gcc/gcc
Richard Sandiford 6e23549157 vect: Fix load costs for SLP permutes
For the following test case (compiled with load/store lanes
disabled locally):

  void
  f (uint32_t *restrict x, uint8_t *restrict y, int n)
  {
    for (int i = 0; i < n; ++i)
      {
	x[i * 2] = x[i * 2] + y[i * 2];
	x[i * 2 + 1] = x[i * 2 + 1] + y[i * 2];
      }
  }

we have a redundant no-op permute on the x[] load node:

   node 0x4472350 (max_nunits=8, refcnt=2)
          stmt 0 _5 = *_4;
          stmt 1 _13 = *_12;
          load permutation { 0 1 }

Then, when costing it, we pick a cost of 1, even though we need 4 copies
of the x[] load to match a single y[] load:

   ==> examining statement: _5 = *_4;
   Vectorizing an unaligned access.
   vect_model_load_cost: unaligned supported by hardware.
   vect_model_load_cost: inside_cost = 1, prologue_cost = 0 .

The problem is that the code only considers the permutation for
the first scalar iteration, rather than for all VF iterations.

This patch tries to fix that by making vect_transform_slp_perm_load
calculate the value instead.

gcc/
	* tree-vectorizer.h (vect_transform_slp_perm_load): Take an
	optional extra parameter.
	* tree-vect-slp.c (vect_transform_slp_perm_load): Calculate
	the number of loads as well as the number of permutes, taking
	the counting loop from...
	* tree-vect-stmts.c (vect_model_load_cost): ...here.  Use the
	value computed by vect_transform_slp_perm_load for ncopies.
2020-10-29 13:38:01 +00:00
..
ada Daily bump. 2020-10-29 00:16:50 +00:00
analyzer Daily bump. 2020-10-29 00:16:50 +00:00
brig
c Daily bump. 2020-10-29 00:16:50 +00:00
c-family Daily bump. 2020-10-29 00:16:50 +00:00
common Enable GCC to support Intel Key Locker ISA 2020-10-29 16:42:47 +08:00
config Enable GCC to support Intel Key Locker ISA 2020-10-29 16:42:47 +08:00
cp Daily bump. 2020-10-29 00:16:50 +00:00
d Daily bump. 2020-10-28 00:16:38 +00:00
doc Enable GCC to support Intel Key Locker ISA 2020-10-29 16:42:47 +08:00
fortran Daily bump. 2020-10-28 00:16:38 +00:00
ginclude
go libgo: handle linking to NetBSD's versioned symbols 2020-10-28 18:20:50 -07:00
jit
lto Daily bump. 2020-10-29 00:16:50 +00:00
objc Daily bump. 2020-10-12 00:16:25 +00:00
objcp
po
testsuite CSE conversions within sincos 2020-10-29 06:30:50 -03:00
ABOUT-GCC-NLS
acinclude.m4
aclocal.m4
addresses.h
adjust-alignment.c
alias.c
alias.h
align.h
alloc-pool.c
alloc-pool.h
array-traits.h
asan.c
asan.h
attr-fnspec.h Extend builtin fnspecs 2020-10-26 20:22:16 +01:00
attribs.c PR middle-end/97552 - missing waning passing null to a VLA argument declared [static] 2020-10-23 12:37:38 -06:00
attribs.h
auto-inc-dec.c
auto-profile.c
auto-profile.h
backend.h
BASE-VER
basic-block.h
bb-reorder.c
bb-reorder.h
bitmap.c
bitmap.h
brig-builtins.def
builtin-attrs.def Fix fnspec of math builtins 2020-10-27 09:51:56 +01:00
builtin-types.def
builtins.c Add string builtins to builtin_fnspec 2020-10-27 09:01:41 +01:00
builtins.def Fix fnspec of math builtins 2020-10-27 09:51:56 +01:00
builtins.h Generalize compute_objsize to return maximum size/offset instead of failing (PR middle-end/97023). 2020-10-12 09:05:55 -06:00
caller-save.c
calls.c Extend builtin fnspecs 2020-10-26 20:22:16 +01:00
calls.h Convert -Wrestrict pass to ranger. 2020-10-20 20:46:08 +02:00
ccmp.c
ccmp.h
cfg-flags.def
cfg.c Add overloaded debug_bb and debug_bb_n with dump flags 2020-10-26 02:52:39 -05:00
cfg.h Add overloaded debug_bb and debug_bb_n with dump flags 2020-10-26 02:52:39 -05:00
cfganal.c
cfganal.h
cfgbuild.c
cfgbuild.h
cfgcleanup.c
cfgcleanup.h
cfgexpand.c Implement no_stack_protector attribute. 2020-10-22 10:10:50 +02:00
cfgexpand.h
cfghooks.c Separate new_edges compute in copy_bbs 2020-10-21 10:45:08 +02:00
cfghooks.h
cfgloop.c
cfgloop.h
cfgloopanal.c
cfgloopmanip.c
cfgloopmanip.h
cfgrtl.c
cfgrtl.h
cgraph.c cgraph: move former_thunk_p out of CHECKING_P macro. 2020-10-24 08:42:33 +02:00
cgraph.h Implement three-level optimize_for_size predicates 2020-10-26 18:19:48 +01:00
cgraphbuild.c
