On Tue, Jul 29, 2025 at 9:28 AM H.J. Lu wrote:
>
> On Mon, Jul 28, 2025 at 01:53:08PM -0700, H.J. Lu wrote:
> > On Mon, Jul 28, 2025 at 04:51:24PM +0800, Hongtao Liu wrote:
> > > On Wed, Jul 23, 2025 at 8:07 AM H.J. Lu wrote:
> > > >
> > > > For TLS calls:
> > > >
> > > > 1. UNSPEC_TLS_GD:
> > >
On Fri 2025-08-01 14:52:07, Gerald Pfeifer wrote:
> On Tue, 29 Jul 2025, Filip Kastl wrote:
> > okay to push? Confirmed that I didn't break anything by running `make
> > html`.
>
> ...and `make pdf`?
Ah, thanks. Also checked make pdf.
> > This patch changes two things. Firstly, we document
>
On Sun, 3 Aug 2025, 19:29 hexne, wrote:
> std::byteswap was added to the header file in c++23 and has been
> implemented, but it was not exported in std.cc.in. Therefore, when
> using import std, std::byteswap cannot be used. Exporting it in
> std.cc.in can solve this problem.
>
> Tested on x86_
вс, 3 авг. 2025 г. в 14:57, Georg-Johann Lay :
>
> This patch removes remains of reload from the avr BE.
>
> No new regressions. Ok for trunk?
Please apply.
Denis
From: Pan Li
Add asm dump check and run test for vec_duplicate + vmerge.vvm
combine to vmerge.vxm, with the GR2VR cost is 0, 2 and 15.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/autovec/vx_vf/vx_binary.h: Add test
helper macros.
* gcc.target/riscv/rvv/autovec/vx_vf/
From: Pan Li
This patch would like to combine the vec_duplicate + vaadd.vv to the
vaadd.vx. From example as below code. The related pattern will depend
on the cost of vec_duplicate from GR2VR. Then the late-combine will
take action if the cost of GR2VR is zero, and reject the combination
if th
From: Pan Li
This patch would like to introduce the combine of vec_dup + vmerge.vv
into vmerge.vx on the cost value of GR2VR. The late-combine will take
place if the cost of GR2VR is zero, or reject the combine if non-zero
like 1, 2, 15 in test.
From:
| ...
| vmv.v.x
| L1:
| vaadd.vv
On Mon, Jul 28, 2025 at 6:25 PM H.J. Lu wrote:
>
> For TLS calls:
>
> 1. UNSPEC_TLS_GD:
>
> (parallel [
> (set (reg:DI 0 ax)
> (call:DI (mem:QI (symbol_ref:DI ("__tls_get_addr")))
> (const_int 0 [0])))
> (unspec:DI [(symbol_ref:DI ("e") [flags 0x50])
>
For TLS calls:
1. UNSPEC_TLS_GD:
(parallel [
(set (reg:DI 0 ax)
(call:DI (mem:QI (symbol_ref:DI ("__tls_get_addr")))
(const_int 0 [0])))
(unspec:DI [(symbol_ref:DI ("e") [flags 0x50])
(reg/f:DI 7 sp)] UNSPEC_TLS_GD)
(clobber (reg:DI 5 di))]
Thanks Jeff.
> OK. Any change we could get run test for this in addition to the scanners?
I tried to add run test target this but seems happen to work here for mulhsu.
Here we only cares the result(high bits of mul) has values or not, looks like
mulhsu will not result in correctness problem.
T
On Fri, Aug 1, 2025 at 8:10 PM H.J. Lu wrote:
>
> Don't hoist non all 0s/1s vector set outside of the loop to avoid extra
> spills.
The patch LGTM.
>
> gcc/
>
> PR target/120941
> * config/i386/i386-features.cc (x86_cse_kind): Moved before
> ix86_place_single_vector_set.
>
On 7/30/25 11:22 PM, pan2...@intel.com wrote:
From: Pan Li
The previous code-gen of scalar unsigned SAT_MUL, aka usmul.
