On Fri, Aug 30, 2019 at 2:08 AM Hongtao Liu wrote:
>
> On Fri, Aug 30, 2019 at 2:09 AM Uros Bizjak wrote:
> >
> > 2019-08-28 Uroš Bizjak
> >
> > * config/i386/i386.c (ix86_register_move_cost): Do not
> > limit the cost of moves to/from XMM register to minimum 8.
> >
> > Bootstrapped an
warning_at() calls with the %G directive rely on the gimple statement
for both their location and the inlining context. When the statement
is not associated with a location, the warning doesn't point at any
line even if the location argument passed to the call was valid.
The attached patch change
On Fri, Aug 30, 2019 at 8:10 AM Hongtao Liu wrote:
>
> On Fri, Aug 30, 2019 at 2:09 AM Uros Bizjak wrote:
> >
> > 2019-08-28 Uroš Bizjak
> >
> > * config/i386/i386.c (ix86_register_move_cost): Do not
> > limit the cost of moves to/from XMM register to minimum 8.
> >
> > Bootstrapped an
On Fri, Aug 30, 2019 at 2:09 AM Uros Bizjak wrote:
>
> 2019-08-28 Uroš Bizjak
>
> * config/i386/i386.c (ix86_register_move_cost): Do not
> limit the cost of moves to/from XMM register to minimum 8.
>
> Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
>
> Actually committe
On 8/29/19 9:44 AM, Maxim Kuvyrkov wrote:
> Hi,
>
> The first patch adds ranking statistics for autoprefetcher heuristic.
>
> The second one makes it easier to diff scheduler debug dumps by adding more
> context lines for diff at clock increments.
>
> OK to commit?
OK for both.
jeff
Hi Mike,
On Mon, Aug 26, 2019 at 04:31:02PM -0400, Michael Meissner wrote:
> (rs6000_asm_output_opcode): New function for prifixed memory.
Typo. Just say "New." or "New function." please.
> --- gcc/config/rs6000/rs6000.c(revision 274871)
> +++ gcc/config/rs6000/rs6000.c(wo
The -Warray-bounds enhancement I added to GCC 9 causes false
positives in languages like Fortran whose first array element
is at a non-zero index. The attached patch has the function
responsible for the warning normalize the array bounds to
always start at zero to avoid these false positives.
Te
On Wed, Aug 28, 2019 at 05:26:55PM -0400, Michael Meissner wrote:
> On Wed, Aug 28, 2019 at 12:14:58PM -0500, Segher Boessenkool wrote:
> > > +/* Enumeration giving the type of traditional addressing that would be
> > > used to
> > > + decide whether an instruction uses prefixed memory or not.
On 8/29/19 11:08 AM, Christophe Lyon wrote:
> On Thu, 29 Aug 2019 at 10:58, Kyrill Tkachov
> wrote:
>>
>> Hi Bernd,
>>
>> On 8/28/19 10:36 PM, Bernd Edlinger wrote:
>>> On 8/28/19 2:07 PM, Christophe Lyon wrote:
Hi,
This patch causes an ICE when building libgcc's unwind-arm.o
w
On 8/28/19 5:56 PM, Marek Polacek wrote:
--- gcc/doc/invoke.texi
+++ gcc/doc/invoke.texi
@@ -3516,6 +3516,19 @@ result in a call to @code{terminate}.
Disable the warning about the case when a conversion function converts an
object to the same type, to a base class of that type, or to void; su
og:
/usr/aarch64-suse-linux/bin/ld: _gotest_.o: in function
`archive..z2ftar.Reader.next':
/opt/gcc/gcc-20190829/Build/aarch64-suse-linux/ilp32/libgo/gotest1086/test/reader.go:72:
undefined reference to `runtime.goPanicExtendSliceAlen'
Andreas.
--
Andreas Schwab, sch...@lin
On Fri, 30 Aug 2019, Akshat Garg wrote:
> > The first question for any new thing that is syntactically a qualifier is:
> > is it intended generally to be counted as a qualifier where the standard
> > refers to qualified type, the unqualified version of a type, etc.? Or is
> > it, like _Atomic, a
Hi Joseph,
Many thanks for giving us your feedback.
On Tue, Aug 20, 2019 at 3:57 AM Joseph Myers
wrote:
> On Tue, 30 Jul 2019, Martin Sebor wrote:
>
> > On 7/30/19 1:13 AM, Akshat Garg wrote:
> > > Hi,
> > > This patch includes C front-end code for a type qualifier
> _Dependent_ptr.
> >
> > Just
Hi Ian,
> This patch to the Go frontend and libgo changes the panic message
> reported for an out of bounds index or slice operation to include the
> invalid values. This makes it easier for the user to see what the
> problem is. This implements https://golang.org/cl/161477 in the
> gofrontend,
Hi
I am having this warning:
/home/fdt/dev/gcc/git/libstdc++-v3/testsuite/util/testsuite_performance.h:170:
attention: ignoring return value of « void* malloc(size_t) » declared
with attribute « warn_unused_result » [-Wunused-result]
170 | malloc(0); // Needed for some implementat
2019-08-29 Uroš Bizjak
* config/i386/i386.c (inline_secondary_memory_needed): Return true
for moves between SSE and non-general registers and between
mask and non-general registers.
(ix86_register_move_cost): Remove stalled comment.
Bootstrapped and regression tested on x86_64-
Hi Alexander,
> So essentially the main issue is not a hardware peculiarity, but rather the
> bad schedule being totally wrong (it could only make sense if loads had
> 1-cycle
> latency, which they do not).
The scheduling is only bad because the specific intrinsics used are mapped
onto asm stat
2019-08-28 Uroš Bizjak
* config/i386/i386.c (ix86_register_move_cost): Do not
limit the cost of moves to/from XMM register to minimum 8.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
Actually committed as r274994 with the wrong ChangeLog.
Uros.
diff --git a/gcc/conf
We reject this test with errors like
nontype1.C:22:14: error: taking address of rvalue [-fpermissive]
22 | A{}> a2;
| ^~~
that happens because for converting "C{}" to int we generate
"C::operator int (&TARGET_EXPR )". The second
template argument is a binary opera
On Thu, 29 Aug 2019, Maxim Kuvyrkov wrote:
> >> r1 = [rb + 0]
> >>
> >> r2 = [rb + 8]
> >>
> >> r3 = [rb + 16]
> >>
> >>
> >> which, apparently, cortex-a53 autoprefetcher doesn't recognize. This
> >> schedule happens because r2= load gets lower priority than the
> >> "irrelevant" due to the
Hi Maxim,
> It appears that cores with autoprefetcher hardware prefer loads and stores
>bundled together, not interspersed with > other instructions to occupy the
>rest of CPU units.
I don't believe it is as simple as that - modern cores have multiple
prefetchers but
won't prefer bund
We currently use default mid-end expanders for logical DImode operations.
These split operations without first splitting off complex immediates or
memory operands. The resulting expansions are non-optimal and allow for
fewer LDRD/STRD opportunities. So add back explicit expanders which ensure
mem
Hi!
On Tue, Aug 20, 2019 at 02:00:31PM -0400, Michael Meissner wrote:
> I
> noticed on power5 that the V1TImode mode is allowed in Altivec registers, even
> though power5 doesn't have Altivec registers.
>
> While it doesn't seem to effect anything (I couldn't create a test case that
> failed), it
> On Aug 29, 2019, at 7:29 PM, Richard Biener
> wrote:
>
> On August 29, 2019 5:40:47 PM GMT+02:00, Maxim Kuvyrkov
> wrote:
>> Hi,
>>
>> This patch tweaks autoprefetcher heuristic in scheduler to better group
>> memory loads and stores together.
