Kenneth Zadeck - Re: [dataflow]: PATCH: clear DF_LR_RUN_DCE flag after df_analyze in sch (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] |
- From: Kenneth Zadeck
- To: Seongbae Park
- Cc: gcc-patches
- Date: Tue, 20 Feb 2007 09:48:23 -0500
- Subject: Re: [dataflow]: PATCH: clear DF_LR_RUN_DCE flag after df_analyze in sched-ebb
- References: <ab3a61990702200646w3b729176ga3c166a4eee86f4a@mail.gmail.com>
Seongbae Park wrote:
Hi,
This is to fix a regression on ia64 - ia64 machine reorg pass inserts many nops, which then get deleted by df_analyze() call within ia64_mach_reorg. Clear the flag here to prevent any df_analyze() calls after sched_ebbs() from accidently triggering DCE.
The same change has been bootstraped and regtested before the latest merge with the mainline on dataflow, on ia64 and i686.
I've restarted the bootstrap and regtest on ia64. OK ?
ChangeLog.dataflow
2007-02-20 Seongbae Park seongbae.park@gmail.com
* sched-ebb.c (schedule_ebbs): Clear DF_LR_RUN_DCE flag after df_analyze.
Index: sched-ebb.c
--- sched-ebb.c (revision 122164) +++ sched-ebb.c (working copy) @@ -541,6 +541,7 @@ schedule_ebbs (void) df_set_flags (DF_LR_RUN_DCE); df_ri_add_problem (DF_RI_LIFE); df_analyze ();
- df_clear_flags (DF_LR_RUN_DCE); sched_init ();
compute_bb_for_insn (); ok to check into the dataflow branch.
kenny
- References:
- [dataflow]: PATCH: clear DF_LR_RUN_DCE flag after df_analyze in sched-ebb
* From: Seongbae Park
- [dataflow]: PATCH: clear DF_LR_RUN_DCE flag after df_analyze in sched-ebb
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |