On Thu, 27 Oct 2016, tbsaunde+...@tbsaunde.org wrote:
> From: Trevor Saunders
>
> gcc/c/ChangeLog:
>
> 2016-10-27 Trevor Saunders
>
> * gimple-parser.c (c_parser_gimple_switch_stmt): Fix
> uninitialized use of loc.
Thanks, applied & pushed.
Richard.
> ---
> gcc/c/gimple-pars
From: Trevor Saunders
gcc/c/ChangeLog:
2016-10-27 Trevor Saunders
* gimple-parser.c (c_parser_gimple_switch_stmt): Fix
uninitialized use of loc.
---
gcc/c/gimple-parser.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/gcc/c/gimple-parser.c b/gcc/c/
On Tue, Oct 25, 2016 at 03:33:36PM +0200, Richard Biener wrote:
>
> Hi,
>
> so I did the massaging to split out the GIMPLE parsing routines out
> to a separate file (quite tricky to get the gengtype issues correctly
> so I thought to help out here and get things started).
actually it looks like
I'm currently working on a builtin expansion of strncmp for powerpc
similar to the one for memcmp I checked recently. One thing I
encountered is that the code in expand_strn_compare will not attempt to
expand the cmpstrnsi pattern at all if neither string parameter is a
constant string. This doesn'
On Tue, Oct 18, 2016 at 01:25:55PM +0200, Bernd Schmidt wrote:
> On 10/17/2016 09:46 PM, tbsaunde+...@tbsaunde.org wrote:
> > From: Trevor Saunders
> >
> > gcc/ChangeLog:
> >
> > 2016-10-17 Trevor Saunders
> >
> > * ccmp.c (expand_ccmp_expr_1): Adjust.
> > (expand_ccmp_expr): Likewis
I forgot to mention, I will be working on a follow-on patch to this that
enables QImode and HImode to go in the vector registers for ISA 3.0, since ISA
3.0 now adds load (with zero extend) and store instructions for those types.
I probably also will update vector extract for the case where the sma
PowerPC GCC has traditionally only allowed DImode to go into FPR registers (and
now VSX registers) in order to allow floating point conversions. Conversions
to/from SImode have always had to deal with special UNSPECs to allow the
generation of the LFIWAX, LXSIWAX, LFIWZX, LXSIWZX, STFIWX, and STXS
Currently, the nvptx backend is only neutering the worker axis when
propagating variables used in conditional expressions across the worker
and vector axes. That's a problem with the worker-state spill and fill
propagation implementation because all of the vector threads in worker 0
all write the t
Jeff Law writes:
> The fallthrus in mips16_constant_cost are pretty obvious. One buffer
> overflow in mips16_build_call_stub and one obvious fallthru comment
> adjustment in mips16_build_call_stub.
>
> The usual stuff. Installing.
Thanks Jeff.
Matthew
On Wed, 26 Oct 2016, Martin Sebor wrote:
> The attached patch implements one such approach by having the pretty
> printer recognize the space format flag to suppress the type suffix,
> so "%E" still prints the suffix but "% E" does not. I did this to
> preserve the existing output but I think it
This adjusts existing fallthru comments and adds some new ones. I
couldn't convince myself the original code in movsicc was correct, It
looks like we properly reverse the condition, but not the operands in
the case of LT/LE/LEU/LTU. But then we do almost the same th ing in
sh_emit_scc_to_t
The fallthrus in mips16_constant_cost are pretty obvious. One buffer
overflow in mips16_build_call_stub and one obvious fallthru comment
adjustment in mips16_build_call_stub.
The usual stuff. Installing.
jeff
commit c03f0de027e1079edd02e3494601c6def50dbac1
Author: law
Date: Wed Oct 26
The following one-liner segfaults on arm-eabi when compiled with
-mfloat-abi=hard -g:
__simd64_float16_t usingit;
The problem is that the pretty printer (in simple_type_specificer()) is
dereferencing a NULL result from c_common_type_for_mode:
int prec = TYPE_PRECISION (t);
On 10/26/16 1:10 PM, Gerald Pfeifer wrote:
On Tue, 25 Oct 2016, Peter Bergner wrote:
Perhaps add a disclaimer at the top of changes.html that this
is still work in progress as part of that commit?
