[LLVMbugs] [Bug 9646] New: fallback to -no-canonical-prefixes if canonicalization fails

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Thu Apr 7 13:28:48 PDT 2011


http://llvm.org/bugs/show_bug.cgi?id=9646

           Summary: fallback to -no-canonical-prefixes if canonicalization
                    fails
           Product: clang
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Driver
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: nobled at dreamwidth.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=6410)
 --> (http://llvm.org/bugs/attachment.cgi?id=6410)
add fall back to -no-canonical-prefixes

GetMainExecutable() may fail, but the driver never checks its return value - so
even if it gets an empty path, the driver invocation goes ahead anyway and it
ends up like this:

 "" -cc1 -triple i386-pc-linux-gnu -emit-obj -mrelax-all -disable-free
-main-file-name test.cpp -mrelocation-model static -mdisable-fp-elim
-masm-verbose -mconstructor-aliases -target-cpu pentium4 -target-linker-version
2.20.51 -momit-leaf-frame-pointer -v -resource-dir ../lib/clang/3.0
-ferror-limit 19 -fmessage-length 80 -fcxx-exceptions -fexceptions
-fgnu-runtime -fdiagnostics-show-option -fcolor-diagnostics -o test.o -x c++
test.cpp
clang++: error: clang frontend command failed due to signal 127 (use -v to see
invocation)

which obviously doesn't work. The attached patch adds a fallback that acts as
if -no-canonical-prefixes was passed.

To reproduce this: it fails on Linux if it can't read
/proc/self/exe; e.g. if it's run in a chroot and
/proc isn't mounted.

This isn't a complete fix, since -no-canonical-prefixes
is still completely broken when running an installed
clang; see PR9576.

-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.



More information about the llvm-bugs mailing list