[PATCH]: update to df for mainline.
Kenneth Zadeck
zadeck@naturalbridge.com
Wed May 17 15:09:00 GMT 2006
Dear Bernd,
Attached is an updated patch. Unless noted below, all of your changes
were made as requested.
Bernd Schmidt wrote:
>> - struct df_ri_problem_data *problem_data =
>> - (struct df_ri_problem_data *) dflow->problem_data;
>
> I think the "=" should be on the second line.
>
I made this change uniformly in all of the df* files. This swells the
size of the patch somewhat.
>> + struct df * df = dflow->df;
>
> "df *df"
>
Ditto.
>> + if (flags & DF_RI_LIFE)
>> + if (df_ignore_stack_reg (REGNO (XEXP (link, 0))))
>> + REG_N_DEATHS (REGNO (XEXP (link, 0)))++;
>
> Add break here for clarity.
>
I did not add the break here. The case is meant to drop thru, note the
/*fallthru*/ comment on the next line. If you are complaining about
having fallthru's, I can duplicate the next case into this one.
>
>
> Lots of code shared between df_set_{dead,unused}_notes_for_mw.
>
These functions were originally one. This is why they look structurally
similar. I had split them when I could no longer follow what was going
on because of the special cases for each.
>
>
>> + /* Kill this register if it is not a subreg store. */
>> + if (!(DF_REF_FLAGS (def) & DF_REF_PARTIAL))
>> + bitmap_clear_bit (live, dregno);
>
> Looks like this depends on another part. I have to admit that I'm not
> convinced yet by the subreg handling in DF as a whole. A normal
> (non-multiword non-strict-low) subreg store _does_ kill the whole reg.
>
The setting of DF_REF_PARTIAL has been redone in df-scan.c so that it
fixes the problems mentioned in all of the mail.
> Bernd
In addition I split DF_REF_CLOBBER into DF_REF_MAY_CLOBBER and
DF_REF_MUST_CLOBBER cases (as you had requested in an earlier email).
The may clobber is set for registers that cross calls. This may change
in the future as the aliasing gets better, we may be be able convert
some of them to must clobbers. The must clobbers are set where we have
an explicit clobber in the code.
The reaching defs and uninitialized register dataflow problems were
rewritten to take this fully into account since a gen is never created
for either may or must clobbers, but a kill is only generated for must
clobbers (assuming that it is not also DF_REF_PARTIAL.)
The code assumes that it is possible to have partial clobber even though
it is not clear if this will ever happen.
This patch has been bootstrapped and regression tested on
x86_64-unknown-linux-gnu
powerpc64-unknown-linux-gnu
i686-pc-linux-gnu
Is this ok?
Kenny (and Danny)
2006-05-17 Daniel Berlin <dberlin@dberlin.org>
Kenneth Zadeck <zadeck@naturalbridge.com>
PR rtl-optimization/26855
* df-scan.c (mw_reg_pool, mw_link_pool): New allocation pools for
multiword refs.
(df_scan_alloc): Added code to properly handle multiword hard
registers and add all_blocks parameter.
(df_scan_free_internal, df_insn_refs_delete, df_ref_record): Added
code to properly handle multiword hard registers.
(df_rescan_blocks): Added code to remove deleted blocks from
bitmap.
(df_ref_create_structure, df_ref_record): Added code to properly
handle subregs.
(df_ref_record_1): Changed DF_REF_CLOBBER into DF_REF_MUST_CLOBBER
and set DF_REF_PARTIAL.
(df_defs_record): Changed DF_REF_CLOBBER into DF_REF_MUST_CLOBBER.
(df_uses_record): Added DF_REF_PARTIAL for subreg.
(df_scan_add_problem): Added flags parameter.
(df_ref_create_structure): Changed switching structure.
(df_bb_refs_record): Fixed case where duplicate artificial refs
were created. Changed location of flags.
(df_record_entry_block_defs): Added code to make stack pointer
live in entry block. Refined cases where frame pointer is needed.
Changed location of flags.
(df_record_exit_block_uses, df_insn_refs_record): Changed location
of flags.
(df_set_state): Removed function.
(df_grow_reg_info, df_reg_chain_unlink, df_ref_remove,
df_insn_create_insn_record, df_insn_refs_delete,
df_ref_create_structure): Formatting changes.
* df-core.c (df_mvs_dump, df_set_flags, df_clear_flags,
df_delete_basic_block): New function.
