| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
|
|
this patch implements the IPA propagation part of EAF flags handling in
ipa-modref. It extends the local analysis to collect lattice consisting of
flags and escape points. SSA name escapes if it is passed directly or
indirectly to a function call.
If useful flags are found for parameter its escape list is stored into escape
summaries. This time each call site is annotated with info on which function
parameters escape to what argument of function call.
At IPA time we then perform iterative dataflow and produce final flags.
ipa-modref is still cheaper than pure-const when running on cc1plus (about 2-3%
that is what accounts every non-trivial passs) and the dataflow converges in 1
or 2 iterations.
Local analysis does some work to avoid streaming escape points when they are
not useful to determine final flags (that is, local escape analysis determined
good enough flags). For cc1plus there are 225k calls with useful escape
summary.
* ipa-modref.c (escape_point): New type.
(modref_lattice): New type.
(escape_entry): New type.
(escape_summary): New type.
(escape_summaries_t): New type.
(escape_summaries): New static variable.
(eaf_flags_useful_p): New function.
(modref_summary::useful_p): Add new check_flags
attribute; check eaf_flags for usefulness.
(modref_summary_lto): Add arg_flags.
(modref_summary_lto::useful_p): Add new check_flags
attribute; check eaf_flags for usefulness.
(dump_modref_edge_summaries): New function.
(remove_modref_edge_summaries): New function.
(ignore_retval_p): New predicate.
(ignore_stores_p): Also ignore for const.
(remove_summary): Call remove_modref_edge_summaries.
(modref_lattice::init): New member function.
(modref_lattice::release): New member unction.
(modref_lattice::dump): New member function.
(modref_lattice::add_escape_point): New member function.
(modref_lattice::merge): Two new member functions.
(modref_lattice::merge_deref): New member functions.
(modref_lattice::merge_direct_load): New member function.
(modref_lattice::merge_direct_store): New member function.
(call_lhs_flags): Rename to ...
(merge_call_lhs_flags): ... this one; reimplement using
modreflattice.
(analyze_ssa_name_flags): Replace KNOWN_FLAGS param by LATTICE;
add IPA parametr; use modref_lattice.
(analyze_parms): New parameter IPA and SUMMARY_LTO; update for
modref_lattice; initialize escape_summary.
(analyze_function): Allocate escape_summaries; update uses of useful_p.
(modref_write_escape_summary): New function.
(modref_read_escape_summary): New function.
(modref_write): Write escape summary.
(read_section): Read escape summary.
(modref_read): Initialie escape_summaries.
(remap_arg_flags): New function.
(update_signature): Use it.
(escape_map): New structure.
(update_escape_summary_1, update_escape_summary): New functions.
(ipa_merge_modref_summary_after_inlining): Merge escape summaries.
(propagate_unknown_call): Do not remove useless summaries.
(remove_useless_summaries): Remove them here.
(modref_propagate_in_scc): Update; do not dump scc.
(modref_propagate_dump_scc): New function.
(modref_merge_call_site_flags): New function.
(modref_propagate_flags_in_scc): New function.
(pass_ipa_modref::execute): Use modref_propagate_flags_in_scc
and modref_propagate_dump_scc; delete escape_summaries.
(ipa_modref_c_finalize): Remove escape_summaries.
* ipa-modref.h (modref_summary): Update prototype of useful_p.
* params.opt (param=modref-max-escape-points): New param.
* doc/invoke.texi (modref-max-escape-points): Document.
|
|
A minimal patch for the EAF flags discovery. It works only in local ipa-modref
and gives up on cyclic SSA graphs. It improves pt_solution_includes
disambiguations twice.
gcc/Changelog:
* gimple.c: Include ipa-modref-tree.h and ipa-modref.h.
(gimple_call_arg_flags): Use modref to determine flags.
* ipa-modref.c: Include gimple-ssa.h, tree-phinodes.h,
tree-ssa-operands.h, stringpool.h and tree-ssanames.h.
