PR target/62631
* config/sparc/sparc.h (TARGET_HARD_MUL): Remove TARGET_V8PLUS.
(TARGET_HARD_MUL32): Rewrite based on TARGET_HARD_MUL.
* config/sparc/sparc.c (sparc_rtx_costs) <MULT>: Return costs based on
int_mulX for integers in 64-bit mode if TARGET_HARD_MUL is not set.
From-SVN: r220369
PR other/63504
* combine.c (reg_n_sets_max): New variable.
(can_change_dest_mode, reg_nonzero_bits_for_combine,
reg_num_sign_bit_copies_for_combine, get_last_value_validate,
get_last_value): Use REG_N_SETS only on pseudos < reg_n_sets_max.
(try_combine): Use INC_REG_N_SETS only on pseudos < reg_n_sets_max.
(rest_of_handle_combine): Initialize reg_n_sets_max.
From-SVN: r220368
Consider the example code mentionned in this PR:
$ cat -n test.c
1 #define C(a, b) a ## b
2 #define L(x) C(L, x)
3 #define M(a) goto L(__LINE__); __LINE__; L(__LINE__):
4 M(a /* --> this is the line of the expansion point of M. */
5 ); /* --> this is the line of the end of the invocation of M. */
$
"cc1 -quiet -E test.c" yields:
goto L5; 5; L4:
;
Notice how we have a 'L4' there, where it should be L5. That is the issue.
My understanding is that during the *second* expansion of __LINE__
(the one between the two L(__LINE__)), builtin_macro() is called by
enter_macro_context() with the location of the expansion point of M
(which is at line 4). Then _cpp_builtin_macro_text() expands __LINE__
into the line number of the location of the last token that has been
lexed, which is the location of the closing parenthesis of the
invocation of M, at line 5. So that invocation of __LINE__ is
expanded into 5.
Now let's see why the last invocation of __LINE__ is expanded into 4.
In builtin_macro(), we have this code at some point:
/* Set pfile->cur_token as required by _cpp_lex_direct. */
pfile->cur_token = _cpp_temp_token (pfile);
cpp_token *token = _cpp_lex_direct (pfile);
/* We should point to the expansion point of the builtin macro. */
token->src_loc = loc;
The first two statements insert a new token in the stream of lexed
token and pfile->cur_token[-1], is the "new" last token that has been
lexed. But the location of pfile->cur_token[-1] is the same location
as the location of the "previous" pfile->cur_token[-1], by courtesy of
_cpp_temp_token(). So normally, in subsequent invocations of
builtin_macro(), the location of pfile->cur_token[-1] should always be
the location of the closing parenthesis of the invocation of M at line
5. Except that that code in master now has the statement
"token->src_loc = loc;" on the next line. That statement actually
sets the location of pfile->cur_token[-1] to 'loc'. Which is the
location of the expansion point of M, which is on line 4.
So in the subsequent call to builtin_macro() (for the last expansion
of __LINE__ in L(__LINE__)), for _cpp_builtin_macro_text(),
pfile->cur_token[-1].src_loc is going to have a line number of 4.
I think the core issue here is that the location that is passed to
builtin_macro() from enter_macro_context() is not correct when we are
in presence of a top-most function-like macro invocation; in that
case, that location should be the location of the closing parenthesis
of the macro invocation. Otherwise, if we are in presence of a a
top-most object-like macro invocation then the location passed down
to builtin_macro should be the location of the expansion point of the
macro.
That way, in the particular case of the input code above, the location
received by builtin_macro() will always have line number 5.
Boostrapped and tested on x86_64-unknown-linux-gnu against trunk.
libcpp/ChangeLog:
* internal.h (cpp_reader::top_most_macro_node): New data member.
* macro.c (enter_macro_context): Pass the location of the end of
the top-most invocation of the function-like macro, or the
location of the expansion point of the top-most object-like macro.
(cpp_get_token_1): Store the top-most macro node in the new
pfile->top_most_macro_node data member.