cgraphclones.c Fix simdclones 2020-10-26 14:11:35 +01:00
cgraphunit.c Fix simdclones 2020-10-26 14:11:35 +01:00
ChangeLog Daily bump. 2020-10-29 00:16:50 +00:00
ChangeLog-1997
ChangeLog-1998
ChangeLog-1999
ChangeLog-2000
ChangeLog-2001
ChangeLog-2002
ChangeLog-2003
ChangeLog-2004
ChangeLog-2005
ChangeLog-2006
ChangeLog-2007
ChangeLog-2008
ChangeLog-2009
ChangeLog-2010
ChangeLog-2011
ChangeLog-2012
ChangeLog-2013
ChangeLog-2014
ChangeLog-2015
ChangeLog-2016
ChangeLog-2017
ChangeLog-2018
ChangeLog-2019
ChangeLog.dataflow
ChangeLog.gimple-classes
ChangeLog.graphite
ChangeLog.jit
ChangeLog.lib
ChangeLog.ptr
ChangeLog.tree-ssa
ChangeLog.tuples
cif-code.def
collect2-aix.c
collect2-aix.h
collect2.c collect-utils.c, lto-wrapper + mkoffload: Improve -save-temps filename 2020-10-20 12:14:03 +02:00
collect2.h
collect-utils.c collect-utils.c, lto-wrapper + mkoffload: Improve -save-temps filename 2020-10-20 12:14:03 +02:00
collect-utils.h collect-utils.c, lto-wrapper + mkoffload: Improve -save-temps filename 2020-10-20 12:14:03 +02:00
color-macros.h
combine-stack-adj.c
combine.c combine: Fix up simplify_shift_const_1 for nested ROTATEs [PR97386] 2020-10-13 19:13:26 +02:00
common.md
common.opt Rename -fevrp-mode= to --param=evrp-mode=. 2020-10-07 18:06:15 +02:00
compare-elim.c
conditions.h
config.build
config.gcc Enable GCC to support Intel Key Locker ISA 2020-10-29 16:42:47 +08:00
config.host
config.in Update check for working assembler --gdwarf-4 option 2020-10-24 09:03:36 -07:00
configure Update check for working assembler --gdwarf-4 option 2020-10-24 09:03:36 -07:00
configure.ac Update check for working assembler --gdwarf-4 option 2020-10-24 09:03:36 -07:00
context.c
context.h
convert.c
convert.h
COPYING
COPYING3
COPYING3.LIB
COPYING.LIB
coretypes.h Implement three-level optimize_for_size predicates 2020-10-26 18:19:48 +01:00
coroutine-builtins.def
coroutine-passes.cc
coverage.c GCOV: do not mangle .gcno files. 2020-10-02 12:10:03 +02:00
coverage.h
cppbuiltin.c
cppbuiltin.h
cppdefault.c
cppdefault.h
cprop.c
cse.c
cselib.c
cselib.h
cstamp-h.in
data-streamer-in.c Add poly_int64 streaming support 2020-10-02 13:01:01 +02:00
data-streamer-out.c Add poly_int64 streaming support 2020-10-02 13:01:01 +02:00
data-streamer.c
data-streamer.h Add poly_int64 streaming support 2020-10-02 13:01:01 +02:00
DATESTAMP Daily bump. 2020-10-29 00:16:50 +00:00
dbgcnt.c dbgcnt: print list after compilation 2020-10-06 15:49:42 +02:00
dbgcnt.def IPA MOD REF: add debug counter. 2020-10-08 13:35:41 +02:00
dbgcnt.h
dbxout.c
dbxout.h
dce.c
dce.h
ddg.c
ddg.h
debug.c
debug.h
defaults.h
DEV-PHASE
df-core.c
df-problems.c
df-scan.c
df.h
dfp.c Fix PR97439 2020-10-22 12:38:01 +02:00
dfp.h
diagnostic-color.c
diagnostic-color.h
diagnostic-core.h
diagnostic-event-id.h
diagnostic-format-json.cc
diagnostic-metadata.h
diagnostic-path.h
diagnostic-show-locus.c
diagnostic-url.h
diagnostic.c
diagnostic.def
diagnostic.h
digraph.cc
digraph.h
dojump.c
dojump.h
dominance.c
dominance.h
domwalk.c
domwalk.h
double-int.c
double-int.h
dse.c
dump-context.h
dumpfile.c
dumpfile.h
dwarf2asm.c
dwarf2asm.h
dwarf2cfi.c
dwarf2out.c Update check for working assembler --gdwarf-4 option 2020-10-24 09:03:36 -07:00
dwarf2out.h
early-remat.c
edit-context.c
edit-context.h
emit-rtl.c
emit-rtl.h
errors.c
errors.h
escaped_string.h
et-forest.c
et-forest.h
except.c
except.h
exec-tool.in
explow.c
explow.h
expmed.c
expmed.h
expr.c middle-end/97521 - always use single-bit bools in mask vector types 2020-10-26 12:28:30 +01:00
expr.h
fibonacci_heap.c
fibonacci_heap.h
file-find.c
file-find.h
file-prefix-map.c
file-prefix-map.h
final.c
fixed-value.c
fixed-value.h
flag-types.h Hybrid EVRP and testcases 2020-10-06 13:03:13 -04:00
flags.h
fold-const-call.c
fold-const-call.h
fold-const.c tree-optimization/97482 - fix split_constant_offset of nop-conversions 2020-10-15 10:54:24 +02:00
fold-const.h
fp-test.c
FSFChangeLog
FSFChangeLog.10
FSFChangeLog.11
function-abi.cc
function-abi.h
function-tests.c
function.c Come up with stack_protector enum. 2020-10-22 10:06:06 +02:00
function.h
