Hi all,
This patch enables -fsanitize-recover for KASan by default. This causes
KASan to continue execution after error in case of inline
instrumentation. This feature is needed because
- reports during early bootstrap won't even be printed
- needed to run all tests w/o rebooting machine for e
On 09/01/14 02:13, Zhenqiang Chen wrote:
To split live-range of register, split_live_ranges_for_shrink_wrap will
introduce additional register copies. If such copies can not be optimized by
later optimizations, it will lead to code size and performance regression.
My tests on ARM THUMB1 code siz
On 09/03/14 18:22, Kaz Kojima wrote:
Jeff Law wrote:
* gcc.c-torture/execute/pr39228.c: Use dg-additional-options
instead of dg-options and remove sh*-*-* from its target list.
Add "inline" keyword to test functions.
Wouldn't we be better off moving this into execute/ie
On 09/04/14 06:11, Richard Biener wrote:
The following two patches each fix PR63148, a wrong-code issue
caused by bogus array indices a-la
&global_data.b[(sizetype) i + 536870911] which have a correct
address when lowered but bogus index.
The case in question can be mitigated by disabling foldi
On 09/01/14 23:17, Bin.Cheng wrote:
For this specific case, I think the reuse of r84 comes from coalescing
during expanding, and this is necessary to remove redundant reg-moves.
Then we need to fix this in coming passes?
Hmmm. Yea, I can see how that might be happening. There's a certain
inher
On 09/01/14 23:14, Bin.Cheng wrote:
On Tue, Sep 2, 2014 at 11:40 AM, Jeff Law wrote:
On 08/31/14 22:18, Bin.Cheng wrote:
Note that i0..i4 need not be consecutive insns, so you'd have to walk the
chain from the location with the death note to the proposed death note
site.
If between those loc
On 09/04/14 20:32, Joey Ye wrote:
Trunk fails to build plugin again due to missing plugin header files. This
patch fixes it.
OK to trunk?
ChangeLog:
PR plugin/59335
* Makefile.in (PLUGIN_HEADERS): Add wide-int.h, signop.h, hash-map.h,
OK.
jeff
Hi All,
For the bswap built-in, there are unnecessary uxts generated as reported
in pr43550. Can we rely on the argument being unsigned and set the
SUBREG promoted accordingly.
At least in ARM ABI, arguments are supposed to be properly zero/sign
extended.
Any thoughts?
Bootstrapped and regressi
On 09/04/14 21:16, Trevor Saunders wrote:
Hi,
Looks like the patch is to big to send uncompressed, so sorry if this
the second mail you get :)
Given picochip has been obsolete for several years at this point, and
when I asked a while back if anyone wanted to keep it nobody responded
I'm going t
On Thu, Sep 4, 2014 at 6:15 PM, Joseph S. Myers wrote:
>
> gcc/c-family:
> 2014-09-05 Joseph Myers
>
> * c-cppbuiltin.c (c_cpp_builtins): Also define
> __LIBGCC_EH_TABLES_CAN_BE_READ_ONLY__,
> __LIBGCC_EH_FRAME_SECTION_NAME__, __LIBGCC_JCR_SECTION_NAME__,
> __LIB
Hi,
Looks like the patch is to big to send uncompressed, so sorry if this
the second mail you get :)
Given picochip has been obsolete for several years at this point, and
when I asked a while back if anyone wanted to keep it nobody responded
I'm going to claim its time to kill it.
bootstrapped +
On 09/04/14 03:42, Richard Biener wrote:
On Wed, Sep 3, 2014 at 7:54 PM, Aldy Hernandez wrote:
Flow-wise I still want to move hand-off to the cgraph code to toplev.c,
out from the FEs final_write_gloabals hook:
/* This must also call finalize_compilation_unit. */
lang_hooks.decls.fina
On 09/04/14 19:27, Easwaran Raman wrote:
I've attached the revised patch. Bootstrapped and no test regressions
on x86_64/linux with 4.9 branch. Ok for 4.9 branch? While the bug
doesn't show up in trunk, seems obvious that this should go to trunk
as well. Is it ok for trunk if tests pass?