Leverage the mulhs by mistake, it should be mulhu for the
hight bit result of mul. Thus, this patch would like to make
it correct.
gcc/ChangeLog:
* config/riscv/
On Thu, Jul 3, 2025 at 4:50 AM Richard Sandiford
wrote:
>
> Karl Meakin writes:
> > This patch series adds support for the CMPBR extension. It includes the
> > new `+cmpbr` option and rules to generate the new instructions when
> > lowering conditional branches.
>
> Thanks for the update, LGTM.
Wrapped stdint.h for AArch64 MinGW32 is useful for bare-matal PE
target e.g. UEFI, as those platform does not provide a system
stdint.h, this would align with x86_64 mingw32 target which provides a
wrapped stdint.h
I have tested this by compiling a AArch64 UEFI Application using gcc's
stdint.h, w
An interesting case to consider is:
bool same11(const std::extents& e1,
const std::extents& e2)
{ return e1 == e2; }
Which has the following properties:
- There's no mismatching static extents, preventing any
short-circuiting.
- There's a comparison between dynamic and
Using __int_traits avoids the need to include from .
This in turn should reduce the size of the pre-compiled .
Similar refactoring was carried out for PR92546. Unfortunately,
./gcc/xgcc -std=c++23 -P -E -x c++ - -include mdspan | wc -l
shows a decrease by 1(!) line. This is due to bits/max_siz
One previous commit optimized fully dynamic extents; and another
refactored __size such that __fwd_prod is valid for __r = 0, ..., rank
(exclusive).
Therefore, by noticing that __rev_prod (and __fwd_prod) never accesses
the first (or last) extent, one can avoid pre-computing partial products
of st
In mdspan related code involving static extents, often the IndexType is
part of the template parameters, even though it's not needed.
This commit extracts the parts of _ExtentsStorage not related to
IndexType into a separate class _StaticExtents.
It also prefers passing the array of static extent
In mdspan related code, for extents with no static extents, i.e. only
dynamic extents, the following simplifications can be made:
- The array of dynamic extents has size rank.
- The two arrays dynamic-index and dynamic-index-inv become
trivial, e.g. k[i] == i.
- All elements of the arrays
The methods layout_{left,right}::mapping::stride are defined
as
\prod_{i = 0}^r E[i]
\prod_{i = r+1}^n E[i]
This is computed as the product of a precomputed static product and the
product of the required dynamic extents.
Disassembly shows that even for low-rank extents, i.e. rank == 1 and
ra
In both fully static and dynamic extents the comparison
static_extent(i) == dynamic_extent
is known at compile time. As a result, extents::extent doesn't
need to perform the check at runtime.
An illustrative example is:
using E = std::extents;
int required_span_size(const typename Layout::m
Let E denote an multi-dimensional extent; n the rank of E; r = 0, ...,
n; E[i] the i-th extent; and D[k] be the (possibly empty) array of
dynamic extents.
The two partial products for r = 0, ..., n:
\prod_{i = 0}^r E[i] (fwd)
\prod_{i = r+1}^n E[i] (rev)
can be computed as the product
The combined effect of this sequence of change is:
* a reduction in the number of template instantiations, by
- avoiding needless dependency of IndexType,
- special formulas for low-rank extents,
- special formulas for (nearly) fully dynamic extents.
* improved code quality, by
On Sun, Aug 03, 2025 at 12:20:24PM +0100, Paul Richard Thomas wrote:
First, the easy one:
> + /* Match the binding name; depending on type (operator / generic) format
> + it for future error messages into bind_name. */
This comment looks muddled.
> + if (op_type == INTERFACE_GENERIC
std::byteswap was added to the header file in c++23 and has been
implemented, but it was not exported in std.cc.in. Therefore, when
using import std, std::byteswap cannot be used. Exporting it in
std.cc.in can solve this problem.
Tested on x86_64-linux.
Signed-off-by: hexne
diff --git a/libstdc+
On Sun, 3 Aug 2025, Jason Merrill wrote:
> On 8/1/25 11:16 AM, Patrick Palka wrote:
> > Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK
> > for trunk and 15/14 after 15.2 is released?
> >
> > -- >8 --
> >
> > r13-3299 changed our internal declaration of __dynamic_cast to res
On 8/1/25 12:52 PM, Jakub Jelinek wrote:
Hi!
