diff options
| author | Jakub Jelinek <jakub@redhat.com> | 2025-04-22 21:27:28 +0200 |
|---|---|---|
| committer | Jakub Jelinek <jakub@gcc.gnu.org> | 2025-04-22 21:27:28 +0200 |
| commit | 4b62cf555b5446cb02fc471519cf1afa09e1a108 (patch) | |
| tree | c6e937430586b493cb64f9d966c022bbebfe2783 /gcc | |
| parent | 49e79b86936373fc70c3bdea3b2d8353cf8957b1 (diff) | |
| download | gcc-4b62cf555b5446cb02fc471519cf1afa09e1a108.zip gcc-4b62cf555b5446cb02fc471519cf1afa09e1a108.tar.gz gcc-4b62cf555b5446cb02fc471519cf1afa09e1a108.tar.bz2 | |
rs6000: Ignore OPTION_MASK_SAVE_TOC_INDIRECT differences in inlining decisions [PR119327]
The following testcase FAILs because the always_inline function can't
be inlined.
The rs6000 backend has similarly to other targets a hook which rejects
inlining which would bring in new ISAs which aren't there in the caller.
And this hook rejects this because of OPTION_MASK_SAVE_TOC_INDIRECT
differences.
This flag is set if explicitly requested or by default depending on
whether the current function looks hot (or at least not cold):
if ((rs6000_isa_flags_explicit & OPTION_MASK_SAVE_TOC_INDIRECT) == 0
&& flag_shrink_wrap_separate
&& optimize_function_for_speed_p (cfun))
rs6000_isa_flags |= OPTION_MASK_SAVE_TOC_INDIRECT;
The target nodes that are being compared here are actually the default
target node (which was created when cfun was NULL) vs. one that was
created for the always_inline function when it wasn't NULL, so one
doesn't have it, the other does.
In any case, this flag feels like a tuning decision rather than hard
ISA requirement and I see no problems why we couldn't inline
even explicit -msave-toc-indirect function into -mno-save-toc-indirect
or vice versa.
We already ignore OPTION_MASK_P{8,10}_FUSION which are also more
like tuning flags.
2025-04-22 Jakub Jelinek <jakub@redhat.com>
PR target/119327
* config/rs6000/rs6000.cc (rs6000_can_inline_p): Ignore also
OPTION_MASK_SAVE_TOC_INDIRECT differences.
* g++.dg/opt/pr119327.C: New test.
Diffstat (limited to 'gcc')
| -rw-r--r-- | gcc/config/rs6000/rs6000.cc | 11 | ||||
| -rw-r--r-- | gcc/testsuite/g++.dg/opt/pr119327.C | 16 |
2 files changed, 23 insertions, 4 deletions
diff --git a/gcc/config/rs6000/rs6000.cc b/gcc/config/rs6000/rs6000.cc index 737c3d6..12dbde2 100644 --- a/gcc/config/rs6000/rs6000.cc +++ b/gcc/config/rs6000/rs6000.cc @@ -25765,10 +25765,13 @@ rs6000_can_inline_p (tree caller, tree callee) } } - /* Ignore -mpower8-fusion and -mpower10-fusion options for inlining - purposes. */ - callee_isa &= ~(OPTION_MASK_P8_FUSION | OPTION_MASK_P10_FUSION); - explicit_isa &= ~(OPTION_MASK_P8_FUSION | OPTION_MASK_P10_FUSION); + /* Ignore -mpower8-fusion, -mpower10-fusion and -msave-toc-indirect options + for inlining purposes. */ + HOST_WIDE_INT ignored_isas = (OPTION_MASK_P8_FUSION + | OPTION_MASK_P10_FUSION + | OPTION_MASK_SAVE_TOC_INDIRECT); + callee_isa &= ~ignored_isas; + explicit_isa &= ~ignored_isas; /* The callee's options must be a subset of the caller's options, i.e. a vsx function may inline an altivec function, but a no-vsx function diff --git a/gcc/testsuite/g++.dg/opt/pr119327.C b/gcc/testsuite/g++.dg/opt/pr119327.C new file mode 100644 index 0000000..598ae1c --- /dev/null +++ b/gcc/testsuite/g++.dg/opt/pr119327.C @@ -0,0 +1,16 @@ +// PR target/119327 +// { dg-do compile { target c++11 } } +// { dg-options "-Os" } + +#pragma GCC optimize "fp-contract=off" + +template <class T> +void +foo (T f) +{ + f (); +} + +struct S { + S () { [] {}; foo ([] __attribute__((always_inline)) {}); } +} s; |