Hello!
> > I am proposing a series of patches which is supposed to address the
> > remaining issues (I am aware of) preventing us from enabling the
> > -ftrack-macro-expansion by default.
> >
> > The idea is to address each issue I notice in the course of trying to
> > bootstrap the compiler and r
Initial implementation of RTX_COSTS target function for rx-elf. Minor
increase in coremark scores, and enables division by multiplication of
reciprocals, tested on trunk and 4.7. Ok for trunk and/or 4.7 branch?
* config/rx/rx.c (TARGET_RTX_COSTS): Define.
(rx_rtx_costs): New.
I
On 04/30/2012 07:37 PM, Paolo Carlini wrote:
Thus, my question would be: is something like the below in the right
direction? The alternate possibility I can see, would be basically
redoing a slightly slimmed version of for_each_template_parm specialized
for our needs (a few less conditionals)
I
This revision splits the medium/large code model toc reference after
reload. I expected this to be more difficult, but it turned out
surprisingly easy. Besides creating the rtl that way, and tweaking
toc_relative_expr_p to match, it was just: Move the 'R' constraint
handling out of various movsi
On 04/30/2012 08:43 PM, Michael Hope wrote:
On 1 May 2012 10:01, Jeff Law wrote:
On 04/30/2012 03:47 PM, Michael Hope wrote:
2012-04-27 Michael Hope
* config/arm/linux-eabi.h (GLIBC_DYNAMIC_LINKER): Pick the loader
using a spec rule.
Michael,
can you try this patch please
Hi,
This patch improves the expansion of COND_EXPR into RTL, directly
using conditional moves.
I had to fix a bug in the x86 backend where emit_conditional_move
could cause a crash as we had a comparison mode of DImode which is not
handled by the 32bit part. can_conditionally_move_p return true
On 1 May 2012 10:01, Jeff Law wrote:
> On 04/30/2012 03:47 PM, Michael Hope wrote:
>
2012-04-27 Michael Hope
* config/arm/linux-eabi.h (GLIBC_DYNAMIC_LINKER): Pick the loader
using a spec rule.
>>>
>>> Michael,
>>>
>>> can you try this patch please. It s
On Thu, Apr 26, 2012 at 1:00 AM, Alan Modra wrote:
> PR target/53038
> * config/rs6000/rs6000.c (load_lr_save, restore_saved_lr,
> load_cr_save, add_crlr_cfa_restore): New functions.
> (rs6000_restore_saved_cr): Rename to..
> (restore_saved_cr): ..this. Add cfa
On Thu, Feb 10, 2011 at 1:44 PM, Jakub Jelinek wrote:
> Hi!
>
> single_nonlooparound_use doesn't ignore debug uses, which results in
> -fcompare-debug failures. Fixed thusly, bootstrapped/regtested on
> x86_64-linux and i686-linux, ok for trunk?
>
> 2011-02-10 Jakub Jelinek
>
> PR debug
I've been reading your patches...
On Apr 30, 2012, at 4:34 PM, DJ Delorie wrote:
> In going through the Apple test cases, I discovered one HUGE
> disadvantage to using __attribute__ to tag structures for bit reversal
> - it doesn't propogate. I.e.:
This is why pragma exists. :-) Certainly, onc
Hi again,
On 04/29/2012 11:42 AM, Paolo Carlini wrote:
This might just be a matter of calling for_each_template_parm and
returning 1 if we see any template parameter.
Good. Today I quickly tried something along these lines (see 'p'
attachment) and got some fails:
Ah, well. I guess for_each_t
> Set up a cron job to ping once a day. :-) Did you ever dig up the
> Apple test cases from the APPLE LOCAL work I pointed you at earlier?
> They will be more comprehensive that any testing you've done, and,
> if you get them to all pass, the work should be closer to being
> complete. The featur
* crossconfig.m4: Since we know that all TPF builds are cross-
builds and cannot run configuration-time link tests, do not
allow it; just go with known supported linker options.
* configure: Regenerate (called as GLIBCXX_CROSSCONFIG).
