On Thu, Dec 17, 2020 at 12:10:14AM +0100, Jakub Jelinek via Gcc-patches wrote:
> The following patch should fix that, ok for trunk if it passes testing?
Tested successfully with build both without and with
--with-gcc-major-version-only and make install for both.
> 2020-12-17 Jakub Jelinek
>
>
Hi Harald,
OK for master? Since we generate wrong code under the given circumstances,
what do people think about backports?
OK for master. The patch is also simple enough that the risk
of regressions is almost nil, so backports are also OK.
Best regards
Thomas
Hi Tobias,
Calling gfc_find_vtab during resolution during parsing
comes too early for finalizers. This patch just moves it
to the resolution, which seems to be the simplest solution.
Ah, finalizers - one area where we still are deficient WRT Fortran 2003.
Thanks for doing things in this area!
Hi Thomas, hi all,
On 17.12.20 10:02, Thomas Koenig via Fortran wrote:
OK. Just one remark: Could you make the test into a run-time test
to check that the finalizers are called correctly?
Done – see attachment.
Thanks for the quick review!
Tobias
-
Mentor Graphics (Deutschla
On Dez 15 2020, Nathan Sidwell wrote:
> diff --git c/libcody/fatal.cc w/libcody/fatal.cc
> new file mode 100644
> index 000..b35094e6b19
> --- /dev/null
> +++ w/libcody/fatal.cc
> @@ -0,0 +1,78 @@
> +// CODYlib -*- mode:c++ -*-
> +// Copyright (C) 2019-2020 Nathan Sidwell, nat...
On Thu, Dec 17, 2020 at 11:01:40AM +0100, Andreas Schwab wrote:
> On Dez 15 2020, Nathan Sidwell wrote:
>
> > diff --git c/libcody/fatal.cc w/libcody/fatal.cc
> > new file mode 100644
> > index 000..b35094e6b19
> > --- /dev/null
> > +++ w/libcody/fatal.cc
> > @@ -0,0 +1,78 @@
> > +// CODYl
The RTL SSA merge broke SPARC bootstrap:
In file included from ./tm_p.h:4,
from /vol/gcc/src/hg/master/local/gcc/rtl-ssa.h:54,
from /vol/gcc/src/hg/master/local/gcc/fwprop.c:29:
/vol/gcc/src/hg/master/local/gcc/config/sparc/sparc-protos.h:45:47: error: use
of enu
ChangeLog:
2020-12-17 Marius Hillenbrand
* MAINTAINERS (Write After Approval): Add myself.
---
MAINTAINERS | 1 +
1 file changed, 1 insertion(+)
diff --git a/MAINTAINERS b/MAINTAINERS
index 32f8a2b7292..48335cc7920 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -428,6 +428,7 @@ George
A cascaded error leads to some confusion in this case, fixed by making
Inherit_Predicate_Flags a no-op pre Ada 2012.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_util.adb (Inherit_Predicate_Flags): No-op before Ada 2012.diff --git a/gcc/ada/sem_util.adb b/gcc/ada/sem_
When mixing the <> in an aggregrate and the default initialization of
the Type involved by the box references Type'Access, GNAT may generate a
spurious error of the form:
"Access" attribute cannot be applied to type
Fixed by only considering nodes that come from source in this legality
check:
These fixes aim at making GNAT messages more readable and uniform, just
by following the current rules generally followed for messages:
- start with a lowercase character
- use ALL CAPITAL for keywords (so that they are properly quoted in the
final message), Capitalized attribute/pragma names
- u
When declaring a component of a protected type as part of a controlled
type, the compiler may crash trying to generate the finalize routine.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* exp_ch7.adb (Make_Final_Call, Make_Init_Call): Take protected
types into accou
An improper run-time error is raised on a bounded synchronized queue
when the container element is a discriminated type that is itself a
bounded container.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/a-cbsyqu.ads (Implementation): Provide a box
initializa
This flips the default for the generation of encodings for fixed-point
types in the debugging information generated by the compiler.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* exp_dbug.adb (Get_Encoded_Name): Generate encodings for fixed
point types only if -fgn
This patch fixes an issue in the compiler whereby a reference to the
current instance of the type occurring within a subtype contraint causes
a crash during backend expansion of associated discriminant checks.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* checks.adb (Build
This changes the implementation of the Value attribute for floating-point
types not to use exponentiation to compute the final value, at least for
the common bases 2, 4, 8, 10 and 16. The powers of two use the Scaling
attribute instead, while a precomputed table of powers is used for 10.
