LLVM Bugzilla is read-only and represents the historical archive of all LLVM issues filled before November 26, 2021. Use github to submit LLVM bugs

Bug 8720 - llvm-gcc testsuite regressions with symbol aliases
Summary: llvm-gcc testsuite regressions with symbol aliases
Status: RESOLVED FIXED
Alias: None
Product: new-bugs
Classification: Unclassified
Component: new bugs (show other bugs)
Version: unspecified
Hardware: All All
: P normal
Assignee: Unassigned LLVM Bugs
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-12-01 16:05 PST by Dan Gohman
Modified: 2010-12-07 09:25 PST (History)
3 users (show)

See Also:
Fixed By Commit(s):


Attachments
alias.patch (3.33 KB, patch)
2010-12-01 22:09 PST, Rafael Ávila de Espíndola
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Dan Gohman 2010-12-01 16:05:29 PST
The following tests appear to have started failing with r119152 in "clang-i386-darwin10":

gcc.dg/alias-5.c (test for excess errors)
gcc.dg/special/alias-1.c (test for excess errors)
gcc.dg/special/alias-2.c (test for excess errors)
gcc.dg/special/wkali-1.c (test for excess errors)
Comment 1 Dale Johannesen 2010-12-01 16:10:49 PST
Yes.  From our internal bug report:
These tests are protected by dg-require-alias, which is computed by running a program containing alias to see what it does.  With gcc and llvm-gcc this program gets an error message:

alias-2.c:13: error: only weak aliases are supported in this configuration

so these tests are not run.   Clang gives no error message, but doesn't implement aliases correctly, so the tests fail.  I have no real opinion on whether the feature should work or not, but it should either work or fail  cleanly.
Comment 2 Rafael Ávila de Espíndola 2010-12-01 22:09:17 PST
Created attachment 5855 [details]
alias.patch

I am not sure if newer darwin systems support alias or not, but the attached patch should produce an error that matches the one that the gcc testsuite expects.
Comment 3 Dale Johannesen 2010-12-06 19:34:51 PST
This appears to work, but doesn't the new test need to conditionalized for non-darwin somehow?
Comment 4 Rafael Ávila de Espíndola 2010-12-07 09:25:43 PST
Fixed in 121137.