The current rs6000 rtx_cost for comparisons against 0 is very high if
TARGET_ISEL && !TARGET_MFCRF, much higher than for reg-reg comparisons,
much higher than a load of 0 and such a reg-reg-comparison. This leads
to infinite recursion in CSE (see PR81288).
This patch removes the too-high cost, also simplifying this code.
PR 81288/target
* config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
combine: Don't split insns if half is unused (PR82621)
If we have a PARALLEL of two SETs, and one half is unused, we currently
happily split that into two instructions (albeit the unused one is
useless). Worse, as PR82621 shows, combine will happily merge this
insn into I3 even if some intervening insn sets the same register
again, which is wrong.
This fixes it by not splitting PARALLELs with REG_UNUSED notes. It
all is handled fine by combine in that case; just the "single set
that is unused" case isn't handled properly.
This also results in better code: combine will now actually throw
away the unused SET. (It still won't do that in an I3).
PR rtl-optimization/82621
* combine.c (try_combine): Do not split PARALLELs of two SETs if the
dest of one of those SETs is unused.
Rainer Orth [Tue, 21 Nov 2017 09:41:34 +0000 (09:41 +0000)]
Adapt Solaris 12 references
libgo:
Backport from mainline
libgo: adapt Solaris 12 references
With the change in the Solaris release model (no more major releases
like Solaris 12 but only minor ones like 11.4), the Solaris 12
references in GCC need to be adapted.
Paul Thomas [Tue, 14 Nov 2017 17:38:38 +0000 (17:38 +0000)]
backport: re PR fortran/82934 (Segfault on assumed character length in allocate)
2017-11-13 Paul Thomas <pault@gcc.gnu.org>
Backport from trunk
PR fortran/82934
* trans-stmt.c (gfc_trans_allocate): Remove the gcc_assert on
null string length for assumed length typespec and set
expr3_esize to NULL_TREE;
2017-11-13 Paul Thomas <pault@gcc.gnu.org>
Backport from trunk
PR fortran/82934
* gfortran.dg/allocate_assumed_charlen_1.f90: New test.
Paul Thomas [Tue, 14 Nov 2017 08:15:52 +0000 (08:15 +0000)]
backport: re PR fortran/78619 (ICE in copy_reference_ops_from_ref, at tree-ssa-sccvn.c:889)
2017-11-13 Paul Thomas <pault@gcc.gnu.org>
Backport from trunk
PR fortran/78619
* check.c (same_type_check): Introduce a new argument 'assoc'
with default value false. If this is true, use the symbol type
spec of BT_PROCEDURE expressions.
(gfc_check_associated): Set 'assoc' true in the call to
'same_type_check'.
2017-11-13 Paul Thomas <pault@gcc.gnu.org>
Backport from trunk
PR fortran/78619
* gfortran.dg/pr78619.f90: New test.
Eric Botcazou [Fri, 10 Nov 2017 08:24:40 +0000 (08:24 +0000)]
utils.c (convert): Add comment and do not fall through to the next case.
* gcc-interface/utils.c (convert) <RECORD_TYPE>: Add comment and do
not fall through to the next case.
<ARRAY_TYPE>: Deal specially with a dereference from another array
type with the same element type.
The result is that gcc 6.4 now refuses to allow -march=armv7ve and
-mcpu=XXX to be used together, even when -mcpu is set to an armv7ve
compatible core:
Fix by defining flags for armv7ve compatible cores directly from
FL_FOR_ARCH7VE, rather than re-creating the armv7ve flags
independently by combining FL_FOR_ARCH7A with the armv7ve specific
FL_THUMB_DIV and FL_ARM_DIV flags.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
2017-11-09 Andre McCurdy <armccurdy@gmail.com>
* config/arm/arm-cores.def: Fix missing FL_LPAE flag for armv7ve
compatible cores.
Backport from mainline
2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
PR rtl-optimization/64682
PR rtl-optimization/69567
PR rtl-optimization/69737
PR rtl-optimization/82683
* combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
register mentioned in the note, drop the note, unless it came from I3,
in which case it should go to I3 again.
Paul Thomas [Mon, 6 Nov 2017 11:50:53 +0000 (11:50 +0000)]
backport: re PR fortran/78641 ([OOP] ICE on polymorphic allocatable function in array constructor)
2017-11-06 Paul Thomas <pault@gcc.gnu.org>
Backported from trunk
PR fortran/78641
* resolve.c (resolve_ordinary_assign): Do not add the _data
component for class valued array constructors being assigned
to derived type arrays.
* trans-array.c (gfc_trans_array_ctor_element): Take the _data
of class valued elements for assignment to derived type arrays.
2017-11-06 Paul Thomas <pault@gcc.gnu.org>
Backported from trunk
PR fortran/78641
* gfortran.dg/class_66.f90: New test.
Paul Thomas [Mon, 6 Nov 2017 11:48:32 +0000 (11:48 +0000)]
backport: re PR fortran/69739 (ICE during array result, allocatable assignment)
2017-11-06 Paul Thomas <pault@gcc.gnu.org>
Backported from trunk
PR fortran/69739
* trans-expr.c (gfc_map_intrinsic_function): Return false for
bounds without the DIM argument instead of ICEing.
2017-11-06 Paul Thomas <pault@gcc.gnu.org>
Backported from trunk
PR fortran/69739
* gfortran.dg/pr69739.f90: New test.
