On Mon, Jan 21, 2019 at 6:03 PM Joseph Myers wrote:
>
> On Mon, 21 Jan 2019, H.J. Lu wrote:
>
> > The testcase at
> >
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88931
> >
> > with -frounding-math. __floattisf and __floattidf from libgcc2.c give
> > the wrong results for FE_UPWARD and FE_TOW
On Tue, 22 Jan 2019, Martin Sebor wrote:
> PS In GCC 10, unless there is an important use case that escapes
> me, I think GCC should warn for zero-length non-member array
> objects, or perhaps even for internal struct members (those followed
> by another member). Not to avoid these sorts of bugs
The enhancement to treat char initializer-lists as STRING_CSTs
committed earlier last year introduced an assumption that array
elements have a non-zero size. As it turns out, the zero-length
array extension that makes it possible to define even multi-
dimensional zero-length array objects breaks
On Tue, 22 Jan 2019, Richard Biener wrote:
> > Or instead just come up with target specific strings to determine the ABI,
> > say i386, x86_64 and x32 for the 3 ABIs on x86, powerpc{,64}{,le} on rs6000
> > etc.
>
> Yeah, I would even suggest to use a target hook multilib_ABI_active_p
> (const cha
A libgomp test demonstrated a problem with reduction code in the Power back end
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87064) for little-endian code.
The
define_insn_and_split at the heart of the problem depends on big-endian data
layout,
and should have been disabled when little-endian
It is enabled at -O3, but the doc currently says -O2. This fixes it.
Committing to trunk.
Segher
2019-01-22 Segher Boessenkool
* doc/invoke.texi (-fsplit-paths): This is enabled by default at -O3.
---
gcc/doc/invoke.texi | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff
Hi Steve,
No we are using aarch64_be-*-* but this is the only one that popped up as a
failure which is why I didn't change the others.
But now I'm wondering why... I'll check the log file manually tomorrow to be
sure.
Cheers,
Tamar
From: Steve Ellcey
Se
On Tue, Jan 22, 2019 at 2:03 PM Tom de Vries wrote:
>
> [ was: Re: [PATCH 7/9] [libbacktrace] Handle DW_FORM_GNU_ref_alt ]
> On 18-01-19 15:25, Ian Lance Taylor wrote:
> > On Thu, Jan 17, 2019 at 6:14 AM Tom de Vries wrote:
> >>
> >> On 17-01-19 01:35, Ian Lance Taylor wrote:
> >>> On Wed, Jan 16
On Mon, 2019-01-21 at 18:00 +, Tamar Christina wrote:
>
> > That would need to be aarch64*-*-* to include big-endian. Fixing that here
> > and in the other tests is OK under the obvious rule.
>
> Ah, true, I didn't look at the testcase. I tested the ILP32 case and
> committed the fix for bot
Hi!
As mentioned in the PR, gimple_build can return some trees that aren't valid
in MEM_REF's operand (in particular, SSA_NAMEs are allowed there, but
only a subset of invariants). The following patch forces the address into
a SSA_NAME if it fails the MEM_REF operand's predicate.
Bootstrapped/re
Hi!
BIT_INSERT_EXPR is only defined for integral and vector types, so if we have
an array type which is bitfield representative if there is no integral type,
this fails to gimplify. This still uses BIT_INSERT_EXPR with those types
to hold the expressions between c-omp.c and gimplification, but fi
Hi!
SCEV can analyze not just integral/pointer IVs, but (scalar) float ones as
well. Calling build_int_cst on such types results in ICE, build_zero_cst
works. Though the loop invariant PHI IVs, if we represent them as using
+0.0 step, aren't correct if honoring signed zeros, as + 0.0 will make
a
Hi!
The following testcase ICEs, because we assert that if during parsing we
haven't found any break; stmts in SWITCH_STMT_BODY, the switch statement
break label will not be used. It is used in this case though, because
while during parsing we expect break stmts only from switch body to go to
the
Hi!
This patch adjusts genericize_cp_loop similarly to the
PR88984 change in genericize_switch_stmt, so that the following testcase now
behaves the same in both C and C++, plus adds the testcase and
documentation.
Bootstrapped/regtested on {x86_64,i686,powerpc64{,le}}-linux, ok for trunk?
2019-0
On Wed, Jan 16, 2019 at 2:32 PM Jason Merrill wrote:
>
> On 1/16/19 4:48 PM, Nathan Sidwell wrote:
> > This PR reports a bug where we select a non-const operator function and
> > then apply it to a const object. That's happening because the
> > expression 'c[0]' is not dependent, so we figure end
[ was: Re: [PATCH 7/9] [libbacktrace] Handle DW_FORM_GNU_ref_alt ]
On 18-01-19 15:25, Ian Lance Taylor wrote:
> On Thu, Jan 17, 2019 at 6:14 AM Tom de Vries wrote:
>>
>> On 17-01-19 01:35, Ian Lance Taylor wrote:
>>> On Wed, Jan 16, 2019 at 4:17 PM Tom de Vries wrote:
this handles DW_FO
On 19-01-19 01:54, Ian Lance Taylor wrote:
> On Fri, Jan 18, 2019 at 4:45 PM Tom de Vries wrote:
>>
>> On 18-01-19 15:23, Ian Lance Taylor wrote:
>>> On Thu, Jan 17, 2019 at 5:59 AM Tom de Vries wrote:
now that the rest of the patch series has been committed, here's an
updated vers
Hi Harald,
here's my revised patch for the treatment of (2**e) ** n and (- 2**e) **
n, trying to take into account the bitwidth of the result.
I've also extended the testcase to sample different integer kinds.
ChangeLogs see below.
Note: I don't have commit rights, so please somebody else tak
On Tue, 22 Jan 2019 at 11:08, Andreas Schwab wrote:
>
> In file included from ../../gcc/d/d-system.h:23,
> from ../../gcc/d/dmd/root/dsystem.h:24,
> from ../../gcc/d/dmd/mtype.c:11:
> ../../gcc/d/dmd/mtype.c: In member function 'Identifier*
> Type::getTypeInfoIde
The above PR had a partial fix by Tobias Burnus for a valid use of
STORAGE_SIZE, but did not properly handle an error situation.
The attached simple fix adds a list of allowed intrinsics for use
in specification inquiries in F2008+. Testcase by Tobias.
Regtested on x86_64-pc-linux-gnu.
OK for t
On Mon, Jan 21, 2019 at 03:14:53PM -0500, Jason Merrill wrote:
> On 1/18/19 9:12 AM, Marek Polacek wrote:
> > On Thu, Jan 17, 2019 at 04:17:29PM -0500, Jason Merrill wrote:
> > > On 1/17/19 2:09 PM, Marek Polacek wrote:
> > > > This patch ought to fix the rest of 78244, a missing narrowing warning
Hello world,
the attached patch moves the packing / unpacking of arrays to the front
end when optimizing, but not for size.
Rationale: internal_pack and internal_unpack are opaque to the compiler.
This can lead to a lot of information loss for inlining and inter-
procedural optimization, and in
On Tue, 22 Jan 2019 at 17:29, Maya Rashish wrote:
>
> libphobos/libdruntime changes were contributed upstream:
> https://github.com/dlang/druntime/pull/2472
> (caveat: pending a change to netbsd/execinfo.d)
>
> One missing patch is needed for GDC to work.
>
> gcc/config.gcc (*-*-netbsd*): add netb
This patch simplifies the expression sinh (x) / cosh (x) to tanh (x).
This rule is mathematically valid.
There's a slight difference in the result when applying this
optimization with x in the interval 0 < x <= 1e-4951. With the
optimization, the result using long double is -0 and without the
opti
Hello!