Index: crossconfig.m4
===
On 04/28/2012 09:56 AM, Bernd Schmidt wrote:
The problem is that resize_reg_info only resizes the data
structure, but leaves it uninitialized.
Argh. Something like this maybe (currently testing on i686-linux, ok if it
passes?)
That patch cleans up the SPEC build failures I was seeing and a
On 04/30/2012 03:47 PM, Michael Hope wrote:
2012-04-27 Michael Hope
* config/arm/linux-eabi.h (GLIBC_DYNAMIC_LINKER): Pick the loader
using a spec rule.
Michael,
can you try this patch please. It should make it possible to then
create linux-eabihf.h containing just
#undef
On 1 May 2012 03:24, Richard Earnshaw wrote:
> On 27/04/12 00:27, Michael Hope wrote:
>> On 27 April 2012 08:20, Carlos O'Donell wrote:
>>> On Mon, Apr 23, 2012 at 5:36 PM, Michael Hope
>>> wrote:
2012-04-24 Michael Hope
Richard Earnshaw
* config/arm/l
Hi again,
On 04/30/2012 07:04 PM, Jason Merrill wrote:
On 04/29/2012 11:42 AM, Paolo Carlini wrote:
This might just be a matter of calling for_each_template_parm and
returning 1 if we see any template parameter.
Good. Today I quickly tried something along these lines (see 'p'
attachment) and g
Hello!
On BMI2 targets, we prefer mulx, so switch places of constraints 0 and 1.
2012-04-30 Uros Bizjak
PR target/53141
* config/i386/i386.md (*umul3_1): Switch places of
constraints 0 and 1.
Tested on x86_64-pc-linux-gnu {,-m32}, committed to mainline SVN.
Uros.
Ind
On 04/30/2012 04:38 PM, Jan Hubicka wrote:
This makes sense to me.
Thanks (and sorry for bogus subject).
I bootstrapped/regtested the change and tested it indeed lets me build kernel
and Mozilla with the sanity check that there are no function<->variable
aliases. Is the change OK?
OK.
Jason
> This makes sense to me.
Thanks (and sorry for bogus subject).
I bootstrapped/regtested the change and tested it indeed lets me build kernel
and Mozilla with the sanity check that there are no function<->variable
aliases. Is the change OK?
Honza
>
> Jason
This makes sense to me.
Jason
Reviewers: xur, davidxl, iant2,
Message:
I backported the following patch:
2012-03-12 Richard Guenther
* gthr.h (__GTHREAD_MUTEX_INIT_FUNCTION): Adjust specification.
* gthr-posix.h (__GTHREAD_MUTEX_INIT_FUNCTION): Define.
(__gthread_mutex_init_function): New function
> Hello!
>
> There is a problem with multiple calls of cgraph_node_name in fprintf
> dumps. Please note that C++ uses caching in
> cxx_printable_name_internal (aka LANG_HOOKS_DECL_PRINTABLE_NAME), so
> when cxx_printable_name_internal is called multiple times from printf
> (i.e. fprintf "%s/%i ->
On 04/30/2012 06:43 AM, Jakub Jelinek wrote:
Hi!
As shown on the testcase below, if REE modifies some sign/zero extension
insn, which is on the candidate vector, as a def_insn of some other
extension, before combine_reaching_defs is called on that insn, we might
ignore the changes done to that i
Hello!
There is a problem with multiple calls of cgraph_node_name in fprintf
dumps. Please note that C++ uses caching in
cxx_printable_name_internal (aka LANG_HOOKS_DECL_PRINTABLE_NAME), so
when cxx_printable_name_internal is called multiple times from printf
(i.e. fprintf "%s/%i -> %s/%i"), it ca
On 30.04.12 18:54, David Edelsohn wrote:
On Sat, Apr 28, 2012 at 3:45 PM, Andreas Tobler wrote:
Hello all,
I did a backport of the powerpc64-freebsd support to the 4.7 branch, here
the results:
http://gcc.gnu.org/ml/gcc-testresults/2012-04/msg02768.html
http://gcc.gnu.org/ml/gcc-testresults/2
This patch to libgo fixes the build on MIPS GNU/Linux. I haven't been
able to fully test gccgo, as the MIPS64 machine in the GCC compile farm
is running a version of glibc that is too old--it does not support
makecontext and friends. However, this patch at least gets past the
immediate build prob
Hi,
> Your patch only looks at immediate subexpressions; I believe what the ABI
> means is any subexpression, which is why I think something involving
> walk_tree might be the way to go.
Ok, I'm going to open code something using walk_tree and quite similar to
for_each_template_parm but checki
On Mon, Apr 30, 2012 at 3:10 PM, Jakub Jelinek wrote:
> On Mon, Apr 30, 2012 at 02:54:05PM +0200, Uros Bizjak wrote:
>> > My recent changes to zero_extend expanders should handle this
>> > automatically, and will undo generation of zero_extend pattern. Please
>> > see zero_extendsi2_and expander,
On 04/29/2012 11:42 AM, Paolo Carlini wrote:
This might just be a matter of calling for_each_template_parm and
returning 1 if we see any template parameter.
Good. Today I quickly tried something along these lines (see 'p'
attachment) and got some fails:
Ah, well. I guess for_each_template_par
On 04/30/2012 01:47 AM, Richard Earnshaw wrote:
> +operands[2] = gen_rtx_REG (SImode, REGNO (target) + 1);
> +return "ldrexd%?\t%0, %2, %C1";
> + }
Use "ldrexd%?\t%0, %H0, %C1", then you don't need to construct operands[2]
Otherwise, OK.
Ah, nice. I also fixed the other instanc
Hello,
"sed s/…/p -e d" as used in s-header-vars doesn't work on
at least ia64-hpux, where s/.../p only prints out if -n was
requested as well.
The attached patch fixes this by using '-n' instead of a
trailing '-e d'
We have been using this on all our gcc-4.5 based configurations
for a couple of
On Sat, Apr 28, 2012 at 3:45 PM, Andreas Tobler wrote:
> Hello all,
>
> I did a backport of the powerpc64-freebsd support to the 4.7 branch, here
> the results:
>
> http://gcc.gnu.org/ml/gcc-testresults/2012-04/msg02768.html
> http://gcc.gnu.org/ml/gcc-testresults/2012-04/msg02767.html
>
> Is the
Some other changes I've been working on needed this const-correctness
adjustment.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit e69aeb777fdbbcecbc603c9b91ef3dc17cf812a5
Author: Jason Merrill
Date: Thu Apr 19 10:02:22 2012 -0400
* dwarf2out.c (comdat_symbol_id): Add const.
(
OK.
Jason
Hello,
as a second step in refactoring this patch introduces a routine
vect_find_single_use to determine whether a defining statement
has one single use within the current vectorization domain.
The helper is then called wherever that check is currently
open-coded. There should be no change in be
Hello,
in working on a fix for PR 52633, I noticed that tree-vect-patterns.c now
contains a number of copies of rather similar code (of which my patch
would have added another copy), so it seems to make sense to do a bit of
refactoring first.
This patch introduced a new helper function "vect_same
On Apr 29, 2012, at 10:38 AM, Dodji Seketeli wrote:
> While bootstrapping the tree again, it appeared that an output
> regression of the objc test objc.dg/foreach-7.m flew below my radar.
> This looks fairly obvious to me, but I am CC-ing Mike Stump, just in
> case.
That's fine.
On 30/04/12 16:36, Georg-Johann Lay wrote:
> Richard Earnshaw schrieb:
>> On 30/04/12 15:07, Richard Sandiford wrote:
>>
>>> Richard Earnshaw writes:
>>>
Jim MacArthur wrote:
> The current code in reg_fits_class_p appears to be incorrect; since
> offset may be negative, it's nec
Richard Earnshaw schrieb:
On 30/04/12 15:07, Richard Sandiford wrote:
Richard Earnshaw writes:
Jim MacArthur wrote:
The current code in reg_fits_class_p appears to be incorrect; since
offset may be negative, it's necessary to check both ends of the range
otherwise an array overrun or unde
On 27/04/12 00:27, Michael Hope wrote:
> On 27 April 2012 08:20, Carlos O'Donell wrote:
>> On Mon, Apr 23, 2012 at 5:36 PM, Michael Hope
>> wrote:
>>> 2012-04-24 Michael Hope
>>>Richard Earnshaw
>>>
>>>* config/arm/linux-eabi.h (GLIBC_DYNAMIC_LINKER_SOFT_FLOAT): Define.