(analyze_ssa_name_flags): Declare.
(modref_summary::useful_p): Summary is also useful if arg flags are
known.
(dump_eaf_flags): New function.
(modref_summary::dump): Use it.
(get_modref_function_summary): Be read for current_function_decl
being NULL.
(memory_access_to): New function.
(deref_flags): New function.
(call_lhs_flags): New function.
(analyze_parms): New function.
(analyze_function): Use it.
* ipa-modref.h (struct modref_summary): Add arg_flags.
* doc/invoke.texi (ipa-modref-max-depth): Document.
* params.opt (ipa-modref-max-depth): New param.
|
|
* ipa-modref.c (modref_summary::dump): Dump writes_errno.
(parm_map_for_arg): Break out from ...
(merge_call_side_effects): ... here.
(get_access_for_fnspec): New function.
(process_fnspec): New function.
(analyze_call): Use it.
(analyze_stmt): Update.
(analyze_function): Initialize writes_errno.
(modref_summaries::duplicate): Duplicate writes_errno.
* ipa-modref.h (struct modref_summary): Add writes_errno.
* tree-ssa-alias.c (call_may_clobber_ref_p_1): Check errno.
|
|
this is largely mechanical patch fixing some suboptimal datastructure decision
in modref. It records three different things
1) optimization_summaries that are used by tree-ssa-alias to disambiguate
(computed by local passes or ipa execute)
2) summaries produced by local analysis and used by the ipa execute
3) summaries_lto produced by analysis when streaming is expected,
streamed, used by ipa execute
All three items are stored in "summaries" datastructure where 1 dn 2
are mixed and differentiated by "finished" flags.
This use extra memory and also makes it impossible to use modref while producing
other IPA summaries (by ipa-prop and ipa-devirt). This patch separates the
summaries into three special purpose datastructures.
There is one fix to propagation in ipa_merge_modref_summary_after_inlining
where check to ignore stores was placed incorrectly. This seems to lead
to increased clobber disambiguations:
Alias oracle query stats:
refs_may_alias_p: 64267006 disambiguations, 74475486 queries
ref_maybe_used_by_call_p: 142119 disambiguations, 65169365 queries
call_may_clobber_ref_p: 22975 disambiguations, 28762 queries
nonoverlapping_component_refs_p: 0 disambiguations, 36803 queries
nonoverlapping_refs_since_match_p: 19401 disambiguations, 55550 must overlaps, 75722 queries
aliasing_component_refs_p: 54714 disambiguations, 759027 queries
TBAA oracle: 23636760 disambiguations 56001742 queries
16112157 are in alias set 0
10614737 queries asked about the same object
125 queries asked about the same alias set
0 access volatile
3994423 are dependent in the DAG
1643540 are aritificially in conflict with void *
Modref stats:
modref use: 11667 disambiguations, 40207 queries
modref clobber: 1508990 disambiguations, 1829697 queries
3916688 tbaa queries (2.140621 per modref query)
623504 base compares (0.340769 per modref query)
PTA query stats:
pt_solution_includes: 967354 disambiguations, 13605701 queries
pt_solutions_intersect: 1032982 disambiguations, 13121107 queries
Bootstrapped/regtested x86_64-linux. I plan to commit it tomorrow if there are
no complains.
gcc/ChangeLog:
2020-10-11 Jan Hubicka <hubicka@ucw.cz>
* ipa-modref.c (modref_summaries): Remove field IPA.
(class modref_summary_lto): New global variable.
(class modref_summaries_lto): New.
(modref_summary::modref_summary): Remove loads_lto and stores_lto.
(modref_summary::~modref_summary): Remove loads_lto and stores_lto.
(modref_summary::useful_p): Do not use lto_useful.
(modref_records_lto): New typedef.
(struct modref_summary_lto): New type.
(modref_summary_lto::modref_summary_lto): New member function.
