Hi,
this patch implements symver attribute which can be use din place of
.symver directive and is LTO friendly. I would welcome feedback for
this, in particular some evidence that codebases can be easily converted
to use it.
There is some discussion in the PR and its duplicate filled by Carlos.
On Thu, Nov 14, 2019 at 10:35 PM Giuliano Belinassi
wrote:
>
> Previously, the suggested patch removed all tree-loop-distributions.c global
> variables moving them into a struct and passing it aroung across the
> functions.
> This patch address this problem by using C++ classes instead, avoiding
Musl does not support initial-exec tls in dynamically loaded shared
libraries.
libgomp/ChangeLog:
2019-11-15 Szabolcs Nagy
* configure.tgt: Avoid IE tls on *-*-musl*.
libitm/ChangeLog:
2019-11-15 Szabolcs Nagy
* configure.tgt: Avoid IE tls on *-*-musl*.
diff --git
microblaze-linux-musl build fails without this.
(This is a rebase of an earlier patch posted on bugzilla.)
gcc/ChangeLog:
2019-11-15 Nick Clifton
Szabolcs Nagy
* config/microblaze/microblaze.c (print_operand): Print value as long.
diff --git
> PASS: gcc.dg/ipa/inline-9.c (test for excess errors)
> gcc.dg/ipa/inline-9.c: dump file does not exist
> UNRESOLVED: gcc.dg/ipa/inline-9.c scan-tree-dump inline "Inlined 1 calls"
>
> but fixing the obvious bug in there, s/scan-tree-dump/scan-ipa-dump/
> doesn't help, nothing is really inlined.
> * Jan Hubicka:
>
> > Internally the patch tries to mimic what happens in ELF. In particular
> >
> > __attribute__ ((__symver__ ("foo@VERS_1"))) int
> > foo_v1 (void)
> > {
> > }
> >
> > creates a special form of alias with symver flag. Its assembler name is
> > foo@VERS_1 since it is what
The new tree-cfg.c checking in r278245 tripped on folds of
ALTIVEC_BUILTIN_VPERM_*, which were using gimple_convert
rather than VIEW_CONVERT_EXPR to reinterpret the contents
of a vector as a different type.
Spot-tested on powerpc64-linux-gnu. OK to install?
Richard
(There are actually two bugs
Hi!
On Thu, Nov 14, 2019 at 09:12:32PM -0600, Xiong Hu Luo wrote:
> P9LE generated instruction is not worse than P8LE.
> mtvsrdd;xxlnot;stxv vs. not;not;std;std.
> Update the test case to fix failures.
So this no longer runs it for p7, and it also doesn't run it for cpus
after p9 anymore. Could
> On Fri, Nov 15, 2019 at 10:16 AM Richard Biener
> wrote:
> >
> > On Fri, Nov 15, 2019 at 9:10 AM Jan Hubicka wrote:
> > >
> > > > next is initialized only in the loop before, it is never updated
> > > > in it's own loop.
> > > >
> > > > gcc/ChangeLog
> > > >
> > > > 2019-11-15 Xiong Hu
On Пт, ноя 15, 2019 at 00:17, Konstantin Kharlamov
wrote:
On 14.11.2019 23:50, Segher Boessenkool wrote:
Hi!
On Thu, Nov 14, 2019 at 07:01:47PM +0300, Konstantin Kharlamov wrote:
Generally, people expect functions to accept arguments directly. But
ones defined in gdbinit did not use the
> On Fri, Nov 15, 2019 at 9:10 AM Jan Hubicka wrote:
> >
> > > next is initialized only in the loop before, it is never updated
> > > in it's own loop.
> > >
> > > gcc/ChangeLog
> > >
> > > 2019-11-15 Xiong Hu Luo
> > >
> > > * ipa-inline.c (inline_small_functions): Update iterator
On Thu, Nov 14, 2019 at 4:41 PM Sudakshina Das wrote:
>
> Hi
>
> This patch is trying to fix PR88398 for AArch64. As discussed in the PR,
> loop unrolling is probably what we can do here. As an easy fix, the
> existing unroll_stupid is unrolling the given example better than the
>
On Fri, Nov 15, 2019 at 9:10 AM Jan Hubicka wrote:
>
> > next is initialized only in the loop before, it is never updated
> > in it's own loop.