Richard Earnshaw writes:
> On 30/04/12 15:39, Richard Sandiford wrote:
>> Richard Earnshaw writes:
>>> On 30/04/12 15:07, Richard Sandiford wrote:
Richard Earnshaw writes:
> On 26/04/12 14:20, Jim MacArthur wrote:
>> The current code in reg_fits_class_p appears to be incorrect; sinc
Ping ?
http://gcc.gnu.org/ml/gcc-patches/2012-04/txt00094.txt
Adding Gabriel in Cc: in case this falls in the diagnostic category.
On Fri, 20 Apr 2012, Marc Glisse wrote:
Hello,
here is a patch that requires parentheses around the argument of sizeof I
am not sure what variadic76.C tests
Ping?
http://gcc.gnu.org/ml/gcc-patches/2012-04/msg01034.html
Since then, I've run a c,c++ bootstrap and:
make -k check RUNTESTFLAGS="--target_board=my-sde-sim"
where my-sde-sim is the dejagnu board posted by H.J. Lu to run tests
inside Intel's simulator, no difference between before and after
On 30/04/12 15:39, Richard Sandiford wrote:
> Richard Earnshaw writes:
>> On 30/04/12 15:07, Richard Sandiford wrote:
>>> Richard Earnshaw writes:
On 26/04/12 14:20, Jim MacArthur wrote:
> The current code in reg_fits_class_p appears to be incorrect; since
> offset may be negative,
Richard Earnshaw writes:
> On 30/04/12 15:07, Richard Sandiford wrote:
>> Richard Earnshaw writes:
>>> On 26/04/12 14:20, Jim MacArthur wrote:
The current code in reg_fits_class_p appears to be incorrect; since
offset may be negative, it's necessary to check both ends of the range
>>>
On Mon, Apr 16, 2012 at 9:09 AM, Jan Hubicka wrote:
> Hi,
> this patch moves cgraph/varpool hashes into symbol table hashes, so the
> symbol table is actually almost a symbol table ;)
> Work done.
>
> Bootstrapped/regtested x86_64-linux. Will commit it after bit of more
> testing - the assembler n
On 30/04/12 15:07, Richard Sandiford wrote:
> Richard Earnshaw writes:
>> On 26/04/12 14:20, Jim MacArthur wrote:
>>> The current code in reg_fits_class_p appears to be incorrect; since
>>> offset may be negative, it's necessary to check both ends of the range
>>> otherwise an array overrun or u
Hello Richard,
Thanks for the constructive exchange :-)
On Apr 26, 2012, at 10:48 , Richard Guenther wrote:
>> In particular, I'm pretty sure that we can get component
>> refs of integral modes that access a smaller range of bits
>> than what the mode conveys. It is common with packing or
>>
On 16/04/12 13:41, Richard Earnshaw wrote:
P.S. This patch can't actually be committed until my "NEON DImode
immediate constants" patch is approved and committed. (Without that the
load #0 needs a constant pool, and loading constants this late has a bug
at -O0.)
neon-neg64.patch
2012-04-12 An
On 30/04/12 15:07, Richard Sandiford wrote:
> Richard Earnshaw writes:
>> On 26/04/12 14:20, Jim MacArthur wrote:
>>> The current code in reg_fits_class_p appears to be incorrect; since
>>> offset may be negative, it's necessary to check both ends of the range
>>> otherwise an array overrun or u
On 27/04/12 16:17, Richard Earnshaw wrote:
This is OK.
Thanks, now committed.
It would be good to merge all the target32 movdi variants into one
pattern and then use alternative enabling to deal with the different
valid alternatives.