(_cpp_pop_context): Clear the new cpp_reader::top_most_macro_node
data member.
gcc/testsuite/ChangeLog:
* gcc.dg/cpp/builtin-macro-1.c: New test case.
Signed-off-by: Dodji Seketeli <dodji@redhat.com>
From-SVN: r220367
* ipa-inline.c (early_inliner): Skip inlining only in always_inlined;
if some always_inline was inlined, apply changes before inlining
heuristically.
* g++.dg/ipa/devirt-37.C: Disable early inlining.
From-SVN: r220359
2015-02-02 Christophe Lyon <christophe.lyon@linaro.org>
* gcc.target/aarch64/advsimd-intrinsics/arm-neon-ref.h
(_ARM_FPSRC): Add DN and AHP fields.
(clean_results): Force DN=1 on AArch64.
* gcc.target/aarch64/advsimd-intrinsics/binary_op_no64.inc: New file.
* gcc.target/aarch64/advsimd-intrinsics/vhadd.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vhsub.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vmax.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vmin.c: New file.
* gcc.target/aarch64/advsimd-intrinsics/vrhadd.c: New file.
From-SVN: r220353
2015-02-02 Tejas Belagod <tejas.belagod@arm.com>
Andrew Pinski <pinskia@gcc.gnu.org>
Jakub Jelinek <jakub@gcc.gnu.org>
PR target/64231
* config/aarch64/aarch64.c (aarch64_classify_symbol): Fix large
integer typing for small model. Use IN_RANGE.
Co-Authored-By: Andrew Pinski <pinskia@gcc.gnu.org>
Co-Authored-By: Jakub Jelinek <jakub@gcc.gnu.org>
From-SVN: r220348
gcc/
* config/rs6000/rs6000.c (rs6000_call_aix): Use unspec rather
than mem for toc_restore.
* config/rs6000/rs6000.md (UNSPEC_TOCSLOT): Define.
(call_indirect_aix, call_value_indirect_aix): Adjust to suit.
(call_indirect_elfv2, call_value_indirect_elfv2): Likewise.
gcc/testsuite/
* gcc.target/powerpc/cprophard.c: New.
From-SVN: r220344
PR libgomp/64635
* configure.tgt (*-*-aix*): Use standard posix plugin-suffix.h.
Link with -lpthread.
* config/aix/plugin-suffix.h: Delete.
From-SVN: r220341
PR debug/64817
* cfgexpand.c (deep_ter_debug_map): New variable.
(avoid_deep_ter_for_debug): New function.
(expand_debug_expr): If TERed SSA_NAME is in
deep_ter_debug_map, use the corresponding DEBUG_EXPR_DECL
instead of trying to expand SSA_NAME's def stmt.
(expand_debug_locations): When expanding debug bind
of a DEBUG_EXPR_DECL to corresponding SSA_NAME,
temporarily remove the DEBUG_EXPR_DECL from deep_ter_debug_map's
value.
(pass_expand::execute): Call avoid_deep_ter_for_debug on
all debug bind stmts. Delete deep_ter_debug_map after
expand_debug_location if non-NULL and clear it.
* gcc.dg/pr64817-1.c: New test.
* gcc.dg/pr64817-2.c: New test.
From-SVN: r220320
PR libstdc++/64883
* include/c_global/cstdio (gets): Use __deprecated__ attribute instead
of deprecated.
* include/c_std/cstdio (gets): Likewise.
* testsuite/17_intro/headers/c++1998/all_attributes.cc: Avoid clashing
with attributes used in darwin headers.
From-SVN: r220318
* haifa-sched.c (INSN_RFS_DEBUG_ORIG_ORDER): New access macro.
(rank_for_schedule_debug): Split from ...
(rank_for_schedule): ... this.
(ready_sort): Sort DEBUG_INSNs separately from normal INSNs.
* sched-int.h (struct _haifa_insn_data): New field rfs_debug_orig_order.