> >
> > gcc/ChangeLog
> >
> > 2019-11-15 Xiong Hu Luo
> >
> > * ipa-inline.c (inline_small_functions): Update iterator of next.
>
> OK,
>
* Jan Hubicka:
> Internally the patch tries to mimic what happens in ELF. In particular
>
> __attribute__ ((__symver__ ("foo@VERS_1"))) int
> foo_v1 (void)
> {
> }
>
> creates a special form of alias with symver flag. Its assembler name is
> foo@VERS_1 since it is what happens in ELF too
The gthr weak reference based single thread detection is unsafe with
static linking and in case of dynamic linking it's ineffective on musl
since pthread symbols are defined in libc.so.
Ideally this should be fixed for all targets, since glibc plans to move
libpthread.so into libc.so too and
Hi!
The following patch is in preparation of the deferred declare variant
resolution, where we need to compute the scores during gimplification
for all variants.
This patch just picks the candidate with highest score if that variant
doesn't need to be deferred and thus is an unconditional win,
The following reverts back to the old unroll heuristics before my
PR91975 changes but implements "consistency" the other way around.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2019-11-15 Richard Biener
PR tree-optimization/92039
PR
I'm going to install it as obvious.
gcc/ChangeLog:
2019-11-15 Luo Xiong Hu
* ipa-comdats.c: Fix comments typo.
* ipa-profile.c: Fix comments typo.
* tree-profile.c (gimple_gen_ic_profiler): Use the new variable
__gcov_indirect_call.counters and
next is initialized only in the loop before, it is never updated
in it's own loop.
gcc/ChangeLog
2019-11-15 Xiong Hu Luo
* ipa-inline.c (inline_small_functions): Update iterator of next.
---
gcc/ipa-inline.c | 15 +--
1 file changed, 9 insertions(+), 6
Hi,
building Firefox I now get:
#0 internal_error (gmsgid=0x28f8597 "in %s, at %s:%d") at
../../gcc/diagnostic.c:1578
#1 0x01e20ae2 in fancy_abort (file=0x2892574 "../../gcc/ipa-cp.c",
line=386, function=0x2
893510
"ipa_get_parm_lattices") at ../../gcc/diagnostic.c:1649
#2
> next is initialized only in the loop before, it is never updated
> in it's own loop.
>
> gcc/ChangeLog
>
> 2019-11-15 Xiong Hu Luo
>
> * ipa-inline.c (inline_small_functions): Update iterator of next.
OK,
thanks!
Honza
> ---
> gcc/ipa-inline.c | 15 +--
> 1 file
On Fri, Nov 15, 2019 at 10:16 AM Richard Biener
wrote:
>
> On Fri, Nov 15, 2019 at 9:10 AM Jan Hubicka wrote:
> >
> > > next is initialized only in the loop before, it is never updated
> > > in it's own loop.
> > >
> > > gcc/ChangeLog
> > >
> > > 2019-11-15 Xiong Hu Luo
> > >
> > >
Avoid ICE when the alias is ill-formed.
gcc/cp/
* pt.c (get_underlying_template): Exit loop if the original type
of the alias is null.
gcc/testsuite/
* g++.dg/cpp2a/pr89913.C: New test.
Andrew Sutton
0001-Don-t-segfault-on-error-PR-c-89913.patch
Description: Binary data
Hi Richard,
> So what do we actually do unpatched with -funroll-loops here?
Yes so it does the insane "fully unrolled trailing loop before the unrolled
loop" thing. One always does the trailing loop last (and typically as an
actual loop of course) and then the code ends up much faster, close to
Pinging with more correct maintainers this time :)
Also would need to backport to gcc7,8,9, but need to get this approved
first!
Thank you,
Stam
Forwarded Message
Subject: Re: [PATCH][GCC][ARM] Arm generates out of range conditional
branches in Thumb2 (PR91816)
Date: Mon,
On powerpc and s390x the musl ABI requires 64 bit and 128 bit long
double respectively, so adjust the default.
gcc/ChangeLog:
2019-11-15 Szabolcs Nagy
* configure.ac: Set gcc_cv_target_ldbl128 for musl targets.
* configure: Regenerate.
diff --git a/gcc/configure
Hi all,
Just looking for some feedback on the approach.
Currently the loop vectorizer can't vectorize the following typical loop
for getting max value and index from an array:
void test_vec(int *data, int n) {
int best_i, best = 0;
for (int i = 0; i < n; i++) {
On 15/11/2019 17:48, Segher Boessenkool wrote:
> On Fri, Nov 15, 2019 at 05:26:41PM +, Szabolcs Nagy wrote:
>> On powerpc and s390x the musl ABI requires 64 bit and 128 bit long
>> double respectively, so adjust the default.
>>
>> gcc/ChangeLog:
>>
>> 2019-11-15 Szabolcs Nagy
>>
>> *
On Fri, 15 Nov 2019, Jan Hubicka wrote:
> I was originaly supporting multiple symvers like:
> __attribute__ ((__symver__ ("foo@VERS_1"))) int
> __attribute__ ((__symver__ ("foo@VERS_2"))) int
> foo_v1 (void)
> {
> }
>
> but then noticed it is rejected by gas.
That's
* Jan Hubicka:
>> It's sometimes useful to define multiple versions for a single symbol.
>> For maximum binutils compatibility, you would have to use intermediate
>> aliases. __attribute__ ((__symver__ ("foo@VERS_1", "foo@@VERS_2")))
>> could turn into this:
>>
>> .set foo_v1.symver.1, foo_v1
Richard Biener writes:
> On Fri, Nov 15, 2019 at 11:02 AM Richard Sandiford
> wrote:
>>
>> The new tree-cfg.c checking in r278245 tripped on folds of
>> ALTIVEC_BUILTIN_VPERM_*, which were using gimple_convert
>> rather than VIEW_CONVERT_EXPR to reinterpret the contents
>> of a vector as a
On Mon, Nov 11, 2019 at 7:51 PM Richard Sandiford
wrote:
>
> This patch rewrites the index-based alias checks to use conditions
> of the form:
>
> (unsigned T) (a - b + bias) <= limit
>
> E.g. before the patch:
>
> struct s { int x[100]; };
>
> void
> f1 (struct s *s1, int a, int b)
> {
Set global epilogue_completed when skipping pro_and_epilogue pass
When compiling RTL functions marked to start at a pass after the reload
pass, `skip_pass` is used to mark the reload pass as having completed
since many patterns use the `reload_completed` variable to determine
whether to run or
Hi Richard
I apologise I should have given more explanation on my cover letter.
Although the bug was filed for vectorization, the conversation on it talked
about loops with two exits not being supported in the vectorizer and being not
being possible without lto and peeling causing more harm
On 14/11/19 23:39 +0100, Jakub Jelinek wrote:
On Thu, Nov 14, 2019 at 10:15:21PM +, Jonathan Wakely wrote:
> namespace std {
> struct source_location {
>struct __impl {
Will this work if the library type is actually in an inline namespace,
such as std::__8::source_location::__impl (as
On 15/11/2019 12:21, Jakub Jelinek wrote:
On Thu, Nov 14, 2019 at 04:36:38PM +, Andrew Stubbs wrote:
This patch adds some necessary bits to enable OpenACC testings for amdgcn
offloading.
The two "check_effective" procedures are not actually needed yet, but later
patches to test cases will
On Fri, 15 Nov 2019, Richard Biener wrote:
>
> The following fixes PR92512, we were failing to disqualify
> _2 < 1 ? _2 : 2 with _2 being part of the reduction chain.
>
> Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
And the following is what I committed.
Richard.
2019-11-15
> I checked update_jump_functions_after_inlining(), and found one suspicious
> place:
>
> for (i = 0; i < count; i++)
> {
> struct ipa_jump_func *dst = ipa_get_ith_jump_func (args, i);
> if (!top)
> {
> ipa_set_jf_unknown (dst);
> <
On 14/11/2019 08:40, Christophe Lyon wrote:
Hi,
Is it OK to backport r25 to gcc8 (Ensure dotproduct is only
enabled on armv8 neon) ?
I've noticed unnecessary failure of gcc.target/arm/simd/vdot-compile.c
after I upgraded to recent binutils.
Thanks,
Christophe
Is there a PR for this?
Rather than reimplement brace elision here, we call reshape_init and then
discard the result. We needed to set CLASSTYPE_NON_AGGREGATE a bit more in
this patch, since outside a template it's set in check_bases_and_members.
Tested x86_64-pc-linux-gnu, applying to trunk.
* pt.c
On 15/11/2019 15:05, Francesco Petrogalli wrote:
> Thank you Szabolcs for working on this.
>
>> On Nov 14, 2019, at 2:23 PM, Szabolcs Nagy wrote:
>>
>> Sorry v2 had a bug.
>>
>> v2: added documentation and tests.
>> v3: fixed expand_simd_clones so different isa variants are actually
>>
On 11/15/19 2:07 AM, Szabolcs Nagy wrote:
microblaze-linux-musl build fails without this.
(This is a rebase of an earlier patch posted on bugzilla.)
gcc/ChangeLog:
2019-11-15 Nick Clifton
Szabolcs Nagy
* config/microblaze/microblaze.c (print_operand): Print value as
On Fri, Nov 15, 2019 at 05:26:41PM +, Szabolcs Nagy wrote:
> On powerpc and s390x the musl ABI requires 64 bit and 128 bit long
> double respectively, so adjust the default.
>
> gcc/ChangeLog:
>
> 2019-11-15 Szabolcs Nagy
>
> * configure.ac: Set gcc_cv_target_ldbl128 for musl
On Thu, Nov 14 2019, Jan Hubicka wrote:
>> On Mon, Nov 04 2019, Jan Hubicka wrote:
>> > Index: ipa-utils.h
>> > ===
>> > --- ipa-utils.h(revision 277780)
>> > +++ ipa-utils.h(working copy)
>> > @@ -47,6 +47,9 @@ void
On Nov 15 2019, Bernd Schmidt wrote:
> Are these with the patch?
Yes.
> Are you on real hardware
No, I'm using aranym.
Andreas.
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1
"And now for something completely different."
> On Nov 15, 2019, at 10:01 AM, Szabolcs Nagy wrote:
>
> On 15/11/2019 15:05, Francesco Petrogalli wrote:
>> Thank you Szabolcs for working on this.
>>
>>> On Nov 14, 2019, at 2:23 PM, Szabolcs Nagy wrote:
>>>
>>> Sorry v2 had a bug.
>>>
>>> v2: added documentation and tests.
>>> v3: fixed
Jason,
using ENUM::v is proving rather tricky, when the memberness of the
enumerator and the context of the using decl differ.
struct B
{
enum E {e};
};
struct D
{
private:
using B::e; // ok
static inline const auto ok1 = e; // #1 accessible
};
struct F : D
{
static inline const auto
Add the dynamic linker name and fix a type name to use the public name
instead of the glibc internal name.
Build tested on m68k-linux-musl and m68k-linux-gnu.
gcc/ChangeLog:
2019-11-15 Szabolcs Nagy
* config/m68k/linux.h (MUSL_DYNAMIC_LINKER): Define.
libgcc/ChangeLog:
2019-11-15
Add the musl dynamic linker names.
Build tested on s390x-linux-musl and s390x-linux-gnu.
gcc/ChangeLog:
2019-11-15 Szabolcs Nagy
* config/s390/linux.h (MUSL_DYNAMIC_LINKER32): Define.
(MUSL_DYNAMIC_LINKER64): Define.
diff --git a/gcc/config/s390/linux.h
Hi Tobias,
I think you need to add at least VOLATILE to this list
Yes, I'll do that.
Any other comments?
Regards
Thomas
Ping...
FYI this jumping into fake inline frames happens rather often
when debugging, and is especially annoying once you know how
easy it is to fix it...
Bernd.
On 11/9/19 11:20 AM, Bernd Edlinger wrote:
> Ping...
>
> On 11/2/19 7:49 AM, Bernd Edlinger wrote:
>> Ping...
>>
>> On 10/27/19
On 15/11/2019 11:32 am, Andrew Stubbs wrote:
On 14/11/2019 15:33, Kwok Cheung Yeung wrote:
The kernel attributes are changed to request at least 64 SGPRs and 24
VGPRs (i.e. the non-kernel maximum, otherwise the callees may not have
enough registers to run in) for non-leaf kernels to take
Hi,
thanks for feedback. Here is updated patch that incorporates Martin's
changes, formatting corrections and makes symvers of aliases work.
Honza
* c-family/c-attribs.c (handle_symver_attribute): New.
(c_common_attribytes): Add symver.
* cgraph.h (symtab_node): Add
Forgot to CC maintainer.
On 15/11/2019 18:03, Joel wrote:
> Hi all,
>
> Just looking for some feedback on the approach.
>
> Currently the loop vectorizer can't vectorize the following typical
> loop for getting max value and index from an array:
>
> void test_vec(int *data, int n) {
> int
On 07/11/2019 16:16, Olivier Hainque wrote:
Hello,
About a year ago we discussed various possibilities regarding
an issue with the aarch64 selection of r18 as the static chain,
problematic in environments where the OS uses r18 for "private"
reasons. VxWorks uses this permission to use r18 as a
On 11/15/19 2:48 PM, Andreas Schwab wrote:
> Here are the results of running the testsuite on m68k-linux:
>
> http://gcc.gnu.org/ml/gcc-testresults/2019-11/msg00908.html
>
> This is a list of regressions:
Are these with the patch? I'm not seeing any of these in my testing with
qemu. Are you on
From: "Dragan Mladjenovic"
This patch tightens the instruction definitions to make sure
that MSA branch instructions cannot be put into delay slots and have their
delay slots eligible for being filled. Also, MSA *div*3 patterns use MSA
branches for zero checks but are not marked as being
Honza,
I made some changes: do not penalize self-recursive function, and add
--param ipa-cp-min-recursive-probability, similar to recursive inline. Please
review this new one.
Thanks,
Feng
From: Jan Hubicka
Sent: Friday, November 15, 2019 4:33 AM
On 15/11/2019 15:51, Kwok Cheung Yeung wrote:
On 15/11/2019 11:32 am, Andrew Stubbs wrote:
On 14/11/2019 15:33, Kwok Cheung Yeung wrote:
The kernel attributes are changed to request at least 64 SGPRs and 24
VGPRs (i.e. the non-kernel maximum, otherwise the callees may not
have enough
On 11/14/19 9:29 PM, Szabolcs Nagy wrote:
The builtin directive allows specifying the simd attribute for a builtin
function. Similarly how the C language simd attribtue got extended to allow
declaring a specific vector variant, update the fortran builtin directive too.
Sounds useful. For the
On Fri, Nov 15, 2019 at 06:03:25PM +, Szabolcs Nagy wrote:
> On 15/11/2019 17:48, Segher Boessenkool wrote:
> > I think it would be much more readable and maintainable if you did
> > something like
> >
> > case "$target" in
> > powerpc*-*-linux-musl)
> > gcc_cv_target_ldbl128=no
> >
On Mon, Nov 11, 2019 at 7:47 PM Richard Sandiford
wrote:
>
> prune_runtime_alias_test_list swapped dr_as between two dr_with_seg_len
> pairs before finally deciding whether to merge them. Bailing out later
> would therefore leave the pairs in an incorrect state.
>
> IMO a better fix would be to
On Mon, Nov 11, 2019 at 7:46 PM Richard Sandiford
wrote:
>
> The two users of tree-data-ref's runtime alias checks both canonicalise
> the order of the dr_with_seg_lens in a pair before passing them to
> prune_runtime_alias_test_list. It's more convenient for later patches
> if
On Thu, Nov 14, 2019 at 04:36:38PM +, Andrew Stubbs wrote:
> This patch adds some necessary bits to enable OpenACC testings for amdgcn
> offloading.
>
> The two "check_effective" procedures are not actually needed yet, but later
> patches to test cases will use them.
>
> OK to commit?
I'm
On 15/11/19 13:28 +0100, Jakub Jelinek wrote:
On Thu, Nov 14, 2019 at 08:34:26PM +0100, Jakub Jelinek wrote:
The following WIP patch implements __builtin_source_location (),
which returns const void pointer to a std::source_location::__impl
struct that is required to contain __file, __function,
> Hi,
>
> On Fri, Nov 15 2019, luoxhu wrote:
> >
> > Sorry to break the bootstrap. This was my lack of consideration when
> > splitting this small piece of code from the previous patch, the line
> > should be in the first line of the second loop. Could you please add
> > some comments that
> The bug in the walker loop here IMO predated C++ conversion, but to
> avoid member function freeing THIS we probably want to have:
>
> cgraph_edge::remove (e),
> symtab_node::remove (e),
> ipa_ref::remove (ref),
> cgraph_node::resolve_speculation (e, target)
Actually thinking of this, perhas
On 15/11/2019 12:43, Jakub Jelinek wrote:
APUs, such as Carizzo are shared memory. DGPUs, such as Fiji and Vega, have
their own memory. A DGPU can access host memory, provided that it has been
set up just so, but that is very slow, and I don't know of a way to do that
without still having to
On Fri, 2019-11-15 at 13:28 +0100, Jakub Jelinek wrote:
> On Thu, Nov 14, 2019 at 08:34:26PM +0100, Jakub Jelinek wrote:
> > The following WIP patch implements __builtin_source_location (),
> > which returns const void pointer to a std::source_location::__impl
> > struct that is required to
On Fri, Nov 15, 2019 at 09:04:17AM -0500, David Malcolm wrote:
> > +/* Helper of fold_builtin_source_location, return the
> > + std::source_location::__impl type after performing verification
> > + on it. */
> > +
> > +static tree
> > +get_source_location_impl (location_t loc)
> > +{
>
>
On Fri, 15 Nov 2019, Matthew Malcomson wrote:
> Set global epilogue_completed when skipping pro_and_epilogue pass
>
> When compiling RTL functions marked to start at a pass after the reload
> pass, `skip_pass` is used to mark the reload pass as having completed
> since many patterns use the
On Thu, Nov 14, 2019 at 08:34:26PM +0100, Jakub Jelinek wrote:
> The following WIP patch implements __builtin_source_location (),
> which returns const void pointer to a std::source_location::__impl
> struct that is required to contain __file, __function, __line and __column
> fields, the first
On Fri, Nov 15, 2019 at 12:38:06PM +, Andrew Stubbs wrote:
> On 15/11/2019 12:21, Jakub Jelinek wrote:
> > On Thu, Nov 14, 2019 at 04:36:38PM +, Andrew Stubbs wrote:
> > > This patch adds some necessary bits to enable OpenACC testings for amdgcn
> > > offloading.
> > >
> > > The two
On 15/11/19 13:49 +0100, Jakub Jelinek wrote:
On Fri, Nov 15, 2019 at 12:35:22PM +, Jonathan Wakely wrote:
On 15/11/19 13:28 +0100, Jakub Jelinek wrote:
> On Thu, Nov 14, 2019 at 08:34:26PM +0100, Jakub Jelinek wrote:
> > The following WIP patch implements __builtin_source_location (),
> >
On 14/11/2019 17:42, Richard Henderson wrote:
What I committed today does in fact ICE for thumb1, as you suspected.