The first
Discrete values in unchecked conversions were only folded when
compiling. Now they are also folded in GNATprove and semantics checking
modes (-gnatc).
This is particularly important for calls to static functions, since
unchecked conversions are likely to appear for their actual parameters
or resu
Reuse Error_Msg_Ada_2020_Feature to complain about a missing -gnat2020
switch when code that would be legal in Ada 2020 is compiled in an
earlier language version.
Also, refer to language construct and not to syntactic category of the
missing feature, i.e. "declare expression" and not "declare_exp
...Append and container aggregates.
This patch implements the above AI, except we do not yet remove
the Append procedures that append vectors (for compatibility
reasons).
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/a-cbdlli.adb, libgnat/a-cbdlli.ads,
lib
Routines that examine the parent chain of a subexpression don't need to
cross the subprogram or package boundaries and go until the very root of
the compilation unit.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_util.adb (In_Check_Node): Add guard and rename Node to
Complain about missing -gnat20xx switches when the code would be
accepted in a newer version of Ada rather than silently rejecting it.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* par-ch12.adb (P_Formal_Derived_Type_Definition): Complain
about formal type with asp
In Analyze_Case_Statement we save the current value of
Unblocked_Exit_Count with the intention to restore it on exit. However,
when returning early due to errors we failed to restore the previous
value. It seems cleaner to only modify this variable after the possible
early exits.
Mainly just a cle
Spotted by running GNATcoverage on GNAT sources, some subprograms are
no longer used.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* checks.adb: Remove, not used.
* checks.ads: Likewise.
* exp_ch6.adb: Likewise.
* exp_ch7.adb: Likewise.
* ex
This temporary code kept 3+ years ago can now be removed.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_util.adb (New_Requires_Transient_Scope): Renamed
Requires_Transient_Scope.
(Requires_Transient_Scope, Old_Requires_Transient_Scope,
Results_D
Spotted by running GNATcoverage on GNAT sources, some files are no
longer used.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* symbols.ads, symbols.adb: Removed no longer used.diff --git a/gcc/ada/symbols.adb /dev/null
deleted file mode 100644
--- a/gcc/ada/symbols.adb
+++
Combining an N_Expression_With_Actions within another
N_Expression_With_Actions is causing inconsistencies and leads to
dropping some actions on the floor. Fixed by creating a transient scope
for declare expressions when needed.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
There is a fallback code in gigi in case the front-end constructs
trees referencing non existing record fields. At this stage, it's
safer and cleaner to ensure that the front-end never generates such
references, and eventually replace the code in gigi by a gcc_unreachable().
We are not quite there
CM_Membership is used implicitly by some tagged related constructs, and
having it never inlined may cause performance issues, so move it to the
spec by taking advantage of the recently added declare expressions.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/a-tags.
With multi unit naming scheme (e.g. "unit.1.ada" and "unit.2.ada"
instead of "unit.ads" and "unit.adb") the Multiple_Unit_Index global
variable keeps the current index number. It is only assigned by routines
that return non-negative numbers.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/a
Hi Tobias,
Thanks for the quick review!
It helps when I'm on holiday :-)
Best regards
Thomas
Hi!
On Thu, Dec 17, 2020 at 02:15:51PM +0530, Senthil Kumar Selvaraj via
Gcc-patches wrote:
> The work on my github branch was not complete - I'd blindly followed
> whatever the CC0 Transition wiki mentioned (the first three steps of
> case #2), and fixed any regression fallout (for ATmega128).
>
With the introduction of C++20 modules and libcody, cc1plus and
cc1objplus gained a dependency on the socket functions. Before those
were merged into libc in Solaris 11.4, one needed to link with -lsocket -lnsl
on Solaris, so that merge broke the Solaris 11.3 build.
While we already have 4 differ
Hi Nikhil,
>> I suspect what is happening here is that godump sees "typedef ucontext_t
>> struct
>> ucontext" and outputs the typedef immediately. Only later does it observe
>> that
>> "struct ucontext" is invalid. At that point it is too late to comment out the
>> typedef for _ucontext_t.
>
> O
On 16/12/20 12:58 +0100, Matthias Kretz wrote:
--- /dev/null
+++ b/libstdc++-v3/scripts/check_simd
@@ -0,0 +1,76 @@
+#!/bin/sh
+
+# check_simd
+# Read config from $CHECK_SIMD_CONFIG file or $target_list
+
+scriptdir="$(cd "${0%/*}" && pwd)"
The ${0%/*} substitution is required by POSIX sh si
On 12/16/20 6:10 PM, Jakub Jelinek wrote:
Hi!
Seems c++tools doesn't honor --with-gcc-major-version-only.
Our distro uses that flag and so everything is installed in
/usr/lib/gcc//11/...
/usr/libexec/gcc//11/...
except
/usr/libexec/gcc//11.0.0/g++-mapper-server
The following patch should fix th
On 12/17/20 5:07 AM, Jakub Jelinek wrote:
On Thu, Dec 17, 2020 at 11:01:40AM +0100, Andreas Schwab wrote:
On Dez 15 2020, Nathan Sidwell wrote:
diff --git c/libcody/fatal.cc w/libcody/fatal.cc
new file mode 100644
index 000..b35094e6b19
--- /dev/null
+++ w/libcody/fatal.cc
@@ -0,0 +1,7
On 12/17/20 7:21 AM, Rainer Orth wrote:
With the introduction of C++20 modules and libcody, cc1plus and
cc1objplus gained a dependency on the socket functions. Before those
were merged into libc in Solaris 11.4, one needed to link with -lsocket -lnsl
on Solaris, so that merge broke the Solaris 1
On 12/17/20 8:25 AM, Nathan Sidwell wrote:
On 12/17/20 5:07 AM, Jakub Jelinek wrote:
On Thu, Dec 17, 2020 at 11:01:40AM +0100, Andreas Schwab wrote:
/tmp/ccabCPZ7.s: Assembler messages:
/tmp/ccabCPZ7.s:28: Error: Wrong number of input operands
Yeah. If it is meant as an optimization barrie
On Thu, Dec 17, 2020 at 08:25:12AM -0500, Nathan Sidwell wrote:
> > Yeah. If it is meant as an optimization barrier, shouldn't it be just
> >__asm__ volatile ("");
> > or
> >__asm__ volatile ("" : : : "memory");
> > ? That said, shouldn't that be guarded on the compiler being GCC
> > (or
Hi Rainer,
> -Original Message-
> From: Gcc-patches On Behalf Of
> Rainer Orth
> Sent: 17 December 2020 10:36
> To: gcc-patches@gcc.gnu.org
> Cc: Richard Sandiford
> Subject: [PATCH] rtl-ssa: Include memmodel.h before tm_p.h
>
> The RTL SSA merge broke SPARC bootstrap:
>
> In file incl
Hi all,
While experimenting with some backend costs for Advanced SIMD and SVE I hit
many cases where GCC would pick SVE for VLA auto-vectorisation even when the
backend very clearly presented cheaper costs for Advanced SIMD.
For a simple float addition loop the SVE costs were:
vec.c:9:21: note:
On Thu, Dec 17, 2020 at 5:36 AM Jakub Jelinek via Gcc-patches
wrote:
>
> On Thu, Dec 17, 2020 at 08:25:12AM -0500, Nathan Sidwell wrote:
> > > Yeah. If it is meant as an optimization barrier, shouldn't it be just
> > >__asm__ volatile ("");
> > > or
> > >__asm__ volatile ("" : : : "memory
Hi Kyryll,
>> Fixed by moving the memmove.h include in rtl-ssa.h before tm_p.h.
