On the PowerPC starting with ISA 2.07 (power8), moving a single precision value
(SFmode) from a vector register to a GPR involves converting the scalar value
in the register from being in double (DFmode) format to the 32-bit
vector/storage format, doing the move to the GPR, and then doing a shift
On 9/19/17 12:38 PM, Bill Schmidt wrote:
> Hi,
>
> https://gcc.gnu.org/PR82255 identifies a problem in the vector cost model
> where a vectorized load is treated as having the cost of a strided load
> in a case where we will not actually generate a strided load. This is
> simply a mismatch
On 09/12/2017 11:06 AM, Jeff Law wrote:
On 08/17/2017 10:03 AM, Martin Sebor wrote:
First, there is the risk that someone will write code based
on the pure declaration even if there's no intervening call
to the function between it and the const one. Code tends to
be sloppy, and it's also not
On 09/19/2017 07:13 AM, Rainer Orth wrote:
> Daniel Santos writes:
>
>> On 09/17/2017 10:53 AM, Uros Bizjak wrote:
>>> OK.
>>>
>>> Thanks,
>>> Uros.
>> Thanks. I should have posted this Friday when my tests finished, but
>> I'll be committing with one minor change so
Bill Schmidt writes:
> Index: gcc/tree-vect-stmts.c
> ===
> --- gcc/tree-vect-stmts.c (revision 252760)
> +++ gcc/tree-vect-stmts.c (working copy)
> @@ -1091,8 +1091,19 @@ vect_model_load_cost
On Sep 19, 2017, at 4:21 PM, Bill Schmidt wrote:
>
> On Sep 19, 2017, at 3:58 PM, Richard Sandiford
> wrote:
>>
>> Bill Schmidt writes:
>>> Index: gcc/tree-vect-stmts.c
>>>
On 07/26/2017 05:20 AM, Richard Biener wrote:
> On Tue, Jul 25, 2017 at 4:50 PM, Andrew MacLeod wrote:
>> On 07/25/2017 03:12 AM, Richard Biener wrote:
>>>
>>> On Fri, Jul 21, 2017 at 9:30 PM, Aldy Hernandez wrote:
On Mon, Jul 17, 2017 at 6:23 AM,
On Tue, Sep 19, 2017 at 4:42 PM, Steven Taschuk wrote:
> The behaviour of echo for arguments containing the two-character
> substring `\0` varies among implementations: in coreutils echo,
> and in the builtins of ash, bash, busybox sh, csh, and fish, the two
> characters `\0`
On 09/06/2017 05:30 PM, Joseph Myers wrote:
On Thu, 17 Aug 2017, Martin Sebor wrote:
+/* Check LAST_DECL and NODE of the same symbol for attributes that are
+ recorded in EXCL to be mutually exclusive with ATTRNAME, diagnose
+ them, and return true if any have been found. NODE can be a
Am 18.09.2017 um 11:50 schrieb Dominique d'Humières:
Warning: Conversion from 'REAL(4)' to 'REAL(8)' at (1) [-Wconversion-extra]
Not me (not in the general case)
even if may allow to detect things such as ‘pi8=acos(-1.0)’?
This one would be interesting to catch (even with -Wall), although
On 09/19/2017 01:58 AM, Jakub Jelinek wrote:
> On Mon, Sep 18, 2017 at 06:10:29PM -0500, Daniel Santos wrote:
>> Mike, can you take a look at this please?
>>
>> On 09/18/2017 10:17 AM, Dominique d'Humières wrote:
>>> This patch (r252896) breaks bootstrap on x86_64-apple-darwin10 configured
>>>
> Le 19 sept. 2017 à 21:59, Thomas Koenig a écrit :
>
> Am 18.09.2017 um 11:50 schrieb Dominique d'Humières:
>> Warning: Conversion from 'REAL(4)' to 'REAL(8)' at (1) [-Wconversion-extra]
>
> Not me (not in the general case)
>
>> even if may allow to detect things such
On Sep 19, 2017, at 3:58 PM, Richard Sandiford
wrote:
>
> Bill Schmidt writes:
>> Index: gcc/tree-vect-stmts.c
>> ===
>> --- gcc/tree-vect-stmts.c(revision 252760)
>>
On Tue, Sep 19, 2017 at 05:24:14PM +0200, Jakub Jelinek wrote:
> These changes broke DWARF-5 support. E.g. in gcc-7 and before this change
> there was:
Here is a fix, make check-g++ RUNTESTFLAGS=dwarf2.exp now passes again
even with !DWARF2_ASM_LINE_DEBUG_INFO compiler. Haven't tried LTO
On Tue, Sep 19, 2017 at 9:01 AM, Peryt, Sebastian
wrote:
>> >> >> > This patch adds options -march=/-mtune=knm for Knights Mill.