fwprop.c
gcc-ar.c
gcc-main.c
gcc-plugin.h
gcc-rich-location.c
gcc-rich-location.h
gcc-symtab.h
gcc.c Update check for working assembler --gdwarf-4 option 2020-10-24 09:03:36 -07:00
gcc.h
gcov-counter.def
gcov-dump.c
gcov-io.c
gcov-io.h gcov-profile: use static pool for TOPN first 2020-10-27 11:49:54 +01:00
gcov-iov.c
gcov-tool.c
gcov.c gcov: fix reading of zero sections. 2020-10-23 16:22:55 +02:00
gcse-common.c
gcse-common.h
gcse.c
gcse.h
gdbasan.in
gdbhooks.py
gdbinit.in
gen-pass-instances.awk
genattr-common.c
genattr.c
genattrtab.c
genautomata.c
gencfn-macros.c
gencheck.c
genchecksum.c
gencodes.c
genconditions.c
genconfig.c
genconstants.c
genemit.c
genenums.c
generic-match-head.c
generic-match.h
genextract.c
genflags.c
gengenrtl.c
gengtype-lex.l
gengtype-parse.c
gengtype-state.c
gengtype.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
gengtype.h
genhooks.c
genmatch.c
genmddeps.c
genmddump.c
genmodes.c
genmultilib
genopinit.c
genoutput.c
genpeep.c
genpreds.c
genrecog.c
gensupport.c
gensupport.h
gentarget-def.c
ggc-common.c
ggc-internal.h
ggc-none.c
ggc-page.c
ggc-tests.c
ggc.h
gimple-array-bounds.cc Correct handling of indices into arrays with elements larger than 1 (PR c++/96511) 2020-10-12 09:04:49 -06:00
gimple-array-bounds.h
gimple-builder.c
gimple-builder.h
gimple-expr.c
gimple-expr.h
gimple-fold.c
gimple-fold.h
gimple-isel.cc
gimple-iterator.c
gimple-iterator.h
gimple-laddress.c
gimple-loop-interchange.cc
gimple-loop-jam.c
gimple-loop-versioning.cc Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
gimple-low.c
gimple-low.h
gimple-match-head.c
gimple-match.h
gimple-predict.h
gimple-pretty-print.c
gimple-pretty-print.h
gimple-range-cache.cc Call infer_non_null() directly when checking for non-null. 2020-10-28 22:02:45 -04:00
gimple-range-cache.h Ranger classes. 2020-10-06 12:47:59 -04:00
gimple-range-edge.cc Do not save hash slots across calls to hash_table::get_or_insert. 2020-10-13 11:02:18 -04:00
gimple-range-edge.h Ranger classes. 2020-10-06 12:47:59 -04:00
gimple-range-gori.cc Tweaks to ranger API routines. 2020-10-27 20:17:29 -04:00
gimple-range-gori.h Ranger classes. 2020-10-06 12:47:59 -04:00
gimple-range.cc Tweaks to ranger API routines. 2020-10-27 20:17:29 -04:00
gimple-range.h Refactor range handling of builtins in vr_values and ranger. 2020-10-20 18:21:23 +02:00
gimple-ssa-backprop.c
gimple-ssa-evrp-analyze.c Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
gimple-ssa-evrp-analyze.h Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
gimple-ssa-evrp.c Don't invoke range_of_expr multiple times. 2020-10-16 15:06:44 -04:00
gimple-ssa-isolate-paths.c
gimple-ssa-nonnull-compare.c
gimple-ssa-split-paths.c
gimple-ssa-sprintf.c Convert sprintf/strlen passes to value query class. 2020-10-01 17:11:17 +02:00
gimple-ssa-store-merging.c PR tree-optimization/97546 Bail out of find_bswap_or_nop on non-INTEGER_CST sizes 2020-10-26 11:43:26 +00:00
gimple-ssa-strength-reduction.c
gimple-ssa-warn-alloca.c Convert -Walloca pass to ranger. 2020-10-20 20:45:14 +02:00
gimple-ssa-warn-restrict.c Convert -Wrestrict pass to ranger. 2020-10-20 20:46:08 +02:00
gimple-ssa-warn-restrict.h Convert -Wrestrict pass to ranger. 2020-10-20 20:46:08 +02:00
gimple-ssa.h
gimple-streamer-in.c
gimple-streamer-out.c
gimple-streamer.h
gimple-walk.c
gimple-walk.h
gimple.c Extend builtin fnspecs 2020-10-26 20:22:16 +01:00
gimple.def
gimple.h SLP vectorize across PHI nodes 2020-10-27 13:17:09 +01:00
gimplify-me.c
gimplify-me.h
gimplify.c openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
gimplify.h
glimits.h
godump.c
graph.c
graph.h
graphds.c
graphds.h
graphite-dependences.c
graphite-isl-ast-to-gimple.c
graphite-optimize-isl.c
graphite-poly.c
graphite-scop-detection.c
graphite-sese-to-poly.c
graphite.c
graphite.h
graphviz.cc
graphviz.h
gsstruct.def
gstab.h
gsyms.h
gsyslimits.h
gtm-builtins.def
haifa-sched.c
hard-reg-set.h
hash-map-tests.c
hash-map-traits.h
hash-map.h
hash-set-tests.c
hash-set.h
hash-table.c
hash-table.h
hash-traits.h
highlev-plugin-common.h
