As we are on patching algos we still have this old one.
From the original patch I only kept the memory optimization part as
the new performance test was not showing good result for the other part
to change pivot value. I also kept the small change in
get_temporary_buffer even if I don't ha
Committed with adding comments for those two functions.
On Thu, Jun 11, 2020 at 5:10 AM Jim Wilson wrote:
>
> On Wed, Jun 10, 2020 at 1:08 AM Kito Cheng wrote:
> > * config/riscv/riscv.c (gpr_save_reg_order): New.
> > (riscv_expand_prologue): Use riscv_gen_gpr_save_insn to gen
>
Committed.
On Thu, Jun 11, 2020 at 5:13 AM Jim Wilson wrote:
>
> On Wed, Jun 10, 2020 at 1:08 AM Kito Cheng wrote:
> > * config/riscv/riscv-protos.h (riscv_output_gpr_save): Remove.
> > * config/riscv/riscv-sr.c (riscv_sr_match_prologue): Update
> > value.
> > * c
On May 26, 2020, Alexandre Oliva wrote:
> On May 19, 2020, Joseph Myers wrote:
>> Allowing a missing executable name is reasonable enough, but I was
>> actually thinking that the messages should print "gcc" or whatever command
>> the user ran in place of "collect2".
> Should we make the rege
On Jun 9, 2020, Richard Biener wrote:
> How about simply unconditionally doing dump_flags | TDF_SLIM here
> to have the whole mem_expr on one line.
SGTM
> OK with that change.
Thanks, here's what I'm installing.
slim up mem exprs to avoid line breaks in -fverbose-asm
From: Alexandre Oliva
Thanks for viewing and pushing this.
Thanks,
Haijian Zhang
> -Original Message-
> From: Richard Sandiford [mailto:richard.sandif...@arm.com]
> Sent: Thursday, June 11, 2020 12:01 AM
> To: Zhanghaijian (A)
> Cc: gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH PR95523] aarch64:ICE in registe
On Thu, 11 Jun 2020, Patrick McGehearty wrote:
> I will study real.c carefully along with the C99 std
> to determine if I can find useful values for RMIN2 and RMINSCAL
> for each format which are within range for all instances
> of that format. A quick skim of real.c shows we have ieee half precis
I will study real.c carefully along with the C99 std
to determine if I can find useful values for RMIN2 and RMINSCAL
for each format which are within range for all instances
of that format. A quick skim of real.c shows we have ieee half precision
and two arm half precision formats, for example.
I
On Jun 9, 2020, Martin Jambor wrote:
>> Before your changes, GCC produced the expected:
>>
>> $ ls -1 build-gcc/gcc/testsuite/brig/variables.*???t.*
>> build-gcc/gcc/testsuite/brig/variables.hsail.brig.004t.original
>> build-gcc/gcc/testsuite/brig/variables.hsail.brig.005t.gimple
>>
>> Are you
On Wed, Jun 10, 2020 at 04:27:27PM -0600, Jeff Law wrote:
> On Mon, 2019-07-22 at 12:39 -0400, Arvind Sankar wrote:
> > The gcc configure script does not use the config/picflag.m4 macro to
> > customize PICFLAG according to the host when using --enable-host-shared.
> >
> > Fix configure.ac to do s
On Wed, 10 Jun 2020, Patrick McGehearty wrote:
> #ifdef L_divhc3
> #define RBIG (correct value for half precision)
> #define RMIN (correct value for half precision)
> #define RMIN2 ... (correct value for half precision)
> #define RMINSCAL ... (correct value for half precision)
> #endif
> #ifdef
Joseph,
Thank you again for your prompt and insightful response.
It's now clear that I was engaged in wishful thinking that I only needed
to handle double precision in my proposed change. I understand now that
the text above the routine:
- - - - -
#if defined(L_divhc3) || defined(L_divsc3) || def
On Mon, 2019-07-22 at 12:39 -0400, Arvind Sankar wrote:
> The gcc configure script does not use the config/picflag.m4 macro to
> customize PICFLAG according to the host when using --enable-host-shared.
>
> Fix configure.ac to do so.
>
> Tested bootstrap on x86_64-linux-gnu.
>
> 2019-07-22 Arvin
On Jun 9, 2020, Thomas Schwinge wrote:
> Are you able to easily create/suggest patches for these? (You're
> probably not set up for offloading compilation...)
I can try, but I can certainly use help, if not in coding, at least with
testing.
> Can you suggest
> how/where to adjust: producer-si
A follow up note relating to use of fused multiply add in complex divide:
While reviewing bugs relating to complex divide in libgcc, I
rediscovered bug 59714 - complex division is surprising on targets
with FMA.
The specific concern was when you divide (1.0 + 3.0i) by itself and
fused multiply a
On Wed, 10 Jun 2020, Jonathan Wakely wrote:
> On 10/06/20 15:32 -0400, Patrick Palka via Libstdc++ wrote:
> > ranges::copy and a number of other ranges algorithms have unwrapping
> > optimizations for iterators of type __normal_iterator, move_iterator and
> > reverse_iterator. But in the checks t
Since r10-7096 convert_like, when called in a template, creates an
IMPLICIT_CONV_EXPR when we're converting to/from array type.
In this test, we have e[f], and we're converting f (of type class A) to
int, so convert_like in build_new_op_1 created the IMPLICIT_CONV_EXPR
that got into cp_build_array
Another indication that perhaps this warning is emitted too early. We
crash because same_type_p gets a null type: we have an enumerator
without a fixed underlying type and finish_enum_value_list hasn't yet
run. So check if the type is null before calling same_type_p.
(This is a regression and th
On Wed, Jun 10, 2020 at 1:08 AM Kito Cheng wrote:
> * config/riscv/riscv.c (gpr_save_reg_order): New.
> (riscv_expand_prologue): Use riscv_gen_gpr_save_insn to gen gpr_save.
> (riscv_gen_gpr_save_insn): New.
> ...
Looks good to me. Though these two new functions should ha
On Wed, Jun 10, 2020 at 1:08 AM Kito Cheng wrote:
> * config/riscv/riscv-protos.h (riscv_output_gpr_save): Remove.
> * config/riscv/riscv-sr.c (riscv_sr_match_prologue): Update
> value.
> * config/riscv/riscv.c (riscv_output_gpr_save): Remove.
> * config/ris
On 10/06/20 15:32 -0400, Patrick Palka via Libstdc++ wrote:
ranges::copy and a number of other ranges algorithms have unwrapping
optimizations for iterators of type __normal_iterator, move_iterator and
reverse_iterator. But in the checks that guard these optimizations we
currently only test that
Hi Jason,
Jason Merrill wrote:
> On Tue, Jun 9, 2020 at 5:04 AM Iain Sandoe wrote:
>
> /* Don't bother reversing an operator with two identical parameters.
> */
> - else if (args->length () == 2 && (flags & LOOKUP_REVERSED))
> + else if (args && args->length () == 2 && (fl
Hi Alexandre,
Alexandre Oliva wrote:
On Jun 9, 2020, Iain Sandoe wrote:
I have an ugly patch that makes this work for Darwin (essentially, by
having two versions of the LTO tests
We could deal with that in a similar way to how .dwo files are handled,
namely, with explicit handling in th
Early ping.
> Gesendet: Donnerstag, 04. Juni 2020 um 20:47 Uhr
> Von: "Harald Anlauf"
> An: "fortran" , "gcc-patches"
> Betreff: [PATCH, PR fortran/95503] [9/10/11 Regression] ICE in
> gfc_is_simply_contiguous, at fortran/expr.c:5844
>
> The following patch fixes an almost obvious ICE in inval
ranges::copy and a number of other ranges algorithms have unwrapping
optimizations for iterators of type __normal_iterator, move_iterator and
reverse_iterator. But in the checks that guard these optimizations we
currently only test that the iterator of the iterator/sentinel pair has
the appropriat
Hi!
On Wed, Jun 10, 2020 at 09:14:07AM -0700, Carl Love wrote:
> On Wed, 2020-06-10 at 10:46 -0500, will schmidt wrote:
> > Compare that to the other predicates (config/rs6000/predicates.md)
> >
> > Those have explicit checks against both ends of the valid range of
> > values. i.e.
> >
> > ;;
Hi!
On Tue, Jun 09, 2020 at 05:01:45PM -0700, Carl Love wrote:
> On Fri, 2020-06-05 at 16:28 -0500, Segher Boessenkool wrote:
> > > +;; Return 1 if op is a 32-bit constant signed integer
> > > +(define_predicate "s32bit_cint_operand"
> > > + (and (match_code "const_int")
> > > + (match_test
Hi Carl,
On Wed, Jun 10, 2020 at 09:05:36AM -0700, Carl Love wrote:
> I committed this patch to mainline and backported to GCC 9.
>
> I have looked at GCC 8. The functional issue is there, i.e. the
> vcmpnez is used instead of vcmpne. However the test case
> builtins-8-p9-runnable.c does not
On Jun 9, 2020, Iain Sandoe wrote:
> That means that the intermediate objects proceed all the way to .s
> output - and thus the ‘final’ pass is run (producing the extra files seen).
> You can mimic this with x86 Linux by appending -ffat-lto-objects to an
> LTO command line.
I see, thanks.
>
On Jun 9, 2020, Rainer Orth wrote:
> this is wrong unfortunately: braces are the Tcl equivalent of single
> quotes so you're setting skip_lto to the string inside.
Aah, thanks. So when $skip_lto is expanded in the ifs, the whole thing
is evaluated, and that's why it works anyway?
> While this
On 10/06/20 18:40 +0200, François Dumont wrote:
On 10/06/20 4:49 pm, Jonathan Wakely wrote:
On 10/06/20 08:18 +0200, François Dumont via Libstdc++ wrote:
On 09/06/20 10:53 pm, Jonathan Wakely wrote:
This reminds me that I was going to extend the condition for using
memcmp to also apply to
Tamar Christina writes:
> Hi All,
>
> The test in check_effective_target_exceptions_enabled uses a C++ keyword
> `throw`
> and the test fails with a syntax error on any non-g++ test. I now tell the
> testsuite driver that this is a C++ input file so it runs it as such in all
> the
> drivers.
N
On Wed, Jun 10, 2020 at 9:57 AM Tamar Christina wrote:
>
> Hi All,
>
> The amdgcn-amdhsa test seems to be running for all targets unconditionally
> while
> only really makes sense for certain targets. This patch adds an opt-out list
> and opts out arm targets.
>
> Regtested on aarch64-none-linux
Hi All,
The test in check_effective_target_exceptions_enabled uses a C++ keyword `throw`
and the test fails with a syntax error on any non-g++ test. I now tell the
testsuite driver that this is a C++ input file so it runs it as such in all the
drivers.
Regtested on aarch64-none-linux-gnu and no
"Kewen.Lin" writes:
> diff --git a/gcc/tree-vect-loop-manip.c b/gcc/tree-vect-loop-manip.c
> index ca68d04a919..1fac5898525 100644
> --- a/gcc/tree-vect-loop-manip.c
> +++ b/gcc/tree-vect-loop-manip.c
> @@ -420,8 +420,8 @@ vect_set_loop_controls_directly (class loop *loop,
> loop_vec_info loop_vi
Hi All,
The amdgcn-amdhsa test seems to be running for all targets unconditionally while
only really makes sense for certain targets. This patch adds an opt-out list
and opts out arm targets.
Regtested on aarch64-none-linux-gnu and no issues.
Ok for master?
Thanks,
Tamar
gcc/testsuite/ChangeL
On 10/06/20 4:49 pm, Jonathan Wakely wrote:
On 10/06/20 08:18 +0200, François Dumont via Libstdc++ wrote:
On 09/06/20 10:53 pm, Jonathan Wakely wrote:
This reminds me that I was going to extend the condition for using
memcmp to also apply to unsigned integers with sizeof(T) > 1 on big
endian t
"Yangfei (Felix)" writes:
> Hi,
>
> PR: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=95570
>
> Here, we are doing loop versioning for alignment. The only dr here is a
> gather-statter operation: x[start][i].
> Scalar evolution analysis for this dr failed, so DR_STEP is NULL_TREE, which
>
The inactive development branches are sorted alphabetically except for the
openacc-gcc-*-branch entries, which are grouped with the gomp*-branch entries.
Would it be better to place devel/omp/gcc-9 alphabetically (in which case it
will go between debuglocus and dwarf4), or by topic (in which cas
On 6/10/20 12:00 PM, Andreas Schwab wrote:
On Jun 10 2020, Nathan Sidwell wrote:
needed 'origin' -- eg 'git merge origin master'
That's an octopus merge. I don't think you want that.
ah. Somehow I convinced myself that was how to merge from master, but I
guess my formulation resulted in o
On Wed, 2020-06-10 at 10:46 -0500, will schmidt wrote:
> > On Fri, 2020-06-05 at 16:28 -0500, Segher Boessenkool wrote:
> > > > +;; Return 1 if op is a 32-bit constant signed integer
> > > > +(define_predicate "s32bit_cint_operand"
> > > > + (and (match_code "const_int")
> > > > + (match_t
On Fri, 2020-05-08 at 15:35 -0400, Lewis Hyatt wrote:
> On Fri, Jan 31, 2020 at 03:31:59PM -0500, David Malcolm wrote:
> > On Fri, 2020-01-31 at 14:31 -0500, Lewis Hyatt wrote:
> > > Hello-
> > >
> > > Here is the second patch that I mentioned when I submitted the
> > > other
> > > related
> > > p
The following avoids allocating stmt info structs for debug stmts.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
2020-06-10 Richard Biener
* tree-vect-loop.c (vect_determine_vectorization_factor):
Skip debug stmts.
(_loop_vec_info::_loop_vec_info): Like
The following avoids leading debug stmts in BB vectorizer regions.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
2020-06-10 Richard Biener
PR tree-optimization/95576
* tree-vect-slp.c (vect_slp_bb): Skip leading debug stmts.
* g++.dg/vect/pr95576.cc: N
Segher, Bill:
I committed this patch to mainline and backported to GCC 9.
I have looked at GCC 8. The functional issue is there, i.e. the
vcmpnez is used instead of vcmpne. However the test case
builtins-8-p9-runnable.c does not exist in GCC 8. The patch consists
of the functional fix:
---
Hi,
"Zhanghaijian (A)" writes:
> This is a simple fix for pr95523.
> When registering the tuple type in register_tuple_type, the TYPE_ALIGN
> (tuple_type) will be changed by -fpack-struct=n.
> We need to maintain natural alignment in handle_arm_sve_h.
> Bootstrap and tested on aarch64 Linux plat
On Jun 10 2020, Nathan Sidwell wrote:
> needed 'origin' -- eg 'git merge origin master'
That's an octopus merge. I don't think you want that.
Andreas.
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1
"And now for something compl
On Wed, Jun 10, 2020 at 05:47:54PM +0200, Thomas Schwinge wrote:
> The automated system runs once a day. I guess it's generally deemed
> acceptable if the source code state (commits) and the 'ChangeLog' files
> state is out of sync for one day. There surely must be some mechanism in
> place to ma
On Wed, Jun 10, 2020 at 8:38 AM Nathan Sidwell wrote:
>
> I'm trying to merge master to devel/c++-modules, as I usually do. But
> there's an ICE creeping in, so I'm attempting incremental merges to
> figure out what happened.
>
> I thought 'git merge --no-commit origin $HASH' would do that, pulli
On 6/10/20 11:38 AM, Nathan Sidwell wrote:
I'm trying to merge master to devel/c++-modules, as I usually do. But
there's an ICE creeping in, so I'm attempting incremental merges to
figure out what happened.
I thought 'git merge --no-commit origin $HASH' would do that, pulling
$HASH out of ma
On Tue, 2020-06-09 at 17:01 -0700, Carl Love wrote:
> Segher:
>
> So I have been looking at the predicate definitions that I had
> created.
>
> On Fri, 2020-06-05 at 16:28 -0500, Segher Boessenkool wrote:
> > > +;; Return 1 if op is a 32-bit constant signed integer
> > > +(define_predicate "s32bi
Hi!
On 2020-06-10T17:15:41+0200, Tobias Burnus wrote:
> On 6/10/20 3:34 PM, Kwok Cheung Yeung wrote:
>> This patch updates the previous entry on the website for
>> devel/omp/gcc-9 to refer to the new branch instead. This removes
>> references to the old branch, as new development should occur on
I'm trying to merge master to devel/c++-modules, as I usually do. But
there's an ICE creeping in, so I'm attempting incremental merges to
figure out what happened.
I thought 'git merge --no-commit origin $HASH' would do that, pulling
$HASH out of master's commit history (git tree --first-pare
On 6/10/20 3:34 PM, Kwok Cheung Yeung wrote:
This patch updates the previous entry on the website for
devel/omp/gcc-9 to refer to the new branch instead. This removes
references to the old branch, as new development should occur on the
newer branch.
Can you move the old entry to "Inactive Devel
Thanks both!
Cheers,
Tamar
> -Original Message-
> From: Martin Liška
> Sent: Wednesday, June 10, 2020 2:41 PM
> To: Tamar Christina ; Jonathan Wakely
>
> Cc: Jakub Jelinek ; g...@gcc.gnu.org; gcc-patches patc...@gcc.gnu.org>
> Subject: Re: [IMPORTANT] ChangeLog related changes
>
> On
On 10/06/20 08:18 +0200, François Dumont via Libstdc++ wrote:
On 09/06/20 10:53 pm, Jonathan Wakely wrote:
This reminds me that I was going to extend the condition for using
memcmp to also apply to unsigned integers with sizeof(T) > 1 on big
endian targets.
This illustrates what I tried to avo
On 6/10/20 2:12 AM, Martin Liška wrote:
PING^1
The exclusion changes are just what I was suggesting, thanks.
Martin
On 5/25/20 3:10 PM, Martin Liška wrote:
On 5/21/20 4:53 PM, Martin Sebor wrote:
On 5/21/20 5:28 AM, Martin Liška wrote:
On 5/18/20 10:37 PM, Martin Sebor wrote:
I know the
This new Ada 202x AI introduces a new package Modular_Arithmetic.
Related discussion also suggested to rename the recently introduced
Arithmetic package -> Integer_Arithmetic, for consistency, so this is
done at the same time.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Arnaud C
Renamed files are listed in the following format:
M gcc/ada/Makefile.rtl
M gcc/ada/impunit.adb
R097gcc/ada/libgnat/s-atopar.adbgcc/ada/libgnat/s-aoinar.adb
R095gcc/ada/libgnat/s-atopar.adsgcc/ada/libgnat/s-aoinar.ads
A gcc/ada/libgnat/s-aomoar.adb
A gcc/ada
Hello.
I've just pushed the documentation improvement about Co-Authored-By.
Martin
---
htdocs/codingconventions.html | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/htdocs/codingconventions.html b/htdocs/codingconventions.html
index 6ae962ea..a08ddcbb 100644
--- a/htdocs/
On 6/10/20 3:34 PM, Tamar Christina wrote:
Hi All,
Hello.
We've been wondering since we no longer list authors in the changelog (at least
mklog doesn't generate it),
You are right, it's preferred solution and it's documented here:
https://gcc.gnu.org/codingconventions.html#ChangeLogs
'''
On Wed, Jun 10, 2020 at 01:34:54PM +, Tamar Christina wrote:
> Hi All,
>
> We've been wondering since we no longer list authors in the changelog (at
> least mklog doesn't generate it),
> How do we handle multi author patches nowadays?
>
> Tried searching for it on the website but couldn’t fi
Working on this AI it appeared that GNAT wasn't implementing the Ada
2012 notion of "require late initialization", so plug this hole and
implement the new rule from AI12-0192 at the same time.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Arnaud Charlet
gcc/ada/
* exp_
Building equivalent record aggregates when the type of the aggregate has
predicate functions can result in Gigi crashes if the type hasn't been
frozen yet. Since Build_Equivalent_Record_Aggregate is an optimization,
it's ok to disable it when encountering aggregates with predicates.
Tested on x86_
This patch enhances the warnings on overlapping actuals of composite
types when only one of them is writable. If these parameters are passed
by reference it is the case that assignment to one could have the
undesirable effect of modifying the other inside the called subprogram.
Tested on x86_64-pc
This Ada 202x AI defines among other things new check names. Recognize
them as no-ops for now.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Arnaud Charlet
gcc/ada/
* snames.ads-tmpl (Name_Characters_Assertion_Check,
Name_Containers_Assertion_Check,
Na
This makes sure that the semantics specified by this AI is observed
by using an expression with actions in order to insert the PE raise
statement in a membership context with multiple choices.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Eric Botcazou
gcc/ada/
* exp_c
This makes sure that a Sloc is put on the dereferences inserted by the
new procedure Copy_And_Maybe_Dereference.
No functional changes.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Eric Botcazou
gcc/ada/
* sem_util.adb (Copy_And_Maybe_Dereference): Temporarily copy
This plugs the only loophole in the front-end through which implicit
dereferences can reach the code generator without having being turned
into explicit ones.
No functional changes.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Eric Botcazou
gcc/ada/
* sem_util.adb (C
This removes a block of code in Resolve_Call that inserts an explicit
dereference for a call whose prefix is an access-to-subprogram type, but
this processing is already done earlier in Analyze_Call.
No functional changes.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Eric Botcaz
The routine Determine_License is brittle and doesn't e.g. handle
properly wide characters. Furthermore this is just a heuristic, which
isn't really needed, so remove it to simplify maintenance and remove
latent issues with wide characters.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06
Folding of Enum_Rep attribute was partly done in evaluation (for
expressions like "Typ'Enum_Rep (Enum_Literal)") and partly in expansion
(for expressions like "Enum_Literal'Enum_Rep". Moreover, some of the
code in evaluation was dead and some of the code in expansion was
violating internal assertio
A workaround for a bug in expansion of Enum_Rep attribute in the
GNATprove mode was to expand First and Last attributes. Now with
handling of Enum_Rep fixed we don't need this workaround anymore.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Piotr Trojanek
gcc/ada/
* e
This patch implements AI12-0220, which adds contracts to
Access_To_Subprogram types so that pre/postconditions are applied to all
indirect calls through such an access type.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Ed Schonberg
gcc/ada/
* einfo.ads (Access_Subprog
This changes the way some assignments of aggregates of dynamic
discriminated record types are expanded by the front-end: they used to
always give rise to the creation of a temporary, which is unnecessary if
the by-copy semantics can be guaranteed.
This also puts the treatment of qualified aggregat
This patch fixes a bug where if pragma Assertion_Policy(Ignore) is in
effect, if the only reference to a given declaration is in an Invariant,
spurious warnings about unused entities are given. For example, if a
compilation unit says "with X;", and the only reference to X is in an
invariant, the co
Membership test operators (i.e. "in" and "not in") have either the right
operand present (when there is just one choice) or the list of
alternatives present (when there are many choices), as documented in
sinfo.ads.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Piotr Trojanek
g
When post-call actions need to be inserted in an expression context, an
N_Expression_With_Actions node is used. That's not done here for the
condition of an if-expression, so the change adds this case.
It also deals with the case of a call to a primitive of a tagged type
written in prefixed notat
Overriding dispatching primitives Initialize, Adjust or Finalize of a
controlled type by means of a subprogram body that has no specification
causes the frontend to initialize incorrectly the dispatch table slots
of these primitives.
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 J
This commit fixes a bug introduced in a previous commit that attempted
to detect illegal body definitions on null procedures but did not
account for the fact that function definitions such as `function F
return Integer with Global => null is` could use the same path. This
would result in an asserti
This patch corrects an issue whereby the compiler would issue incorrect
accessibility errors and checks for objects initialized by functions
returning anonymous access types or type conversions where the operand
is a function returning an anonymous access type.
Tested on x86_64-pc-linux-gnu, commi
The `if Token = ...` condition was added when aspect specifications were
introduced in GNAT: aspect specification parsing is done as part of
subprogram declaration parsing, and so once aspect specifications are
parsed, it is necessary to re-start the subprogram declaration-parsing
autotmata. But re
Routine Get_Low_Bound was added to frontend while moving utility
routines from the GNATprove backend, but it is no longer referenced.
(Surprisingly, there is no Get_High_Bound routine anywhere).
Tested on x86_64-pc-linux-gnu, committed on trunk
2020-06-10 Piotr Trojanek
gcc/ada/
* se
Hi All,
We've been wondering since we no longer list authors in the changelog (at least
mklog doesn't generate it),
How do we handle multi author patches nowadays?
Tried searching for it on the website but couldn’t find anything.
Thanks,
Tamar
> -Original Message-
> From: Gcc-patches
Hello
The devel/omp/gcc-10 branch was pushed yesterday as the new development branch
for OpenMP/OpenACC/offloading functionality, based on the GCC 10 release.
This patch updates the previous entry on the website for devel/omp/gcc-9 to
refer to the new branch instead. This removes references t
The following avoids leading debug stmts in BB vectorizer regions.
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
2020-06-10 Richard Biener
PR tree-optimization/95576
* tree-vect-slp.c (vect_slp_bb): Skip leading debug stmts.
* g++.dg/vect/pr95576.cc: New
On 6/10/20 2:27 PM, Martin Liška wrote:
/home/marxin/Programming/testcases/vect-low.c: In function ‘v2df foo(v2df,
v2df, v2df, v2df)’:
/home/marxin/Programming/testcases/vect-low.c:3:6: error: BB 2 is missing an EH
edge
Ok, I was missing copying of the EH edges:
FOR_EACH_EDGE (e, ei, g
V3: Update the define_expand as optab changes.
gcc/ChangeLog:
2020-MM-DD Kewen Lin
* config/rs6000/vsx.md (lenload): New define_expand.
(lenstore): Likewise.
---
gcc/config/rs6000/vsx.md | 32
1 file changed, 32 insertions(+)
diff --git a/
on 2020/6/10 下午5:22, Richard Sandiford wrote:
> "Kewen.Lin" writes:
>> @@ -2497,6 +2499,9 @@ expand_mask_load_optab_fn (internal_fn, gcall *stmt,
>> convert_optab optab)
>>
>>if (optab == vec_mask_load_lanes_optab)
>> icode = get_multi_vector_move (type, optab);
>> + else if (optab ==
On 6/10/20 12:50 PM, Richard Biener wrote:
with -fnon-call-exceptions should trigger it.
Thanks, that works!
We start with:
foo (v2df a, v2df b, v2df c, v2df d)
Eh tree:
1 try land:{1,} catch:{}
{
void * _1;
v2df _2;
v2df _8;
[local count: 1073741824]:
[LP 1] _8 = VEC_COND_EXPR
On 10/06/20 13:13 +0100, Jonathan Wakely wrote:
On 10/06/20 09:43 +0200, Martin Liška wrote:
On 6/9/20 9:46 PM, Jonathan Wakely wrote:
Is this worth adding to contrib/prepare-commit-msg?
I like the idea and I would make it conditional based on
an environment variable? Similarly to GCC_GIT_DIF
This makes {SLP_TREE,STMT_VINFO}_VEC_STMTS a vector of gimple * and
not allocate a stmt_vec_info for vectorizer generated stmts since
this is now possible after removing the only use which was chaining
of vector stmts via STMT_VINFO_RELATED_STMT.
This also removes all stmt_vec_info allocations d
On 10/06/20 13:13 +0100, Jonathan Wakely wrote:
On 10/06/20 09:43 +0200, Martin Liška wrote:
On 6/9/20 9:46 PM, Jonathan Wakely wrote:
Is this worth adding to contrib/prepare-commit-msg?
I like the idea and I would make it conditional based on
an environment variable? Similarly to GCC_GIT_DIF
On 10/06/20 09:43 +0200, Martin Liška wrote:
On 6/9/20 9:46 PM, Jonathan Wakely wrote:
Is this worth adding to contrib/prepare-commit-msg?
I like the idea and I would make it conditional based on
an environment variable? Similarly to GCC_GIT_DIFF_FILE?
With this patch you can use the gcc-con
On Wed, Jun 10, 2020 at 01:14:59PM +0200, Martin Liška wrote:
> >From 4d2e0b1e87b08ec21fd82144f00d364687030706 Mon Sep 17 00:00:00 2001
> From: Martin Liska
> Date: Tue, 19 May 2020 16:57:56 +0200
> Subject: [PATCH] Add missing store in emission of asan_stack_free.
>
> gcc/ChangeLog:
>
> 2020-05
On 6/10/20 12:08 PM, Jakub Jelinek wrote:
On Wed, Jun 10, 2020 at 11:49:01AM +0200, Martin Liška wrote:
- store_by_pieces (shadow_mem, sz, builtin_memset_read_str, &c,
-BITS_PER_UNIT, true, RETURN_BEGIN);
+ {
+ /* Emit:
+ memset(ShadowBase
On Wed, Jun 10, 2020 at 10:51 AM Martin Liška wrote:
>
> On 6/9/20 3:42 PM, Richard Biener wrote:
> > On Mon, Jun 8, 2020 at 1:04 PM Martin Liška wrote:
> >>
> >> Hello.
> >>
> >> Thank you for the approval. There's the patch that defines 4 new
> >> DEF_INTERNAL_OPTAB_FN.
> >>
> >> Patch can boo
On Tue, Jun 9, 2020 at 9:16 PM Martin Liška wrote:
>
> On 6/9/20 6:32 PM, Richard Biener wrote:
> > On Tue, Jun 9, 2020 at 10:09 AM Martin Liška wrote:
> >>
> >> On 6/8/20 4:53 PM, Martin Liška wrote:
> >>> Hi.
> >>>
> >>> Thank you for the report. It's caused by fact that LLVM switch in
> >>> 4
On Wed, Jun 10, 2020 at 11:49:01AM +0200, Martin Liška wrote:
> - store_by_pieces (shadow_mem, sz, builtin_memset_read_str, &c,
> - BITS_PER_UNIT, true, RETURN_BEGIN);
> + {
> + /* Emit:
> +memset(ShadowBase, kAsanStackAfterReturnMagic, ShadowSize);
>
On 6/10/20 10:42 AM, Jakub Jelinek wrote:
E.g. we just shouldn't reuse
MEMs (even after adjusting them) from different indirection levels because
we risk some attributes (alias set, MEM_EXPR, whatever else) will stay
around from the different indirection level.
All right, what about the updated
1 - 100 of 121 matches
Mail list logo
