[llvm-bugs] [Bug 47607] New: TestCases/Posix/no_asan_gen_globals.c FAILs on Solaris
via llvm-bugs
llvm-bugs at lists.llvm.org
Tue Sep 22 04:17:16 PDT 2020
https://bugs.llvm.org/show_bug.cgi?id=47607
Bug ID: 47607
Summary: TestCases/Posix/no_asan_gen_globals.c FAILs on Solaris
Product: compiler-rt
Version: unspecified
Hardware: All
OS: Solaris
Status: NEW
Severity: normal
Priority: P
Component: asan
Assignee: unassignedbugs at nondot.org
Reporter: ro at gcc.gnu.org
CC: llvm-bugs at lists.llvm.org, vitalybuka at google.com
The TestCases/Posix/no_asan_gen_globals.c test FAILs on Solaris.
$ nm no_asan_gen_globals.c.tmp.exe |grep ___asan_gen
0808056b r .L___asan_gen_.1
0808232b r .L___asan_gen_.2
08078df7 r .L___asan_gen_.4
0808232b r .L___asan_gen_.5
08082387 r .L___asan_gen_.7
0808232b r .L___asan_gen_.8
Analysis has shown two issues so far:
* Unlike GNU ld, the Solaris ld doesn't strip local symbols (i.e. those
starting
with ".L") from executables on output. I'll follow up with the linker
engineers about this. However, one might argue that the linker has no
business stripping them since they should never have been in the input in the
first place.
Simply relinking the input objects with gld lets those symbols above vanish.
* However, they are present in the input object, too:
$ nm no_asan_gen_globals.o |grep ___asan_gen00000000 r .L___asan_gen_.1
00000002 r .L___asan_gen_.2
0000005e r .L___asan_gen_.4
00000060 r .L___asan_gen_.5
000000bc r .L___asan_gen_.7
000000be r .L___asan_gen_.8
This is a bug in LLVM's integrated assebler, as can be seen when using
-fno-integrated-as where those symbols are not present, as they shouldn't
have
been in the first place.
This issue isn't Solaris-specfic: I see those local symbols in the object
file
on Linux/x86_64, too.
I have not yet found where to address the second issue. Maybe the test should
be XFAILed in the meantime?
--
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20200922/feca7f60/attachment-0001.html>
More information about the llvm-bugs
mailing list