force-pushed the win-backtrace branch from b6c0d66 to 38a0d15 Compare August 28, 2015 14:40

Diggsey force-pushed the win-backtrace branch from 9173b93 to 3aec908 Compare September 3, 2015 22:22

Diggsey and others added 3 commits September 4, 2015 01:25

Update libbacktrace to r227145

3427760

Reapply rust-specific changes to libbacktrace

ac12c7f

This includes the changes from commits: - cd8f317 - fcb30a0

Diggsey force-pushed the win-backtrace branch from 3aec908 to d4fc3ec Compare September 4, 2015 00:31

bors merged commit d4fc3ec into rust-lang:master Sep 4, 2015

alexcrichton mentioned this pull request Sep 20, 2015

Stack traces / backtraces in Windows show everything in "main" #20351

Closed

Jonesey13 mentioned this pull request Dec 29, 2015

Backtraces only showing unknowns on Windows 10 #30625

Closed

Add line numbers to windows-gnu backtraces #28004

Add line numbers to windows-gnu backtraces #28004

Conversation

Diggsey commented Aug 26, 2015

rust-highfive commented Aug 26, 2015

alexcrichton Aug 26, 2015

Choose a reason for hiding this comment

alexcrichton commented Aug 26, 2015

Diggsey commented Aug 27, 2015

alexcrichton Aug 27, 2015

Choose a reason for hiding this comment

alexcrichton commented Aug 27, 2015

Diggsey commented Aug 27, 2015

alexcrichton commented Aug 27, 2015

alexcrichton commented Aug 27, 2015

Diggsey commented Aug 27, 2015

alexcrichton commented Aug 27, 2015

bors commented Aug 28, 2015

bors commented Aug 28, 2015

aturon commented Aug 28, 2015

bors commented Aug 28, 2015

bors commented Aug 28, 2015

bors commented Aug 28, 2015

eddyb commented Aug 28, 2015

alexcrichton commented Aug 31, 2015

bors commented Aug 31, 2015

bors commented Sep 1, 2015

Diggsey commented Sep 3, 2015

alexcrichton commented Sep 3, 2015

Diggsey commented Sep 3, 2015

bors commented Sep 4, 2015

Diggsey commented Sep 4, 2015

eddyb commented Sep 4, 2015

bors commented Sep 4, 2015

bors commented Sep 4, 2015

bors commented Sep 4, 2015

Diggsey commented Sep 4, 2015

alexcrichton commented Sep 4, 2015

bors commented Sep 4, 2015

bors commented Sep 4, 2015

retep998 commented Sep 4, 2015

jonas-schievink commented Sep 5, 2015

Diggsey commented Sep 5, 2015