(modref_summary_lto::~modref_summary_lto): New member function.
(modref_summary_lto::useful_p): New member function.
(modref_summary::dump): Do not handle lto.
(modref_summary_lto::dump): New member function.
(get_modref_function_summary): Use optimization_summary.
(merge_call_side_effects): Use optimization_summary.
(analyze_call): Use optimization_summary.
(struct summary_ptrs): New struture.
(analyze_load): Update to handle separate lto and non-lto summaries.
(analyze_store): Likewise.
(analyze_stmt): Likewise.
(remove_summary): Break out from ...
(analyze_function): ... here; update to handle seprated summaries.
(modref_summaries::insert): Do not handle lto summary.
(modref_summaries_lto::insert): New member function.
(modref_summaries::duplicate): Do not handle lto summary.
(modref_summaries_lto::duplicate): New member function.
(read_modref_records): Expect nolto_ret or lto_ret to be NULL>
(modref_write): Write lto summary.
(read_section): Handle separated summaries.
(modref_read): Initialize separated summaries.
(modref_transform): Handle separated summaries.
(pass_modref::execute): Turn summary to optimization_summary; handle
separate summaries.
(ignore_edge): Handle separate summaries.
(ipa_merge_modref_summary_after_inlining): Likewise.
(collapse_loads): Likewise.
(modref_propagate_in_scc): Likewise.
(pass_ipa_modref::execute): Likewise.
(ipa_modref_c_finalize): Likewise.
* ipa-modref.h (modref_records_lto): Remove typedef.
(struct modref_summary): Remove stores_lto, loads_lto and finished
fields; remove lto_useful_p member function.
|
|
cc1plus stats are now:
Alias oracle query stats:
refs_may_alias_p: 62971744 disambiguations, 73160711 queries
ref_maybe_used_by_call_p: 141176 disambiguations, 63867883 queries
call_may_clobber_ref_p: 23573 disambiguations, 29322 queries
nonoverlapping_component_refs_p: 0 disambiguations, 37720 queries
nonoverlapping_refs_since_match_p: 19432 disambiguations, 55659 must overlaps, 75860 queries
aliasing_component_refs_p: 54724 disambiguations, 753570 queries
TBAA oracle: 24124230 disambiguations 56228428 queries
16058141 are in alias set 0
10338303 queries asked about the same object
125 queries asked about the same alias set
0 access volatile
3919230 are dependent in the DAG
1788399 are aritificially in conflict with void *
Modref stats:
modref use: 10408 disambiguations, 46993 queries
modref clobber: 1418549 disambiguations, 1951251 queries
4898707 tbaa queries (2.510547 per modref query)
396878 base compares (0.203397 per modref query)
PTA query stats:
pt_solution_includes: 975364 disambiguations, 13604284 queries
pt_solutions_intersect: 1026606 disambiguations, 13181198 queries
So compared to
https://gcc.gnu.org/pipermail/gcc-patches/2020-September/554692.html we get 25%
use disambiguations and 91% more clobber disambiguations.
Tramp3d is
Alias oracle query stats:
refs_may_alias_p: 2056905 disambiguations, 2317461 queries
ref_maybe_used_by_call_p: 7137 disambiguations, 2093762 queries
call_may_clobber_ref_p: 234 disambiguations, 234 queries
nonoverlapping_component_refs_p: 0 disambiguations, 4313 queries
nonoverlapping_refs_since_match_p: 329 disambiguations, 10200 must overlaps, 10616 queries
aliasing_component_refs_p: 858 disambiguations, 34600 queries
TBAA oracle: 894996 disambiguations 1695991 queries
138346 are in alias set 0
470668 queries asked about the same object
0 queries asked about the same alias set
0 access volatile
191666 are dependent in the DAG
315 are aritificially in conflict with void *
Modref stats:
modref use: 842 disambiguations, 2265 queries
modref clobber: 14833 disambiguations, 28900 queries
34884 tbaa queries (1.207059 per modref query)
5041 base compares (0.174429 per modref query)
PTA query stats:
pt_solution_includes: 313372 disambiguations, 525724 queries
pt_solutions_intersect: 130374 disambiguations, 415138 queries
So about twice many use and 40% clobber disambiguations.