>> >> >> >
>> >> >> > 2017-09-14 Sebastian Peryt gcc/
>> >> >> >
>> >> >> > * config.gcc: Support
The behaviour of echo for arguments containing the two-character
substring `\0` varies among implementations: in coreutils echo,
and in the builtins of ash, bash, busybox sh, csh, and fish, the two
characters `\0` are emitted literally; the builtins of tcsh and zsh
emit a null character and
In the 1.9 upgrade of libgo I took out the word "goroutine" from a
traceback showing a goroutine running in C code, to let
TestCgoNumGoroutine pass. However, it turns out that some code is
actually checking for that string; for example,
On Sep 18 2017, Joseph Myers wrote:
> Building glibc for many different configurations and running the
> compilation parts of the testsuite runs into failures of the
> elf/check-execstack test for hppa, ia64 and microblaze.
ia64 is non-execstack by default, so it
The following plugs some holes in extract_affine which failed
to modulo-reduce signed values in conversions, failed to
interpret pointer-plus offsets as signed (yeah, stupid GIMPLE IL...)
and mishandled BIT_NOT_EXPR.
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
Ok?
Thanks,
On Sep 18 2017, Jeff Law wrote:
> On 09/18/2017 10:09 AM, Andreas Schwab wrote:
>> On Sep 18 2017, Jeff Law wrote:
>>
>>> Can you confirm if the probe was in the red zone vs the live areas on
>>> the stack?
>>
>> It overwrites a nearby variable. sp + 8
On 09/18/2017 10:09 PM, Eric Botcazou wrote:
I'm not sure anyone really cares so it's up to you I'd say.
Ok, thanks. Done! Committed as r252971.
--
Pierre-Marie de Rodat
On 09/18/2017 03:05 PM, Jakub Jelinek wrote:
> On Mon, Sep 18, 2017 at 03:01:53PM +0200, Martin Liška wrote:
>> As discussed here:
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81224
>>
>> We have fallout caused by the patch and it's backport to active branches.
>> I'm planning to revert the
On Mon, Sep 18, 2017 at 06:10:29PM -0500, Daniel Santos wrote:
> Mike, can you take a look at this please?
>
> On 09/18/2017 10:17 AM, Dominique d'Humières wrote:
> > This patch (r252896) breaks bootstrap on x86_64-apple-darwin10 configured
> > with
> >
> > ../work/configure
On Tue, Sep 19, 2017 at 1:10 AM, Daniel Santos wrote:
> Mike, can you take a look at this please?
>
> On 09/18/2017 10:17 AM, Dominique d'Humières wrote:
>> This patch (r252896) breaks bootstrap on x86_64-apple-darwin10 configured
>> with
>>
>> ../work/configure
On lundi 18 septembre 2017 13 h 20 min 25 s CEST Martin Sebor wrote:
> On 09/18/2017 12:26 PM, Steve Ellcey wrote:
> > This patch is for PR target/79868, where some aarch64 diagnostics are
> > said to be not translatable due to how they are implemented. See the
> > bug report for more details on
Richard Biener writes:
> On Mon, 18 Sep 2017, Richard Sandiford wrote:
>> Richard Biener writes:
>> > The following is said to fix a 482.sphinx3 regression.
>> >
>> > Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
>> >
>> > Richard.
>> >
>> >
OK, thanks.
On Tue, Sep 19, 2017 at 9:03 AM, Jakub Jelinek wrote:
> Hi!
>
> In make check-c++-all I've noticed some UNSUPPORTED tests and some
> failures, the following patch attempts to fix those.
> The failures are due to new not emitting operator new (...) != NULL
>
Hi,
This patch fixes the aarch64 bug 81422
https://gcc.gnu.org/PR81422
Before adding REG_EQUIV notes in the TLS symbol handling code,
we should check whether the "dest" is a REG or NOT (sometimes,
it's a SUBREG as in this bug). Only when the “dest” is a REG, the note
will be added.
a new small
On Tue, Sep 19, 2017 at 10:26:52PM +0200, Dominique d'Humières wrote:
>
> I am really upset by the time spent on warning at the expense
> of more serious problems.
>
https://tinyurl.com/y6wunnk9
https://gcc.gnu.org/ml/fortran/2017-09/msg00065.html
or this is a consequence
On Tue, Sep 19, 2017 at 04:11:52PM +0200, Jakub Jelinek wrote:
> Will try now following plus testcase, the rest of constants I believe end up
> being DW_FORM_block encoded and so is pretty much what we emit even for the
> initializer_constant_valid_p tree fallback case.
I've
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
> ow...@gcc.gnu.org] On Behalf Of Uros Bizjak
> Sent: Tuesday, September 19, 2017 6:13 PM
> To: Tsimbalist, Igor V
> Cc: gcc-patches@gcc.gnu.org
> Subject: Re:
On Tue, 19 Sep 2017, Martin Sebor wrote:
> > In general, the data structures where you need to ensure manually that if
> > attribute A is listed in EXCL for B, then attribute B is also listed in
> > EXCL for A, seem concerning. I'd expect either data structures that make
> > such asymmetry
With the 32-bit SVR4 ABI we don't have a red zone, so we have to restore
the callee-saved registers before we restore the stack pointer.
The previous fix for this PR failed in two ways, for huge frames: first,
we use a negative offset from r11 in that case, so the (mem:BLK 11) access
does no
Hi Kelvin,
This is in quite good shape. In addition to Segher's comments, I have a few
questions/requests below.
> On Sep 15, 2017, at 4:04 PM, Kelvin Nilsen
> wrote:
>
> @@ -385,6 +396,25 @@ const_load_sequence_p (swap_web_entry *insn_entry,
>
Hello!
As mentioned in PR 82259. These are similar to existing BT jcc patterns.
2017-09-19 Uros Bizjak
* config/i386/i386.md (*scc_bt): New insn_and_split pattern.
(*scc_bt_1): Ditto.
(*scc_bt_mask): Ditto.
testsuite/ChangeLog:
2017-09-19 Uros Bizjak
On 09/19/2017 07:06 AM, Alexander Monakov wrote:
Hi,
After recent changes, the member_name_cmp qsort comparator can indicate
A < B < A (i.e. lacks anti-commutativity) for distinct TYPE_DECL nodes
that have the same source location. If their order doesn't matter, the
comparator should return 0.
Hello,
The autopref_rank_for_schedule qsort sub-comparator is not actually a proper
comparator. For instance, it lacks transitivity: if there's insns A, B, C
such that B has AUTOPREF_MULTUPASS_DATA_IRRELEVANT status, but A and C
compare such that C < A, we can have A == B == C < A according to
On Tue, Sep 19, 2017 at 8:25 AM, Kugan Vivekanandarajah
wrote:
> Hi Richard,
>
> On 18 September 2017 at 17:50, Richard Biener
> wrote:
>> On Mon, Sep 18, 2017 at 3:36 AM, Kugan Vivekanandarajah
>>
On 09/19/2017 07:25 AM, Alexander Monakov wrote:
On Tue, 19 Sep 2017, Nathan Sidwell wrote:
On 09/19/2017 07:06 AM, Alexander Monakov wrote:
Hi,
After recent changes, the member_name_cmp qsort comparator can indicate
A < B < A (i.e. lacks anti-commutativity) for distinct TYPE_DECL nodes
that
Hi,
After recent changes, the member_name_cmp qsort comparator can indicate
A < B < A (i.e. lacks anti-commutativity) for distinct TYPE_DECL nodes
that have the same source location. If their order doesn't matter, the
comparator should return 0.
Invoking qsort with improper comparator at best
On Tue, 19 Sep 2017, Nathan Sidwell wrote:
> On 09/19/2017 07:06 AM, Alexander Monakov wrote:
> > Hi,
> >
> > After recent changes, the member_name_cmp qsort comparator can indicate
> > A < B < A (i.e. lacks anti-commutativity) for distinct TYPE_DECL nodes
> > that have the same source location.
The following fixes a bug in VRP where on removal of asserts it ended
up propagating a constant in place of abnormal SSA uses. That doesn't
work when the use is in an abnormal PHI thus instead replace the
assert with a copy in that case.