From-SVN: r220316
Similar to the issue with fatal_error that I fixed in
<https://gcc.gnu.org/ml/gcc-patches/2015-01/msg02690.html>, the
overloads of gfc_warning and gfc_warning_now (with and without a first
argument for an option number) also break gcc.pot regeneration because
xgettext expects the translated string argument to be in a fixed
position for a given function name. This patch applies the
corresponding fix of always passing a first argument (option number or
0), just like the core diagnostic functions warning and warning_at,
and removing the problem overloads without it.
Bootstrapped with no regressions on x86_64-unknown-linux-gnu.
* error.c (gfc_warning (const char *, ...), gfc_warning_now (const
char *, ...)): Remove functions.
* gfortran.h (gfc_warning (const char *, ...), gfc_warning_now
(const char *, ...)): Remove declarations.
* arith.c, check.c, data.c, decl.c, frontend-passes.c,
interface.c, intrinsic.c, io.c, matchexp.c, module.c, openmp.c,
options.c, parse.c, primary.c, resolve.c, scanner.c, symbol.c,
trans-common.c, trans-const.c, trans-stmt.c: All callers of
gfc_warning and gfc_warning_now changed to pass 0 or option number
as first argument.
From-SVN: r220313
2015-01-31 Sandra Loosemore <sandra@codesourcery.com>
gcc/
* doc/extend.texi: Use "x86", "x86-32", and "x86-64" as the
preferred names of the architecture and its 32- and 64-bit
variants.
* doc/invoke.texi: Likewise.
* doc/md.texi: Likewise.
From-SVN: r220309
* config/pa/linux-atomic.c (__kernel_cmpxchg2): Change declaration of
oldval and newval to const void *. Fix typo.
(FETCH_AND_OP_2): Use __atomic_load_n to load value.
(FETCH_AND_OP_WORD): Likewise.
(OP_AND_FETCH_WORD): Likewise.
(COMPARE_AND_SWAP_2): Likewise.
(__sync_val_compare_and_swap_4): Likewise.
(__sync_lock_test_and_set_4): Likewise.
(SYNC_LOCK_RELEASE_2): Likewise.
Remove support for long long atomic operations.
From-SVN: r220307
2015-01-30 Vladimir Makarov <vmakarov@redhat.com>
PR target/64688
* lra-constraints.c (original_subreg_reg_mode): New.
(simplify_operand_subreg): Try to simplify subreg of const. Use
original_subreg_reg_mode for it.
(swap_operands): Update original_subreg_reg_mode.
(curr_insn_transform): Set up original_subreg_reg_mode.
2015-01-30 Vladimir Makarov <vmakarov@redhat.com>
PR target/64688
* g++.dg/pr64688-2.C: New.
From-SVN: r220297
2015-01-30 Vladimir Makarov <vmakarov@redhat.com>
PR target/64617
* lra-constraints.c (prohibited_class_reg_set_mode_p): New
function.
(process_alt_operands): Use it.
(curr_insn_transform): Check the optional reload pseudo class is
ok for the mode.
2015-01-30 Vladimir Makarov <vmakarov@redhat.com>
PR target/64617
* gcc.target/i386/pr64617.c: New test.
From-SVN: r220294
The patch <https://gcc.gnu.org/ml/gcc-patches/2014-11/msg00698.html>
adding an overload for fatal_error that passes an explicit location
broke gcc.pot regeneration because xgettext cannot handle function
overloads with the diagnostic string argument in different positions.
As the desired direction is for all diagnostics to have explicit
locations, this patch addresses the regression by removing the version
of fatal_error that does not pass a location, passing explicit
input_location everywhere (in the hope that those will incrementally
be changed to other locations, much as with the addition of a location
argument to pedwarn some time ago - a lot of cases aren't meaningfully
associated with a source file so UNKNOWN_LOCATION may be better).
Note that this patch does not attempt to fix any existing issues with
these diagnostics (such as wrongly starting with capital letters or
ending with '.' or '\n'); it just adds the input_location argument.
Bootstrapped with no regressions for x86_64-unknown-linux-gnu (Go
excluded).
gcc:
* diagnostic.c (fatal_error (const char *, ...)): Remove function.