hooks.c
hooks.h
host-default.c
hosthooks-def.h
hosthooks.h
hw-doloop.c
hw-doloop.h
hwint.c
hwint.h
ifcvt.c
ifcvt.h
inchash.c
inchash.h
incpath.c
incpath.h
init-regs.c
input.c
input.h
insn-addr.h
insn-notes.def
int-vector-builder.h
internal-fn.c SLP: fix SVE issues 2020-10-12 15:23:20 +02:00
internal-fn.def Perforate fnspec strings 2020-10-02 15:56:12 +02:00
internal-fn.h
intl.c
intl.h
ipa-comdats.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa-cp.c LTO: get_section: add new argument 2020-10-29 14:32:48 +01:00
ipa-devirt.c
ipa-fnsummary.c LTO: get_section: add new argument 2020-10-29 14:32:48 +01:00
ipa-fnsummary.h Inline functions with builtin_constant_p more agressively. 2020-10-21 20:00:22 +02:00
ipa-icf-gimple.c make use of CALL_FROM_NEW_OR_DELETE_P 2020-10-02 11:22:20 +02:00
ipa-icf-gimple.h
ipa-icf.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa-icf.h
ipa-inline-analysis.c Make default duplicate and insert methods of summaries abort; fix fallout 2020-10-26 11:24:33 +01:00
ipa-inline-transform.c Fix simdclones 2020-10-26 14:11:35 +01:00
ipa-inline.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa-inline.h
ipa-modref-tree.c ipa-modref-tree.c: fix selftest leaks 2020-10-22 06:44:27 -04:00
ipa-modref-tree.h Fix ipa-modref signature updates 2020-10-27 16:25:12 +01:00
ipa-modref.c Fix ipa-modref signature updates 2020-10-27 16:25:12 +01:00
ipa-modref.h Cleanup ipa-modref 2020-10-12 16:17:10 +02:00
ipa-param-manipulation.c Materialize clones on demand 2020-10-22 17:32:32 +02:00
ipa-param-manipulation.h
ipa-polymorphic-call.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa-predicate.c Turn offset_map to HOST_WIDE_INT 2020-10-14 16:07:07 +02:00
ipa-predicate.h Turn offset_map to HOST_WIDE_INT 2020-10-14 16:07:07 +02:00
ipa-profile.c
ipa-prop.c Make default duplicate and insert methods of summaries abort; fix fallout 2020-10-26 11:24:33 +01:00
ipa-prop.h Make default duplicate and insert methods of summaries abort; fix fallout 2020-10-26 11:24:33 +01:00
ipa-pure-const.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa-ref.c
ipa-ref.h
ipa-reference.c Make default duplicate and insert methods of summaries abort; fix fallout 2020-10-26 11:24:33 +01:00
ipa-reference.h
ipa-split.c
ipa-sra.c Make default duplicate and insert methods of summaries abort; fix fallout 2020-10-26 11:24:33 +01:00
ipa-utils.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa-utils.h
ipa-visibility.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ipa.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
ira-build.c
ira-color.c
ira-conflicts.c
ira-costs.c Extend special_memory_constraint. 2020-10-22 10:28:56 +08:00
ira-emit.c
ira-int.h
ira-lives.c
ira.c Extend special_memory_constraint. 2020-10-22 10:28:56 +08:00
ira.h
is-a.h
json.cc
json.h
jump.c
langhooks-def.h
langhooks.c LTO: get_section: add new argument 2020-10-29 14:32:48 +01:00
langhooks.h
LANGUAGES
lcm.c
lcm.h
libfuncs.h
limitx.h
limity.h
lists.c
lock-and-run.sh
loop-doloop.c
loop-init.c
loop-invariant.c
loop-iv.c
loop-unroll.c
loop-unroll.h
lower-subreg.c
lower-subreg.h
lra-assigns.c
lra-coalesce.c
lra-constraints.c Extend special_memory_constraint. 2020-10-22 10:28:56 +08:00
lra-eliminations.c
lra-int.h
lra-lives.c
lra-remat.c
lra-spills.c
lra.c
lra.h
lto-cgraph.c lto: LTO cgraph support for late declare variant resolution [PR96680] 2020-10-28 10:29:09 +01:00
lto-compress.c
lto-compress.h
lto-opts.c
lto-section-in.c
lto-section-names.h
lto-section-out.c
lto-streamer-in.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
lto-streamer-out.c lto: LTO cgraph support for late declare variant resolution [PR96680] 2020-10-28 10:29:09 +01:00
lto-streamer.c
lto-streamer.h lto: LTO cgraph support for late declare variant resolution [PR96680] 2020-10-28 10:29:09 +01:00
lto-wrapper.c lto: no sub-make when --jobserver-auth= is missing 2020-10-27 08:29:46 +01:00
machmode.def
machmode.h
main.c
Makefile.in analyzer: move svalue and region decls to their own header files 2020-10-28 20:09:04 -04:00