Bootstrapped/regtested x86_64-linux, I plan to commit it later today after
more testing.
2020-09-26 Jan Hubicka <hubicka@ucw.cz>
* ipa-inline-transform.c: Include ipa-modref-tree.h and ipa-modref.h.
(inline_call): Call ipa_merge_modref_summary_after_inlining.
* ipa-inline.c (ipa_inline): Do not free summaries.
* ipa-modref.c (dump_records): Fix formating.
(merge_call_side_effects): Break out from ...
(analyze_call): ... here; record recursive calls.
(analyze_stmt): Add new parameter RECURSIVE_CALLS.
(analyze_function): Do iterative dataflow on recursive calls.
(compute_parm_map): New function.
(ipa_merge_modref_summary_after_inlining): New function.
(collapse_loads): New function.
(modref_propagate_in_scc): Break out from ...
(pass_ipa_modref::execute): ... here; Do iterative dataflow.
* ipa-modref.h (ipa_merge_modref_summary_after_inlining): Declare.
|
|
this patch fixes bug in tracking memory stats and also I have noticed that while
the pass takes care to stop traking things when things are obviously out of hand
it still keeps summaries that have no useful info for loads or stores and also
many summaries are just copying const/pure attributes. This patch thus also
adds logic to detect if summary is useful and drop it early otherwise. This
reduces number of queries to the oracle and saves memory/lto streaming.
For cc1plus LTO build (configured with --disable-plugin
--enable-checking=release --with-build-config=lto) I now get:
Alias oracle query stats:
refs_may_alias_p: 62488734 disambiguations, 72660949 queries
ref_maybe_used_by_call_p: 128863 disambiguations, 63393551 queries
call_may_clobber_ref_p: 16013 disambiguations, 21776 queries
nonoverlapping_component_refs_p: 0 disambiguations, 37628 queries
nonoverlapping_refs_since_match_p: 19397 disambiguations, 55370 must overlaps, 75516 queries
aliasing_component_refs_p: 54741 disambiguations, 752198 queries
TBAA oracle: 21632692 disambiguations 52565147 queries
15656420 are in alias set 0
10108172 queries asked about the same object
124 queries asked about the same alias set
0 access volatile
3640460 are dependent in the DAG
1527279 are aritificially in conflict with void *
Modref stats:
modref use: 5712 disambiguations, 31221 queries
modref clobber: 684316 disambiguations, 1010000 queries
1779717 tbaa queries (1.762096 per modref query)
PTA query stats:
pt_solution_includes: 947334 disambiguations, 13601373 queries
pt_solutions_intersect: 1011662 disambiguations, 13139565 queries
The number of queries should change, but the number of disambiguations should
not. However comparing with stats here
https://gcc.gnu.org/pipermail/gcc-patches/2020-September/554309.html
I see about 50% drop in clobber disambiguations. There is however same drop in
other alias oracle stats. I suppose someting changed in meanwhile on mainline
because I was basing that on older tree. I tried to proofread changes between
mainline and branch and they seem all quite obvious.