Btw, is
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* config/arc/arc.c (arc_print_operand): Use insn method of
final_sequence for type-safety.
* config/bfin/bfin.c (bfin_hardware_loop): Strengthen param
"insn" from rtx to rtx_insn *.
* config/frv/frv.c (frv_print_
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* reorg.c (relax_delay_slots): Move declaration of "trial_seq"
above the conditional, and convert the check on GET_CODE to a
dyn_cast, so that "trial_seq" is available as an rtx_sequence * in
the conditional. Simplify t
Trunk fails to build plugin again due to missing plugin header files. This
patch fixes it.
OK to trunk?
ChangeLog:
PR plugin/59335
* Makefile.in (PLUGIN_HEADERS): Add wide-int.h, signop.h, hash-map.h,
hash-set.h.
diff --git a/gcc/Makefile.in b/gcc/Makefile.in
index 63124f8..8e7aada 10064
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* haifa-sched.c (check_clobbered_conditions): Strengthen local
"link" from rtx to rtx_insn_list *, and use its methods for
clarity and type-safety.
(toggle_cancelled_flags): Likewise.
(restore_last_backtrack_poin
From: Trevor Saunders
just changing argument type so pre approved.
bootstrap + regtested on x86_64-linux-gnu.
Trev
gcc/ChangeLog:
2014-09-04 Trevor Saunders
* valtrack.c (dead_debug_insert_temp): Take an rtx_insn * instead of
an rtx.
* valtrack.h: Adjust.
---
gcc/C
From: Trevor Saunders
just variable type changes so pre approved committing to trunk.
bootstrap + regtest on x86_64-linux-gnu, and config-list.mk
Trev
gcc/ChangeLog:
2014-09-04 Trevor Saunders
* emit-rtl.c (emit_insn_before_noloc): Take an rtx_insn * instead of
an rtx.
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* sched-deps.c (sched_get_condition_with_rev_uncached): Strengthen
param "insn" from const_rtx to const rtx_insn *.
(sched_get_reverse_condition_uncached): Likewise.
(sched_get_condition_with_rev): Likewise.
(sch
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* recog.c (peep2_attempt): Strengthen return type from rtx to
rtx_insn *.
(peep2_update_life): Likewise for params "last", "prev", removing
a checked cast made redundant by this.
(peephole2_optimize): Likewise fo
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* basic-block.h (set_block_for_insn): Eliminate this macro in
favor of...
* rtl.h (set_block_for_insn): New inline function, imposing the
requirement that the "insn" param is an rtx_insn *.
OK.
jeff
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* caller-save.c (setup_save_areas): Strengthen local "insn" from
rtx to rtx_insn *.
* final.c (get_call_reg_set_usage): Likewise for first param,
eliminating a checked cast.
* regs.h (get_call_reg_set_usage): Lik
On 09/04/14 19:52, David Malcolm wrote:
gcc/
* except.c (for_each_eh_label): Within param "callback",
strengthen param from rtx to rtx_code_label *. Strengthen local
"lab" from rtx to rtx_code_label *. Remove redundant check on
LABEL_P (lab), since this is known
On 09/04/14 19:51, David Malcolm wrote:
gcc/
* output.h (final_scan_insn): Strengthen first param from rtx to
rtx_insn *.
* final.c (final_scan_insn): Likewise, renaming it back from
"uncast_insn" to "insn", eliminating the checked cast.
* config/h8300/h
On 09/04/14 19:51, David Malcolm wrote:
gcc/
* jump.c (delete_related_insns): Introduce a new local "table" by
replacing JUMP_TABLE_DATA_P with a dyn_cast, then use the
get_labels method of "table" to simplify access to the labels in
the jump table.
OK.
Jeff
gcc/
* haifa-sched.c (check_clobbered_conditions): Strengthen local
"link" from rtx to rtx_insn_list *, and use its methods for
clarity and type-safety.
(toggle_cancelled_flags): Likewise.
(restore_last_backtrack_point): Likewise.