Do you mean like the following? If so, we'd have to remember to
remove the last hunk when GCC 7 i
On 10/26/2016 08:50 PM, David Malcolm wrote:
The following patch adds back in the INSN_UID for all insns.
Successfully bootstrapped®rtested on x86_64-pc-linux-gnu.
OK for trunk?
Yes.
Bernd
On Thu, 2016-10-20 at 16:11 +0200, Bernd Schmidt wrote:
> On 10/20/2016 03:55 PM, David Malcolm wrote:
> > Currently the jump insn in question looks like this:
> >
> > (cjump_insn (set (pc)
> > (label_ref 20))
> > (nil))
> >
> > With explicit INSN_UIDs i
On Tue, 25 Oct 2016, Peter Bergner wrote:
>> Perhaps add a disclaimer at the top of changes.html that this
>> is still work in progress as part of that commit?
> Do you mean like the following? If so, we'd have to remember to
> remove the last hunk when GCC 7 is released.
Yep, something like tha
Functions with interrupt or no_caller_saved_registers attribute should
have different function pointer types from those without the attribute
since they require different prologue and epilogue. 2 functions with
different interrupt or no_caller_saved_registers attribute, which are
otherwise equival
On 10/26/2016 07:27 PM, Segher Boessenkool wrote:
On Wed, Oct 26, 2016 at 03:13:06PM +0200, Bernd Schmidt wrote:
On 10/26/2016 03:05 PM, Segher Boessenkool wrote:
2) We do not necessarily have all notes yet, if the port lets dwarf2cfi
create notes by itself. Most (or even all?) ports do.
Hmm
On Wed, Oct 26, 2016 at 03:13:06PM +0200, Bernd Schmidt wrote:
> On 10/26/2016 03:05 PM, Segher Boessenkool wrote:
> >2) We do not necessarily have all notes yet, if the port lets dwarf2cfi
> >create notes by itself. Most (or even all?) ports do.
>
> Hmm. Are you recording the insns in prologue_c
Tested on Linux-PPC64.
2016-10-26 Ville Voutilainen
Use constexpr addressof in optional, SFINAE housekeeping
for any, optional and tuple.
* include/std/any (__do_emplace(_Args&&...)): New.
(__do_emplace(initializer_list<_Up>, _Args&&...)): Likewise.
(__any_constructible): L
This should have been in the microblaze.c changes, but was missed.
Essentially it had a fallthru path, but using fallthru didn't really
simplify the code in a meaningful way. So I just added the return to
avoid the fallthru path.
Installed on the trunk.
Jeff
commit 370f943246c079efebe323b4
On 10/26/2016 04:46 PM, Georg-Johann Lay wrote:
+if { [istarget avr-*-*] } {
+ # If the value of a label does not fit into 16 bits, the linker
+ # will generate a stub (containing a direct jump) and we end up
+ # with the address of the stub instead of the address of the ver
Hi,
For stmt defining reduction, GCC vectorizer assumes that the reduction variable
is always the last (second) operand. Another fact is that vectorizer doesn't
swap operands for cond_reduction during analysis stage. The problem is GCC
middle-end may canonicalize cond_expr into a form that red
Hi,
This patch adds support for the Cortex-M23 processor launched by ARM [1]. The
patch adds support for the name and wires it up to the ARMv8-M Baseline
architecture and arm_v6m_tune tuning parameters for the time being. It also
updates documentation to mention this new processor.
[1] http:
Hi,
This patch adds support for the Cortex-M33 processor launched by ARM [1]. The
patch adds support for the name and wires it up to the ARMv8-M Mainline with DSP
extensions architecture and arm_v7m_tune tuning parameters for the time being.
It also updates documentation to mention this new pr
When formatting an integer constant using the %E directive GCC
includes a suffix that indicates its type. This can perhaps be
useful in some situations but in my experience it's distracting
and gets in the way when writing tests.
Here's an example:
$ cat b.c && gcc b.c
constexpr __SIZE_TYPE
The usuall fallthru and sprintf stuff. These required a bit more
thought, but nothing terribly taxing.
Installing on the trunk.
jeff
commit 7227c03164300c5f5188a1261b3c57f0b7072394
Author: law
Date: Wed Oct 26 16:33:22 2016 +
* config/arc/arc.c (acr_print_operand): Adjust fal
On 26/10/16 16:26 +, Bernd Edlinger wrote:
* libsupc++/new_opa.cc: Don't include in a free standing
environmnet. Declare memalign directly in that case.
Typo "environmnet".
OK for trunk, thanks.
We might actually want to avoid in other cases, as some
systems have a #war
On 26/10/16 17:26, Andre Vieira (lists) wrote:
On 26/10/16 13:51, Kyrill Tkachov wrote:
Hi Andre,
On 25/10/16 17:29, Andre Vieira (lists) wrote:
On 24/08/16 12:01, Andre Vieira (lists) wrote:
On 25/07/16 14:23, Andre Vieira (lists) wrote:
This patch extends support for the ARMv8-M Security
On 26/10/16 17:28, Andre Vieira (lists) wrote:
On 26/10/16 10:33, Kyrill Tkachov wrote:
+static tree
+arm_handle_cmse_nonsecure_entry (tree *node, tree name,
+ tree /* args */,
+ int /* flags */,
+ bool *no_add_attrs)
+{
+ tree fndecl;
+
+ if (!
On 26/10/16 10:33, Kyrill Tkachov wrote:
>
> +static tree
> +arm_handle_cmse_nonsecure_entry (tree *node, tree name,
> + tree /* args */,
> + int /* flags */,
> + bool *no_add_attrs)
> +{
> + tree fndecl;
> +
> + if (!use_cmse)
> +{
> + *n
Clearly a desired fallthru in cr16_pritn_operand handling of 'g'.
Shortly thereafter there's another fallthru, but AFAICT it shouldn't
ever happen. I've added a gcc_unreachable in that case.
Installing on the trunk.
Jeff
commit 24e65f049fd63325296b4172fc175633a2cde8d0
Author: law
Date:
On 10/25/16 12:50 PM, Peter Bergner wrote:
> On 10/25/16 12:17 PM, Gerald Pfeifer wrote:
>> Perhaps add a disclaimer at the top of changes.html that this
>> is still work in progress as part of that commit?
>
> Do you mean like the following? If so, we'd have to remember to remove
> the last hunk
On 26/10/16 13:51, Kyrill Tkachov wrote:
> Hi Andre,
>
> On 25/10/16 17:29, Andre Vieira (lists) wrote:
>> On 24/08/16 12:01, Andre Vieira (lists) wrote:
>>> On 25/07/16 14:23, Andre Vieira (lists) wrote:
This patch extends support for the ARMv8-M Security Extensions
'cmse_nonsecure_entr
Hi,
this patch avoids including malloc.h in free standing builds with
eCos/newlib, and declares the memalign directly.
Successfully built a cross compiler for eCos.
Is it OK for trunk?
Thanks
Bernd.
2016-10-26 Bernd Edlinger
* libsupc++/new_opa.cc: Don't include in a free standing
envir
Hi!
The Fortran omp_finish_clause langhook can add OMP_CLAUSE_MAP clauses with
VAR_DECL OMP_CLAUSE_SIZE (or change one with constant into non-constant
one), if that appears in another OpenMP region, we need to notice the
variables there so that they can be properly shared/firstprivatized etc.
Boo
And the same code again, this time in the epiphany port...
Installing on the trunk.
Jeff
commit c4bc8d8761d31e34e26958bbc3c33f371690edf2
Author: law
Date: Wed Oct 26 16:23:27 2016 +
* config/epiphany/epiphany.c (epiphany_print_operand): Adjust
fallthru comment.
g
Exact same code as in another port. Just adjusting the comment so we
don't warn on a desired fallthru. Installing on the trunk.
Jeff
commit f1203ee28fdad66269ad0d0b761629f50dc3449b
Author: law
Date: Wed Oct 26 16:19:55 2016 +
* config/fr30/fr30.c (fr30_print_operand): Adjust fa
We will return garbage from comparison_string in the event of an error
in an ASM due to unintentional fallthru. My first through was to have
output_operand_lossage declared as non-returning, but returning in the
case of a user ASM is what it's designed to do AFAICT.
So we just avoid the fal
So the comment change in iq2000_function_arg is trivial.
expand_one_builtin is pretty obvious once you look at the code. We're
currently falling through from the zero-operand case to the one operand
case. That results in reading from op[0], which is almost certainly
wrong. So rather than
In encode_pattern_1, we stuff the 'm' into the pattern for the memory
operand, then want to recurse on the address. So A fallthru seems
reasonable.
In m32c_legitimate_address_p we have a special case for certain
registers, but which does not apply to A0. So we've got
case ...
case ..
Hi!
The following patch starts emitting another new DWARF5 feature - DW_AT_rank
and DW_TAG_generic_subrange for Fortran assumed-rank arrays.
Unlike DW_TAG_subrange_type, the expressions in DW_TAG_generic_subrange
attributes have a magic 0 to rank-1 values pushed onto the DWARF stack
first; rather
And the same for m32r. Again, it's got the fallthru marked, just not in
a way we handle. Installing on the trunk.
Jeff
commit 024d7f4c44ddf36bf76658937abf9da039228ce8
Author: law
Date: Wed Oct 26 16:02:58 2016 +
* config/m32r/m32r.c (m32r_print_operand): Adjust fallthru comme
Hi Thomas,
On 24/10/16 09:05, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 14/10/16 14:50, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 03/10/16 17:45, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:46, Thomas Preudhomme wrote:
Hi,
This patch
And similarly for mcore. Installing on the trunk.
Jeff
commit 71ab5ec1fc6e513112580bfe9c14a6cc1a2ad649
Author: law
Date: Wed Oct 26 15:59:34 2016 +
* config/mcore/mcore.c (mcore_gen_compare): Adjust fallthru comments.
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@2415
Just tweaking a fallthru comment. Installing on the trunk.
Jeff
commit b357502f77b0f6fbc79e216479bee47eab1741a8
Author: law
Date: Wed Oct 26 15:52:41 2016 +
* config/microblaze/microblaze.c (microblaze_function_arg): Adjust
fallthru comment.
git-svn-id: svn+ssh:
This patch just tweaks a comment so that we no longer warn. Installing
on the trunk.
Jeff
commit 7aa35b20124373889e039a054076dd8d5288c9e2
Author: law
Date: Wed Oct 26 15:49:25 2016 +
* config/msp430/msp430.c (msp430_legitimate_address_p): Adjust
fallthru comment.
Inserts a suitable return rather than falling through. Installing on
the trunk.
Jeff
commit 7a2cc4bd5ebbc3133fa1c89320ca82bb4b3b4a71
Author: law
Date: Wed Oct 26 15:47:48 2016 +
* config/nios2/nios2.c (nios2_rtx_costs): Avoid fallthru.
git-svn-id: svn+ssh://gcc.gnu.or
Hi Thomas,
On 24/10/16 09:05, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 14/10/16 14:50, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 03/10/16 17:44, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:44, Thomas Preudhomme wrote:
Hi,
This patch
rl78_calclate_death_nodes, with one exception wants to treat JUMP_INSNs
and CALL_INSNs the same. Naturally it's implemented with a case
statement like
case JUMP_INSN:
case CALL_INSN:
We're just missing the fallthru comment to silence the warning.
Like the stormy16, rl78 gets
On 10/26/2016 01:34 AM, Ulrich Weigand wrote:
Jeff Law wrote:
First, there's a missing fallthru comment in spu_sched_reorder for
TYPE_LOAD/TYPE_STORE cases. If I'm reading the SPU docs properly a
load/store insn is handled by pipe_1 and we're also trying to model some
aspects of the load-store
On 10/26/2016 05:01 AM, Jonathan Wakely wrote:
On 25/10/16 08:20 -0400, Ed Smith-Rowland wrote:
+explicit
+param_type(_RealType __radius = _RealType(1))
+: _M_radius(__radius)
+{
+ _GLIBCXX_DEBUG_ASSERT(_M_radius > _RealType(0));
Nowadays we're able to do cheaper assertion
[ There's at least one more port that has the exact same problem. ]
stormy16's ctor/dtor output routines have this:
char buf[16];
sprintf (buf, ".ctors.%.5u",
/* Invert the numbering so the linker puts us in the proper
order; constructors are run from r
On 26/10/16 16:19 +0100, Jonathan Wakely wrote:
Another Filesystem TS change, implementing LWG 2706.
* include/experimental/bits/fs_dir.h (recursive_directory_iterator):
Overload pop (LWG 2706).
* src/filesystem/dir.cc (recursive_directory_iterator::pop): Define
n
If we have enough thunks we could potentially overwrite the output
buffer in pa_asm_output_mi_thunk:
ASM_GENERATE_INTERNAL_LABEL (label, "LTHN", current_thunk_number);
In reality we're unlikely to get enough thunks to cause a problem, but
fixing this is easy by just increasing the siz
This fixes problems building the H8 port with the current trunk. Just a
trivial fix to the format of a fallthru comment. Installed on the trunk.
Jeff
commit 356e50be8376f57e4f872ebc5a67b5ed41c98ab7
Author: law
Date: Wed Oct 26 15:20:33 2016 +
* config/h8300/h8300.c (h8300_pri
On Wed, Oct 26, 2016 at 4:05 PM, Marc Glisse wrote:
> On Wed, 26 Oct 2016, Bin.Cheng wrote:
>
>> On Wed, Oct 26, 2016 at 3:10 PM, Bin.Cheng wrote:
>>>
>>> On Wed, Oct 26, 2016 at 3:04 PM, Marc Glisse
>>> wrote:
On Wed, 26 Oct 2016, Bin.Cheng wrote:
> Thanks for reviewing, upda
Another Filesystem TS change, implementing LWG 2706.
* include/experimental/bits/fs_dir.h (recursive_directory_iterator):
Overload pop (LWG 2706).
* src/filesystem/dir.cc (recursive_directory_iterator::pop): Define
new overload.
* testsuite/experimental/fil
On Wed, 26 Oct 2016, Bin.Cheng wrote:
On Wed, Oct 26, 2016 at 3:10 PM, Bin.Cheng wrote:
On Wed, Oct 26, 2016 at 3:04 PM, Marc Glisse wrote:
On Wed, 26 Oct 2016, Bin.Cheng wrote:
Thanks for reviewing, updated patch attached. Is it OK?
+/* (convert (minmax ((convert (x) c -> minmax (
Hi Christophe,
Here's the updated patch.
Cheers,
Tamar
From: Christophe Lyon
Sent: Wednesday, October 19, 2016 11:23:56 AM
To: Tamar Christina
Cc: GCC Patches; Kyrylo Tkachov; nd
Subject: Re: [PATCH v2][AArch32][NEON] Implementing vmaxnmQ_ST and vminnmQ_S
On Okt 26 2016, Fritz Reese wrote:
> If so, I am not sure how to narrow down the issue without more debug
> info like a traceback or memory dump at the point of error.
Tell me more, I don't know anything about fortran.
Andreas.
--
Andreas Schwab, SUSE Labs, sch...@suse.de
GPG Key fingerprint
On Wed, Oct 26, 2016 at 10:32 AM, Andreas Schwab wrote:
> On Okt 26 2016, Fritz Reese wrote:
>
>> I can't seem to reproduce this on x86-64. Did you fully apply the
>> patch?
>
> I don't have any patches.
>
Sorry for the confusion, I meant the originally attached
dec_bitwise_ops.diff. I am just w
There are targets that support taking values of labels but where any arithmetic
on such values might produce garbage.
This patch introduces new dg-require-effective-target label_offsets which is a
subset of label_values, and adjusts respective test cases to the more
restricted predicate.
Run
On Okt 26 2016, Fritz Reese wrote:
> I can't seem to reproduce this on x86-64. Did you fully apply the
> patch?
I don't have any patches.
Andreas.
--
Andreas Schwab, SUSE Labs, sch...@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something compl
On Wed, Oct 26, 2016 at 3:10 PM, Bin.Cheng wrote:
> On Wed, Oct 26, 2016 at 3:04 PM, Marc Glisse wrote:
>> On Wed, 26 Oct 2016, Bin.Cheng wrote:
>>
>>> Thanks for reviewing, updated patch attached. Is it OK?
>>
>>
>> +/* (convert (minmax ((convert (x) c -> minmax (x c) if x is promoted
>> +
Hi Andre,
On 25/10/16 17:29, Andre Vieira (lists) wrote:
On 24/08/16 12:01, Andre Vieira (lists) wrote:
On 25/07/16 14:23, Andre Vieira (lists) wrote:
This patch extends support for the ARMv8-M Security Extensions
'cmse_nonsecure_entry' attribute to safeguard against leak of
information throug
All,
Attached is a patch to the GNU Fortran front-end and runtime library
(libgfortran) which accepts real numbers with missing exponents as if
'0' was given as the exponent when the compile flag -fdec is given,
for further compatibility with legacy compilers. By default, GNU
Fortran will reject r
On 10/26/2016 04:10 PM, David Malcolm wrote:
The following patch moves various global state in print-rtl.c into
a new "rtx_writer" class, giving us a place to stash additional state
relating to dumping (and the possibility of putting extra
setup/cleanup in ctor/dtor).
I didn't bother renaming th
On Wed, Oct 26, 2016 at 3:04 PM, Marc Glisse wrote:
> On Wed, 26 Oct 2016, Bin.Cheng wrote:
>
>> Thanks for reviewing, updated patch attached. Is it OK?
>
>
> +/* (convert (minmax ((convert (x) c -> minmax (x c) if x is promoted
> + and the outer convert demotes the expression back to x's t
On Wed, Oct 26, 2016 at 08:57:11AM -0500, Bill Schmidt wrote:
> On Oct 26, 2016, at 8:29 AM, Segher Boessenkool
> wrote:
> >
> > So, you do not want to create the builtins that we expand to machine insns
> > that are not supported with the -mcpu= (or other flags) in use. What does
> > the ABI h
On Wed, 26 Oct 2016, Bin.Cheng wrote:
Thanks for reviewing, updated patch attached. Is it OK?
+/* (convert (minmax ((convert (x) c -> minmax (x c) if x is promoted
+ and the outer convert demotes the expression back to x's type. */
+(for minmax (min max)
+ (simplify
+ (convert (minmax
On Tue, Oct 25, 2016 at 1:00 PM, Richard Biener
wrote:
> On Tue, Oct 25, 2016 at 1:21 PM, Bin Cheng wrote:
>> Hi,
>> This is an update patch for
>> https://gcc.gnu.org/ml/gcc-patches/2016-10/msg00738.html . In this version,
>> existing pattern (convert (op:s (convert@2 @0) (convert?@3 @1))) is
On Oct 26, 2016, at 8:29 AM, Segher Boessenkool
wrote:
>
> So, you do not want to create the builtins that we expand to machine insns
> that are not supported with the -mcpu= (or other flags) in use. What does
> the ABI have to say about this?
The ABI is silent on this point. The appendix of
On Tue, Oct 25, 2016 at 12:48 PM, Richard Biener
wrote:
> On Tue, Oct 25, 2016 at 1:21 PM, Bin Cheng wrote:
>> Hi,
>> This is a patch set adding various match.pd patterns in order to generate
>> simplified MIN/MAX_EXPR mostly from COND_EXPR. This is the first one
>> optimizing (convert1 (minma
t for excess errors)
> Excess errors:
> /usr/local/gcc/gcc-20161026/gcc/testsuite/gfortran.dg/dec_bitwise_ops_1.f90:33:16:
> Error: Can't convert INTEGER(4) to INTEGER(4) at (1)
> /usr/local/gcc/gcc-20161026/gcc/testsuite/gfortran.dg/dec_bitwise_ops_1.f90:34:16:
> Error: Can'
On Wed, Oct 26, 2016 at 03:13:06PM +0200, Bernd Schmidt wrote:
> On 10/26/2016 03:05 PM, Segher Boessenkool wrote:
> >2) We do not necessarily have all notes yet, if the port lets dwarf2cfi
> >create notes by itself. Most (or even all?) ports do.