* diagnostic-core.h (fatal_error (const char *, ...)): Remove
prototype.
* toplev.h (init_asm_output): Update comment on use of
UNKNOWN_LOCATION with fatal_error.
* cgraph.c, collect-utils.c, collect2.c, config/arc/arc.c,
config/arc/arc.md, config/avr/avr.c, config/c6x/c6x.h,
config/darwin.c, config/host-darwin.c, config/i386/host-cygwin.c,
config/i386/intelmic-mkoffload.c, config/nios2/nios2.c,
config/nvptx/mkoffload.c, config/nvptx/nvptx.h,
config/rs6000/host-darwin.c, config/rs6000/rs6000.c,
config/s390/s390.c, gcc.c, gcov-io.h, gcov-tool.c, ggc-common.c,
ggc-page.c, graph.c, ipa-inline-analysis.c, ipa-reference.c,
lto-cgraph.c, lto-section-in.c, lto-streamer-in.c, lto-streamer.c,
lto-wrapper.c, objc/objc-act.c, opts.c, passes.c, plugin.c,
tlink.c, toplev.c, tree-streamer-in.c, varpool.c: All callers of
fatal_error changed to pass input_location as first argument.
gcc/c-family:
* c-opts.c, c-pch.c, cppspec.c: All callers of fatal_error changed
to pass input_location as first argument.
gcc/cp:
* class.c, except.c, parser.c, pt.c: All callers of fatal_error
changed to pass input_location as first argument.
gcc/fortran:
* f95-lang.c, gfortranspec.c, trans-const.c, trans-expr.c: All
callers of fatal_error changed to pass input_location as first
argument.
gcc/java:
* class.c, expr.c, jcf-parse.c, jvspec.c: All callers of
fatal_error changed to pass input_location as first argument.
gcc/lto:
* lto-object.c, lto-symtab.c, lto.c: All callers of fatal_error
changed to pass input_location as first argument.
libcc1:
* plugin.cc: All callers of fatal_error changed to pass
input_location as first argument.
From-SVN: r220293
The GCC function is changing.
The error functions need to move to a compiler-indepent
approach so that changes to the GCC functions are not a
concern here.
From-SVN: r220292
This adds the pkgpath symbol to the export data whenever it is
needed--whenever the pkgpath symbol is not the obvious
transformation of the pkgpath. This lets us determine the
correct symbol name for indirectly imported packages compiled
with -fgo-prefix.
From-SVN: r220291
2015-01-30 Robert Dewar <dewar@adacore.com>
* a-assert.adb: Minor reformatting.
* sem_ch13.adb: Minor comment clarification.
* types.ads: Minor comment update.
* sem_eval.adb (Real_Or_String_Static_Predicate_Matches): Avoid blow up
when we have a predicate that is nothing but an inherited dynamic
predicate.
From-SVN: r220290
2015-01-30 Robert Dewar <dewar@adacore.com>
* einfo.ads: Minor comment fix.
* freeze.adb (Freeze_Profile): Add test for suspicious import
in pure unit.
* sem_prag.adb (Process_Import_Or_Interface): Test for suspicious
use in Pure unit is now moved to Freeze (to properly catch
Pure_Function exemption).
2015-01-30 Bob Duff <duff@adacore.com>
* sem_res.ads: Minor comment fix.
* sem_type.adb: sem_type.adb (Remove_Conversions): Need to
check both operands of an operator.
2015-01-30 Yannick Moy <moy@adacore.com>
* a-assert.ads, a-assert.adb: Mark package spec in SPARK. Set assertion
policy for Pre to Ignore.
(Assert): Add precondition.
From-SVN: r220288
2015-01-30 Robert Dewar <dewar@adacore.com>
* sem_prag.adb (Process_Import_Or_Interface): Warn if used in
Pure unit.
* s-valllu.ads (Scan_Raw_Long_Long_Unsigned): Clarify
documentation for some special cases of invalid attempts at
based integers.
From-SVN: r220287
