This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Fixing struct-by-value-[567] on solaris2 with -fpic
- From: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>
- To: janis187 at us dot ibm dot com
- Cc: gcc at gcc dot gnu dot org
- Date: Sat, 3 Apr 2004 09:37:10 -0500 (EST)
- Subject: Fixing struct-by-value-[567] on solaris2 with -fpic
Hi Janis,
I'm seeing some failures on sparc-sun-solaris2.7 for
struct-by-value-[567] when running a testsuite pass with -fpic. E.g.
http://gcc.gnu.org/ml/gcc-testresults/2004-04/msg00142.html
The logfile shows this for all three tests when linking:
> [...]
> ld: fatal: too many symbols require `small' PIC references:
> have 1283, maximum 1024 -- recompile some modules -K PIC.
> collect2: ld returned 1 exit status
> compiler exited with status 1
It doesn't indicate a bug in GCC, but it creates testsuite noise. So
I'd like to silence it somehow.
I don't think having all the cases in one executable is what you were
trying to test, so I'd like to split these three tests each into two
files. E.g. I was thinking split testcase 5 into 5a and 5b and put
half of the stuff in each one. Ditto for cases 6 and 7 respectively.
Thoughts?
Thanks,
--Kaveh
--
Kaveh R. Ghazi ghazi@caip.rutgers.edu