| Rev |
Log message |
Author |
Age |
Path |
| 1009 |
Modify the gcommondefs to enable boot strap on X8664
Use opencc/openCC/openf90 to build the libraries. |
laijx |
2187d 20h |
/ |
| 1008 |
PBO bug fix, see bug 262 & 282 |
zx |
2187d 21h |
/ |
| 1007 |
fix for PBO bug, see bug252 |
zx |
2187d 21h |
/ |
| 1006 |
fix for bug300 only for x8664 |
hucheng |
2188d 02h |
/ |
| 1005 |
optimization about __attribute__((noreturn)) semantic |
syang |
2189d 02h |
/ |
| 1004 |
Modify the default build option to -O2 -fno-strict-aliasing, r.e., BUILD_OPTIMIZE=NODEBUG. |
hucheng |
2189d 18h |
/ |
| 1003 |
fix this assertion for the time being. it break the bootstrap compiler when adding Is_True_On flags. |
tianwei |
2189d 21h |
/ |
| 1002 |
Use -OPT:alias=no_typed to temporarily work around s_data.c file so that we can also bootstrap fortran fe. |
fengzhou |
2189d 21h |
/ |
| 1001 |
Fix the error in #1000, the PU_size after inlining should be Orig_Caller.PU_Size + Orig_Callee.PU_Size - 1, Because OPR_FUNC_ENTRY is not calculated into PU_Size. |
dehao |
2190d 00h |
/ |
| 1000 |
Fix the pu_size estimation of a PU. And it also fix the size update while doing inlining. We can now disable the "operator" heuristic which is aimed at making the inlining of MatrixInversion work well. |
dehao |
2190d 13h |
/ |
| 999 |
Pass CFG to summary phase in IPL. |
jianian |
2190d 19h |
/ |
| 998 |
At the end of CG_Generate_Code(), CXX_DELETE() is called upon region_tree. However, this pointer is not set properly if region hierarchy is not constructed |
syang |
2190d 23h |
/ |
| 997 |
1. ipo_inline_util.cxx for fixing inliner bug #277
2. ipa_cg.cxx for fix a bug when doing IPA tracing. Some other phases (e.g. IPA_Convert_Icalls) rely on the IPA_NODE's visited flag, so clear it after printing the trace
3. fix bug 296 (x86 targ has this fix, merge it to ia64) |
fengzhou |
2191d 23h |
/ |
| 996 |
Fix for another WHIRL writing error on Google code.
This one only materialized at -m32 and - I think - only
with DEBUG compiler.
The fix was provided by Gautam (Pathscale). Thanks a ton. |
rhundt |
2192d 11h |
/ |
| 995 |
add new files to the points-to analysis |
syang |
2193d 00h |
/ |
| 994 |
Tim (Pathscale) fixed the -m32 register allocator bug.
Bugzilla: #259
Allow CG to use ASM "m" operands directly from memory instead
of first loading it into a register. Always do this
optimization for local symbol operands, even if -fPIC.
Restricts fix to bug 5575 to fix bug 12605. Related to bug 3111.
Thanks! |
rhundt |
2193d 08h |
/ |
| 993 |
fix some extern "C" conflicts with no extern"C" caught by the latest g++ 4.1.1, provided by Fred. |
tianwei |
2193d 14h |
/ |
| 992 |
Revert the changes on kpro64/kgccfe/gnu/libgcc2.h in rev 987. |
laijx |
2194d 02h |
/ |
| 991 |
Bug fix for #274 - WHIRL writing error.
This fix was contributed by Fred. Thanks a ton! |
rhundt |
2194d 03h |
/ |
| 990 |
Bug fix for #286
This fix works around a problem in the gcc FE, where
a symbol assember name is being checked for being set,
and it is, but the actual name is still NULL.
Contributed by Fred |
rhundt |
2194d 03h |
/ |