Bootstrapped and tested on x86_64-unknown-linux-gnu,
Daniel Santos writes:
> On 09/17/2017 10:53 AM, Uros Bizjak wrote:
>> OK.
>>
>> Thanks,
>> Uros.
>
> Thanks. I should have posted this Friday when my tests finished, but
> I'll be committing with one minor change so tests don't run on m32 or mx32:
>
> ---
This patch introduces the new option -fstack-clash-protection and a
couple params that can be used to control its behavior. I think the
only change of significance since V3 is the params are in powers of 2
rather than byte counts per Richi's request.
The controlling predicate for the testsuite
The only change since V3 of this patch was to fix a typo in the sparc.c
changes.
The purpose of this patch is to allow targets that are not going to have
stack-clash protected prologues to use their -fstack-check=specific
prologue support to provide a degree of protection for their prologue
This patch introduces a new note we can add to insns. REG_STACK_CHECK
which is added to a stack allocation to tell to the optimizers that the
allocation should not be combined with other allocations or otherwise
modified. Since V3 the scheduler was modified to honor REG_STACK_CHECK
as well.
The main purpose of this patch is to protect dynamic stack allocations
from stack-clash attacks. I think the only notable change since V3 was
to use a target hook instead of a target macro to control a bit of
special behavior we're going to want on aarch64.
Bootstrapped and regression tested on
On Mon, Sep 18, 2017 at 4:13 PM, Ian Lance Taylor wrote:
> On Mon, Sep 18, 2017 at 3:24 PM, Ian Lance Taylor wrote:
>> I merged revision 252949 from trunk to the gccgo branch.
>
> Missed a patch. Merged revision 252954 to the gccgo branch.
And now merged
This patch just provides some generic logging facilities that will be
used by the backends to verify their behavior at a high level. I don't
think it changed since V3.
This patch was bootstrapped and regression tested with the next patch
(x86 bits). It was not bootstrapped and regression
This patch introduces the x86 stack clash protected prologue support as
well as the tests. I believe the only change since V3 was the more
aggressive introduction of scheduling barriers. It also enables the
stack-clash tests for x86 targets.
The scheduling barriers prevent the memory
Hi!
The following patch implements P0386R1 - NSDMIs for bit-fields.
While working on that, I've discovered our parser mishandles attributes
on bitfields, already C++11 says:
identifier[opt] attribute-specifier-seq[opt] : constant-expression
in the grammar, but we actually parsed
identifier[opt] :
On Tue, Sep 19, 2017 at 08:59:23AM -0400, Tim Song wrote:
> On Tue, Sep 19, 2017 at 8:54 AM, Jakub Jelinek wrote:
> > Hi!
> >
> > The following patch implements P0386R1 - NSDMIs for bit-fields.
>
> It's P0683R1.
Oops, got it right in the testcases, just not in the ChangeLog
Here is an updated patch (version #2). The main differences are:
- Change attribute and option names;
- Add additional parameter to gimple_build_call_from_tree by adding a type
parameter and
use it 'nocf_check' attribute propagation;
- Reimplement fixes in expand_call_stmt to propagate
On Thu, Oct 13, 2016 at 09:16:07AM +0200, Richard Biener wrote:
>
> This merges a few more bits guarding stuff with ! early_dwarf, mostly
> to avoid creating locations that involve addresses of decls early
> but also to avoid wasting work for BLOCK_NONLOCALIZED_VARs.
>
> Bootstrapped and tested
On Tue, 19 Sep 2017, Jakub Jelinek wrote:
> On Thu, Oct 13, 2016 at 09:16:07AM +0200, Richard Biener wrote:
> >
> > This merges a few more bits guarding stuff with ! early_dwarf, mostly
> > to avoid creating locations that involve addresses of decls early
> > but also to avoid wasting work for
On Tue, 19 Sep 2017, Maxim Kuvyrkov wrote:
> How about the following:
> 1. if both instructions are "irrelevant", then return "0".
> 2. if one instruction is "relevant" and another is "irrelevant", then
> "relevant" instruction is always greater (or lesser) than the non-relevant.
> 3. if both
Hi!
In make check-c++-all I've noticed some UNSUPPORTED tests and some
failures, the following patch attempts to fix those.
The failures are due to new not emitting operator new (...) != NULL
comparison with -std=c++17 anymore, so there is nothing to fold away
(first 2 hunks). The rest is about
Ah thanks!