match.pd [PATCH] fold x << (n % C) to x << (n & C-1) if C meets power2 2020-10-19 17:26:41 +08:00
mcf.c
mem-stats-traits.h
mem-stats.h
memmodel.h
memory-block.cc
memory-block.h
mkconfig.sh
mode-classes.def
mode-switching.c
modulo-sched.c
multiple_target.c
omp-builtins.def
omp-expand.c openmp: Improve composite triangular loop lowering and expansion 2020-10-13 09:30:47 +02:00
omp-expand.h
omp-general.c lto: LTO cgraph support for late declare variant resolution [PR96680] 2020-10-28 10:29:09 +01:00
omp-general.h
omp-low.c openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
omp-low.h
omp-offload.c openmp: Implicitly discover declare target for variants of declare variant calls 2020-10-28 10:36:31 +01:00
omp-offload.h
omp-simd-clone.c
omp-simd-clone.h
ONEWS
opt-functions.awk
opt-gather.awk
opt-include.awk
opt-problem.cc
opt-problem.h
opt-read.awk
opt-suggestions.c
opt-suggestions.h
optabs-libfuncs.c
optabs-libfuncs.h
optabs-query.c
optabs-query.h
optabs-tree.c
optabs-tree.h
optabs.c
optabs.def
optabs.h
optc-gen.awk opts: Sanity check for param names. 2020-10-29 11:53:59 +01:00
optc-save-gen.awk options: Avoid unused variable mask warning [PR97305] 2020-10-07 10:52:47 +02:00
opth-gen.awk options: Save and restore opts_set for Optimization and Target options fallout 2020-10-05 09:34:42 +02:00
optinfo-emit-json.cc
optinfo-emit-json.h
optinfo.cc
optinfo.h
opts-common.c Add -fdiagnostics-path-format=separate-events to -fdiagnostics-plain-output 2020-10-07 09:37:11 -04:00
opts-diagnostic.h
opts-global.c dbgcnt: print list after compilation 2020-10-06 15:49:42 +02:00
opts.c dbgcnt: print list after compilation 2020-10-06 15:49:42 +02:00
opts.h
ordered-hash-map-tests.cc
ordered-hash-map.h
output.h LTO: get_section: add new argument 2020-10-29 14:32:48 +01:00
params.opt opts: Sanity check for param names. 2020-10-29 11:53:59 +01:00
pass_manager.h
passes.c lto: LTO cgraph support for late declare variant resolution [PR96680] 2020-10-28 10:29:09 +01:00
passes.def Materialize clones on demand 2020-10-22 17:32:32 +02:00
plugin.c
plugin.def
plugin.h
poly-int-types.h
poly-int.h
postreload-gcse.c
postreload.c
predict.c Implement three-level optimize_for_size predicates 2020-10-26 18:19:48 +01:00
predict.def
predict.h Implement three-level optimize_for_size predicates 2020-10-26 18:19:48 +01:00
prefix.c
prefix.h
pretty-print.c
pretty-print.h
print-rtl-function.c
print-rtl.c Add overloaded debug_bb and debug_bb_n with dump flags 2020-10-26 02:52:39 -05:00
print-rtl.h
print-tree.c
print-tree.h
profile-count.c IPA: fix profile handling in IRA 2020-10-15 09:56:55 +02:00
profile-count.h
profile.c
profile.h
range-op.cc Handle signed 1-bit ranges in irange::invert. 2020-10-26 19:05:53 +01:00
range-op.h
range.cc
range.h
read-md.c
read-md.h
read-rtl-function.c
read-rtl-function.h
read-rtl.c
README.Portability
real.c
real.h
realmpfr.c
realmpfr.h
recog.c Extend special_memory_constraint. 2020-10-22 10:28:56 +08:00
recog.h
ree.c
reg-notes.def
reg-stack.c
regcprop.c
regcprop.h
reginfo.c
regrename.c
regrename.h
regs.h IPA: fix profile handling in IRA 2020-10-15 09:56:55 +02:00
regset.h
regstat.c
reload1.c
reload.c
reload.h
reorg.c
resource.c
resource.h
rtl-error.c
rtl-error.h
rtl-iter.h
rtl-tests.c
rtl.c
rtl.def
rtl.h Extend special_memory_constraint. 2020-10-22 10:28:56 +08:00
rtlanal.c
rtlhash.c
rtlhash.h
rtlhooks-def.h
rtlhooks.c
rtx-vector-builder.c
rtx-vector-builder.h
run-rtl-passes.c
run-rtl-passes.h
sancov.c
sanitizer.def
sanopt.c
sbitmap.c middle-end/97554 - avoid overflow in alloc size compute 2020-10-26 11:33:50 +01:00
sbitmap.h
sched-deps.c
sched-ebb.c
sched-int.h
sched-rgn.c
sel-sched-dump.c
sel-sched-dump.h
sel-sched-ir.c
sel-sched-ir.h
sel-sched.c
sel-sched.h
selftest-diagnostic.c
selftest-diagnostic.h
selftest-rtl.c
selftest-rtl.h
selftest-run-tests.c
selftest.c
selftest.h
sese.c
sese.h
shortest-paths.h
shrink-wrap.c
shrink-wrap.h
signop.h
simplify-rtx.c Simplify vec_select of a subreg of X to just a vec_select of X. 2020-10-22 11:37:11 +08:00
sort.cc
sparseset.c
sparseset.h