This is consistent with what I get on tramp3d:
Alias oracle query stats:
refs_may_alias_p: 2051320 disambiguations, 2312132 queries
ref_maybe_used_by_call_p: 7058 disambiguations, 2088222 queries
call_may_clobber_ref_p: 232 disambiguations, 232 queries
nonoverlapping_component_refs_p: 0 disambiguations, 4339 queries
nonoverlapping_refs_since_match_p: 329 disambiguations, 10200 must overlaps, 10616 queries
aliasing_component_refs_p: 857 disambiguations, 34639 queries
TBAA oracle: 886768 disambiguations 1670635 queries
131572 are in alias set 0
461689 queries asked about the same object
0 queries asked about the same alias set
0 access volatile
190291 are dependent in the DAG
315 are aritificially in conflict with void *
Modref stats:
modref use: 430 disambiguations, 1885 queries
modref clobber: 9657 disambiguations, 16076 queries
19027 tbaa queries (1.183566 per modref query)
PTA query stats:
pt_solution_includes: 311756 disambiguations, 524179 queries
pt_solutions_intersect: 129689 disambiguations, 415878 queries
In both cases the number of disambiguations should be same (queries are not
comparable).
Bootstrapped/regtested x86_64-linux, comitted.
gcc/ChangeLog:
2020-09-23 Jan Hubicka <hubicka@ucw.cz>
* ipa-modref.c (modref_summary::lto_useful_p): New member function.
(modref_summary::useful_p): New member function.
(analyze_function): Drop useless summaries.
(modref_write): Skip useless summaries.
(pass_ipa_modref::execute): Drop useless summaries.
* ipa-modref.h (struct GTY): Declare useful_p and lto_useful_p.
* tree-ssa-alias.c (dump_alias_stats): Fix.
(modref_may_conflict): Fix stats.
|
|
* ipa-modref-tree.c: Add namespace selftest.
(modref_tree_c_tests): Rename to ...
(ipa_modref_tree_c_tests): ... this.
* ipa-modref.c (pass_modref): Remove destructor.
(ipa_modref_c_finalize): New function.
* ipa-modref.h (ipa_modref_c_finalize): Declare.
* selftest-run-tests.c (selftest::run_tests): Call
ipa_modref_c_finalize.
* selftest.h (ipa_modref_tree_c_tests): Declare.
* toplev.c: Include ipa-modref-tree.h and ipa-modref.h
(toplev::finalize): Call ipa_modref_c_finalize.
|
|
2020-09-19 David Cepelik <d@dcepelik.cz>
Jan Hubicka <hubicka@ucw.cz>
* Makefile.in: Add ipa-modref.c and ipa-modref-tree.c.
* alias.c: (reference_alias_ptr_type_1): Export.
* alias.h (reference_alias_ptr_type_1): Declare.
* common.opt (fipa-modref): New.
* gengtype.c (open_base_files): Add ipa-modref-tree.h and ipa-modref.h
* ipa-modref-tree.c: New file.
* ipa-modref-tree.h: New file.
* ipa-modref.c: New file.
* ipa-modref.h: New file.
* lto-section-in.c (lto_section_name): Add ipa_modref.
* lto-streamer.h (enum lto_section_type): Add LTO_section_ipa_modref.
* opts.c (default_options_table): Enable ipa-modref at -O1+.
* params.opt (-param=modref-max-bases, -param=modref-max-refs,
-param=modref-max-tests): New params.
* passes.def: Schedule pass_modref and pass_ipa_modref.
* timevar.def (TV_IPA_MODREF): New timevar.
(TV_TREE_MODREF): New timevar.
* tree-pass.h (make_pass_modref): Declare.
(make_pass_ipa_modref): Declare.
* tree-ssa-alias.c (dump_alias_stats): Include ipa-modref-tree.h
and ipa-modref.h
(alias_stats): Add modref_use_may_alias, modref_use_no_alias,
modref_clobber_may_alias, modref_clobber_no_alias, modref_tests.
(dump_alias_stats): Dump new stats.
(nonoverlapping_array_refs_p): Fix formating.
(modref_may_conflict): New function.
(ref_maybe_used_by_call_p_1): Use it.
(call_may_clobber_ref_p_1): Use it.
(call_may_clobber_ref_p): Update.
(stmt_may_clobber_ref_p_1): Update.
* tree-ssa-alias.h (call_may_clobber_ref_p_1): Update.
|