The CWG1709 just codifies existing GCC (and clang) behavior, so this
just adds a testcase for that.
Tested on x86_64-linux, ok for trunk?
OK.
2025-08-01 Jakub Jelinek
PR preprocessor/120778
* g++.dg/DRs/dr1709.C: New test.
--
On 7/31/25 1:58 PM, Jakub Jelinek wrote:
On Thu, Jul 31, 2025 at 11:04:56AM -0400, Jason Merrill wrote:
If the current backup handling is doing the wrong thing in this case, it
seems better to fix it rather than add more earlier in the function. Would
it be enough to move the existing CPP_PRAGMA
On 8/1/25 11:16 AM, Patrick Palka wrote:
Bootstrapped and regtested on x86_64-pc-linux-gnu, does this look OK
for trunk and 15/14 after 15.2 is released?
-- >8 --
r13-3299 changed our internal declaration of __dynamic_cast to reside
inside the abi / __cxxabiv1 namespace instead of the global na
Hi Paul,
On 7/31/2025 6:02 AM, Paul Richard Thomas wrote:
That's exactly how I had a mind to do it. You beat me to it :-(
Just get on, polish the patch and add more tests.
I've updated the patch with improvements to both the functionality and
test cases. It should now work well for simple sc
On 7/31/2025 1:45 AM, Joseph Myers wrote:
On Sun, 6 Jul 2025, Yuao Ma wrote:
+#ifndef HAVE_COSPI
+#define HAVE_COSPI 1
+double cospi (double);
+
+double
+cospi (double x)
+{
+ return cos (x * pihi_d + x * pilo_d);
For reasonable results for large x you should first reduce mod 2 to the
ran
Users might be using a space in their build directory path. To allow
specifying such a root for the module mapper started by GCC, we need the
command to allow quotes. Previously quoting a path passed to the module
mapper was not possible, so replace the custom argv parsing with
the argv parsing log
Hi Mikael,
All three patches are OK for mainline. About half of the chunks I recognise
and blames should point to me :-(
Thanks for the patch.
Paul
On Sun, 3 Aug 2025 at 14:30, Mikael Morin wrote:
> From: Mikael Morin
>
> The fortran frontend has getter and setter functions to generate code
From: Mikael Morin
Regression-tested on x86_64-pc-linux-gnu.
OK for master?
-- >8 --
In some places, a write to an array descriptor offset field was
generated simply by adding a modification of a reference to it that
was already available. This change uses the existing setter
f
From: Mikael Morin
The fortran frontend has getter and setter functions to generate code
either to read array descriptor fields or to write to them.
In some places, those functions are not used, and the access to the
field is generated manually.
The following patches replace those direct accesse
From: Mikael Morin
Regression-tested on x86_64-pc-linux-gnu.
OK for master?
-- >8 --
The function gfc_conv_descriptor_data_addr generates an address to the
data field of an array descriptor. It is only used once, and in the
single place where it is used, the address is immediately dereferenced
From: Mikael Morin
Regression-tested on x86_64-pc-linux-gnu.
OK for master?
-- >8 --
In some places, a write to an array descriptor data field was generated
simply by adding a modification of a reference to it that was already
available. This change uses the existing setter function instead in
Hi Anna,
thanks for your patch! I did not have to time look at this
(and I can't approve it), but here is something I noticed.
You have an error for jumping into a defer block
>
> @@ -4244,10 +4274,29 @@ check_earlier_gotos (tree label, struct c_label_vars*
> label_vars)
>
>if (
On Sun, 3 Aug 2025, 14:54 Jonathan Wakely, wrote:
>
>
> On Sun, 3 Aug 2025, 01:55 Andrew Pinski, wrote:
>
>> On Thu, Dec 7, 2023 at 2:07 PM Jonathan Wakely
>> wrote:
>> >
>> > Tested x86_64-linux. Pushed to trunk.
>> >
>> > -- >8 --
>> >
>> > The changes in r14-6198-g5e8a30d8b8f4d7 were broken,
On Sun, 3 Aug 2025, 01:55 Andrew Pinski, wrote:
> On Thu, Dec 7, 2023 at 2:07 PM Jonathan Wakely wrote:
> >
> > Tested x86_64-linux. Pushed to trunk.
