You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
With clang-upc as it exists in the repo currently, I have no problems with clang and the default ILP32 ABI on Solaris-11 for x86-64, except the already reported issue with stdarg.h.
With -m64, however, the system linker is unable to deal with the unwind tables that clang is generating:
{phargrov@pcp-j-20 ~}$ clang -m64 hello.c -fasynchronous-unwind-tables
ld: fatal: unwind table: file /tmp/hello-f265e5.o: section .eh_frame: bad cie version 0: offset 0x4
clang: error: linker command failed with exit code 1 (use -v to see invocation)
On a per-compilation basis I can work-around with -fno-asynchronous-unwind-tables.
However, I am currently building on Solaris-11/x86-64 with the following patch instead, which may or may not be "overkill".
Sorry, I committed the issue before I had proof-read it (and got the title wrong as a result).
I should have mentioned for clarity that -fasynchronous-unwind-tables in the example is required to reproduce the error message with the patched build. Prior to the one-line patch the error message would appear when invoking clang with just -m64 and a filename.
With clang-upc as it exists in the repo currently, I have no problems with clang and the default ILP32 ABI on Solaris-11 for x86-64, except the already reported issue with
stdarg.h
.With
-m64
, however, the system linker is unable to deal with the unwind tables that clang is generating:On a per-compilation basis I can work-around with
-fno-asynchronous-unwind-tables
.However, I am currently building on Solaris-11/x86-64 with the following patch instead, which may or may not be "overkill".
The text was updated successfully, but these errors were encountered: