Hi!
On (at least some of these) tests and on some targets, GCC prints messages
like:
./cc1 -quiet -O3 -fopt-info-vec-optimized -mvis vect-ivdep-1.c
vect-ivdep-1.c:11:3: optimized: loop vectorized using 8 byte vectors
vect-ivdep-1.c:11:3: optimized: loop versioned for vectorization to enhance
ali
Hi!
If we consider -fno-delete-null-pointer-checks as a way to support e.g. AVR
and other targets which can validly place objects at NULL rather than a way
to workaround UBs in code, I believe the following testcase must pass if
there is e.g.
char a[32]; // And this object ends up at a
Hi!
This is my proposal for fixing this PR, just a heuristics when optimizing
successive divides might not be a good idea (first testcase), and includes
Marc's testcases which showed cases where optimizing successive divides is a
good idea even if the inner divide is not a single use.
Unfortunate
Hi!
The following patch handles clobber stmts the same how we handle them in
convert_local_reference_stmt, if it is the clobber with decl on the lhs and
that lhs needs to be replaced by a field inside of a structure, the clobber
is replaced by GIMPLE_NOP.
Bootstrapped/regtested on x86_64-linux an
On Thu, Dec 6, 2018 at 9:31 AM wuyuan (E) wrote:
>
> Hi ARM maintainers:
> The taishanv110 core uses generic pipeline scheduling, which
> restricted the performance of taishanv110 core. By adding the pipeline
> scheduling of taishanv110 core in GCC,The performance of taishanv110 has been
The __builtin_has_attribute function fails with an ICE when
its first argument is an expression such as INDIRECT_REF, or
many others. The code simply assumes it's either a type or
a decl. The attached patch corrects this oversight.
While testing the fix I also noticed the C++ front end expects
Hi!
On Tue, Dec 04, 2018 at 03:30:47PM +, Richard Sandiford wrote:
> Segher Boessenkool writes:
> Hmm, so we allow top-level "asm volatile" in C++ but not C?
Apparently. Evert top-level asm is effectively volatile, so this doesn't
really matter; but should we try to resolve the difference?
On 12/4/18 9:04 AM, Fritz Reese wrote:
On Tue, Dec 4, 2018 at 10:12 AM Jakub Jelinek wrote:
Just a couple of random comments.
-fdec-pad-with-spaces option name doesn't look right, because it doesn't say
what the option affects. So perhaps have transfer in the option name?
[...]
Wouldn't it
Hi ARM maintainers:
The taishanv110 core uses generic pipeline scheduling, which restricted
the performance of taishanv110 core. By adding the pipeline scheduling of
taishanv110 core in GCC,The performance of taishanv110 has been improved.
The patch as follows, please join.
dif
On 11/16/18 5:32 PM, Alexandre Oliva wrote:
When binding an object to a template parameter of reference type, we
take the address of the object and dereference that address. The type
of the address may still carry (template) typedefs, but
verify_unstripped_args_1 rejects such typedefs other than
On 11/16/18 5:32 PM, Alexandre Oliva wrote:
When a member template is redeclared as a friend, we enter the context
of the member before looking it up, and then we check that the decls
are compatible. However, when the member template references template
types of the enclosing context, say an enc
On 11/16/18 3:50 PM, Jakub Jelinek wrote:
On Fri, Nov 16, 2018 at 10:01:05AM -0500, Nathan Sidwell wrote:
2018-11-16 Nathan Sidwell
PR c++/87269
* parser.c (lookup_literal_operator): Mark overload for keeping
when inside template. Refactor.
* g++.dg/lookup/p
On 11/16/18 9:16 PM, Alexandre Oliva wrote:
We used to preserve deferred access check along with resolved template
ids, but a tentative parsing firewall introduced additional layers of
deferred access checks, so that we don't preserve the checks we
want to any more.
This patch collapses the acce
On 12/3/18 5:54 PM, David Malcolm wrote:
I was going to ping this patch:
https://gcc.gnu.org/ml/gcc-patches/2018-11/msg00875.html
but it has bit-rotted somewhat, so here's a refreshed version of it.
Successfully bootstrapped & regrtested on x86_64-pc-linux-gnu.
OK for trunk?
Thanks
Dave
B
PR85770 is fixed by Segher's combiner patch to avoid combining hard
regs. Presumably it helps because it gives the allocators more freedom.
I'm adding the testcase from the PR to the regression suite.
Jeff
commit 40fc691eac0ea9414f7908826c91afc70ff78617
Author: law
Date: Thu Dec 6 00:40:08 2
On Wed, Dec 05, 2018 at 04:48:28PM -0500, Fritz Reese wrote:
> On Wed, Dec 5, 2018 at 12:00 AM Steve Kargl
> wrote:
> >
> > I intend to commit the attached patch on Saturday.
>
> Thanks for the work. I assume the patch bootstraps and passes
> regression tests?
The patch passed regression testing
On 12/5/18 1:56 PM, Uros Bizjak wrote:
> This macro is always defined through defaults.h. We can remove #ifdef
> check from dwarf2out.c
>
> 2018-12-05 Uros Bizjak
>
> * dwarf2out.c (field_byte_offset): Remove
> #ifdef PCC_BITFIELD_TYPE_MATTERS check.
>
> Bootstrapped and regression te
On Tue, 4 Dec 2018 14:55:03 +0100
Richard Biener wrote:
> On Tue, 4 Dec 2018, Jakub Jelinek wrote:
>
> > On Mon, Dec 03, 2018 at 11:40:39PM +, Julian Brown wrote:
> > > Jakub asked in the following email at the time of the patch
> > > submission for the gomp4 branch what the difference was
Hi, Honza,
I have one question relate to whether to disable nothrow for -flive-patching as
following:
actually, there are two passes here:
1. local nothrow pass: in this pass, nothrow attribute is set locally after
analyzing every stmt of the function
body:
unsigned int
pass_nothrow::execut
On 11/29/18 4:43 PM, Martin Sebor wrote:
> On 11/29/18 4:07 PM, Jeff Law wrote:
>> On 11/29/18 1:34 PM, Martin Sebor wrote:
>>> On 11/16/2018 02:07 AM, Richard Biener wrote:
On Fri, Nov 16, 2018 at 4:12 AM Martin Sebor wrote:
>
> Ping: https://gcc.gnu.org/ml/gcc-patches/2018-08/msg018
This libgo patch by Cherry Zhang adds support for precise stack
scanning to the Go runtime. This uses per-function stack maps stored
in the exception tables in the language-specific data area. The
compiler needs to generate these stack maps; currently this is only
done by a version of LLVM, not b
Hi!
On Wed, Dec 05, 2018 at 04:47:37PM -0500, Jason Merrill wrote:
> On 12/2/18 11:38 AM, Segher Boessenkool wrote:
> >PR55681 observes that currently only one qualifier is allowed for
> >inline asm, so that e.g. "volatile asm" is allowed, "const asm" is also
> >okay (with a warning), but "const v
On 12/5/18 5:34 PM, Paolo Carlini wrote:
Hi,
On 05/12/18 20:31, Jason Merrill wrote:
On 12/5/18 7:45 AM, Paolo Carlini wrote:
Hi,
as mentioned in one of my last patches, we can now improve this
location. Note: in the same function there are a few further issues
which I mean to incrementally
On 11/21/18 7:19 AM, Tobias Burnus wrote:
Hello all,
if a class contains any 'virtual ... = 0', it's an abstract class and for an
abstract class, the destructor not added to the vtable.
For a normal
virtual ~class() { }
that's not a problem as the class::~class() destructor will be generated
Hi,
On 05/12/18 20:31, Jason Merrill wrote:
On 12/5/18 7:45 AM, Paolo Carlini wrote:
Hi,
as mentioned in one of my last patches, we can now improve this
location. Note: in the same function there are a few further issues
which I mean to incrementally fix (eg, the diagnostics for
warn_if_not
On Wed, Dec 5, 2018 at 12:00 AM Steve Kargl
wrote:
>
> I intend to commit the attached patch on Saturday.
Thanks for the work. I assume the patch bootstraps and passes regression tests?
RE:
> PR fortran/88228
> * expr.c (check_null, check_elemental): Work around -fdec and
>
Ping.
This is a ping of my patch set to implement the SIMD ABI on Aarch64.
https://gcc.gnu.org/ml/gcc-patches/2018-11/msg00636.html
https://gcc.gnu.org/ml/gcc-patches/2018-11/msg00637.html
https://gcc.gnu.org/ml/gcc-patches/2018-11/msg00639.html
https://gcc.gnu.org
On 12/2/18 11:38 AM, Segher Boessenkool wrote:
PR55681 observes that currently only one qualifier is allowed for
inline asm, so that e.g. "volatile asm" is allowed, "const asm" is also
okay (with a warning), but "const volatile asm" gives an error. Also
"goto" has to be last.
This patch changes
On Tue, 4 Dec 2018 15:02:15 +0100
Jakub Jelinek wrote:
> On Tue, Aug 28, 2018 at 03:19:19PM -0400, Julian Brown wrote:
> > 2018-08-28 Julian Brown
> > Cesar Philippidis
> >
> > PR middle-end/70828
> >
> > gcc/
> > * gimplify.c (gimplify_omp_ctx): Add decl_data_clause ha
This macro is always defined through defaults.h. We can remove #ifdef
check from dwarf2out.c
2018-12-05 Uros Bizjak
* dwarf2out.c (field_byte_offset): Remove
#ifdef PCC_BITFIELD_TYPE_MATTERS check.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
OK for mainline?
Uros
On Wed, Dec 05, 2018 at 03:49:26PM -0500, Jason Merrill wrote:
> On 11/28/18 3:42 AM, Jakub Jelinek wrote:
> > Whenever we need to clone a cdtor (either because the target doesn't support
> > aliases the way we need, e.g. initlist105.C testcase on darwin, where it has
> > been originally reported,
On 11/28/18 3:42 AM, Jakub Jelinek wrote:
Whenever we need to clone a cdtor (either because the target doesn't support
aliases the way we need, e.g. initlist105.C testcase on darwin, where it has
been originally reported, or when it has virtual bases, like the made up
initlist106.C on x86_64-linu
GROUP_BITFIELDS_BY_ALIGN was long gone, PCC_BITFIELD_TYPE_MATTERS
defaults to 1 in i386.h for all x86 targets.
2018-12-05 Uros Bizjak
* config/i386/cygming.h (PCC_BITFIELD_TYPE_MATTERS): Remove.
(GROUP_BITFIELDS_BY_ALIGN): Ditto.
Tested by building crosscompiler to x86_64-w64-mingw32.
PR c++/88320 reports that the C and C++ FEs can offer bogus suggestions
for misspelled identifiers within initializers, where the thing being
initialized is suggested. If the user follows these suggestions,
it will lead to a -Wuninitialized warning. For example:
test.c:9:19: error: 'aresults' wa
On 12/5/18 12:13 PM, Jeff Law wrote:
On 12/5/18 11:59 AM, David Edelsohn wrote:
Jeff,
Thanks for the patch.
I continue to see a failure on AIX 32 bit mode (2 byte wchar).
FAIL: gcc.dg/strlenopt-58.c scan-tree-dump-times optimized
"call_in_true_branch_not_eliminated" 0
I'm not certain if this
On 12/5/18 7:45 AM, Paolo Carlini wrote:
Hi,
as mentioned in one of my last patches, we can now improve this
location. Note: in the same function there are a few further issues
which I mean to incrementally fix (eg, the diagnostics for
warn_if_not_aligned ICEs for unnamed bit-fields). Tested
On 12/5/18 11:59 AM, David Edelsohn wrote:
> Jeff,
>
> Thanks for the patch.
>
> I continue to see a failure on AIX 32 bit mode (2 byte wchar).
>
> FAIL: gcc.dg/strlenopt-58.c scan-tree-dump-times optimized
> "call_in_true_branch_not_eliminated" 0
>
> I'm not certain if this is AIX-specific or
Jeff,
Thanks for the patch.
I continue to see a failure on AIX 32 bit mode (2 byte wchar).
FAIL: gcc.dg/strlenopt-58.c scan-tree-dump-times optimized
"call_in_true_branch_not_eliminated" 0
I'm not certain if this is AIX-specific or more fallout from wchar
size. I could ignore with target power
Hi Chung-Lin,
> > +module openacc_c_string
> > + implicit none
> > +
> > + interface
> > +function strlen (s) bind (C, name = "strlen")
> > + use iso_c_binding, only: c_ptr, c_size_t
> > + type (c_ptr), intent(in), value :: s
> > + integer (c_size_t) :: strlen
> > +end fun
On Wed, Dec 05, 2018 at 05:10:07PM +, Iain Sandoe wrote:
> Hi,
>
> The PR is about unnecessary saves of the pic base register, it shows on m32
> Linux and m32/m64 Darwin.
>
> The fix is to check that we are in a pic mode and that the picbase has
> actually been used.
> As a bonus, some #ifd
On Wed, Dec 05, 2018 at 11:37:27AM -0600, Segher Boessenkool wrote:
> On Wed, Dec 05, 2018 at 10:04:56AM +, Richard Sandiford wrote:
> > Martin Sebor writes:
> > > Martin suggested we update the Coding Conventions to describe
> > > the expected style for function declarations with a pointer
>
On 12/5/18 3:04 AM, Richard Sandiford wrote:
Thanks for doing this,
Martin Sebor writes:
Martin suggested we update the Coding Conventions to describe
the expected style for function declarations with a pointer
return types, and for overloaded operators. Below is the patch.
As an aside, rega
On Wed, 5 Dec 2018, Segher Boessenkool wrote:
> Patches are usually ignored because everyone thinks someone else will
> handle it.
And in this case, it looks like this patch would best be reviewed first in
the GDB context - then once committed to binutils-gdb, the committer could
post to gcc-pa
On Wed, Dec 05, 2018 at 10:04:56AM +, Richard Sandiford wrote:
> Martin Sebor writes:
> > Martin suggested we update the Coding Conventions to describe
> > the expected style for function declarations with a pointer
> > return types, and for overloaded operators. Below is the patch.
> >
> > A
On 12/5/18 7:58 AM, Richard Biener wrote:
> On Wed, 5 Dec 2018, Jeff Law wrote:
>
>> On 12/4/18 6:16 AM, Richard Biener wrote:
>>>
>>> This tries to make bugs like that in PR88317 harder to create by
>>> introducing a bitmap_release function that can be used as
>>> pendant to bitmap_initialize for
On Wed, 5 Dec 2018, Дилян Палаузов wrote:
> Can you please make it clear, when contributions will be ignored, or
> agree on some procedures, where all contributions are handled in
> reasonable time withot reminders, in a way that the processes work
> reliably.
1. You do not have a contract in pla
On 12/4/18 8:49 PM, Sandra Loosemore wrote:
On 12/4/18 8:13 PM, Martin Sebor wrote:
On 12/4/18 2:04 PM, Sandra Loosemore wrote:
On 12/4/18 9:26 AM, Martin Sebor wrote:
[snip]
+The keyword @code{__attribute__} allows you to specify various special
+properties of types. Some type attributes a
Hi!
On Wed, Dec 05, 2018 at 07:36:09AM +, Дилян Палаузов wrote:
> on 27th October I sent to gcc-patches a mail with the subject “Don’t
> build gdb/readline/libreadline.a, when --with-system-readline is
> supplied”
The patch looks fine to me, fwiw.
> and on 14th November I sent a reminder. I
Hi,
The PR is about unnecessary saves of the pic base register, it shows on m32
Linux and m32/m64 Darwin.
The fix is to check that we are in a pic mode and that the picbase has actually
been used.
As a bonus, some #ifdef’d TARGET_MACHO code is no longer required.
Tested on power7, bootstrapped
On Wed, 2018-12-05 at 11:03 -0500, Jason Merrill wrote:
> On Mon, Dec 3, 2018 at 5:14 PM Joseph Myers
> wrote:
> >
> > On Sat, 1 Dec 2018, Jason Merrill wrote:
> >
> > > Hmm, it looks like the C front-end comptypes will return 1 for
> > > e.g. enum and
> > > int. It seems to me that what you wa
On 12/5/18 2:09 AM, Tom Honermann wrote:
On 12/3/18 5:01 PM, Jason Merrill wrote:
On 12/3/18 4:51 PM, Jason Merrill wrote:
On 11/5/18 2:39 PM, Tom Honermann wrote:
This patch adds support for the P0482R5 core language changes. This
includes:
- The -fchar8_t and -fno_char8_t command line opti
Hi Martin,
>>> PS I'm not happy about duplicating the same test across all those
>>> targets. It would be much nicer to have a single test somewhere
>>> in dg.exp #include a target-specific header with macros describing
>>> the target-specific parameters.
>>
>> why so complicated? Just have a si
On 12/5/18 6:09 AM, Rainer Orth wrote:
Hi Martin,
The tests for the new __builtin_has_attribute function have been
failing on a number of targets because of a couple of assumptions
that only hold on some.
First, they expect that it's safe to apply attribute aligned with
a smaller alignment tha
On Mon, Dec 3, 2018 at 5:14 PM Joseph Myers wrote:
>
> On Sat, 1 Dec 2018, Jason Merrill wrote:
>
> > Hmm, it looks like the C front-end comptypes will return 1 for e.g. enum and
> > int. It seems to me that what you want for this warning is actually to
> > check
> > for the same type. Perhaps
OK, thanks.
On Wed, Dec 5, 2018 at 1:32 AM Alexandre Oliva wrote:
>
> On Nov 29, 2018, Jason Merrill wrote:
>
> > Let's go with this. And remove the comment.
>
> > And the !processing_template_decl is also redundant, since that's
> > checked at the top of value_dependent_expression_p.
>
> I've t
> unfortunately, the new tests FAIL to compile, both 32 and 64-bit:
>
> +FAIL: gcc.target/sparc/20181129-1.c (test for excess errors)
> +UNRESOLVED: gcc.target/sparc/20181129-1.c compilation failed to produce
> executable +FAIL: gcc.target/sparc/20181129-2.c (test for excess errors)
> +UNRESOLVED:
On Tue, 4 Dec 2018 at 14:17, Richard Biener wrote:
>
> On Tue, 4 Dec 2018, Christophe Lyon wrote:
>
> > Hi Richard,
> > On Mon, 3 Dec 2018 at 14:37, Richard Biener wrote:
> > >
> > >
> > > This fixes a missed optimization in EVRP by teaching the code
> > > figuring out conditional asserts about c
On Wed, Dec 5, 2018 at 2:10 AM Tom Honermann wrote:
>
> On 12/3/18 2:59 PM, Jason Merrill wrote:
> > On 11/5/18 2:39 PM, Tom Honermann wrote:
> >> This patch adds documentation for new -fchar8_t and -fno-char8_t
> >> options.
> >>
> >> gcc/ChangeLog:
> >>
> >> 2018-11-04 Tom Honermann
> >>
On Wed, 5 Dec 2018, Jeff Law wrote:
> On 12/4/18 6:16 AM, Richard Biener wrote:
> >
> > This tries to make bugs like that in PR88317 harder to create by
> > introducing a bitmap_release function that can be used as
> > pendant to bitmap_initialize for non-allocated bitmap heads.
> > The function
Committed.
Richard.
2018-12-05 Richard Biener
PR middle-end/63184
* c-c++-common/pr19807-2.c: New testcase.
* c-c++-common/pr19807-3.c: Likewise.
diff --git a/gcc/testsuite/c-c++-common/pr19807-2.c
b/gcc/testsuite/c-c++-common/pr19807-2.c
new file mode 100644
index
On 12/4/18 6:16 AM, Richard Biener wrote:
>
> This tries to make bugs like that in PR88317 harder to create by
> introducing a bitmap_release function that can be used as
> pendant to bitmap_initialize for non-allocated bitmap heads.
> The function makes sure to poison the bitmaps obstack member
>
This fixes a cut & paste oversight in udivmodhi4 (which is currently used only
by the pdp11 target) reported by Stefan Kanthak.
Committed as obvious.
paul
ChangeLog:
2018-12-05 Paul Koning
* udivmodhi4.c (__udivmodhi4): Fix loop end check.
Index: libgcc/udivmodhi4.c
==
Hi Chung-Lin!
On Mon, 19 Nov 2018 16:33:30 +0900, Chung-Lin Tang
wrote:
> Hi Thomas,
> actually the current version of the acc_get/set_default_async patch is
> combined into:
> https://gcc.gnu.org/ml/gcc-patches/2018-09/msg01426.html
>
> This patch you're referring here was a version from early
Hi Chung-Lin!
On Mon, 19 Nov 2018 16:33:30 +0900, Chung-Lin Tang
wrote:
> On 2018/11/18 10:36 AM, Thomas Schwinge wrote:
> > Generally, I envision test cases running a few "acc_get_cuda_stream"
> > calls with relevant argument values, to see whether the expected
> > queues/streames are being use
Rainer Orth writes:
> Hi Andreas,
>
>> On Okt 01 2018, David Malcolm wrote:
>>
>>> Is there a link to the .log files somewhere so I can see the precise
>>> messages in question? (e.g. are they all "loop versioned for
>>> vectorization to enhance alignment"?).
>>
>> Yes, they are all the same me
On 12/5/18 6:09 AM, Rainer Orth wrote:
> Hi Martin,
>
>> The tests for the new __builtin_has_attribute function have been
>> failing on a number of targets because of a couple of assumptions
>> that only hold on some.
>>
>> First, they expect that it's safe to apply attribute aligned with
>> a sma
On 12/5/18 6:15 AM, Rainer Orth wrote:
> Hi Eric,
>
>> This started as a simple fix for a small issue (passing floating-point
>> vectors
>> to variadic functions in 64-bit mode) and then evolved into a small cleanup
>> of
>> the code implementing the calling conventions of the 2 SPARC ABIs.
>>
Hi Eric,
> This started as a simple fix for a small issue (passing floating-point
> vectors
> to variadic functions in 64-bit mode) and then evolved into a small cleanup
> of
> the code implementing the calling conventions of the 2 SPARC ABIs.
>
> Tested and compat-regtested on SPARC/Solaris 1
Hi Martin,
> The tests for the new __builtin_has_attribute function have been
> failing on a number of targets because of a couple of assumptions
> that only hold on some.
>
> First, they expect that it's safe to apply attribute aligned with
> a smaller alignment than the target provides when GCC
Hi,
as mentioned in one of my last patches, we can now improve this
location. Note: in the same function there are a few further issues
which I mean to incrementally fix (eg, the diagnostics for
warn_if_not_aligned ICEs for unnamed bit-fields). Tested x86_64-linux.
Thanks, Paolo.
//
SLP vectorization forgot to reset vect_location which then "leaks"
to autopar eventually accessing stale data.
Committed to trunk.
Richard.
2018-12-05 Richard Biener
PR tree-optimization/86637
* tree-vectorizer.c (pass_slp_vectorize::execute): Reset
vect_location at
Hi!
Sorry for my late follow-up; had a lot of catch up to do back then.
On Thu, 27 Sep 2018 11:47:31 +0200, Richard Biener
wrote:
> On Mon, Sep 24, 2018 at 9:14 PM Indu Bhagat wrote:
> >
> > Done. Attached is updated patch.
> >
> > Patch is tested on x86_64
>
> You obviously did _not_ properl
Hi!
On Wed, Nov 28, 2018 at 09:42:47AM +0100, Jakub Jelinek wrote:
> 2018-11-28 Jakub Jelinek
>
> PR c++/86669
> * optimize.c (clone_body_copy_decl): New function.
> (clone_body): Use it for base cdtors. Remap here only
> DECL_INITIAL of decls that don't have DECL_CONT
Hi Maciej, please see below:
On 2018/12/4 12:51 AM, Maciej W. Rozycki wrote:
+module openacc_c_string
+ implicit none
+
+ interface
+function strlen (s) bind (C, name = "strlen")
+ use iso_c_binding, only: c_ptr, c_size_t
+ type (c_ptr), intent(in), value :: s
+ integer (c_s
Thanks for doing this,
Martin Sebor writes:
> Martin suggested we update the Coding Conventions to describe
> the expected style for function declarations with a pointer
> return types, and for overloaded operators. Below is the patch.
>
> As an aside, regarding the space convention in casts: a
On Wed, Dec 5, 2018 at 9:39 AM Richard Sandiford
wrote:
>
> As Jakub pointed out, if we narrow a plus, minus or mult operation based
> on the number of bits that consumers need, we have to convert a signed
> operation to an unsigned one in order to avoid new undefined behaviour.
> This patch does
On Tue, Dec 4, 2018 at 10:36 PM David Malcolm wrote:
>
> When debugging a failing test, I typically invoke DejaGnu at
> verbosity level 2 (via RUNTESTFLAGS="-v -v dg.exp=something"),
> so that DejaGnu prints the command line used to invoke the
> compiler; specifically these two sites:
> target.e
On Wed, 5 Dec 2018, Jakub Jelinek wrote:
> Hi!
>
> The following testcases ICE, because tree_loop_unroll_and_jam optimizes one
> loop and on another one after it fails to analyze data dependencies and
> returns. The end effect of that is that neither the code at the end of the
> function to do f
On Wed, Dec 05, 2018 at 04:34:14AM -0200, Alexandre Oliva wrote:
> for gcc/cp/ChangeLog
>
> PR c++/85569
> * constexpr.c (adjust_temp_type): Test for type equality with
> same_type_p.
> (constexpr_call_hasher::equal): Likewise.
>
> for gcc/testsuite
>
> PR c++/855
As Jakub pointed out, if we narrow a plus, minus or mult operation based
on the number of bits that consumers need, we have to convert a signed
operation to an unsigned one in order to avoid new undefined behaviour.
This patch does that and generalises vect_convert_input and
vect_recog_over_widenin
On Wed, Dec 05, 2018 at 04:50:19AM -0200, Alexandre Oliva wrote:
> Jeff tested this with a cross compiler to h8300-elf, and several other
> native and cross toolchains IIUC. I'm regstrapping it myself on i686-
> and x86_64-linux-gnu. Ok to install?
>
>
> for gcc/ChangeLog
>
> * cselib.c
81 matches
Mail list logo