>
> Hmm. Are you recording the insns in prologue_c
On Tue, 2016-10-25 at 14:47 +0200, Bernd Schmidt wrote:
> On 10/21/2016 10:27 PM, David Malcolm wrote:
> > Thanks. I attemped to use those fields of recog_data, but it
> > doesn't
> > seem to be exactly what's needed here.
>
> Yeah, I may have been confused. I'm not sure that just looking at
> SC
On 24/10/16 17:50 +0100, Jonathan Wakely wrote:
Make directory iterators become end iterator on error
* src/filesystem/dir.cc (open_dir): Return same value for errors
whether ignored or not.
(_Dir::advance(error_code*, directory_options)): Return false on
error.
(d
Hi Kelvin,
On Wed, Oct 26, 2016 at 02:41:43PM +0200, Segher Boessenkool wrote:
> PR target/78056
> * config/rs6000/rs6000.c (spe_init_builtins): Modify loops to not
> define builtin functions from the bdesc_spe_predicates or
> bdesc_spe_evsel arrays if the builtin mask is n
On 16/10/16 07:00, Prathamesh Kulkarni wrote:
Hi,
This patch overrides expand_divmod_libfunc hook for ARM port.
I separated the SImode tests into separate file from DImode tests
because certain arm configs (cortex-15) have hardware div insn for
SImode but not for DImode, and for that config we w
This fixes parsing and dumping of switch stmts as well as dumping of
PHI nodes in case a src BB has a label. It also fixes PHI lowering
(the gsi_remove already does the gsi_next).
Tested on x86_64-unknown-linux-gnu.
New testcase is
/* { dg-do run } */
/* { dg-options "-O -fgimple" } */
int __
On 10/26/2016 03:05 PM, Segher Boessenkool wrote:
2) We do not necessarily have all notes yet, if the port lets dwarf2cfi
create notes by itself. Most (or even all?) ports do.
Hmm. Are you recording the insns in prologue_contains etc.? You could
add dependencies for a prologue insn following
On Wed, Oct 26, 2016 at 02:30:04PM +0200, Bernd Schmidt wrote:
> On 10/26/2016 01:19 PM, Segher Boessenkool wrote:
> >Separate shrink-wrapping creates this not-all-that-smart code because
> >the profile is quite messed up, and spread_components isn't super smart
> >to begin with. I'll have a patch
On Wed, Oct 26, 2016 at 12:11:44PM +, Wilco Dijkstra wrote:
> Add a SHA1H pattern with a V2SI input. This avoids unnecessary
> DUPs when using intrinsics like vsha1h_u32 (vgetq_lane_u32 (x, 0)).
I think this is incorrect for big endian - element 0 of a vec_select in
big-endian for V4SImode is
On 26/10/16 10:12, Kyrill Tkachov wrote:
Hi Andre, thanks for resending them.
On 25/10/16 17:26, Andre Vieira (lists) wrote:
On 24/08/16 12:00, Andre Vieira (lists) wrote:
On 25/07/16 14:19, Andre Vieira (lists) wrote:
This patch adds the support of the '-mcmse' option to enable ARMv8-M's
Se
Hi Andre,
On 25/10/16 17:29, Andre Vieira (lists) wrote:
On 24/08/16 12:01, Andre Vieira (lists) wrote:
On 25/07/16 14:23, Andre Vieira (lists) wrote:
This patch extends support for the ARMv8-M Security Extensions
'cmse_nonsecure_entry' attribute to safeguard against leak of
information throug
t for excess errors)
> Excess errors:
> /usr/local/gcc/gcc-20161026/gcc/testsuite/gfortran.dg/dec_bitwise_ops_1.f90:33:16:
> Error: Can't convert INTEGER(4) to INTEGER(4) at (1)
...