spellcheck-tree.c
spellcheck-tree.h
spellcheck.c
spellcheck.h
sreal.c
sreal.h
ssa-iterators.h
ssa.h
stab.def
stack-ptr-mod.c
statistics.c
statistics.h
stmt.c
stmt.h
stor-layout.c stor-layout: Reject forming arrays with elt sizes not divisible by elt alignment [PR97164] 2020-10-23 10:07:36 +02:00
stor-layout.h
store-motion.c
streamer-hooks.c
streamer-hooks.h
stringpool.c
stringpool.h
substring-locations.c
substring-locations.h
symbol-summary.h call_summary: move hooks to base. 2020-10-27 08:25:51 +01:00
symtab-thunks.cc Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
symtab-thunks.h Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
symtab.c lto: LTO cgraph support for late declare variant resolution [PR96680] 2020-10-28 10:29:09 +01:00
sync-builtins.def
system.h
target-def.h
target-globals.c
target-globals.h
target-hooks-macros.h
target-insns.def
target.def
target.h
targhooks.c
targhooks.h
timevar.c
timevar.def
timevar.h
toplev.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
toplev.h
tracer.c [gimple] Move can_duplicate_bb_p to gimple_can_duplicate_bb_p 2020-10-14 14:37:03 +02:00
tracer.h
trans-mem.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
trans-mem.h
tree-affine.c
tree-affine.h
tree-call-cdce.c
tree-cfg.c Avoid changing PHIs in GIMPLE split_edge 2020-10-20 14:21:01 +02:00
tree-cfg.h
tree-cfgcleanup.c
tree-cfgcleanup.h
tree-chrec.c
tree-chrec.h
tree-complex.c cplxlower: Avoid a transform when looking at a default definition 2020-10-19 19:21:10 +02:00
tree-core.h openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
tree-data-ref.c tree-optimization/97482 - fix split_constant_offset of nop-conversions 2020-10-15 10:54:24 +02:00
tree-data-ref.h
tree-dfa.c
tree-dfa.h
tree-diagnostic-path.cc
tree-diagnostic.c
tree-diagnostic.h
tree-dump.c
tree-dump.h
tree-eh.c
tree-eh.h
tree-emutls.c
tree-hash-traits.h
tree-hasher.h
tree-if-conv.c
tree-if-conv.h
tree-inline.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
tree-inline.h
tree-into-ssa.c Commonize handling of attr-fnspec 2020-10-02 13:31:05 +02:00
tree-into-ssa.h
tree-iterator.c
tree-iterator.h
tree-loop-distribution.c
tree-nested.c openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
tree-nested.h Move nested function info out of cgraph_node 2020-10-22 06:33:34 +02:00
tree-nrv.c Disable TBAA in some uses of call_may_clobber_ref_p 2020-10-08 17:23:16 +02:00
tree-object-size.c
tree-object-size.h
tree-outof-ssa.c
tree-outof-ssa.h
tree-parloops.c
tree-parloops.h
tree-pass.h Materialize clones on demand 2020-10-22 17:32:32 +02:00
tree-phinodes.c
tree-phinodes.h
tree-predcom.c
tree-pretty-print.c openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
tree-pretty-print.h
tree-profile.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
tree-scalar-evolution.c
tree-scalar-evolution.h
tree-sra.c
tree-sra.h
tree-ssa-address.c
tree-ssa-address.h
tree-ssa-alias.c Re-enable fnspec checking once fortran frontend is fixed. 2020-10-27 10:23:46 +01:00
tree-ssa-alias.h Disable TBAA in some uses of call_may_clobber_ref_p 2020-10-08 17:23:16 +02:00
tree-ssa-ccp.c Use EAF_RETURN_ARG in tree-ssa-ccp.c 2020-10-27 09:03:45 +01:00
tree-ssa-ccp.h
tree-ssa-coalesce.c
tree-ssa-coalesce.h
tree-ssa-copy.c Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
tree-ssa-dce.c c++: Set CALL_FROM_NEW_OR_DELETE_P on more calls. 2020-10-02 11:22:20 +02:00
tree-ssa-dce.h
tree-ssa-dom.c Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
tree-ssa-dom.h
tree-ssa-dse.c
tree-ssa-dse.h
tree-ssa-forwprop.c
tree-ssa-ifcombine.c
tree-ssa-live.c
tree-ssa-live.h
tree-ssa-loop-ch.c [tree-ssa-loop-ch] Add missing NULL test for dump_file 2020-10-07 08:06:47 +02:00
tree-ssa-loop-im.c
tree-ssa-loop-ivcanon.c
tree-ssa-loop-ivopts.c Do not use doloop pattern with pragma Unroll 2020-10-23 09:58:33 +02:00
tree-ssa-loop-ivopts.h
tree-ssa-loop-manip.c
tree-ssa-loop-manip.h
tree-ssa-loop-niter.c random memory leak fixes 2020-10-09 10:40:44 +02:00
tree-ssa-loop-niter.h
tree-ssa-loop-prefetch.c