>>
>> From https://gcc.gnu.org/bugzilla/show_b
On August 29, 2019 5:40:47 PM GMT+02:00, Maxim Kuvyrkov
wrote:
>Hi,
>
>This patch tweaks autoprefetcher heuristic in scheduler to better group
>memory loads and stores together.
>
>From https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91598:
>
>There are two separate changes, both related to instruct
Hi,
On Fri, Jul 12 2019, Feng Xue OS wrote:
> Current IPA-cp only generates cost-evaluating predicate for conditional
> statement like
> "if (param cmp const_val)", it is too simple and conservative. This patch
> generalizes the
> process to handle the form as T(param), a mathematical transforma
Hi,
The first patch adds ranking statistics for autoprefetcher heuristic.
The second one makes it easier to diff scheduler debug dumps by adding more
context lines for diff at clock increments.
OK to commit?
--
Maxim Kuvyrkov
www.linaro.org
0002-Add-missing-entry-for-rank_for_schedule-stat
Hi,
This patch tweaks autoprefetcher heuristic in scheduler to better group memory
loads and stores together.
From https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91598:
There are two separate changes, both related to instruction scheduler, that
cause the regression. The first change in r253235
On 29/08/2019 15:57, Christophe Lyon wrote:
Hi,
On 15/05/2019 14:39, Christophe Lyon wrote:
Hello,
This patch series implements the GCC contribution of the FDPIC ABI for
ARM targets.
This ABI enables to run Linux on ARM MMU-less cores and supports
shared libraries to reduce the memory footpri
> Am 22.08.2019 um 15:45 schrieb Ilya Leoshkevich :
>
> Bootstrap and regtest running on x86_64-redhat-linux and
> s390x-redhat-linux.
>
> This patch series adds signaling FP comparison support (both scalar and
> vector) to s390 backend.
I'm running into a problem on ppc64 with this patch, and i
Hi Christophe,
On 5/15/19 1:39 PM, Christophe Lyon wrote:
We call __aeabi_read_tp() to get the thread pointer. Since this is a
function call, we have to restore the FDPIC register afterwards.
2019-XX-XX Christophe Lyon
Mickaël Guêné
gcc/
* config/arm/arm.c (arm_load
Here is an updated version that makes use of the helper
gen_restore_pic_register_after_call
Christophe
On 15/05/2019 14:39, Christophe Lyon wrote:
We call __aeabi_read_tp() to get the thread pointer. Since this is a
function call, we have to restore the FDPIC register afterwards.
2019-XX-XX
On 31/07/2019 16:44, Christophe Lyon wrote:
On Tue, 16 Jul 2019 at 14:42, Kyrill Tkachov
wrote:
On 7/16/19 12:18 PM, Kyrill Tkachov wrote:
Hi Christophe
On 5/15/19 1:39 PM, Christophe Lyon wrote:
In FDPIC mode, the trampoline generated to support pointers to nested
functions looks like:
On 12/07/2019 08:06, Richard Sandiford wrote:
Christophe Lyon writes:
In FDPIC, we need to make sure __do_global_dtors_aux and frame_dummy
are referenced by their address, not by pointers to the function
descriptors.
2019-XX-XX Christophe Lyon
Mickaël Guêné
* libgcc/crtstu
On 16/07/2019 13:58, Richard Sandiford wrote:
Christophe Lyon writes:
The FDPIC register is hard-coded to r9, as defined in the ABI.
We have to disable tailcall optimizations if we don't know if the
target function is in the same module. If not, we have to set r9 to
the value associated with t
Hi Christophe,
On 5/15/19 1:39 PM, Christophe Lyon wrote:
Without this, when we are unwinding across a signal frame we can jump
to an even address which leads to an exception.
This is needed in __gnu_persnality_sigframe_fdpic() when restoring the
PC from the signal frame since the PC saved by t
This patch adds a port for the Linux kernel eBPF architecture to GCC.
ChangeLog:
* configure.ac: Support for bpf-*-* targets.