RTL store motion pass transforms:
(insn 19 18 20 4 (parallel [
(set (mem/c:SI (symbol_ref:SI ("bar_arg") [flags 0x2]
) [1 bar_arg+0 S4 A32])
(fix:SI (reg:DF 89)))
(clobber (scratch:XF))
]) "stdarg-3.c":63:7 156 {fix_truncsi_i387_fisttp}
Hi Jakub,
On Tue, Jan 22, 2019 at 05:31:58PM +0100, Jakub Jelinek wrote:
> The following patch disables *vsx_reduc__v2df_scalar
> define_insn_and_split for little endian, because it both generates
> wrong code there (fixable), but also even if fixed, emit much worse code
> than when it is not pres
On nios2, PIC relocations are defined only for the GNU/Linux ABI, so
nios2-elf gives a "sorry" with -fpic. That was causing some failures in
some new testcases explicitly passing -fpic or -fPIC in dg-options.
I've fixed these tests by adding dg-require-effective-target fpic, which
is how I've
The nios2-elf target defaults to -fno-delete-null-pointer-checks, at the
request of Altera; they wanted to be able to manipulate pointers to
memory at address zero. From time to time I find new tests that have an
implicit dependence on -fdelete-null-pointer-checks, which I've been
fixing for t
On 22/01/2019 14:50, Jakub Jelinek wrote:
> On Tue, Jan 22, 2019 at 02:10:59PM +, Richard Earnshaw (lists) wrote:
>> @@ -6630,6 +6633,13 @@ arm_needs_doubleword_align (machine_mode mode,
>> const_tree type)
>> Make sure we can warn about that with -Wpsabi. */
>>ret = -1;
>>
Hi!
I've committed following addition to porting_to.html to document another
user visible change.
--- htdocs/gcc-9/porting_to.html11 Jan 2019 18:21:45 - 1.1
+++ htdocs/gcc-9/porting_to.html22 Jan 2019 17:40:12 -
@@ -56,6 +56,61 @@ and provide solutions. Let us know if
Fix a failing test - changes in Combine mean the test now fails
eventhough the generated code is the same. Given there are several
AArch64-specific tests for vec-select, remove the scanning of Combine
output. Committed as trivial fix.
ChangeLog:
2019-01-22 Wilco Dijkstra
PR rtl-opti
PR driver/88911 reports an issue where the user couldn't remember the spelling
of the "-dumpspecs" option, and attempted various similar spellings, but many
of them failed to contain a suggestion of the correct spelling:
$ gcc q.c -dump-spec
cc1: warning: unrecognized gcc debugging option: u
I have committed the following patch to the gcc-8-branch as r268158 after a
regtest.
Dominique
Index: gcc/testsuite/ChangeLog
===
--- gcc/testsuite/ChangeLog (revision 268156)
+++ gcc/testsuite/ChangeLog (working copy)
@@
Hi!
The following patch disables *vsx_reduc__v2df_scalar
define_insn_and_split for little endian, because it both generates
wrong code there (fixable), but also even if fixed, emit much worse code
than when it is not present at all.
The comment on the pattern says:
;; Combiner patterns with the ve
libphobos/libdruntime changes were contributed upstream:
https://github.com/dlang/druntime/pull/2472
(caveat: pending a change to netbsd/execinfo.d)
One missing patch is needed for GDC to work.
gcc/config.gcc (*-*-netbsd*): add netbsd-d.o
gcc/config/t-netbsd: add netbsd-d.o
gcc/d/d-builtins.cc:
Messages printed to stdout might be lost when the process terminates
via abort(), because output streams aren't flushed. Write to stderr
instead (if stderr is defined by ).
PR libstdc++/88740
* testsuite/util/testsuite_hooks.h [stderr] (VERIFY): Use fprintf to
write to std
On Tue, Jan 22, 2019 at 02:43:38PM +, Richard Earnshaw (lists) wrote:
> PR target/88469
> * profile-count.h (profile_count): Add dummy file with 64-bit alignment
> on arm-based systems using gcc-6/7/8.
>
> diff --git a/gcc/profile-count.h b/gcc/profile-count.h
> index c83fa3
On 22/01/2019 15:20, Richard Biener wrote:
> On Tue, 22 Jan 2019, Richard Earnshaw (lists) wrote:
>
>> This patch, for gcc 8/9 is a mitigation patch for PR target/88469 where
>> gcc-6/7/8 miscompile a structure whose alignment is dominated by a
>> 64-bit bitfield member. Since the PCS rules for s
On Tue, Jan 22, 2019 at 12:46 AM Uros Bizjak wrote:
>
> > I have committed a patch to update libgo to the Go 1.12beta2 release.
> >
> > As usual this sort of update is too large to include all changes in
> > this e-mail. I've included changes to gccgo-specific files below.
> >
> > Bootstrapped an
Am 22.01.2019 um 14:21 schrieb Dominique d'Humières:
>
>
>> Le 22 janv. 2019 à 10:02, Manfred Schwarb a écrit :
>>
>> Dominique, thanks a lot.
>>
>> I got the request to share my script, so I attached it to this mail.
>> Be aware that the script reports numerous false positives.
>> As one can ex
On Tue, 22 Jan 2019, Richard Earnshaw (lists) wrote:
> This patch, for gcc 8/9 is a mitigation patch for PR target/88469 where
> gcc-6/7/8 miscompile a structure whose alignment is dominated by a
> 64-bit bitfield member. Since the PCS rules for such a type must ignore
> any overalignment of the
Hi Tom,
the nvptx parts are mostly still the same since the last update, except the
GOMP_OFFLOAD_version() hook has been updated to return the new
GOMP_PLUGIN_IF_VERSION symbol.
Thanks,
Chung-Lin
* plugin/plugin-nvptx.c (struct cuda_map): Remove.
(struct ptx_stream): Remove.
(struct
Hi Thomas,
These are the testsuite changes. After the last review round, the amount of
additional changes have been greatly reduced.
Thanks,
Chung-Lin
* testsuite/libgomp.oacc-c-c++-common/asyncwait-2.c: New testcase.
* testsuite/libgomp.oacc-c-c++-common/lib-81.c: Adjust testcase.
*
Hi Thomas,
This patch adds the OpenACC 2.5 new APIs of
acc_get_default_async/acc_set_default_async,
contained are the modifications for Fortran in libgomp/openacc.f90,
libgomp/openacc_lib.h.
Not much is changed since the last submission of this part, except that the
libgomp.map
part is moved t
Hi Jakub,
Since the last revision, the main change is that I've added at your request,
__builtin_expect (aq != NULL, 0) as the if-condition in
gomp_copy_host2dev/dev2host.
Stashing the aq pointer into the coalesce-buf didn't match very well, as
the cbuf argument was sometimes null and didn't alw
Hi Thomas,
in this patch, the issue on the synchronize/serialize hooks should be resolved;
all call sites should now handle the failing case.
As for the host _construct hook issue, I went your suggestion and used 0x...
as the dummy, the NULL pointer is still used for error here. I mentioned o
This patch updates the 'version' libgomp plugin hook for intelmic and hsa to
return GOMP_PLUGIN_IF_VERSION instead of GOMP_VERSION.
Thanks,
Chung-Lin
liboffloadmic/
* plugin/libgomp-plugin-intelmic.cpp (GOMP_OFFLOAD_version):
Update to return GOMP_PLUGIN_IF_VERSION.
libgomp/
This patch separates out the header interface changes.