>>
>> Tested on sparc-sun-solaris2.11 and i386-pc-solaris2.11 (both into stage
>> 3 now, so the compilation error is gone).
>>
>> Ok for master?
>
> AFAIK simple patches like this that fix build/bootstrap are viewed
On 12/16/20 6:24 PM, Patrick Palka wrote:
On Wed, 16 Dec 2020, Jason Merrill wrote:
On 12/14/20 3:29 PM, Patrick Palka wrote:
On Mon, 14 Dec 2020, Jason Merrill wrote:
On 12/14/20 1:07 PM, Patrick Palka wrote:
This makes tracking of potentially unstable satisfaction results more
precise by
> -Original Message-
> From: Przemyslaw Wirkus
> Sent: 08 December 2020 14:05
> To: gcc-patches@gcc.gnu.org
> Cc: ni...@redhat.com; Richard Earnshaw ;
> Ramana Radhakrishnan ; Kyrylo
> Tkachov
> Subject: [PATCH][GCC] arm: Add support for Cortex-A78C
>
> This patch adds support for -mc
this addresses the first tranch of windows build problems, pushing to trunk.
When breaking out the sample server from the gcc/cp directory, it lost
its check for mmap, and the sample resolver just assumed it was there.
Fixed thusly. The non-mapping paths in module.cc weren't (recently)
excercise
Rainer Orth writes:
> Hi Kyryll,
>
>>> Fixed by moving the memmove.h include in rtl-ssa.h before tm_p.h.
>>>
>>> Tested on sparc-sun-solaris2.11 and i386-pc-solaris2.11 (both into stage
>>> 3 now, so the compilation error is gone).
>>>
>>> Ok for master?
>>
>> AFAIK simple patches like this that
On 16/12/20 13:38 +, Jonathan Wakely wrote:
This fixes a bug caused by a mismatch between the macros defined by
when GCC is built and the macros defined by when
users include . If the user code is compiled with
_XOPEN_SOURCE defined to 500 or 600, Darwin suppresses the
ENOTRECOVERABLE and E
On 27/11/20 17:46 +, Jonathan Wakely wrote:
On 20/11/20 16:30 -0800, Thomas Rodgers wrote:
From: Thomas Rodgers
Should include all discussion on and off list to date.
Most of the comments in
https://gcc.gnu.org/pipermail/gcc-patches/2020-November/558090.html
still apply to this version.
libstdc++-v3/ChangeLog:
PR libstdc++/98344
* include/bits/semaphore_base.h: Fix preprocessor condition.
Tested powerpc64le-linux. Committed to trunk.
commit 8dc63f13f03facc49b777195c9068432477b5dcd
Author: Jonathan Wakely
Date: Thu Dec 17 12:09:20 2020
libstdc++: Fix prep
Sadly strsignal is nonportable, so signal numbers it is then.
c++tools/
* server.cc (crash_signal): Don't use strsignal.
pushing to trunk
--
Nathan Sidwell
diff --git i/c++tools/server.cc w/c++tools/server.cc
index 6457dc5b878..8514ef6293b 100644
--- i/c++tools/server.cc
+++ w/c
As noted in PR 66146 comment 35, there is a new warning in the new
std::call_once implementation.
libstdc++-v3/ChangeLog:
* src/c++11/mutex.cc (std::once_flag::_M_finish): Add
maybe_unused attribute to variable used in assertion.
Tested powerpc64le-linux. Committed to trunk.
com
Kyrylo Tkachov via Gcc-patches writes:
> Hi all,
>
> While experimenting with some backend costs for Advanced SIMD and SVE I hit
> many cases where GCC would pick SVE for VLA auto-vectorisation even when the
> backend very clearly presented cheaper costs for Advanced SIMD.