(df_init): Changed location of flags.
(df_add_problem): Added flags parameter and the way flags are
processed.
(df_insn_uid_debug, df_ref_debug, debug_df_defno, debug_df_ref,
debug_df_chain): Improved debugging output.
(df_insn_debug, df_insn_uid_debug): Added multiword reg support.
(df_refs_chain_dump): Removed df parameter.
(df_iterative_dataflow): Added consistency check.
(df_prune_to_subcfg): Made public.
(df_analyze_problem): Added blocks_to_init parameter and made
public.
* df.h: Some reordering to remove forward references.
(df_ref_flags.DF_REF_MW_HARDREG, DF_REF_PARTIAL,
DF_REF_MUST_CLOBBER, DF_REF_MAY_CLOBBER): New fields.
(df_ref_flags.DF_REF_CLOBBER): Deleted field.
(dataflow.flags): New field.
(df.flag): Deleted field.
(df_alloc_function): Added additional bitmap parameter.
(df_dependent_problem_function): New type.
(df_problem.changeable_flags): New field.
(df_ref_flags.DF_REF_DIES_AFTER_THIS_USE, DF_SCAN_INITIAL,
DF_SCAN_GLOBAL, DF_SCAN_POST_ALLOC, df_state): Removed.
(df_mw_hardreg): New struct.
(DF_INSN_UID_MWS): New macro.
(df_refs_chain_dump, df_ref_debug, df_chain_dump): Removed df
parameter.
(df_add_problem, df_ru_add_problem, df_rd_add_problem,
df_lr_add_problem, df_ur_add_problem, df_urec_add_problem,
df_ri_add_problem, df_scan_add_problem): Added flags parameter.
(df_set_state): Removed function.
(df_set_flags, df_clear_flags, df_delete_basic_block) New functions.
* df-problems.c (df_chain_dump): Removed df parameter.
(df_ru_alloc, df_rd_alloc, df_lr_alloc, df_ur_alloc,
df_urec_alloc, df_chain_alloc, df_ri_alloc): Added all blocks
parameter.
(df_ru_alloc, df_rd_alloc): Now resets all blocks.
(df_rd_bb_local_compute_process_def, df_ur_bb_local_compute,
df_chain_create_bb, df_create_unused_note, df_ri_bb_compute):
Split DF_REF_CLOBBER into DF_REF_MAY_CLOBBER and
DF_REF_MUST_CLOBBER cases.
(df_ru_bb_local_compute_process_def,
df_rd_bb_local_compute_process_def, df_lr_bb_local_compute,
df_lr_bb_local_compute, df_ur_bb_local_compute,
df_chain_create_bb): Made subreg aware.
(df_ru_bb_local_compute, df_rd_bb_local_compute,
df_lr_bb_local_compute, df_lr_bb_local_compute,
df_chain_create_bb): Cleanup to use proper macros.
(df_ur_local_finalize, df_urec_local_finalize): Removed unnecessary
code to fixup bitvectors.
(df_ri_alloc): Cleared lifetime.
(df_ignore_stack_reg, df_kill_notes, df_set_notes_for_mw,
df_create_unused_note): New function.
(df_ri_bb_compute, df_ri_compute): Added code to create/update
REG_DEAD and REG_UNUSED notes as well as register information.
(df_ru_dump, df_rd_dump, df_lr_dump, df_ur_dump, df_urec_dump,
df_chains_dump): Fixed crash if problem was never run.
(df_ru_add_problem, df_rd_add_problem, df_lr_add_problem,
df_ur_add_problem, df_urec_add_problem, df_chain_add_problem,
df_ri_add_problem): Processes flags in uniform manner.
(df_ru_alloc, df_ru_local_compute, df_ru_confluence_n, df_ru_free,
df_ru_dump, df_rd_local_compute, df_rd_confluence_n, df_rd_free,
df_rd_dump, df_urec_free_bb_info): Formatting changes.
* modulo-sched.c (sms_schedule): Added flag parameter to calls.
* see.c (see_initialize_data): Ditto.
* final.c (rest_of_clean_state) Added regstack_completed.
* rtl.h (regstack_completed): Ditto.
* reg-stack.c (regstack_completed): Ditto.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: update3b.diff
Type: text/x-patch
Size: 109718 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20060517/50730b20/attachment.bin>
More information about the Gcc-patches
mailing list