A new GOMP_PLUGIN_IF_VERSION symbol has been defined in
include/gomp-constants.h,
specifically for versioning the plugin-interface. Corresponding places in
libgomp
has been updated, leaving GOMP_VERSION* only for the libgomp proper/offload
Hi, this is a rebase to current trunk and re-submission of the OpenACC Async
re-organization work, aiming to commit when stage1 re-opens. This is technically
the 2nd time I'm sending this whole patch series, but because I've named
partial revisions up to v4 by now, for clarity I will just call thi
On Tue, Jan 22, 2019 at 02:10:59PM +, Richard Earnshaw (lists) wrote:
> @@ -6630,6 +6633,13 @@ arm_needs_doubleword_align (machine_mode mode,
> const_tree type)
>Make sure we can warn about that with -Wpsabi. */
> ret = -1;
>}
> +else if (TREE_CODE (field) == F
On Tue, Jan 22, 2019 at 6:10 AM Bernd Edlinger
wrote:
>
> Hi,
>
> unfortunately there are some more issues with -Waddress-of-packed-member,
> that show that my previous
> patch was not yet complete.
>
> That is a bogus warning, in C and C++ (see addition in test case 1), and a
> couple of missin
On Tue, Jan 22, 2019 at 1:27 AM Uros Bizjak wrote:
>
> On Tue, Jan 22, 2019 at 7:57 AM Hongtao Liu wrote:
>
> > This is adjusted patch: Modify BDESC to accept two flags and merge
> > ARGS2/SPEICAL_ARGS2 into ARGS/SPECIAL_ARGS.
> > Bootstrap test is ok, no regression on x86 backend testcases.
> >
This patch, for gcc 8/9 is a mitigation patch for PR target/88469 where
gcc-6/7/8 miscompile a structure whose alignment is dominated by a
64-bit bitfield member. Since the PCS rules for such a type must ignore
any overalignment of the base type we cannot address this by simply
adding a larger ali
After discussions with Jakub on IRC, I've committed this patch along
with a couple of other tweaks to the testsuite. New ChangeLog below.
This issue has existed since GCC-6 but has only come to light following
the release of gcc-8 where code was added to the compiler sources that
exposed the bug.
Hi,
unfortunately there are some more issues with -Waddress-of-packed-member, that
show that my previous
patch was not yet complete.
That is a bogus warning, in C and C++ (see addition in test case 1), and a
couple of missing warnings
in C++ (test case 2).
The latter warning regressions were a
On Tue, Jan 22, 2019 at 2:41 PM H.J. Lu wrote:
>
> With noplt attribute, we load the external function address via the GOT
> slot so that linker won't create an PLT entry for extern function address.
>
> gcc/
>
> PR target/88954
> * config/i386/i386.c (ix86_force_load_from_GOT_p):
With noplt attribute, we load the external function address via the GOT
slot so that linker won't create an PLT entry for extern function address.
gcc/
PR target/88954
* config/i386/i386.c (ix86_force_load_from_GOT_p): Also check
noplt attribute.
gcc/testsuite/
P
On Tue, Jan 22, 2019 at 4:08 AM Richard Biener
wrote:
>
> On Mon, Jan 21, 2019 at 10:27 PM H.J. Lu wrote:
> >
> > On Mon, Jan 21, 2019 at 10:54 AM Jeff Law wrote:
> > >
> > > On 1/7/19 6:55 AM, H.J. Lu wrote:
> > > > On Sun, Dec 30, 2018 at 8:40 AM H.J. Lu wrote:
> > > >> On Wed, Nov 28, 2018 a
> Le 22 janv. 2019 à 10:02, Manfred Schwarb a écrit :
>
> Dominique, thanks a lot.
>
> I got the request to share my script, so I attached it to this mail.
> Be aware that the script reports numerous false positives.
> As one can expect, also gcc.dg would appreciate some love…
I had a quick
On Tue, Jan 22, 2019 at 12:15 PM Jakub Jelinek wrote:
>
> On Tue, Jan 22, 2019 at 12:01:29PM +0100, Martin Liška wrote:
> > I discussed with Jakub possibility of generating canonical triples that will
> > be then used in Fortran FE to filter the 'GCC$ builtin' directives. However,
> > the approach
Hi Chung-Lin!