(queue_to_ready): Us
gcc/
* basic-block.h (set_block_for_insn): Eliminate this macro in
favor of...
* rtl.h (set_block_for_insn): New inline function, imposing the
requirement that the "insn" param is an rtx_insn *.
---
gcc/basic-block.h | 2 --
gcc/rtl.h | 5 +
2 files chan
gcc/
* caller-save.c (setup_save_areas): Strengthen local "insn" from
rtx to rtx_insn *.
* final.c (get_call_reg_set_usage): Likewise for first param,
eliminating a checked cast.
* regs.h (get_call_reg_set_usage): Likewise for first param.
* resource.
gcc/
* sched-deps.c (sched_get_condition_with_rev_uncached): Strengthen
param "insn" from const_rtx to const rtx_insn *.
(sched_get_reverse_condition_uncached): Likewise.
(sched_get_condition_with_rev): Likewise.
(sched_has_condition_p): Likewise.
(sc
gcc/
* recog.c (peep2_attempt): Strengthen return type from rtx to
rtx_insn *.
(peep2_update_life): Likewise for params "last", "prev", removing
a checked cast made redundant by this.
(peephole2_optimize): Likewise for local "last".
---
gcc/recog.c | 9 +
gcc/
* config/arc/arc.c (arc_print_operand): Use insn method of
final_sequence for type-safety.
* config/bfin/bfin.c (bfin_hardware_loop): Strengthen param
"insn" from rtx to rtx_insn *.
* config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
* co
gcc/
* reorg.c (relax_delay_slots): Move declaration of "trial_seq"
above the conditional, and convert the check on GET_CODE to a
dyn_cast, so that "trial_seq" is available as an rtx_sequence * in
the conditional. Simplify the conditional by using methods of
gcc/
* except.c (for_each_eh_label): Within param "callback",
strengthen param from rtx to rtx_code_label *. Strengthen local
"lab" from rtx to rtx_code_label *. Remove redundant check on
LABEL_P (lab), since this is known from the type rtx_code_label *.
*
gcc/
* jump.c (delete_related_insns): Introduce a new local "table" by
replacing JUMP_TABLE_DATA_P with a dyn_cast, then use the
get_labels method of "table" to simplify access to the labels in
the jump table.
---
gcc/jump.c | 12 ++--
1 file changed, 6 inse
gcc/
* output.h (final_scan_insn): Strengthen first param from rtx to
rtx_insn *.
* final.c (final_scan_insn): Likewise, renaming it back from
"uncast_insn" to "insn", eliminating the checked cast.
* config/h8300/h8300.md (define_insn "jump"): Replace loca
The following patches are followups to the 236-patch kit; they
use the new classes in more places, in some cases eliminating
checked casts, and should enable some bigger cleanups that I'm
working towards (e.g. converting the params of single_set and
recog_memoized to require rtx_insn * rather than
The attached trivial patch strengthens a local variable in cse.c from
rtx to rtx_insn *, allowing the elimination of a checked cast, and thus
falls under the pre-approval granted by Jeff here:
https://gcc.gnu.org/ml/gcc-patches/2014-08/msg01310.html
Bootstrapped on x86_64-unknown-linux-gnu (Fedo
>> Here is an attempt to do the value range computation in promoted_mode's
>> type when it is overflowing. Bootstrapped on x86-84.
>
> Err - I think you misunderstood this as a suggestion to do this ;)
> value-ranges should be computed according to the type not according
> to the (promoted) mode.
On Thu, Sep 04, 2014 at 10:05:38PM +0100, Maciej W. Rozycki wrote:
> On Thu, 4 Sep 2014, Alan Modra wrote:
>
> > > In fact if you look at the corresponding location expression for
> > > -fno-section-anchors code you get something quite screwy too.
> > >
> > > 018f 0108 0114 (DW_OP_bre
The attached patch strengthens the 2nd and 3rd params of
modified_between_p from const_rtx to const rtx_insn * (eliminating a
checked cast), and thus falls under the pre-approval granted by Jeff
here:
https://gcc.gnu.org/ml/gcc-patches/2014-08/msg01310.html
Bootstrapped on x86_64-unknown-linux-g
I've attached the revised patch. Bootstrapped and no test regressions
on x86_64/linux with 4.9 branch. Ok for 4.9 branch? While the bug
doesn't show up in trunk, seems obvious that this should go to trunk
as well. Is it ok for trunk if tests pass?
Btw, is g++.dg/opt the right directory for the tes
The attached patch strengthens both params of fixup_args_size_notes from
rtx to rtx_insn * (eliminating a checked cast), and thus falls under the
pre-approval granted by Jeff here:
https://gcc.gnu.org/ml/gcc-patches/2014-08/msg01310.html
Bootstrapped on x86_64-unknown-linux-gnu (Fedora 20), and
It is desirable for various reasons to stop target libraries from
using host-side tm.h: making the headers work on both host and target
(and on build as well) is fragile (especially when dealing with macros
depending on command-line options, which need different definitions
when being used for the
The attached trivial patch strengthens a return type from rtx to
rtx_insn *,
and thus falls under the pre-approval granted by Jeff here:
https://gcc.gnu.org/ml/gcc-patches/2014-08/msg01310.html
Bootstrapped on x86_64-unknown-linux-gnu (Fedora 20), and has been
rebuilt as part of a config-list.mk
Hi,
when opening a file with a name containing an ASCII null character,
the name is truncated at the first null since the IO API's take
null-terminated C strings. By storing such a C string in gfc_unit
instead of the original Fortran string INQUIRE for the file name will
return the correct name, a
On Sep 4, 2014, at 5:22 PM, Kaz Kojima wrote:
> BTW, Jeff's comment on pr39228.c makes me think that this test
> can be moved into execute/ieee. Thought?
I don’t see a down side, specially if we know that the test case is picky about
ieee.
The patch below removes a sh*-*-* specific dg-options line
from gcc.c-torture/execute/pr44683.c. That line is
unnecessary because now -mieee is a default option on sh*.
I've applied it as an obvious change specific to sh.
Tested on i686-linux and sh-linux.
BTW, Jeff's comment on pr39228.c makes m
My extend* pattern patches exposed a problem elsewhere. I don't build
java so I didn't see it. This fixes it. Also handles the zero-extend
case, that had the same problem. Pre-approved by David, committed.
Segher
2014-09-04 Segher Boessenkool
PR target/63165
* config/rs6
This adds an option to allow programs and libraries built
*without* inhibit_libc to stay compatible with system libraries
(really: libgcc_s.so.1) built *with* inhibit_libc, at the cost
of the registration. As mentioned, that's a one-way
compatibility barrier.
While it's nice to avoid the overhead
The directory at $target_header_dir is already inspected in
gcc/configure, for e.g. glibc version and stack protector
support, but not for setting inhibit_libc. This is just
inconsistent and the obvious resolution to me is to inhibit
inhibit_libc when a target *does* "have its own set of headers",
The conditions for inhibit_libc to activate (i.e. for library
headers to be absent) are IMO a bit too automatic and the effect
is too subtle and serious in some situations. For example, if
you pre-install target headers in $target_header_dir, gcc will
find them and use them, but still inhibit_libc
On Thu, 2014-09-04 at 22:22 +0200, Jakub Jelinek wrote:
> On Thu, Sep 04, 2014 at 04:04:17PM -0400, David Malcolm wrote:
> > --- a/gcc/rtl.h
> > +++ b/gcc/rtl.h
> > @@ -195,6 +195,7 @@ union rtunion
> >unsigned int rt_uint;
> >const char *rt_str;
> >rtx rt_rtx;
> > + rtx_insn *rt_insn;
The attached patch strengthens both params of the generated insn_latency
function from rtx to rtx_insn *, and thus falls under the pre-approval
granted by Jeff here:
https://gcc.gnu.org/ml/gcc-patches/2014-08/msg01310.html
Bootstrapped on x86_64-unknown-linux-gnu (Fedora 20), and has been
rebuil
On Thu, 4 Sep 2014, Alan Modra wrote:
> > In fact if you look at the corresponding location expression for
> > -fno-section-anchors code you get something quite screwy too.
