On Thu, Aug 18, 2022 at 4:46 PM Eric Botcazou wrote:
>
> > Hmm :/ But that means we _should_ force a sign extension but only
> > from ptrofftype_p ()? That is, your test above should maybe read
> >
> >signop sgn = TYPE_SIGN (type);
> >if (ptrofftype_p (type))
> > sgn = SIGNED;
> >
>
On Thu, Aug 18, 2022 at 6:13 PM Aldy Hernandez wrote:
>
> The path_range_query constructor that takes an edge is really a
> convenience function for the loop-ch pass. It feels wrong to pollute
> the API with such a specialized function that could be done with
> a small inline function closer to i
On Thu, Aug 18, 2022 at 1:10 PM Aldy Hernandez wrote:
>
> On Thu, Aug 18, 2022, 11:34 Richard Biener wrote:
> >
> > On Thu, Aug 18, 2022 at 9:58 AM Aldy Hernandez wrote:
> > >
> > > FWIW, here is a rebased version with the latest trunk.
> > >
> > > There are no regressions, or differences in thr
On Thu, Aug 18, 2022 at 2:43 PM Martin Liška wrote:
>
> That handles systems that don't have O_NONBLOCK, in that case
> WPA streaming is not using jobserver if --jobserver-auth uses 'fifo'.
>
> Patch can bootstrap on x86_64-linux-gnu and survives regression tests.
> Tested with mingw cross compile
juzhe.zh...@rivai.ai writes:
> From: zhongjuzhe
>
> Hello. This patch is preparing for following RVV support.
>
> Both ARM SVE and RVV (RISC-V 'V' Extension) support length-agnostic vector.
> The minimum vector length of ARM SVE is 128-bit and the runtime invariant of
> ARM SVE is always 128-bit
Thank you so much. Address your comment. I think "maybe_gt (nunits, 1)" is a
more solid solution than I do.
I will send a patch to fix this.
juzhe.zh...@rivai.ai
From: Richard Sandiford
Date: 2022-08-19 16:03
To: juzhe.zhong
CC: gcc-patches; rguenther; kito.cheng
Subject: Re: [PATCH] middle-e
Status
==
The gcc-12 branch is again open for regression and documentation fixes.
Quality Data
Priority # Change from last report
--- ---
P1 0
P2 434 + 2
P3 64+ 2
P4
Hi, Richard. I tried the codes:
if (!nunits.is_constant () && maybe_gt (nunits, 1))
test_vector_subregs_modes (x, nunits - min_nunits, count);
It still failed. For nunits = (1,1) , maybe_gt (nunits, 1) return true value.
But I tried:
if (!nunits.is_constant () && known_gt (nunits, 1))
test_vecto
"juzhe.zh...@rivai.ai" writes:
> Hi, Richard. I tried the codes:
> if (!nunits.is_constant () && maybe_gt (nunits, 1))
> test_vector_subregs_modes (x, nunits - min_nunits, count);
>
> It still failed. For nunits = (1,1) , maybe_gt (nunits, 1) return true value.
Ah, right, sorry for the bogus sug
> Ah, right, sorry for the bogus suggestion.
> In that case, what specifically goes wrong? Which test in
> test_vector_subregs_modes fails, and what does the ICE look like?
Because may_be (nunits, 1) return true. For nunits = (1,1), it will execute
test_vector_subregs_modes.
The fail ICE:
../../
Hello,
This patch teaches the DWARF support in gcc about RA_AUTH_CODE pseudo
hard-register and also
.save {ra_auth_code} and .cfi_offset ra_auth_code dwarf directives for
the PAC feature
in Armv8.1-M architecture.
RA_AUTH_CODE register number is 107 and it's dwarf register number is 143.
Whe
On Fri, 2022-08-19 at 09:56 +0800, Lulu Cheng wrote:
> v1 -> v2:
> - Modify some description information.
> - Add options -W[no]extreme-plt, warn about code model extreme not support
> plt mode,
> and then disable plt.
I think we can make -mcmodel=large imply -fno-plt, and reject if user
uses "-f
On 19/08/2022 11:04, Srinath Parvathaneni via Gcc-patches wrote:
Hello,
This patch teaches the DWARF support in gcc about RA_AUTH_CODE pseudo
hard-register and also
.save {ra_auth_code} and .cfi_offset ra_auth_code dwarf directives for
the PAC feature
in Armv8.1-M architecture.
RA_AUTH_CO
"juzhe.zh...@rivai.ai" writes:
>> Ah, right, sorry for the bogus suggestion.