* configure: Regenerate.
contrib/ChangeLog:
* config-list.mk (LIST): Disable go in bpf-*-* targets.
gcc/ChangeLog:
* config.gcc: Support for bpf-*-* targets.
Hi all,
Advice and help needed!
This patch makes changes to the the complex division in GCC. The algorithm
used is same as in https://gcc.gnu.org/ml/gcc-patches/2019-08/msg01629.html -
same problems, same improvement in robustness, same loss in accuracy.
Since Baudin adds another underflow c
The function integer_range_info makes sure that, if provided, the
initial value fills in the especified range. However, it is necessary
to convert the values to a numerical context before comparing, to make
sure awk is using arithmetical order and not lexicographical order.
gcc/ChangeLog:
[Differences from V5:
. Use TARGET_BIG_ENDIAN instead of TARGET_LITTLE_ENDIAN, and make sure
ASM_SPEC always passes an endianness selector argument to the
assembler.
. De-obfuscate the usage of arg.type_size_in_bytes and
arg_aggregate-type_p.
. Increase the cummulative argument unconditionall
This patch adds an eBPF port to libgcc.
As of today, compiled eBPF programs do not support a single-entry
point schema. Instead, a BPF "executable" is a relocatable ELF object
file containing multiple entry points, in certain named sections.
Also, the BPF loaders in the kernel do not execute .in
This patch makes many tests in gcc.dg and gcc.c-torture to be skipped
in bpf-*-* targets. This is due to the many limitations imposed by
eBPF to what would be perfectly valid C code: no support for more than
5 arguments to function calls, no support for indirect jumps, a very
limited range for dir
* config.sub: Import upstream version 2019-06-30.
* config.guess: Import upstream version 2019-07-24.
---
ChangeLog| 5 ++
config.guess | 264 +++
config.sub | 50 +--
3 files changed, 240 insertions(+), 79 de
ChangeLog:
* MAINTAINERS: Add myself as the maintainer of the eBPF port.
Remove myself from Write After Approval section.
---
ChangeLog | 5 +
MAINTAINERS | 2 +-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index 5d8402949bc..5d69d
gcc/ChangeLog:
* doc/invoke.texi (Option Summary): Cover eBPF.
(eBPF Options): New section.
* doc/extend.texi (BPF Built-in Functions): Likewise.
(BPF Kernel Helpers): Likewise.
---
gcc/ChangeLog | 7 +++
gcc/doc/extend.texi | 171 ++
This patch makes the several effective target checks in
target-supports.exp to be aware of eBPF targets.
gcc/testsuite/ChangeLog:
* lib/target-supports.exp
(check_effective_target_trampolines): Adapt to eBPF.
(check_effective_target_stack_size): Likewise.
(dg-effec
This patch adds a new testsuite to gcc.target, with eBPF specific
tests.
Tests are included for:
- Target specific diagnostics.
- All built-in functions.
testsuite/ChangeLog:
* gcc.target/bpf/bpf.exp: New file.
* gcc.target/bpf/builtin-load.c: Likewise.
* cc.target/bpf/co
This patch adds a new dg_require_effective_target procedure to the
testsuite infrastructure: indirect_calls. This new function tells
whether a target supports calls to non-constant call targets.
This patch also annotates the tests in the gcc.c-torture testuite that
require support for indirect ca
This patch annotates tests that make use of a significant a mount of
stack space. Embedded and other restricted targets may have problems
compiling and running these tests. Note that the annotations are in
many cases not exact.
testsuite/ChangeLog:
* gcc.c-torture/compile/2609-1.c:
Hi!
I've backported following 12 commits from trunk to 8.4,
bootstrapped/regtested on x86_64-linux and i686-linux and committed
to gcc-8-branch.
