Jack Howarth - Re: fix trampolines on powerpc-darwin9 (original) (raw)

This is the mail archive of the gcc-patches@gcc.gnu.orgmailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Geoff, Speaking of reducing noise on the Darwin testsuite results... has there been any progress toward identifying the cause and a fix for the warnings "can't find atom for N_GSYM stabs" that occur with builds using '-m64 -g' on PPC? Jack ps The biggest impact of this one is the huge number of false failures when running the fortran testsuite at -m64.

On Mon, Feb 05, 2007 at 04:33:15PM -0800, Geoffrey Keating wrote:

This fixes the trampoline-related testcases on powerpc-darwin9. With this and the cleanup fixes, darwin9 has better results than darwin8, two new problems (apparently) and three old problems fixed.

Bootstrapped & tested on powerpc-darwin9.

I'll commit this to 4.2 as well, even though it's not technically a regression; it causes a lot of testsuite noise and it affects the shared libgcc, and I think it's pretty safe.

--

===File ~/patches/gcc-darwin-trampolinewriteable.patch====== 2007-02-05 Geoffrey Keating geoffk@apple.com

* config/rs6000/darwin-tramp.asm (__trampoline_setup): Call
__enable_execute_stack on completion.

Index: config/rs6000/darwin-tramp.asm

--- config/rs6000/darwin-tramp.asm (revision 121440) +++ config/rs6000/darwin-tramp.asm (working copy) @@ -96,11 +96,13 @@ addi r3,r3,4 bdnz Lcache - /* Finally synchronize things & return / + / Ensure cache-flushing has finished. / sync isync - blr + / Make stack writeable. */ + b ___enable_execute_stack + Labort: #ifdef DYNAMIC bl L_abort$stub

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]