I'm currently testing the following vs
arm-sim/
arm-sim/-mthumb
arm-sim/-mcpu=cortex-a15/-mthumb.
which, with the default cpu for arm-elf-eabi, should test all of arm,
On Fri, 15 Nov 2019 at 15:27, Richard Earnshaw (lists)
wrote:
>
> On 14/11/2019 08:40, Christophe Lyon wrote:
> > Hi,
> >
> > Is it OK to backport r25 to gcc8 (Ensure dotproduct is only
> > enabled on armv8 neon) ?
> > I've noticed unnecessary failure of gcc.target/arm/simd/vdot-compile.c
> >
On 13/11/19 17:59 -0800, Thomas Rodgers wrote:
+ // TODO verify the standard requires this
+#if 0
+ // register another callback
+ bool cb3called{false};
+ std::stop_callback scb3{stok, [&]
+{
+ cb3called = true;
+
On Fri, Nov 15, 2019 at 11:02 AM Richard Sandiford
wrote:
>
> The new tree-cfg.c checking in r278245 tripped on folds of
> ALTIVEC_BUILTIN_VPERM_*, which were using gimple_convert
> rather than VIEW_CONVERT_EXPR to reinterpret the contents
> of a vector as a different type.
>
> Spot-tested on
On Mon, Nov 11, 2019 at 7:49 PM Richard Sandiford
wrote:
>
> This patch dumps the final (merged) list of alias pairs. It also adds:
>
> - WAW and RAW versions of vect-alias-check-8.c
> - a "well-ordered" version of vect-alias-check-9.c (i.e. all reads
> before any writes)
> - a test with mixed
On Mon, Nov 11, 2019 at 7:50 PM Richard Sandiford
wrote:
>
> This patch prints a message to say how an alias check is being
> implemented.
OK.
>
> 2019-11-11 Richard Sandiford
>
> gcc/
> * tree-data-ref.c (create_intersect_range_checks_index)
>
On 14/11/2019 15:32, Kwok Cheung Yeung wrote:
This patch restricts non-kernel functions to using a maximum of 64 SGPRs
and 24 VGPRs.
Kernels can request various pieces of information from the HSA runtime,
and these will be loaded into the registers consecutively before the
kernel executes.
On Fri, Nov 15, 2019 at 12:35:22PM +, Jonathan Wakely wrote:
> On 15/11/19 13:28 +0100, Jakub Jelinek wrote:
> > On Thu, Nov 14, 2019 at 08:34:26PM +0100, Jakub Jelinek wrote:
> > > The following WIP patch implements __builtin_source_location (),
> > > which returns const void pointer to a
While investigating some other failure, I saw that 'gcn' was missing
from that function. – It is only used in the error case and I failed to
trigger it in a quick way. – Hence, only compiled and regtested w/o
triggering this issue. If needed be, I can try harder, though.
OK?
Tobias
PS: On
> +++ b/gcc/testsuite/gfortran.dg/simd-builtins-10.h
> @@ -0,0 +1,6 @@
> +!GCC$ builtin (sinf) attributes simd (notinbranch, 4, 'n', 'vec_sinf')
> if('aarch64')
> +!GCC$ builtin (sinf) attributes simd (notinbranch, 4, 'n', 'vec_sinf')
> if('aarch64_be')
> +!GCC$ builtin (sinf) attributes simd
Thank you Szabolcs for working on this.
> On Nov 14, 2019, at 2:23 PM, Szabolcs Nagy wrote:
>
> Sorry v2 had a bug.
>
> v2: added documentation and tests.
> v3: fixed expand_simd_clones so different isa variants are actually
>generated.
>
> GCC currently supports two ways to declare the
I checked update_jump_functions_after_inlining(), and found one suspicious
place:
for (i = 0; i < count; i++)
{
struct ipa_jump_func *dst = ipa_get_ith_jump_func (args, i);
if (!top)
{
ipa_set_jf_unknown (dst);
< we should also
This patch allows comparison of numeric values with Holleriths. This
feature is not guarded by a compiler option as it is preferred that
extra options should avoided, this seems reasonable as current Hollerith
support does not have such an option.