On Mon, 14 Jan 2019 21:56:05 +0800, Chung-Lin Tang
wrote:
> this version of the wait-clause-with-no-args patch revises the following:
>
> (1) The way the Fortran FE parts are implemented, which essentially is your
> code.
> (I'll reflect that in the final ChangeLog)
>
> (2) Inst
On Mon, Jan 21, 2019 at 10:27 PM H.J. Lu wrote:
>
> On Mon, Jan 21, 2019 at 10:54 AM Jeff Law wrote:
> >
> > On 1/7/19 6:55 AM, H.J. Lu wrote:
> > > On Sun, Dec 30, 2018 at 8:40 AM H.J. Lu wrote:
> > >> On Wed, Nov 28, 2018 at 12:17 PM Jeff Law wrote:
> > >>> On 11/28/18 12:48 PM, H.J. Lu wrote
On Mon, Jan 21, 2019 at 9:23 PM Martin Sebor wrote:
>
> On 1/21/19 1:18 AM, Richard Biener wrote:
> > On Sat, Jan 19, 2019 at 12:25 AM Martin Sebor wrote:
> >>
> >> Some of the -Warray-bounds and -Wrestrict tests are prone to failing
> >> on targets like arm-eabi and others that use different par
Hi Wilco,
> On 23 Dec 2018, at 12:09, Olivier Hainque wrote:
>
>> Remember these are just temporaries for use in the prolog and epilog -
>> there is no need to save/restore the static base. Setting static chain
>> to x9 and the temporaries to x10/x11 is the simplest solution. We
>> can separatel
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2019-01-22 Richard Biener
PR tree-optimization/88862
* graphite-scop-detection.c
(scop_detection::graphite_can_represent_scev): Reject ADDR_EXPR.
Index: gcc/graphite-scop-detection.c
On Tue, Jan 22, 2019 at 12:01:29PM +0100, Martin Liška wrote:
> I discussed with Jakub possibility of generating canonical triples that will
> be then used in Fortran FE to filter the 'GCC$ builtin' directives. However,
> the approach bring some possibilities and thus I enhanced the previous
> vers
On 1/21/19 3:25 PM, Joseph Myers wrote:
> On Mon, 21 Jan 2019, Martin Liška wrote:
>
>> I like the if('lp64'), if('ilp32') approach and I'm sending patch
>> candidate for that. Would it be accepted by glibc folks?
>
> Since glibc supports libmvec for x86_64, both 64-bit and x32, but not for
> 3
On 21/01/2019 18:03, Jeff Law wrote:
2019-01-18 Andrew Stubbs
gcc/
* doc/extend.tex (AMD GCN Function Attributes): New section.
* doc/install.texi (amdgcn-unknown-amdhsa): New instructions.
* doc/invoke.texi (AMD GCN Options): New section.
* doc/md.texi
There were a couple of issues in the code for mov{si,di}_pic_label_ref in
sparc_delegitimize_address: it wouldn't recognize the GOT register and it
wouldn't recognize a decoded HIGH/LO_SUM combination.
Tested on SPARC/Solaris 11, applied on the mainline.
2019-01-22 Eric Botcazou
*
In file included from ../../gcc/d/d-system.h:23,
from ../../gcc/d/dmd/root/dsystem.h:24,
from ../../gcc/d/dmd/mtype.c:11:
../../gcc/d/dmd/mtype.c: In member function 'Identifier*
Type::getTypeInfoIdent()':
../../gcc/d/dmd/mtype.c:2329:33: error: comparison of inte
On Mon, 7 Jan 2019, Jakub Jelinek wrote:
> Apparently my recent patch turned on many non-delegitimized UNSPEC notes
> (it is checking only note, goes away in release builds, but anyway).
I've been seeing the following in my testsuite runs which I didn't
get a month or so ago. Are these due to you
The following patch from Nidal Faour fixes LTO issues on windows.