Jakub
2019-08-29 Jakub Jelinek
Backported from mainline
2019-04-19 Jakub Jelinek
PR middle-end/90139
* t
On 16/07/2019 12:34, Richard Sandiford wrote:
Christophe Lyon writes:
On 22/05/2019 10:45, Christophe Lyon wrote:
On Wed, 22 May 2019 at 10:39, Szabolcs Nagy wrote:
On 21/05/2019 16:28, Christophe Lyon wrote:
--- a/gcc/config/arm/linux-eabi.h
+++ b/gcc/config/arm/linux-eabi.h
@@ -89,7 +89,
Hi,
On Fri, Jul 12 2019, Feng Xue OS wrote:
> IPA does not construct executability predicate for default case of switch
> statement.
> So execution cost of default case is not properly evaluated in IPA-cp, this
> might
> prevent function clone for function containing switch statement, if certain
On 12/07/2019 08:49, Richard Sandiford wrote:
Christophe Lyon writes:
The new arm-uclinuxfdpiceabi target behaves pretty much like
arm-linux-gnueabi. In order the enable the same set of features, we
have to update several configure scripts that generally match targets
like *-*-linux*: in most p
On 16/07/2019 12:11, Richard Sandiford wrote:
[This isn't really something that should be reviewed under global
reviewership, but if it's either that or nothing, I'll do it anyway...]
Christophe Lyon writes:
2019-XX-XX Christophe Lyon
Mickaël Guêné
gcc/
* config/a
Hi Martin,
> The recent sprintf+strlen integration doesn't handle unbounded
> string lengths entirely correctly for ILP32 targets and causes
> -Wformat-overflow false positives in some common cases, including
> during GCC bootstrap targeting such systems The attached patch
> fixes that mistake.
This patch refactors pieces of the include stacking routines. Again
from the modules branch, so these happen to provide the right slots to
make that work, but I think they're good in their own right.
_cpp_find_file excitingly hid an if as the loop-invariant conditional of
a for. Refactored t
Hi,
On 15/05/2019 14:39, Christophe Lyon wrote:
Hello,
This patch series implements the GCC contribution of the FDPIC ABI for
ARM targets.
This ABI enables to run Linux on ARM MMU-less cores and supports
shared libraries to reduce the memory footprint.
Without MMU, text and data segments rela
Hi,
On Thu, Aug 29 2019, Richard Biener wrote:
> On Thu, Aug 29, 2019 at 11:04 AM Martin Jambor wrote:
>>
>> Hi,
>>
>> when turning a tail-recursive call into a loop, the tail-call pass
>> creates a phi node for each gimple_reg function parameter that has any
>> use at all, even when the value pa
This patch allows for more efficient SVE2 vectorization of Multiply High with
Round and Scale (MULHRS) patterns.
The example snippet:
uint16_t a[N], b[N], c[N];
void foo_round (void)
{
for (int i = 0; i < N; i++)
a[i] = int32_t)b[i] * (int32_t)c[i]) >> 14) +
PR libstdc++/91067
* acinclude.m4 (libtool_VERSION): Bump to 6:28:0.
* configure: Regenerate.
* config/abi/pre/gnu.ver (GLIBCXX_3.4.28): Add new version. Export
missing symbols.
* testsuite/27_io/filesystem/iterators/91067.cc: Test move
const
Hi Robin,
On Thu, Aug 29, 2019 at 11:08:11AM +0200, Robin Dapp wrote:
> >> PR37451. Not clear what target that regressed on, btw.
> >
> > And PR55190 and PR67288 and probably more.
>
> Thanks for finding those. So the hope is to get this fixed or rather
> move towards a fix with the patch seri
On Thu, 29 Aug 2019, Richard Biener wrote:
> On Thu, 29 Aug 2019, Jakub Jelinek wrote:
>
> > On Thu, Aug 29, 2019 at 12:04:53PM +0200, Richard Biener wrote:
> > > + else
> >
> > Perhaps use
> > else if (MAY_HAVE_DEBUG_BIND_INSNS)
> > instead so that you don't walk it once again if there can'
On 8/29/19 3:57 AM, Bernhard Reutner-Fischer wrote:
On Wed, 28 Aug 2019 14:42:57 -0400
Nathan Sidwell wrote:
/* If opened with #import or contains #pragma once. */
- bool once_only;
+ bool once_only : 1;
I'm curious why you have them as bool and not unsigned?