> For a simple float add
> "Richard" == Richard Sandiford via Gcc-patches
> writes:
Richard> +// A class that stores a choice "A or B", where A has type T1 * and B
has
Richard> +// type T2 *. Both T1 and T2 must have an alignment greater than 1,
since
Richard> +// the low bit is used to identify B over A. T1
While this doesn't fix 98324, it was an omission. Cribbed code from
libcpp to build libcody as PIC.
libcody/
* configure.ac: Add --enable-host-shared.
* Makefile.in: Add FLAGPIC.
* configure: Regenerated.
pushing to trunk
--
Nathan Sidwell
diff --git i/libcody/M
On 19/08/20 17:57 -0400, Patrick Palka via Libstdc++ wrote:
On Wed, 22 Jul 2020, Patrick Palka wrote:
On Mon, 20 Jul 2020, Patrick Palka wrote:
> On Mon, 20 Jul 2020, Jonathan Wakely wrote:
>
> > On 20/07/20 08:53 -0400, Patrick Palka via Libstdc++ wrote:
> > > On Mon, 20 Jul 2020, Jonathan Wa
On 14/07/20 15:41 -0400, Patrick Palka via Libstdc++ wrote:
This imports the source files from the Ryu library that define
d2s_buffered_n, f2s_buffered_n, d2fixed_buffered_n and d2exp_buffered_n
and generic_binary_to_decimal, which we're going to need as part of our
std::to_chars implementation.
On 15/07/20 14:12 -0400, Patrick Palka via Libstdc++ wrote:
On Tue, 14 Jul 2020, Patrick Palka wrote:
This performs the following modifications to our local copy of Ryu in
order to make it more easily usable for our std::to_chars implementation:
* Remove all #includes
* Remove copy_special
On 14/07/20 15:41 -0400, Patrick Palka via Libstdc++ wrote:
libstdc++-v3/ChangeLog:
* testsuite/20_util/to_chars/double.cc: New test, consisting of
testcases imported from the MSVC STL testsuite.
* testsuite/20_util/to_chars/float.cc: Likewise.
OK, thanks.
operator_cast::op1_range() does some masking and calculations when a
higher precision value is cast to a lower precision one.
ie, it fills in the ranges for the upper bits and trys to produce more
usable results for the wind-back process. Part of that calculation
involved using PLUS_EXPR to th
On Wed, Dec 16, 2020 at 09:29:31AM +0100, Richard Biener wrote:
> I think it probably makes sense to have some helper split out that
> collects & classifies vector constructor components we can use from
> both forwprop (where matching the V_C_E from integer could be done
> as well IMHO) and bswap (
Hi!
We don't try to optimize for signed x, y (int) (x - 1U) * y + y
into x * y, we can't do that with signed x * y, because the former
is well defined for INT_MIN and -1, while the latter is not.
We could perhaps optimize it during isel or some very late optimization
where we'd turn magically flag
Hi all,
I'd like to submit the following simple patch to fix bootstrap on arm.
Apparently 'cff6dac28a0 arm: Replace calls to __builtin_vcreate* in arm_neon.h
[PR66791]'
forgot to remove one call to '__builtin_neon_vcreatedi'.
Bootstraped arm-linux-gnu, regression ongoing.
Okay for trunk?
Than
> -Original Message-
> From: Andrea Corallo
> Sent: 17 December 2020 14:56
> To: gcc-patches@gcc.gnu.org
> Cc: Kyrylo Tkachov ; Richard Earnshaw
> ; nd ; Prathamesh Kulkarni
>
> Subject: [PATCH] arm: Fix bootstrap
>
> Hi all,
>
> I'd like to submit the following simple patch to fix b
Hi!
Seems the ggc_remove ppc_nx 3 operand member relies on the hash tables to
contain pointers in the first element, which is not the case for
source_location_table* hash table, which has location_t and unsigned as
first two members and pointer somewhere else.
I've tried to change:
static void
> On Dec 17, 2020, at 6:21 AM, Segher Boessenkool
> wrote:
>
> Hi!