Thanks for tracking this down.
Aplied to trunk and GCC 8 branch.
Richard.
2019-01-22 Nidal Faour
PR lto/88422
* simple-object.c (O_BINARY): Define if not already defined.
(simple_object_copy_lto_debu
On Fri, 18 Jan 2019, Jakub Jelinek wrote:
> Hi!
>
> As mentioned in the PR, on the transfer_intrinsic_3.f90 testcase at -O3
> on a few targets we have in number_of_iterations_cond:
> code LE_EXPR
> iv0->base 0
> iv0->step 0
> iv1->base -1
> iv1->step 1
> every_iteration false
>
> The loop starts
On Tue, Jan 22, 2019 at 7:57 AM Hongtao Liu wrote:
> This is adjusted patch: Modify BDESC to accept two flags and merge
> ARGS2/SPEICAL_ARGS2 into ARGS/SPECIAL_ARGS.
> Bootstrap test is ok, no regression on x86 backend testcases.
>
> Changelog:
> ---
> gcc/
>
> 2019-01-21 Hongtao Liu
> H.J.
Dominique, thanks a lot.
I got the request to share my script, so I attached it to this mail.
Be aware that the script reports numerous false positives.
As one can expect, also gcc.dg would appreciate some love...
While cleaning up the script, I found some refinements, and another
7 glitches, see
On Fri, Jan 18, 2019 at 5:43 PM David Malcolm wrote:
> (1) What does GCC mean by "ebb" in this context?
In this context, the EBB is what Muchnick would call a trace.
Ciao!
Steven
Hello!
> I have committed a patch to update libgo to the Go 1.12beta2 release.
>
> As usual this sort of update is too large to include all changes in
> this e-mail. I've included changes to gccgo-specific files below.
>
> Bootstrapped and ran Go testsuite on x86_64-pc-linux-gnu. Committed
> to
From: carll
backport from trunk to gcc-7-branch.
gcc/ChangeLog:
2017-12-11 Carl Love
* config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
vec_extract_fp32_from_shortl]): Add #defines.
* config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
*
On Mon, 21 Jan 2019, Jakub Jelinek wrote:
> Hi!
>
> The following patch is miscompiled on arm thumb1, where there is
> (jump_insn 32 31 33 4 (parallel [
> (set (pc)
> (if_then_else (ne (zero_extract:SI (reg:SI 3 r3 [orig:127
> ret+8 ] [127])
>
On Mon, 21 Jan 2019, Jakub Jelinek wrote:
> Hi!
>
> add_equal_note has special code to handle properly doubleword POPCOUNT and
> similar builtins, e.g. for POPCOUNT if we don't have DImode POPCOUNT, it can
> be expanded as 2 SImode POPCOUNTs added together, but for the REG_EQUAL note
> we want lo
On Mon, 21 Jan 2019, Jakub Jelinek wrote:
> Hi!
>
> Apparently RTL DSE doesn't perform any kind of escape analysis for e.g.
> frame pointer related addresses, instead relies on TREE_ADDRESSABLE being
> set on MEM_EXPR VAR_DECLs if the address could leak to other functions.
>
> Since PR49454 expr
On Mon, 21 Jan 2019, Richard Sandiford wrote:
> Richard Biener writes:
> > @@ -5742,8 +5740,14 @@ vectorize_fold_left_reduction (stmt_vec_info
> > stmt_info,
> >auto_vec vec_oprnds0;
> >if (slp_node)
> > {
> > - vect_get_vec_defs (op0, NULL_TREE, stmt_info, &vec_oprnds0, NULL,
On Mon, 21 Jan 2019, Tamar Christina wrote:
> Hi All,
>
> The testcase for PR/tree-optimization 88903 hits
> undefined behavior for the overwide shifts. This is
> causing the testcases to fail for all Arm platforms.
>
> This updates the testcase to add an explicit & such
> that the behavior is
83 matches
Mail list logo