In addition it also allows comparison of
The following fixes PR92512, we were failing to disqualify
_2 < 1 ? _2 : 2 with _2 being part of the reduction chain.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
Richard.
2019-11-15 Richard Biener
PR tree-optimization/92512
* tree-vect-loop.c
Hi,
On Fri, Nov 15 2019, luoxhu wrote:
>
> Sorry to break the bootstrap. This was my lack of consideration when
> splitting this small piece of code from the previous patch, the line
> should be in the first line of the second loop. Could you please add
> some comments that edge may be freed
Hi.
After I removed ::add_type hashing in r278207, we don't hash
types of SSA names. Moreover, compare_gimple_assign just compares
all operands and ICF does not compare types of SSA names in order
to not be too much conservative. So that, I'm suggesting to compare
LHS type of NOP_EXPRs, which is
On Fri, Nov 15, 2019 at 2:40 PM Martin Liška wrote:
>
> Hi.
>
> After I removed ::add_type hashing in r278207, we don't hash
> types of SSA names. Moreover, compare_gimple_assign just compares
> all operands and ICF does not compare types of SSA names in order
> to not be too much conservative.
Hi Richard,
On Fri, Nov 15, 2019 at 10:02:15AM +, Richard Sandiford wrote:
> The new tree-cfg.c checking in r278245 tripped on folds of
> ALTIVEC_BUILTIN_VPERM_*, which were using gimple_convert
> rather than VIEW_CONVERT_EXPR to reinterpret the contents
> of a vector as a different type.
>
On 14/11/2019 15:30, Kwok Cheung Yeung wrote:
The set of fixed registers is adjusted by the
TARGET_CONDITIONAL_REGISTER_USAGE hook, but this needs to be done on a
per-function basis, whereas the hook is normally called once during GCC
initialization before any functions have been processed
On 2019/11/15 17:19, Jan Hubicka wrote:
>> On Fri, Nov 15, 2019 at 9:10 AM Jan Hubicka wrote:
>>>
next is initialized only in the loop before, it is never updated
in it's own loop.
gcc/ChangeLog
2019-11-15 Xiong Hu Luo
* ipa-inline.c
On Fri, Nov 15, 2019 at 11:59 AM Richard Sandiford
wrote:
>
> Richard Biener writes:
> > On Fri, Nov 15, 2019 at 11:02 AM Richard Sandiford
> > wrote:
> >>
> >> The new tree-cfg.c checking in r278245 tripped on folds of
> >> ALTIVEC_BUILTIN_VPERM_*, which were using gimple_convert
> >> rather
On 14/11/2019 15:33, Kwok Cheung Yeung wrote:
The kernel attributes are changed to request at least 64 SGPRs and 24
VGPRs (i.e. the non-kernel maximum, otherwise the callees may not have
enough registers to run in) for non-leaf kernels to take advantage of
the reduced number of registers used
Richard Biener writes:
> On Mon, Nov 11, 2019 at 7:47 PM Richard Sandiford
> wrote:
>>
>> This patch adds a bunch of flags to dr_with_seg_len_pair_t,
>> for use by later patches. The update to tree-loop-distribution.c
>> is conservatively correct, but might be tweakable later.
>
> Does this all
The documentation for __RTL tests (see "(gccint) RTL Tests" info node) has the
following snippet.
```
The parser expects the RTL body to be in the format emitted by this
dumping function:
DEBUG_FUNCTION void
print_rtx_function (FILE *outfile, function *fn, bool compact);
when
On Fri, 15 Nov 2019, Matthew Malcomson wrote:
> The documentation for __RTL tests (see "(gccint) RTL Tests" info node) has the
> following snippet.
>
> ```
> The parser expects the RTL body to be in the format emitted by this
> dumping function:
>
> DEBUG_FUNCTION void
>
r278235 broke conversions of vector/scalar shifts into vector/vector
shifts on targets that only provide the latter. We need to record
whether a conversion is required in that case too.
Also, the old useless_type_conversion_p condition seemed unnecessarily
strong, since the shift amount can have
1 - 100 of 229 matches
Mail list logo