tree-ssa-loop-split.c tree-optimization/97357 - avoid abnormals in loop splitting conditions 2020-10-12 10:27:27 +02:00
tree-ssa-loop-unswitch.c
tree-ssa-loop.c
tree-ssa-loop.h
tree-ssa-math-opts.c CSE conversions within sincos 2020-10-29 06:30:50 -03:00
tree-ssa-operands.c
tree-ssa-operands.h
tree-ssa-phiopt.c phiopt: Optimize x ? __builtin_clz (x) : 32 in GIMPLE fallout [PR97503] 2020-10-22 09:36:25 +02:00
tree-ssa-phiprop.c
tree-ssa-pre.c
tree-ssa-propagate.c Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
tree-ssa-propagate.h Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
tree-ssa-reassoc.c
tree-ssa-sccvn.c Disable TBAA in some uses of call_may_clobber_ref_p 2020-10-08 17:23:16 +02:00
tree-ssa-sccvn.h
tree-ssa-scopedtables.c
tree-ssa-scopedtables.h
tree-ssa-sink.c tree-optimization/97330 - fix bad load sinking 2020-10-08 14:22:28 +02:00
tree-ssa-strlen.c Generalize compute_objsize to return maximum size/offset instead of failing (PR middle-end/97023). 2020-10-12 09:05:55 -06:00
tree-ssa-strlen.h Convert sprintf/strlen passes to value query class. 2020-10-01 17:11:17 +02:00
tree-ssa-structalias.c Fix simdclones 2020-10-26 14:11:35 +01:00
tree-ssa-tail-merge.c
tree-ssa-ter.c
tree-ssa-ter.h
tree-ssa-threadbackward.c
tree-ssa-threadedge.c Convert vr-values to value query class. 2020-10-01 17:10:47 +02:00
tree-ssa-threadedge.h
tree-ssa-threadupdate.c
tree-ssa-threadupdate.h
tree-ssa-uncprop.c
tree-ssa-uninit.c
tree-ssa.c
tree-ssa.h
tree-ssanames.c
tree-ssanames.h
tree-stdarg.c
tree-stdarg.h
tree-streamer-in.c
tree-streamer-out.c
tree-streamer.c
tree-streamer.h
tree-switch-conversion.c
tree-switch-conversion.h
tree-tailcall.c Disable TBAA in some uses of call_may_clobber_ref_p 2020-10-08 17:23:16 +02:00
tree-vect-data-refs.c dump reason for throwing away SLP instance 2020-10-28 14:15:37 +01:00
tree-vect-generic.c
tree-vect-loop-manip.c SLP vectorize across PHI nodes 2020-10-27 13:17:09 +01:00
tree-vect-loop.c SLP vectorize across PHI nodes 2020-10-27 13:17:09 +01:00
tree-vect-patterns.c SLP: fix SVE issues 2020-10-12 15:23:20 +02:00
tree-vect-slp.c vect: Fix load costs for SLP permutes 2020-10-29 13:38:01 +00:00
tree-vect-stmts.c vect: Fix load costs for SLP permutes 2020-10-29 13:38:01 +00:00
tree-vector-builder.c
tree-vector-builder.h
tree-vectorizer.c SLP vectorize across PHI nodes 2020-10-27 13:17:09 +01:00
tree-vectorizer.h vect: Fix load costs for SLP permutes 2020-10-29 13:38:01 +00:00
tree-vrp.c Move simplify_cond_using_ranges_2 to tree-vrp.c 2020-10-21 09:47:27 +02:00
tree-vrp.h
tree.c openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
tree.def
tree.h openmp: Parsing and some semantic analysis of OpenMP allocate clause 2020-10-28 10:38:01 +01:00
treestruct.def
tristate.cc
tristate.h
tsan.c
tsan.h
tsystem.h
typeclass.h
typed-splay-tree.c
typed-splay-tree.h
ubsan.c
ubsan.h
unique-ptr-tests.cc
valtrack.c
valtrack.h
value-prof.c Move thunks out of cgraph_node 2020-10-23 21:44:23 +02:00
value-prof.h
value-query.cc Check for undefined before not returning a constant value 2020-10-21 20:02:22 -04:00
value-query.h Initial implementation of value query class. 2020-10-01 14:55:08 +02:00
value-range-equiv.cc
value-range-equiv.h
value-range.cc value-range: Give up on POLY_INT_CST ranges [PR97457] 2020-10-28 19:05:49 +00:00
value-range.h Simplify and split irange::copy_legacy_range into two functions. 2020-10-20 15:53:22 +02:00
var-tracking.c
varasm.c LTO: get_section: add new argument 2020-10-29 14:32:48 +01:00
varasm.h
varpool.c
vec-perm-indices.c
vec-perm-indices.h
vec.c
vec.h Remove STMT_VINFO_SAME_ALIGN_REFS 2020-10-13 13:35:11 +02:00
vector-builder.h
version.c
version.h
vmsdbg.h
vmsdbgout.c
vr-values.c Adjust overflow for invariants in bounds_of_var_in_loop. 2020-10-21 10:44:10 +02:00
vr-values.h Move simplify_cond_using_ranges_2 to tree-vrp.c 2020-10-21 09:47:27 +02:00
vtable-verify.c
vtable-verify.h
web.c
wide-int-bitmask.h
wide-int-print.cc
wide-int-print.h
wide-int.cc wide-int: Fix up set_bit_large 2020-10-28 10:24:20 +01:00
wide-int.h
xcoff.h
xcoffout.c
xcoffout.h