because they're true/fa
On Thu, 29 Aug 2019, Uros Bizjak wrote:
> On Thu, Aug 29, 2019 at 12:04 PM Richard Biener wrote:
> >
> >
> > The following fixes the bootstrap-debug miscompare caused by STV
> > where we ended up with chain-to-scalar copies just because of
> > debug uses. Instead we have to avoid that, eventuall
On Thu, 29 Aug 2019, Jakub Jelinek wrote:
> On Thu, Aug 29, 2019 at 12:04:53PM +0200, Richard Biener wrote:
> > + else
>
> Perhaps use
> else if (MAY_HAVE_DEBUG_BIND_INSNS)
> instead so that you don't walk it once again if there can't be DEBUG_INSNs?
Sure - will do as followup to unbreak
On 29/08/19 12:17 +0200, Jakub Jelinek wrote:
On Wed, Aug 28, 2019 at 10:14:35PM +0100, Jonathan Wakely wrote:
* config/abi/post/i386-linux-gnu/baseline_symbols.txt: Update.
* config/abi/post/i486-linux-gnu/baseline_symbols.txt: Update.
* config/abi/post/powerpc-linux-gnu
On Wed, Aug 28, 2019 at 10:14:35PM +0100, Jonathan Wakely wrote:
> * config/abi/post/i386-linux-gnu/baseline_symbols.txt: Update.
> * config/abi/post/i486-linux-gnu/baseline_symbols.txt: Update.
> * config/abi/post/powerpc-linux-gnu/baseline_symbols.txt: Update.
> * config/a
On Thu, Aug 29, 2019 at 12:04:53PM +0200, Richard Biener wrote:
> + else
Perhaps use
else if (MAY_HAVE_DEBUG_BIND_INSNS)
instead so that you don't walk it once again if there can't be DEBUG_INSNs?
Jakub
On Thu, Aug 29, 2019 at 12:04 PM Richard Biener wrote:
>
>
> The following fixes the bootstrap-debug miscompare caused by STV
> where we ended up with chain-to-scalar copies just because of
> debug uses. Instead we have to avoid that, eventually substituting
> into debug uses or resetting debug s
The following fixes the bootstrap-debug miscompare caused by STV
where we ended up with chain-to-scalar copies just because of
debug uses. Instead we have to avoid that, eventually substituting
into debug uses or resetting debug stmts when there are reaching
defs from both inside and outside of
And the patch...
On Thu, Aug 29, 2019 at 12:00 PM Uros Bizjak wrote:
>
> As usual with costing changes, the patch exposes latent problem. The
> patched compiler tries to generate non-existing DImode move from mask
> register to XMM register, and ICEs during reload [1]. Attached patch
> tightens s
As usual with costing changes, the patch exposes latent problem. The
patched compiler tries to generate non-existing DImode move from mask
register to XMM register, and ICEs during reload [1]. Attached patch
tightens secondary_reload_needed condition and fixes the issue.
I'm bootstrapping and regr
On Thu, Aug 29, 2019 at 11:04 AM Martin Jambor wrote:
>
> Hi,
>
> when turning a tail-recursive call into a loop, the tail-call pass
> creates a phi node for each gimple_reg function parameter that has any
> use at all, even when the value passed to the original call is the same
> as the received
> +/* Return true if an argument at the position indicated by CUM should
> + be passed by reference. If the hook returns true, a copy of that
> + argument is made in memory and a pointer to the argument is passed
> + instead of the argument itself. */
> +
> +static
On Thu, Aug 29, 2019 at 10:41 AM Jakub Jelinek wrote:
>
> Hi!
>
> The following patch improves especially V8SFmode permutations for
> AVX (non-AVX2) ISA, where we punted way too often, even when we can handle
> it.