>
> On Thu, Dec 17, 2020 at 02:15:51PM +0530, Senthil Kumar Selvaraj via
> Gcc-patches wrote:
>> The work on my github branch was not complete - I'd blindly followed
>> whatever the CC0 Transition wiki mentioned (the first t
On Thu, Dec 17, 2020 at 09:06:57AM -0500, Nathan Sidwell wrote:
>
> Sadly strsignal is nonportable, so signal numbers it is then.
>
> c++tools/
> * server.cc (crash_signal): Don't use strsignal.
Strange. It is portable, we even have a fallback implementation for it in
libibe
Kyrylo Tkachov writes:
>> -Original Message-
>> From: Andrea Corallo
>> Sent: 17 December 2020 14:56
>> To: gcc-patches@gcc.gnu.org
>> Cc: Kyrylo Tkachov ; Richard Earnshaw
>> ; nd ; Prathamesh Kulkarni
>>
>> Subject: [PATCH] arm: Fix bootstrap
>>
>> Hi all,
>>
>> I'd like to submit t
Tom Tromey writes:
>> "Richard" == Richard Sandiford via Gcc-patches
>> writes:
>
> Richard> +// A class that stores a choice "A or B", where A has type T1 * and
> B has
> Richard> +// type T2 *. Both T1 and T2 must have an alignment greater than
> 1, since
> Richard> +// the low bit
On 12/17/20 10:38 AM, Richard Sandiford via Gcc-patches wrote:
Tom Tromey writes:
"Richard" == Richard Sandiford via Gcc-patches writes:
Richard> +// A class that stores a choice "A or B", where A has type T1 * and B
has
Richard> +// type T2 *. Both T1 and T2 must have an alignment greater
On 10/29/20 8:11 PM, H.J. Lu via Binutils wrote:
> diff --git a/Makefile.tpl b/Makefile.tpl
> index a280a1498c..38f0b021f4 100644
> --- a/Makefile.tpl
> +++ b/Makefile.tpl
> +@if pgo-build
> + && $(MAKE) $(RECURSE_FLAGS_TO_PASS) \
shouldn't make called with -i here? you're not interested lett
The refactoring in r11-5500 altered the condition for the gthreads-timed
test from #if to #ifdef. For some reason that macro is always defined,
rather than being defined to 1 or undefined like most of our autoconf
macros. That means the test always passes now, even for targets where
the macro is de
On 17/12/20 14:05 +, Jonathan Wakely wrote:
On 16/12/20 13:38 +, Jonathan Wakely wrote:
This fixes a bug caused by a mismatch between the macros defined by
when GCC is built and the macros defined by when
users include . If the user code is compiled with
_XOPEN_SOURCE defined to 500 or
On 12/17/20 7:28 AM, Rainer Orth wrote:
I first tried with the new version included, but that broke badly:
cc1 -fpreprocessed sysinfo.i -quiet -O -std=gnu99
-fdump-go-spec=tmp-gen-sysinfo.go -o sysinfo.s
now SEGVs with either infinite or very deep recursion:
Thread 2 received signal SIGSEGV,
On Thu, 17 Dec 2020 at 21:04, Andrea Corallo wrote:
>
> Kyrylo Tkachov writes:
>
> >> -Original Message-
> >> From: Andrea Corallo
> >> Sent: 17 December 2020 14:56
> >> To: gcc-patches@gcc.gnu.org
> >> Cc: Kyrylo Tkachov ; Richard Earnshaw
> >> ; nd ; Prathamesh Kulkarni
> >>
> >> Subj
On 10/15/20 5:05 PM, Tom de Vries wrote:
> On 10/2/20 3:21 PM, Tom de Vries wrote:
>> Hi,
>>
>> Consider the test-case libgomp.c/pr81778.c added in this commit, with
>> this core loop (note: CANARY_SIZE set to 0 for simplicity):
>> ...
>> int s = 1;
>> #pragma omp target simd
>> for (int i =
The attached patch implements delinearization of array accesses in the
Fortran front end, something that has been discussed for a long time.