> >
> > 018f 0108 0114 (DW_OP_breg10 (r10): 0; DW_OP_addr: 144;
> > DW_OP_plus; DW_OP_stack_value
> >
> > whic
On Thu, Sep 04, 2014 at 04:04:17PM -0400, David Malcolm wrote:
> --- a/gcc/rtl.h
> +++ b/gcc/rtl.h
> @@ -195,6 +195,7 @@ union rtunion
>unsigned int rt_uint;
>const char *rt_str;
>rtx rt_rtx;
> + rtx_insn *rt_insn;
>rtvec rt_rtvec;
>enum machine_mode rt_type;
>addr_diff_vec
The attached trivial patch strengthens a param from rtx to rtx_insn *,
and thus falls under the pre-approval granted by Jeff here:
https://gcc.gnu.org/ml/gcc-patches/2014-08/msg01310.html
Bootstrapped on x86_64-unknown-linux-gnu (Fedora 20), and has been
rebuilt as part of a config-list.mk build
On Thu, Sep 04, 2014 at 03:22:15PM +0200, Richard Biener wrote:
> On Thu, Sep 4, 2014 at 3:06 PM, Yury Gribov wrote:
> > Hi all,
> >
> > This patch adds a Vim config (.local.vimrc) to root folder. This would allow
> > automatic setup of GNU formatting for C/C++/Java/Lex files in GCC (similar
> > t
On Tue, 2014-09-02 at 19:50 +0200, Andi Kleen wrote:
> > I suspect the bulk of them currently are coming from the safe_as_a
> > calls within NEXT_INSN and PREV_INSN; do you happen to have
> > information handy on that?
>
> Yes that's right:
>
> - 1.03% lto1[.] bool
> is_a
On Thu, 2014-09-04 11:34:57 -0700, Aldy Hernandez wrote:
> On 09/04/14 11:23, Richard Biener wrote:
> > On September 4, 2014 7:54:14 PM CEST, Aldy Hernandez
> > wrote:
> > > > I think this merge caused bootstrap failure on Linux/i686
> > > > https://gcc.gnu.org/ml/gcc-regression/2014-09/msg00010
Matthew Fortune writes:
> Ping!
For the record, I agree this is the right fix.
Thanks,
Richard
>> -Original Message-
>> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-ow...@gcc.gnu.org]
>> On Behalf Of Matthew Fortune
>> Sent: 18 August 2014 14:25
>> To: 'gcc-patches@gcc.gnu.or
Gerald Pfeifer writes:
> On Thu, 4 Sep 2014, Richard Sandiford wrote:
>> Is this stage 1 libgcc or a later libgcc? Seems unlikely if stage 1.
>
> Stage 1. I kicked off a non-parallel build, which just failed as
> follows:
>
> /scratch2/tmp/gerald/OBJ-0904-1719/./gcc/xgcc
> -B/scratch2/tmp/gerald
Dunno, if I'm allowed to make calls on what an obvious patch is,
considering I completely broke trunk with my last commit to the
debug-early branch, but I'll take my chances...
Yet another *.def file that needs handling in the TAGS file.
Committed to trunk as obvious.
Aldy
commit be2a57c0df76
On Thu, 4 Sep 2014, Richard Sandiford wrote:
> Is this stage 1 libgcc or a later libgcc? Seems unlikely if stage 1.
Stage 1. I kicked off a non-parallel build, which just failed as
follows:
/scratch2/tmp/gerald/OBJ-0904-1719/./gcc/xgcc
-B/scratch2/tmp/gerald/OBJ-0904-1719/./gcc/
-B/home/geral
> On Thu, Sep 4, 2014 at 5:58 AM, Andi Kleen wrote:
> >>
> >> Anyway, removing !optimize checks in favor of flag_no_inline checks
> >> and initializing that properly is a cleanup as well.
> >
> > Patch looks good to me.