>> In that case, what specifically goes wrong? Which test in
>> test_vector_subregs_modes fails, and what does the ICE look like?
>
> Because may_be (nunits, 1) return true. For nunits = (1,1), it will execute
> test_ve
The following avoids adding PHIs to the worklist for uninit processing
if we reach them following backedges. That confuses predicate analysis
because it assumes the use is happening in the same iteration as the the
definition. For the testcase in the PR the situation is like
void foo (int val)
{
Rather obvious, once found. I forgot to add some cleanup, cluttering
/tmp with ".mkoffload.omp_requires".
The same issue exists for GCN also for ".mkoffload.dbg"
OK for mainline? – For the dbg issue, OK also for GCC 11 and 12 backport?
Follow-up to the 'OpenMP: Move omp requires checks to libgo
On Fri, Aug 19, 2022 at 04:01:10PM +0200, Tobias Burnus wrote:
> Rather obvious, once found. I forgot to add some cleanup, cluttering
> /tmp with ".mkoffload.omp_requires".
>
> The same issue exists for GCN also for ".mkoffload.dbg"
>
> OK for mainline? – For the dbg issue, OK also for GCC 11 and
As you mentioned. For poly_uint16 (1, 1), the pattern should be:
shape: nelts_per_pattern == 2, npatterns == 1
elements: a, b[, b, b, b, b, ...]
I tried to print out the rtx that tests create to test:
static void
test_vector_subregs_fore_back (machine_mode inner_mode)
{
poly_uint64 nunits = GE
I rewrite test_vector_subregs_fore_back as follows:
static void
test_vector_subregs_fore_back (machine_mode inner_mode)
{
poly_uint64 nunits = GET_MODE_NUNITS (inner_mode);
unsigned int min_nunits = constant_lower_bound (nunits);
scalar_mode int_mode = GET_MODE_INNER (inner_mode);
unsigned
On 8/17/2022 1:31 AM, Aldy Hernandez via Gcc-patches wrote:
I just have a few high level comments.
On Tue, Aug 16, 2022 at 4:05 PM Richard Biener wrote:
The following refactors profitable_path_p in the backward threader,
splitting out parts that can be computed once the exit block is known,
On Wed, 17 Aug 2022, Iain Buclaw wrote:
> This patch belatedly adds the new features and changes to the D
> front-end during the GCC 12 development cycle, as well as a bullet in
> the caveat section for D's new bootstrapping requirements.
Nice!
> +D:
> +Building and bootstrapping GDC, t
I saw that files such as /tmp/ccxFKYeS.target.o kept accumulating. Not a high
number,
but still several.
I turned out that when compiling an offloading program successfully, they were
removed.
But when it failed, those remained. (Example: See PR.)
This patch fixes this by storing the file name
On Thu, Aug 18, 2022 at 5:56 PM Hongtao Liu via Gcc-patches
wrote:
>
> On Thu, Aug 18, 2022 at 3:36 PM Haochen Jiang via Gcc-patches
> wrote:
> >
> > Hi all,
> >
> > This patch aims to add bf16 abi test after the whole __bf16 type is added.
> >
> > Regtested on x86_64-pc-linux-gnu. Ok for trunk?
Hi all,
Would love some feedback on this patch!
Thanks,
Paul
On Mon, Aug 01, 2022 at 05:18:40AM +, Paul Hollinsky wrote:
> Rather than traversing the all_files linked list for every include,
> this factors out the quick idempotency checks (modification time
> and size) to be the keys in a ha
On Thu, Aug 18, 2022 at 08:33:47PM -0400, Jason Merrill wrote:
> On 8/18/22 13:19, Marek Polacek wrote:
> > On Mon, Aug 15, 2022 at 03:54:05PM -0400, Jason Merrill wrote:
> > > On 8/9/22 09:37, Marek Polacek wrote:
> > > > + /* We're looking for *std::move ((T &) &arg), or
> > > > + *std::move
After deep analysis and several tries. I have made a analysis and conclusion as
follows.
I really appreciate if you could spend some time take a look at the following
analysis that I made:
According to the codes in test_vector_subregs_fore_back:
~~~
poly_uint64 nunits = GET_MODE_NUNITS (inner
v1 -> v2:
- Modify some description information.
- Add options -W[no]extreme-plt, warn about code model extreme not support plt
mode,
and then disable plt.
v2 -> v3:
- When -mcmodel=extreme, default set to -fno-plt mode, if the user forces to use
'-mcmodel=extreme -fplt', an error will be reporte
27 matches
Mail list logo