Andreas Tobler [Sun, 5 Nov 2017 19:33:44 +0000 (20:33 +0100)]
backport: re PR libgcc/82635 (std::thread's join broken on FreeBSD with all GCCs >= 5)
2017-11-05 Andreas Tobler <andreast@gcc.gnu.org>
Backport from mainline
2017-11-04 Andreas Tobler <andreast@gcc.gnu.org>
PR libgcc/82635
* config/i386/freebsd-unwind.h (MD_FALLBACK_FRAME_STATE_FOR): Use a
sysctl to determine whether we're in a trampoline.
Keep the pattern matching method for systems without
KERN_PROC_SIGTRAMP sysctl.
Steven G. Kargl [Sat, 4 Nov 2017 17:16:39 +0000 (17:16 +0000)]
re PR fortran/82796 (Private+equivalence in used module breaks compilation of pure function)
2017-11-01 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/82796
* resolve.c (resolve_equivalence): An entity in a common block within
a module cannot appear in an equivalence statement if the entity is
with a pure procedure.
2017-11-01 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/82796
* gfortran.dg/equiv_pure.f90: New test.
Paul Thomas [Wed, 1 Nov 2017 17:24:35 +0000 (17:24 +0000)]
[multiple changes]
2017-11-01 Andre Vehreschild <vehre@gcc.gnu.org>
Paul Thomas <pault@gcc.gnu.org>
PR fortran/78293
* trans-expr.c (gfc_conv_procedure_call): Prepend deallocation
of alloctable components to post, rather than adding to
se->post.
* trans-stmt.c (gfc_trans_allocate): Move deallocation of expr3
allocatable components so that all expr3s are visited.
2017-11-01 Paul Thomas <pault@gcc.gnu.org>
PR fortran/78293
* gfortran.dg/allocatable_function_10.f90: New test.
* gfortran.dg/class_array_15.f03: Increase builtin_free count
from 11 to 12.
Jonathan Wakely [Mon, 23 Oct 2017 17:47:05 +0000 (18:47 +0100)]
Fix path::iterator post-increment and post-decrement
Backport from mainline
2017-10-19 Jonathan Wakely <jwakely@redhat.com>
* include/experimental/bits/fs_path.h (path::iterator++(int))
(path::iterator--(int)): Fix for paths with only one component.
* testsuite/experimental/filesystem/path/itr/traversal.cc: Test
post-increment and post-decrement.
Bill Schmidt [Fri, 20 Oct 2017 21:35:41 +0000 (21:35 +0000)]
backport: rs6000.c (rs6000_gen_le_vsx_permute): Use rotate instead of vec_select for V1TImode.
2017-10-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
Backport from mainline
2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
* config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
instead of vec_select for V1TImode.
* conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
longer needed.
(VSX_LE_128): Add V1TI to this mode iterator.
(*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
(*vsx_le_perm_store_<mode>): Likewise.
(pre-reload splitter for VSX stores): Likewise.
(post-reload splitter for VSX stores): Likewise.
(*vsx_xxpermdi2_le_<mode>): Likewise.
(*vsx_lxvd2x2_le_<mode>): Likewise.
(*vsx_stxvd2x2_le_<mode>): Likewise.
Peephole patterns exist in the arm backend to spot load/store
operations to adjacent memory operations in order to convert them into
ldrd/strd instructions. However, when we have strict alignment
enforced, then we can only do this if the accesses are known to be
64-bit aligned; this is unlikely to be the case for most loads. The
patch adds some alignment checking to the code that validates the
addresses for use in the peephole patterns. This should also fix
incorrect generation of ldrd/strd with unaligned accesses that could
previously have occurred on ARMv5e where all such operations must be
64-bit aligned.
I've added some new tests as well. In doing so I discovered that the
ldrd/strd peephole tests could never fail since they would match the
source file name in the scanned assembly as well as any instructions
of the intended type. I've fixed those by tightening the scan results
slightly.
gcc:
* config/arm/arm.c (align_ok_ldrd_strd): New function.
(mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of the
mem into it.
(gen_operands_ldrd_strd): Validate the alignment of the accesses.
testsuite:
* gcc.target/arm/peep-ldrd-1.c: Tighten test scan pattern.
* gcc.target/arm/peep-strd-1.c: Likewise.
* gcc.target/arm/peep-ldrd-2.c: New test.
* gcc.target/arm/peep-strd-2.c: New test.
Jakub Jelinek [Fri, 13 Oct 2017 20:14:34 +0000 (22:14 +0200)]
re PR target/82274 (__builtin_mul_overflow fails to detect overflow for int64_t when compiled with -m32)
PR target/82274
* internal-fn.c (expand_mul_overflow): If both operands have
the same highpart of -1 or 0 and the topmost bit of lowpart
is different, overflow is if res <= 0 rather than res < 0.
* libgcc2.c (__mulvDI3): If both operands have
the same highpart of -1 and the topmost bit of lowpart is 0,
multiplication overflows even if both lowparts are 0.
* gcc.dg/pr82274-1.c: New test.
* gcc.dg/pr82274-2.c: New test.
Andreas Tobler [Tue, 10 Oct 2017 18:47:57 +0000 (20:47 +0200)]
config.gcc: (armv7*-*-freebsd*): New target.
2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
* config.gcc: (armv7*-*-freebsd*): New target.
Define TARGET_FREEBSD_ARMv7
* config/arm/freebsd.h: Use TARGET_FREEBSD_ARMv7 to define the
default cpu for armv7*-*-freebsd*.
projects / thirdparty / gcc.git / log