From: Martin Liška
Sent: Tuesday, September 19, 2017 2:31 PM
To: Tamar Christina; Dmitry Vyukov; Kostya Serebryany
Cc: 吴潍浠(此彼); Jakub Jelinek; Jeff Law; wishwu007; Alexander Potapenko;
andreyknvl; nd; GCC Patches
Subject: Re:
On Tue, Sep 19, 2017 at 04:09:20PM +0200, Richard Biener wrote:
> > The problem is that add_const_value_attribute has lots of smarts to handle
> > various kinds of constants, which the
> > if (CHAR_BIT == 8 && BITS_PER_UNIT == 8
> > && initializer_constant_valid_p (init, type))
> > block
Hi!
(Sorry for the resent, forgot to CC gcc-patches).
Another easy C++2A patch. Here I wonder if we don't want
a -Wc++2a-compat warning and whether & (const | volatile)) == const
is best (for the case if there are other quals in the future),
or if we just want == const.
For the compat warning,
Adding the list back in.
From: Tamar Christina
Sent: Tuesday, September 19, 2017 2:11 PM
To: Dmitry Vyukov; Kostya Serebryany
Cc: 吴潍浠(此彼); Jakub Jelinek; Jeff Law; wishwu007; Alexander Potapenko;
andreyknvl; nd
Subject: Re: Add support to trace comparison
Uros Bizjak writes:
>> However, they do FAIL on 64-bit Solaris/x86:
>>
>> +FAIL: gcc.target/i386/pr82196-1.c (test for excess errors)
>>
>> Excess errors:
>> /vol/gcc/src/hg/trunk/local/gcc/testsuite/gcc.target/i386/pr82196-1.c:14:1:
>> error: bp cannot be used in asm here
>>
On Tue, Sep 19, 2017 at 8:54 AM, Jakub Jelinek wrote:
> Hi!
>
> The following patch implements P0386R1 - NSDMIs for bit-fields.
It's P0683R1.
On Tue, 19 Sep 2017, Nathan Sidwell wrote:
> > > > After recent changes, the member_name_cmp qsort comparator can indicate
> > > > A < B < A (i.e. lacks anti-commutativity) for distinct TYPE_DECL nodes
> > > > that have the same source location. If their order doesn't matter, the
> > > >
On 09/19/2017 03:14 PM, Tamar Christina wrote:
> it's fine at O1, O2 and O3 though. Should the test be running for O0?
It's a known issue:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82183
Martin
Here is an updated patch (version #2). Mainly attribute and option names were
changed.
The test for ICF will be introduced in x86 specific tests (patch 0006-Part-6)
as the implementation
checks if the CF instrumentation is on to adjust a hash based on 'nocf'_check'
attribute presence.
In
On Tue, Sep 19, 2017 at 2:13 PM, Rainer Orth
wrote:
> Daniel Santos writes:
>
>> On 09/17/2017 10:53 AM, Uros Bizjak wrote:
>>> OK.
>>>
>>> Thanks,
>>> Uros.
>>
>> Thanks. I should have posted this Friday when my tests finished, but
>> I'll
> On Sep 19, 2017, at 2:21 PM, Alexander Monakov wrote:
>
> Hello,
>
> The autopref_rank_for_schedule qsort sub-comparator is not actually a proper
> comparator. For instance, it lacks transitivity: if there's insns A, B, C
> such that B has
Here is an updated patch (version #2). Mainly attribute and option names were
changed.
gcc/doc/
* extend.texi: Add 'nocf_check' documentation.
* gimple.texi: Add second parameter to gimple_build_call_from_tree.
* invoke.texi: Add -fcf-protection documentation.
*
Uros
Please, find the new patch attached.
Previously, I thought to change integer part of the " ix86_preferred_simd_mode"
function in the same way
as it done with floating point part. I haven't done it to disturb less code by
the patch.
I completely agree with your proposal, currently the code
The following forces scev analyzable SESE liveouts to be handed as
defs. Otherwise we end up forgetting to code-generate them. We
naturally expect SCEV-cprop to handle them but that pass has cost
cutoffs (and one can disable it).
Bootstrap and regtest running on x86_64-unknown-linux-gnu, ok?