>
> Unfortunately it doesn't pass bootstrap (inline-summary re-use
> between
On 03-09-2014 11:01, Maciej W. Rozycki wrote:
> On Tue, 2 Sep 2014, Adhemerval Zanella wrote:
>
>> Ping.
>>
>> On 19-08-2014 13:54, Adhemerval Zanella wrote:
>>> Ping.
>>>
>>> On 06-08-2014 17:21, Adhemerval Zanella wrote:
On 01-08-2014 12:31, Joseph S. Myers wrote:
> On Thu, 31 Jul 2014,
Hi, FWIW I saw the same problem on arm* and aarch64 targets.
On 4 September 2014 20:34, Aldy Hernandez wrote:
> On 09/04/14 11:23, Richard Biener wrote:
>>
>> On September 4, 2014 7:54:14 PM CEST, Aldy Hernandez
>> wrote:
>>>
>>>
I think this merge caused bootstrap failure on Linux/i686
>
On 09/04/14 11:23, Richard Biener wrote:
On September 4, 2014 7:54:14 PM CEST, Aldy Hernandez wrote:
I think this merge caused bootstrap failure on Linux/i686
https://gcc.gnu.org/ml/gcc-regression/2014-09/msg00010.html
I incorrectly pushed my branch onto trunk. It's being reverted.
To
On September 4, 2014 7:54:14 PM CEST, Aldy Hernandez wrote:
>
>> I think this merge caused bootstrap failure on Linux/i686
>>
>> https://gcc.gnu.org/ml/gcc-regression/2014-09/msg00010.html
>>
>>
>
>I incorrectly pushed my branch onto trunk. It's being reverted.
Too bad!
;)
Richard.
Hi James,
Yes we can just mark operand 3 as "&r".
PFB, the updated patch. Ok for trunk?
regards,
Venkat.
gcc/ChangeLog
2014-09-04 Venkataramanan Kumar
* config/aarch64/aarch64.md (stack_protect_test_) Add register
constraint for operand 0 and remove write only constraint from o
I think this merge caused bootstrap failure on Linux/i686
https://gcc.gnu.org/ml/gcc-regression/2014-09/msg00010.html
I incorrectly pushed my branch onto trunk. It's being reverted.
On Wed, Sep 3, 2014 at 10:54 AM, Aldy Hernandez wrote:
> [Jason, Richard]: Is it useful for my patches to contain ChangeLog entries?
> I find them mildly annoying for something that will inevitably be rewritten
> multiple times, but if it aids in reviewing my WIP, I am more than happy to
> continu
Gerald Pfeifer writes:
> On Tue, 2 Sep 2014, Richard Sandiford wrote:
>> Also, i386 was robust against uses of %& in inline asm. The patch
>> makes sure the other ports are too. Using %& in inline asm would
>> often be a mistake, but it should at least trigger a proper error
>> rather than an IC
On Wed, 3 Sep 2014, Uros Bizjak wrote:
> > So I think the default timeout that's used for really quick tests should
> > be extended a bit. I propose a factor of 2, just not to make it too
> > excessive, at least for the beginning (maybe it'll have to be higher
> > eventually).
>
> Or you can ju
On Sep 4, 2014, at 4:14 AM, Ramana Radhakrishnan
wrote:
>> --- a/gcc/testsuite/lib/target-supports.exp
>> +++ b/gcc/testsuite/lib/target-supports.exp
>> @@ -2293,8 +2293,8 @@ proc check_effective_target_arm_little_endian { } {
>> proc check_effective_target_arm_vect_no_misalign { } {
>> retur
On 09/04/2014 07:04 AM, Ramana Radhakrishnan wrote:
> gcc/Changelog
>
> 2014-09-04 Marcus Shawcroft
> Ramana Radhakrishnan
>
>* config/aarch64/aarch64-elf-raw.h (ENDFILE_SPEC): Add
> crtfastmath.o.