> On the
> typedef float __v8sf __attribute__((vector_size (32)));
> typedef double
On Thu, 29 Aug 2019 at 10:58, Kyrill Tkachov
wrote:
>
> Hi Bernd,
>
> On 8/28/19 10:36 PM, Bernd Edlinger wrote:
> > On 8/28/19 2:07 PM, Christophe Lyon wrote:
> >> Hi,
> >>
> >> This patch causes an ICE when building libgcc's unwind-arm.o
> >> when configuring GCC:
> >> --target arm-none-linux-g
>> PR37451. Not clear what target that regressed on, btw.
>
> And PR55190 and PR67288 and probably more.
Thanks for finding those. So the hope is to get this fixed or rather
move towards a fix with the patch series that's currently reviewed which
injects some doloop knowledge into ivopts?
As s
Hi,
when turning a tail-recursive call into a loop, the tail-call pass
creates a phi node for each gimple_reg function parameter that has any
use at all, even when the value passed to the original call is the same
as the received one, when it is the parameter's default definition.
This results in
On Thu, 29 Aug 2019, Jakub Jelinek wrote:
> Hi!
>
> switchconv uses unsigned_type_for to get unsigned type to perform
> computations in, which is fine if you just do a comparison in that type or
> similar, but not when actually constructing range-like checks or doing
> further arithmetics on the
Hi Bernd,
On 8/28/19 10:36 PM, Bernd Edlinger wrote:
On 8/28/19 2:07 PM, Christophe Lyon wrote:
Hi,
This patch causes an ICE when building libgcc's unwind-arm.o
when configuring GCC:
--target arm-none-linux-gnueabihf --with-mode thumb --with-cpu
cortex-a15 --with-fpu neon-vfpv4:
The build wo
jema...@gnu.org (Jose E. Marchesi) writes:
> +/* Return true if an argument at the position indicated by CUM should
> + be passed by reference. If the hook returns true, a copy of that
> + argument is made in memory and a pointer to the argument is passed
> + instead of the argument itself.
Hi!
switchconv uses unsigned_type_for to get unsigned type to perform
computations in, which is fine if you just do a comparison in that type or
similar, but not when actually constructing range-like checks or doing
further arithmetics on the type.
The reassoc and fold-const range test optimizatio
Hi!
The following patch improves especially V8SFmode permutations for
AVX (non-AVX2) ISA, where we punted way too often, even when we can handle
it.
On the
typedef float __v8sf __attribute__((vector_size (32)));
typedef double __v4df __attribute__((vector_size (32)));
typedef int __v8si __attribut
On Wed, 28 Aug 2019, Richard Sandiford wrote:
> Richard Biener writes:
> > When making the SLP tree a graph I messed up and in some cases we can
> > fail to propagate a higher max_nunits upwards when re-using a subtree.
> > For the testcase, but only on the gcc-9-branch, this causes us to
> > mis
On Wed, 28 Aug 2019 14:42:57 -0400
Nathan Sidwell wrote:
>/* If opened with #import or contains #pragma once. */
> - bool once_only;
> + bool once_only : 1;
I'm curious why you have them as bool and not unsigned?
thanks,
On Wed, Aug 28, 2019 at 5:12 PM Uros Bizjak wrote:
>
> Attached patch improves costing for STV shifts and corrects reject
> condition for out of range shift count operands.
>
> 2019-08-28 Uroš Bizjak
>
> * config/i386/i386-features.c
> (general_scalar_chain::compute_convert_gain):
>
On 8/28/19 10:19 PM, Jason Merrill wrote:
> On 8/28/19 12:29 PM, Martin Liška wrote:
>> The patch restores behavior before r265711 where we used
>> cxx_printable_name for __PRETTY_FUNCTION__.
>>
>> Patch can bootstrap on x86_64-linux-gnu and survives regression tests.
>>
>> Ready to be installed?
>
87 matches
Mail list logo