I've been asked to try to get this patch committed on the OG10 branch
since it is blocking some further optimization work with Graphite for
OpenACC kerne
On Fri, Dec 11, 2020 at 4:55 PM Patrick Palka wrote:
>
> On Fri, 11 Dec 2020, Jason Merrill wrote:
>
> > On 12/11/20 10:31 AM, Patrick Palka wrote:
> > > On Thu, 10 Dec 2020, Patrick Palka wrote:
> > >
> > > > On Thu, 10 Dec 2020, Jason Merrill wrote:
> > > >
> > > > > On 12/10/20 11:21 AM, Patric
On 10/29/20 8:11 PM, H.J. Lu via Binutils wrote:
> diff --git a/configure.ac b/configure.ac
> index 7c4bdff0fa..eea9a21099 100644
> --- a/configure.ac
> +++ b/configure.ac
> + if test "$enable_pgo_build" = "lto"; then
> +AC_MSG_CHECKING([whether the compiler supports -flto=jobserver])
> +o
On Thu, 17 Dec 2020, Patrick Palka wrote:
> On Fri, Dec 11, 2020 at 4:55 PM Patrick Palka wrote:
> >
> > On Fri, 11 Dec 2020, Jason Merrill wrote:
> >
> > > On 12/11/20 10:31 AM, Patrick Palka wrote:
> > > > On Thu, 10 Dec 2020, Patrick Palka wrote:
> > > >
> > > > > On Thu, 10 Dec 2020, Jason Me
This patch adds new movmisalign_mve_load and store patterns for
MVE to help vectorization. They are very similar to their Neon
counterparts, but use different iterators and instructions.
Indeed MVE supports less vectors modes than Neon, so we use
the MVE_VLD_ST iterator where Neon uses VQX.
Since
This patch enables MVE vshlq instructions for auto-vectorization.
The existing mve_vshlq_n_ is kept, as it takes a single
immediate as second operand, and is used by arm_mve.h.
We move the vashl3 insn from neon.md to an expander in
vec-common.md, and the mve_vshlq_ insn from mve.md to
vec-common.
This patch enables MVE vshr instructions for auto-vectorization. New
MVE patterns are introduced that take a vector of constants as second
operand, all constants being equal.
The existing mve_vshrq_n_ is kept, as it takes a single
immediate as second operand, and is used by arm_mve.h.
The vashr3
Clang didn't like sizeot (uintset::value) in a templated context. Not
sure where the problem lies -- ambiguous std, gcc erroneous accept or
clang erroneous reject. Anyway, this avoids that construct.
PR c++/98340
gcc/cp/
* module.c (uintset::hash::add): Use uintset
On Thu, Dec 17, 2020 at 8:31 AM Nikhil Benesch via Gcc-patches
wrote:
>
> On 12/17/20 7:28 AM, Rainer Orth wrote:
> > I first tried with the new version included, but that broke badly:
> >
> > cc1 -fpreprocessed sysinfo.i -quiet -O -std=gnu99
> > -fdump-go-spec=tmp-gen-sysinfo.go -o sysinfo.s
> >
On Fri, Dec 11, 2020 at 01:51:44PM -0600, Segher Boessenkool wrote:
> Hi!
>
> On Thu, Nov 19, 2020 at 07:05:24PM -0500, Michael Meissner wrote:
> > If the glibc is not 2.32 or later, this code just compiles to using abort.
>
> That is the compile-time target glibc. That is often *not* the glibc
On Mon, Dec 14, 2020 at 05:26:03PM -0600, Segher Boessenkool wrote:
> Hi!
>
> On Thu, Dec 03, 2020 at 10:57:56PM -0500, Michael Meissner wrote:
> > --- a/gcc/testsuite/gcc.target/powerpc/pr70117.c
> > +++ b/gcc/testsuite/gcc.target/powerpc/pr70117.c
> > @@ -1,5 +1,6 @@
> > -/* { dg-do run { target
On 12/17/20 1:05 PM, Ian Lance Taylor wrote:
Thanks for looking into this. I've gotten confused now, though.
Which patch fixes the problem on Solaris? Thanks.