Looking into this.
---
Fritz Reese
On 10/26/2016 01:19 PM, Segher Boessenkool wrote:
Separate shrink-wrapping creates this not-all-that-smart code because
the profile is quite messed up, and spread_components isn't super smart
to begin with. I'll have a patch for that soon, but even with it (and
also without separate shrink-wrapp
On Tue, Oct 25, 2016 at 11:36 PM, Jerry DeLisle wrote:
> On 10/25/2016 11:52 AM, Fritz Reese wrote:
>>
>> All,
>>
>> Here's the big one. This patch proposes an extension to both the GNU
>> Fortran front-end and runtime library (libgfortran) to support three
>> additional I/O specifiers: CARRIAGECO
Add a SHA1H pattern with a V2SI input. This avoids unnecessary
DUPs when using intrinsics like vsha1h_u32 (vgetq_lane_u32 (x, 0)).
ChangeLog:
2016-10-26 Wilco Dijkstra
* config/aarch64/aarch64-simd.md (aarch64_crypto_sha1hv4si): New
pattern.
--
diff --git a/gcc/config/aarch64/aarch64
Hi Paul,
looks ok to me. At least I couldn't break it with the most devious codes I
could fathom.
Ok for trunk and if applicable to gcc-6.
- Andre
On Wed, 26 Oct 2016 13:28:15 +0200
Paul Richard Thomas wrote:
> Dear All,
>
> The comment in the patch more than adequately describes how this p
The configure test for realpath() is missing a header, and the
fallback version of filesystem::canonical() that gets used if
realpath() fails was not setting an error.
Tested powerpc64le-linux, committed to trunk.
commit f36dd7f817590c618884e14433dd70c03334c288
Author: Jonathan Wakely
Date:
Dear All,
The comment in the patch more than adequately describes how this patch
works. The first testcase checks that correctly functioning code is
produced, when the spurious error is suppressed, and the second checks
that genuine errors are caught.
Bootstraps and regtests on FC21/x86_64 - OK f
On Wed, Oct 26, 2016 at 12:48:54PM +0200, Bernd Schmidt wrote:
> On 10/25/2016 06:57 PM, Segher Boessenkool wrote:
> >This patch makes scheduling not reorder any RTX_FRAME_RELATED_P insns
> >(relative to each other), to fix PR78029. I originally was a bit worried
> >that this would degrade code qu
On Wed, 26 Oct 2016, Jakub Jelinek wrote:
> On Wed, Oct 26, 2016 at 11:54:36AM +0200, Georg-Johann Lay wrote:
> > gen-pass-instances.awk is sensitive to the order in which directives are
> > written down, e.g. in target-pass.def: If a pass that runs first is added
> > first, then the last pass is
On 10/25/2016 06:57 PM, Segher Boessenkool wrote:
This patch makes scheduling not reorder any RTX_FRAME_RELATED_P insns
(relative to each other), to fix PR78029. I originally was a bit worried
that this would degrade code quality, but it seems to even improve it:
more other insns are scheduled b
On Wed, Oct 26, 2016 at 11:54:36AM +0200, Georg-Johann Lay wrote:
> gen-pass-instances.awk is sensitive to the order in which directives are
> written down, e.g. in target-pass.def: If a pass that runs first is added
> first, then the last pass is skipped and not added to pass-instances.def.
>
> W
On Wed, 26 Oct 2016, Prathamesh Kulkarni wrote:
> On 25 October 2016 at 18:47, Richard Biener wrote:
> > On Tue, 25 Oct 2016, Prathamesh Kulkarni wrote:
> >
> >> On 25 October 2016 at 16:17, Richard Biener wrote:
> >> > On Tue, 25 Oct 2016, Prathamesh Kulkarni wrote:
> >> >
> >> >> On 25 October
This adds a dump modifier, -gimple (TDF_GIMPLE) and guards the dumping
changes with it (in the case of dump_function_header simply not calling
it). It also adjusts label dumping in another place (when dumping
switch statements which btw do not yet parse correctly it seems).
Tested on x86_64-unk
1 - 100 of 117 matches
Mail list logo