>* config/aarch64/aarch64-linux.h
> (GNU_USER_TARGET_MATH
On Tue, 2 Sep 2014, Richard Sandiford wrote:
> Also, i386 was robust against uses of %& in inline asm. The patch
> makes sure the other ports are too. Using %& in inline asm would
> often be a mistake, but it should at least trigger a proper error
> rather than an ICE.
>
> Tested on x86_64-linux
On Sep 4, 2014, at 12:22 , Richard Biener wrote:
> So I tend to think that removing / delaying traps is ok unless you
> can catch them within the active EH scheme.
> Certainly that Oliver excludes loads/stores (but not integer
> division!) points at the fact that preserving the order of
> traps
On 4 September 2014 15:47, Wilco Dijkstra wrote:
> This patch adds regmove_costs for Cortex-A57 and A53, and sets the cost of
> GP2FP/FP2GP higher than
> memory cost to block the register allocator allocating integer values in FP
> registers.
>
> Overall these patches give 2-3% speedup on SPEC.
On 4 September 2014 16:41, Wilco Dijkstra wrote:
>> From: Marcus Shawcroft [mailto:marcus.shawcr...@gmail.com]
>> > - NAMED_PARAM (FP2FP, 4)
>> > + NAMED_PARAM (FP2FP, 2)
>>
>> This is not directly related to the change below and it is missing
>> from the ChangeLog. Originally this number had
Ping!
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-ow...@gcc.gnu.org]
> On Behalf Of Matthew Fortune
> Sent: 22 August 2014 10:43
> To: 'gcc-patches@gcc.gnu.org' (gcc-patches@gcc.gnu.org); Eric Christopher
> (echri...@gmail.com)
> Cc: Moore, Catherine (cat
Ping!
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-ow...@gcc.gnu.org]
> On Behalf Of Matthew Fortune
> Sent: 18 August 2014 14:25
> To: 'gcc-patches@gcc.gnu.org' (gcc-patches@gcc.gnu.org)
> Cc: Richard Sandiford; Eric Christopher (echri...@gmail.com)
> Sub
On 4 September 2014 16:39, Marcus Shawcroft wrote:
> On 4 September 2014 15:45, Wilco Dijkstra wrote:
>> Cleanup inconsistent use of __extension__.
>>
>> ChangeLog:
>> 2014-09-04 Wilco Dijkstra
>>
>> * gcc/config/aarch64/aarch64.c: Cleanup use of __extension__.
>
> Write a proper Chang
> From: Marcus Shawcroft [mailto:marcus.shawcr...@gmail.com]
> > - NAMED_PARAM (FP2FP, 4)
> > + NAMED_PARAM (FP2FP, 2)
>
> This is not directly related to the change below and it is missing
> from the ChangeLog. Originally this number had to be > 2 in order
> for secondary reload to kick in.
On 4 September 2014 15:45, Wilco Dijkstra wrote:
> Cleanup inconsistent use of __extension__.
>
> ChangeLog:
> 2014-09-04 Wilco Dijkstra
>
> * gcc/config/aarch64/aarch64.c: Cleanup use of __extension__.
Write a proper ChangeLog entry please.
/Marcus
On 4 September 2014 15:45, Wilco Dijkstra wrote:
> This patch fixes a bug in aarch64_register_move_cost(): GET_MODE_SIZE is in
> bytes not bits. As a
> result the FP2FP cost doesn't need to be set to 4 to catch the special case
> for Q register moves.
>
> ChangeLog:
> 2014-09-04 Wilco Dijkstra
On 4 September 2014 15:44, Wilco Dijkstra wrote:
> Hi,
>
> This is a set of patches improving register costs on AArch64. The first fixes
> aarch64_register_move_cost() to support CALLER_SAVE_REGS and POINTER_REGS so
> costs are calculated
> correctly in the register allocator.
>
> ChangeLog:
> 20
On Thu, Sep 4, 2014 at 7:08 AM, Bin Cheng wrote:
> Hi,
> This patch is posted/approved before at
> https://gcc.gnu.org/ml/gcc-patches/2014-07/msg01398.html
> Unfortunately, it was reverted because a latent bug revealed causing glibc
> build failure and I didn't have enough time to fix back at that
On Thu, Sep 4, 2014 at 6:33 AM, Bin Cheng wrote:
> Hi,
> This patch is to fix a potential bug in arm pattern "arm_movqi_insn".