Copyright (C) 2000-2020 Free Software Foundation, Inc.

This file is intended to contain a few notes about writing C code
within GCC so that it compiles without error on the full range of
compilers GCC needs to be able to compile on.

The problem is that many ISO-standard constructs are not accepted by
either old or buggy compilers, and we keep getting bitten by them.
This knowledge until now has been sparsely spread around, so I
thought I'd collect it in one useful place.  Please add and correct
any problems as you come across them.

I'm going to start from a base of the ISO C90 standard, since that is
probably what most people code to naturally.  Obviously using
constructs introduced after that is not a good idea.

For the complete coding style conventions used in GCC, please read
http://gcc.gnu.org/codingconventions.html


String literals
---------------

Some compilers like MSVC++ have fairly low limits on the maximum
length of a string literal; 509 is the lowest we've come across.  You
may need to break up a long printf statement into many smaller ones.


Empty macro arguments
---------------------

ISO C (6.8.3 in the 1990 standard) specifies the following:

If (before argument substitution) any argument consists of no
preprocessing tokens, the behavior is undefined.

This was relaxed by ISO C99, but some older compilers emit an error,
so code like

#define foo(x, y) x y
foo (bar, )

needs to be coded in some other way.


Avoid unnecessary test before free
----------------------------------

Since SunOS 4 stopped being a reasonable portability target,
(which happened around 2007) there has been no need to guard
against "free (NULL)".  Thus, any guard like the following
constitutes a redundant test:

  if (P)
    free (P);

It is better to avoid the test.[*]
Instead, simply free P, regardless of whether it is NULL.

[*] However, if your profiling exposes a test like this in a
performance-critical loop, say where P is nearly always NULL, and
the cost of calling free on a NULL pointer would be prohibitively
high, consider using __builtin_expect, e.g., like this:

  if (__builtin_expect (ptr != NULL, 0))
    free (ptr);



Trigraphs
---------

You weren't going to use them anyway, but some otherwise ISO C
compliant compilers do not accept trigraphs.


Suffixes on Integer Constants
-----------------------------

You should never use a 'l' suffix on integer constants ('L' is fine),
since it can easily be confused with the number '1'.


			Common Coding Pitfalls
			======================

errno
-----

errno might be declared as a macro.


Implicit int
------------

In C, the 'int' keyword can often be omitted from type declarations.
For instance, you can write

  unsigned variable;

as shorthand for

  unsigned int variable;

There are several places where this can cause trouble.  First, suppose
'variable' is a long; then you might think

  (unsigned) variable

would convert it to unsigned long.  It does not.  It converts to
unsigned int.  This mostly causes problems on 64-bit platforms, where
long and int are not the same size.

Second, if you write a function definition with no return type at
all:

  operate (int a, int b)
  {
    ...
  }

that function is expected to return int, *not* void.  GCC will warn
about this.

Implicit function declarations always have return type int.  So if you
correct the above definition to

  void
  operate (int a, int b)
  ...

but operate() is called above its definition, you will get an error
about a "type mismatch with previous implicit declaration".  The cure
is to prototype all functions at the top of the file, or in an
appropriate header.

Char vs unsigned char vs int
----------------------------

In C, unqualified 'char' may be either signed or unsigned; it is the
implementation's choice.  When you are processing 7-bit ASCII, it does
not matter.  But when your program must handle arbitrary binary data,
or fully 8-bit character sets, you have a problem.  The most obvious
issue is if you have a look-up table indexed by characters.

For instance, the character '\341' in ISO Latin 1 is SMALL LETTER A
WITH ACUTE ACCENT.  In the proper locale, isalpha('\341') will be
true.  But if you read '\341' from a file and store it in a plain
char, isalpha(c) may look up character 225, or it may look up
character -31.  And the ctype table has no entry at offset -31, so
your program will crash.  (If you're lucky.)

It is wise to use unsigned char everywhere you possibly can.  This
avoids all these problems.  Unfortunately, the routines in <string.h>
take plain char arguments, so you have to remember to cast them back
and forth - or avoid the use of strxxx() functions, which is probably
a good idea anyway.

Another common mistake is to use either char or unsigned char to
receive the result of getc() or related stdio functions.  They may
return EOF, which is outside the range of values representable by
char.  If you use char, some legal character value may be confused
with EOF, such as '\377' (SMALL LETTER Y WITH UMLAUT, in Latin-1).
The correct choice is int.

A more subtle version of the same mistake might look like this:

  unsigned char pushback[NPUSHBACK];
  int pbidx;
  #define unget(c) (assert(pbidx < NPUSHBACK), pushback[pbidx++] = (c))
  #define get(c) (pbidx ? pushback[--pbidx] : getchar())
  ...
  unget(EOF);

which will mysteriously turn a pushed-back EOF into a SMALL LETTER Y
WITH UMLAUT.


Other common pitfalls
---------------------

o Expecting 'plain' char to be either sign or unsigned extending.

o Shifting an item by a negative amount or by greater than or equal to
  the number of bits in a type (expecting shifts by 32 to be sensible
  has caused quite a number of bugs at least in the early days).

o Expecting ints shifted right to be sign extended.

o Modifying the same value twice within one sequence point.

o Host vs. target floating point representation, including emitting NaNs
  and Infinities in a form that the assembler handles.

o qsort being an unstable sort function (unstable in the sense that
  multiple items that sort the same may be sorted in different orders
  by different qsort functions).

o Passing incorrect types to fprintf and friends.

o Adding a function declaration for a module declared in another file to
  a .c file instead of to a .h file.