Ian
The patch I submitted upstream is all that is needed to fix the
compilation failures on Solaris:
https://go-review.googlesou
It seems users are confused by the lack of standard library header
units.
gcc/
* doc/invoke.texi (C++ Modules): Document lack of std
library header units.
--
Nathan Sidwell
diff --git i/gcc/doc/invoke.texi w/gcc/doc/invoke.texi
index 054b8371593..8766bcdfc18 100644
--- i/
Add the --enable-pgo-build[=lto] configure option. When binutils+gdb
is not built together with GCC, --enable-pgo-build enables the PGO build:
1. First build with -fprofile-generate.
2. Use "make maybe-check-*" to generate profiling data and pass -i to make
to ignore errors when generating profil
Add the --enable-pgo-build[=lto] configure option. When binutils+gdb
is not built together with GCC, --enable-pgo-build enables the PGO build:
1. First build with -fprofile-generate.
2. Use "make maybe-check-*" to generate profiling data and pass -i to make
to ignore errors when generating profil
Detect GCC LTO plugin. Pass --plugin to AR and RANLIB to support LTO
build.
bfd/
* configure: Regenerated.
binutils/
* configure: Regenerated.
gas/
* configure: Regenerated.
gprof/
* configure: Regenerated.
ld/
* configure: Regenerated.
libctf/
Detect GCC LTO plugin. Pass --plugin to AR and RANLIB to support LTO
build.
* Makefile.tpl (AR): Add @AR_PLUGIN_OPTION@
(RANLIB): Add @RANLIB_PLUGIN_OPTION@.
* configure.ac: Include config/gcc-plugin.m4.
AC_SUBST AR_PLUGIN_OPTION and RANLIB_PLUGIN_OPTION.
*
On Linux/x86_64,
c25b504636fec7bf8f181a84af83a52757ba7e89 is the first bad commit
commit c25b504636fec7bf8f181a84af83a52757ba7e89
Author: Andrew MacLeod
Date: Thu Dec 17 09:24:11 2020 -0500
Fix trap in pointer conversion in op1_range.
caused
FAIL: gcc.dg/pr97750.c (test for excess errors
On Thu, Dec 17, 2020 at 6:16 AM Richard Sandiford via Gcc-patches
wrote:
>
> Kyrylo Tkachov via Gcc-patches writes:
> > Hi all,
> >
> > While experimenting with some backend costs for Advanced SIMD and SVE I hit
> > many cases where GCC would pick SVE for VLA auto-vectorisation even when the
> >
The top of trunk fails to build with the error below. I haven't
spent any time debugging it except to look at git log where
the description for r11-6238 mentions the function also referenced
in the error message. Could the patch be responsible for this?
https://gcc.gnu.org/pipermail/gcc-cvs/2020
From: Thomas Rodgers
Let's see if this one sticks...
Adds
libstdc++/ChangeLog:
* doc/doxygen/user.cfg.in: Add new header.
* include/Makefile.am (std_headers): likewise.
* include/Makefile.in: Regenerate.
* include/precompiled/stdc++.h: Add new header.
*
On Thu, Dec 17, 2020 at 11:07 AM Martin Sebor via Gcc-patches
wrote:
>
> The top of trunk fails to build with the error below. I haven't
> spent any time debugging it except to look at git log where
> the description for r11-6238 mentions the function also referenced
> in the error message. Coul
On 12/17/20 10:04 AM, Jakub Jelinek wrote:
Hi!
Seems the ggc_remove ppc_nx 3 operand member relies on the hash tables to
contain pointers in the first element, which is not the case for
source_location_table* hash table, which has location_t and unsigned as
first two members and pointer somewher
I'd used reg_raw_mode[regno] for general registers, even though
the array is only valid for hard registers. This patch uses
regno_reg_rtx instead.
Tested on i686-linux-gnu, committed as obvious.
Richard
gcc/
PR rtl-optimization/98347
* rtl-ssa/access-utils.h (full_register): Us
1 - 100 of 111 matches
Mail list logo