> >
> > -- >8 --
> >
> > The changes in r14-6198-g5e8a30d8b8f4d7 were broken, as I used
> > _GLIBCXX17_CONSTEXPR for the 'if _GLIBCXX17_CONSTEXPR
On Sat, Aug 2, 2025 at 5:12 PM Gerald Pfeifer wrote:
>
> H.J., I assume you also wanted to activate the respective section?
>
Yes.
Thanks.
> Pushed.
>
> Gerald
> ---
> htdocs/gcc-16/changes.html | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/htdocs/gcc-16/changes.h
On Sat, 2 Aug 2025, 21:06 Jakub Jelinek, wrote:
> Hi!
>
> I've tried compiling
> #include
> with -std=c++26 -fdump-lang-all
> and
> for i in `grep ^Class.std::[^_] *.C.001l.class | sed 's/^Class //;s/[<
> ].*$//' | sort -u | grep -v ::.*::`; do grep -q $i
> /usr/src/gcc/libstdc++-v3/src/c++23/st
On Sun, Aug 3, 2025 at 12:45 AM Uros Bizjak wrote:
>
> On Sat, Aug 2, 2025 at 8:56 PM H.J. Lu wrote:
> >
> > On Fri, Aug 1, 2025 at 10:32 PM Uros Bizjak wrote:
> > >
> > > On Sat, Aug 2, 2025 at 3:22 AM H.J. Lu wrote:
> > > >
> > > > After
> > > >
> > > > commit 965564eafb721f813a3112f1bba8
Hi Harald,
Please find attached an updated patch for the generic statement.
The module attribute in real module function realg is required by all other
brands. gfortran compiles happily with it and so I have added it to the
testcase. This is now PR121379. I have also posted a submodule meta-bug
P
This patch removes remains of reload from the avr BE.
No new regressions. Ok for trunk?
Johann
--
AVR: target/121359: Remove -mlra and remains of reload.
PR target/121359
gcc/
* config/avr/avr.h: Remove -mlra and remains of reload.
* config/avr/avr.cc: Same.
*
The following removes the fixup we apply to pattern stmt operands
before code generating vector epilogues. This isn't necessary anymore
since the SLP graph now exclusively records the data flow. Similarly
fixing up of SSA references inside DR_REF of gather/scatter isn't
necessary since we now rec
The restriction no longer applies, so remove it.
Re-bootstrap and regtest running on x86_64-unknown-linux-gnu.
* tree-vect-data-refs.cc (vect_check_gather_scatter):
Remove restriction on epilogue of epilogue vectorization.
---
gcc/tree-vect-data-refs.cc | 7 ---
1 file change
The following is a prototype-quality patch to make us record the
get_load_store_info results from load/store analysis and re-use
them during transform. In particular this moves where
SLP_TREE_MEMORY_ACCESS_TYPE is stored.
Rather than mass-replacing references to variables I've kept the
locals but
On Sat, Aug 2, 2025 at 8:56 PM H.J. Lu wrote:
>
> On Fri, Aug 1, 2025 at 10:32 PM Uros Bizjak wrote:
> >
> > On Sat, Aug 2, 2025 at 3:22 AM H.J. Lu wrote:
> > >
> > > After
> > >
> > > commit 965564eafb721f813a3112f1bba8d8fae32b
> > > Author: Richard Sandiford
> > > Date: Tue Jul 29 15:58
On Sun, Aug 3, 2025 at 7:31 AM Andi Kleen wrote:
>
> "H.J. Lu" writes:
>
> > Don't hoist non all 0s/1s vector set outside of the loop to avoid extra
> > spills.
>
> It seems this could be a loss if there are actually enough registers.
> So you need to make it depend on the register pressure?
IMO
пт, 1 авг. 2025 г. в 18:07, Georg-Johann Lay :
>
> There are many post-reload define_insn_and_split's that just append
> a (clobber (reg:CC REG_CC)) to the pattern. Instead of repeating
> the original patterns, avr_add_ccclobber (curr_insn) is used to do
> that job.
>
> This avoids repeating pat
51 matches
Mail list logo