>
> For the pattern,
> (define_insn "*arm_movqi_insn"
> [(set (match_operand:QI 0 "nonimmediate_operand" "=r,r,r,l,r,l,Uu,r,m")
> (match_operand:QI 1 "general_op
On 2014-09-04 3:37 AM, Tom de Vries wrote:
On 03-09-14 18:58, Tom de Vries wrote:
I've build the patch and ran the fuse-caller-save tests, and I'm
currently
bootstrapping and reg-testing it on x86_64.
Vladimir,
This patch fixes a problem (found on s390) in one of the committed
fuse-caller-sa
This patch adds regmove_costs for Cortex-A57 and A53, and sets the cost of
GP2FP/FP2GP higher than
memory cost to block the register allocator allocating integer values in FP
registers.
Overall these patches give 2-3% speedup on SPEC.
This passes all regression tests (with this fix
https://gcc.
Cleanup inconsistent use of __extension__.
ChangeLog:
2014-09-04 Wilco Dijkstra
* gcc/config/aarch64/aarch64.c: Cleanup use of __extension__.
---
gcc/config/aarch64/aarch64.c | 38 +++---
1 file changed, 11 insertions(+), 27 deletions(-)
diff --git a/
This patch fixes a bug in aarch64_register_move_cost(): GET_MODE_SIZE is in
bytes not bits. As a
result the FP2FP cost doesn't need to be set to 4 to catch the special case for
Q register moves.
ChangeLog:
2014-09-04 Wilco Dijkstra
* gcc/config/aarch64/aarch64.c (aarch64_register_mov
Hi,
This is a set of patches improving register costs on AArch64. The first fixes
aarch64_register_move_cost() to support CALLER_SAVE_REGS and POINTER_REGS so
costs are calculated
correctly in the register allocator.
ChangeLog:
2014-09-04 Wilco Dijkstra
* gcc/config/aarch64/aarch64.c
currently, the instruction sink in "prepare_shrink_wrap" is a bit conservative
that some further optimization opportunities have been missed.
given the prologue use register A by:
(store A, [sp + offset])
then given the entry_basic_block contains a simply register copy like:
(move A, B)
c
this patch enabled stack shrink-wrap support on AArch64.
no regression on aarch64-none-elf bare-metal.
aarch64 bootstrap OK.
ok to install?
2014-09-04 Renlin Li
gcc/
* config/aarch64/aarch64.md (return): New expand.
(simple_return): Likewise.
* config/aarch64/aarch64.c (aarch64_use_retur
Hi,
While changing register costs for AArch64 (patches to follow), the test
gcc.target/aarch64/vect-mult.c fails. This is caused by ree inserting a TI mode
copy of a DI
register. Since TI mode requires 2 registers, this results in silent corruption
of the 2nd register.
After split2:
(insn 149
On Thu, Sep 4, 2014 at 3:06 PM, Jakub Jelinek wrote:
> On Thu, Sep 04, 2014 at 02:57:47PM +0200, Richard Biener wrote:
>> Note that I think we arrived at the point where the loop structure
>> has annotations that are required for correctness :/ (simduid
>> for example - if that goes away we do ..
Hi,
It's worth setting up the FPCR to flush to zero, round to nearest modes
especially under Ofast and ffast-math.
This patch attempts to do so and additionally cleans up ENDFILE_SPEC for
AArch64 GNU/Linux targets.
Built and tested cross with aarch64-none-elf. A native bootstrap and
test
On Thu, Sep 4, 2014 at 12:28 AM, Bill Schmidt
wrote:
> Hi,
>
> This patch adds more special handling to analyze_swaps to allow us to
> improve more computations. Previously I had disallowed VEC_SELECT in
> all cases. This is now changed to allow a select of a single lane,
> either for an extract
1 - 100 of 136 matches
Mail list logo