Hi Richard,
On 18 September 2017 at 17:50, Richard Biener
wrote:
> On Mon, Sep 18, 2017 at 3:36 AM, Kugan Vivekanandarajah
> wrote:
>> Hi Richard,
>>
>> On 15 September 2017 at 19:31, Richard Biener
>>
On Mon, 18 Sep 2017, Richard Sandiford wrote:
> Richard Biener writes:
> > The following is said to fix a 482.sphinx3 regression.
> >
> > Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
> >
> > Richard.
> >
> > 2017-09-18 Richard Biener
> >
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
> ow...@gcc.gnu.org] On Behalf Of Uros Bizjak
> Sent: Monday, September 18, 2017 9:10 PM
> To: Peryt, Sebastian
> Cc: gcc-patches@gcc.gnu.org; Kirill Yukhin
On 09/19/2017 09:54 AM, Steve Ellcey wrote:
On Tue, 2017-09-19 at 09:50 +0200, Frédéric Marchal wrote:
On lundi 18 septembre 2017 13 h 20 min 25 s CEST Martin Sebor wrote:
I haven't looked at all of them but from the few I have seen it
seems that rephrasing the messages along the following
On Tue, 19 Sep 2017, Andreas Schwab wrote:
> On Sep 18 2017, Joseph Myers wrote:
>
> > Building glibc for many different configurations and running the
> > compilation parts of the testsuite runs into failures of the
> > elf/check-execstack test for hppa, ia64 and
On 19/09/17 15:38 +0100, Jonathan Wakely wrote:
On 18/09/17 16:54 -0700, Tim Shen wrote:
On Mon, Sep 18, 2017 at 4:01 PM, Jonathan Wakely wrote:
On 18/09/17 10:58 -0700, Tim Shen via libstdc++ wrote:
On Mon, Sep 18, 2017 at 10:26 AM, Jonathan Wakely
Hi,
https://gcc.gnu.org/PR82255 identifies a problem in the vector cost model
where a vectorized load is treated as having the cost of a strided load
in a case where we will not actually generate a strided load. This is
simply a mismatch between the conditions tested in the cost model and
those
On Tue, Sep 19, 2017 at 4:00 PM, Shalnov, Sergey
wrote:
> Uros
> Please, find the new patch attached.
> Previously, I thought to change integer part of the "
> ix86_preferred_simd_mode" function in the same way
> as it done with floating point part. I haven't done it to
On 18/09/17 16:54 -0700, Tim Shen wrote:
On Mon, Sep 18, 2017 at 4:01 PM, Jonathan Wakely wrote:
On 18/09/17 10:58 -0700, Tim Shen via libstdc++ wrote:
On Mon, Sep 18, 2017 at 10:26 AM, Jonathan Wakely
wrote:
We need to rewrite this to check the
On 09/19/2017 03:08 AM, Andreas Schwab wrote:
> On Sep 18 2017, Jeff Law wrote:
>
>> On 09/18/2017 10:09 AM, Andreas Schwab wrote:
>>> On Sep 18 2017, Jeff Law wrote:
>>>
Can you confirm if the probe was in the red zone vs the live areas on
the stack?
On 09/18/2017 03:44 PM, Joseph Myers wrote:
On Mon, 18 Sep 2017, Martin Sebor wrote:
It's meant as an escape hatch. It allows declaring compatibility
symbols, for example by the libstdc++ _GLIBCXX_3_4_SYMVER macro
defined in libstdc++-v3/src/c++98/compatibility.cc. The macro is
used to
On Tue, Sep 19, 2017 at 5:18 PM, Tsimbalist, Igor V
wrote:
>> -Original Message-
>> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
>> ow...@gcc.gnu.org] On Behalf Of Uros Bizjak
>> Sent: Monday, September 18, 2017 12:17 PM
>> To:
The failures that need to be fixed can be seen at
https://gcc.gnu.org/ml/gcc-testresults/2017-09/msg01633.html
Uros, thank you for the approval. Based on the approval of the first 3 patches
(I've submitted them today), I need to adjust option and attribute names. I
will resubmit the patch when I fix option and attribute names.
Thanks,
Igor
> -Original Message-
> From: Uros Bizjak
> On Sep 19, 2017, at 5:25 PM, Alexander Monakov wrote:
>
> On Tue, 19 Sep 2017, Maxim Kuvyrkov wrote:
>> How about the following:
>> 1. if both instructions are "irrelevant", then return "0".
>> 2. if one instruction is "relevant" and another is "irrelevant", then
>>
> On Sep 19, 2017, at 6:20 PM, Alexander Monakov wrote:
>
>> I'd like to keep read/write processing balanced. In the above "read"
>> analysis
>> has greater weight than "write" analysis. Also, autopref_rank_data() should
>> not be called if !rtx_equal_p (data1->base,
On Fri, Aug 04, 2017 at 02:21:29PM +0200, Richard Biener wrote:
> ! /* Initialize the various sections and labels for dwarf output. */
>
> static void
> ! init_sections_and_labels (void)
...
These changes broke DWARF-5 support. E.g. in gcc-7 and before this change
there was:
> ! if
On 09/19/2017 12:17 AM, Andreas Schwab wrote:
On Sep 18 2017, Joseph Myers wrote:
Building glibc for many different configurations and running the
compilation parts of the testsuite runs into failures of the
elf/check-execstack test for hppa, ia64 and microblaze.
On Tue, Sep 19, 2017 at 3:30 AM, Richard Biener wrote:
>
> The following plugs some holes in extract_affine which failed
> to modulo-reduce signed values in conversions, failed to
> interpret pointer-plus offsets as signed (yeah, stupid GIMPLE IL...)
> and mishandled
I noticed this typo while working on PR77729. Trivial, obvious, and
committed.
Segher
2017-09-17 Segher Boessenkool
* simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
---
gcc/simplify-rtx.c | 2 +-
1 file changed, 1 insertion(+), 1
I forgot to make is_nothrow_invocable_r check whether the conversion
to the result type can throw. Fixed like so.
Tested powerpc64le-linux, committed to trunk. gcc-7 backport to
follow.
commit 067cdd4fba31f82316f71954335b622e3b6da58a
Author: Jonathan Wakely
Date: Tue Sep
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
> ow...@gcc.gnu.org] On Behalf Of Uros Bizjak
> Sent: Monday, September 18, 2017 12:17 PM
> To: gcc-patches@gcc.gnu.org
> Cc: Tsimbalist, Igor V ; Tsimbalist, Igor V
>
> I'd like to keep read/write processing balanced. In the above "read" analysis
> has greater weight than "write" analysis. Also, autopref_rank_data() should
> not be called if !rtx_equal_p (data1->base, data2->base).
I'm afraid this doesn't work. Consider you have insns A, B, C such that all
Ping #3: https://gcc.gnu.org/ml/gcc-patches/2017-08/msg00912.html
Thanks
Martin
On 08/28/2017 08:34 PM, Martin Sebor wrote:
Ping #2: https://gcc.gnu.org/ml/gcc-patches/2017-08/msg00912.html
On 08/23/2017 01:46 PM, Martin Sebor wrote:
Ping:
On Mon, Sep 18, 2017 at 1:48 PM, Mike Stump wrote:
> I was hoping an RM would approve this as it seems just a hair beyond a normal
> darwin approval. I'm fine with this, and it does help darwin. Other ports
> should not care.
Mike,
Unless I am misreading this...
-- sorry for the duplicate, forgot to post to list as well first time --
Hi Richard,
The testcase seems to fail on aarch64-none-elf when -O1 or -O2,
-O0, -Os and -O3 seem to work fine.
dc[0] ends up being 0 for the cases that fail.
Kind regards,
Tamar
On Tue, Sep 19, 2017 at 9:17 AM, Richard Biener wrote:
>
> The following forces scev analyzable SESE liveouts to be handed as
> defs. Otherwise we end up forgetting to code-generate them. We
> naturally expect SCEV-cprop to handle them but that pass has cost
> cutoffs (and
On Tue, 2017-09-19 at 09:50 +0200, Frédéric Marchal wrote:
> On lundi 18 septembre 2017 13 h 20 min 25 s CEST Martin Sebor wrote:
> >
> > I haven't looked at all of them but from the few I have seen it
> > seems that rephrasing the messages along the following lines would
> > be a way to get
Hello!
"Convert TLS location to DEFAULT_TLS_SEG_REG address space" patch [1]
allows to remove special handling of LEA addresses from
ix86_split_long_move. The address space is the property of MEM RTX,
not the address itself.
2017-09-19 Uros Bizjak
* config/i386/i386.c
100 matches
Mail list logo