Yes, I'll take a look.
Andrew
Richard Earnshaw writes:
> On 26/04/12 14:20, Jim MacArthur wrote:
>> The current code in reg_fits_class_p appears to be incorrect; since
>> offset may be negative, it's necessary to check both ends of the range
>> otherwise an array overrun or underrun may occur when calling
>> in_hard_reg_set
On Mon, Apr 30, 2012 at 3:34 PM, Uros Bizjak wrote:
> On Mon, Apr 30, 2012 at 3:10 PM, Jakub Jelinek wrote:
>> On Mon, Apr 30, 2012 at 02:54:05PM +0200, Uros Bizjak wrote:
>>> > My recent changes to zero_extend expanders should handle this
>>> > automatically, and will undo generation of zero_ext
On 26/04/12 14:20, Jim MacArthur wrote:
> The current code in reg_fits_class_p appears to be incorrect; since
> offset may be negative, it's necessary to check both ends of the range
> otherwise an array overrun or underrun may occur when calling
> in_hard_reg_set_p. in_hard_reg_set_p should als
On 24/04/12 17:31, Greta Yorsh wrote:
> Generalize ldm_stm_operation_p with additional parameters that will be used
> by epilogue patterns:
> * machine mode to support both SImode and DFmode registers
> * flag to request consecutive registers in the register list
> * flag to indicate whether
On 24/04/12 17:26, Greta Yorsh wrote:
> Move the code of the special predicates load_multiple_operation and
> store_multiple_operation into a separate function. No change in
> functionality.
>
> gcc/ChangeLog
>
> 2012-04-24 Ian Bolton
> Sameera Deshpande
> Greta Yors
Hi!
As shown on the testcase below, if REE modifies some sign/zero extension
insn, which is on the candidate vector, as a def_insn of some other
extension, before combine_reaching_defs is called on that insn, we might
ignore the changes done to that insn and just assume it is only an
extension to
Hi!
compute_all_dependences in 4.7+ can fail, and cond_if_else_store_replacement
isn't prepared to handle the chrec_dont_know DDR added there in case of
failure (with NULL DDR_A/DDR_B).
Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux,
ok for trunk/4.7?
BTW, tree-ssa-loop-pref
On Mon, Apr 30, 2012 at 3:10 PM, Jakub Jelinek wrote:
> On Mon, Apr 30, 2012 at 02:54:05PM +0200, Uros Bizjak wrote:
>> > My recent changes to zero_extend expanders should handle this
>> > automatically, and will undo generation of zero_extend pattern. Please
>> > see zero_extendsi2_and expander,
On 28/04/12 11:04, Alexey Kravets wrote:
> Hi guys,
> Please, take a look at this patch. It fixes the invalid star symbol
> processing in validate_switches function reported in GCC bugzilla:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51020
>
> With this patch invalid options are no longer acce
On Mon, Apr 30, 2012 at 02:54:05PM +0200, Uros Bizjak wrote:
> > My recent changes to zero_extend expanders should handle this
> > automatically, and will undo generation of zero_extend pattern. Please
> > see zero_extendsi2_and expander, and how it handles
> > TARGET_ZERO_EXTEND_WITH_AND targets.
On Thu, Apr 26, 2012 at 5:34 PM, Uros Bizjak wrote:
> On Thu, Apr 26, 2012 at 5:28 PM, Jakub Jelinek wrote:
>
>>> > We have a splitter for reg1 = reg2 & 0x, but only if regnums
>>> > are different. But movl %edi, %edi is a cheaper variant of
>>> > andq $0x, %rdi even with the sam
Hi,
This is a ping for the patch that add 'function_vector' attribute for
H8300 targets:
http://gcc.gnu.org/ml/gcc-patches/2012-04/msg00250.html
The patch has been updated to address Jeff's comments which can be referred
at: http://gcc.gnu.org/ml/gcc-patches/2012-03/msg01884.html
Please review
Dodji Seketeli writes:
> I am proposing a series of patches which is supposed to address the
> remaining issues (I am aware of) preventing us from enabling the
> -ftrack-macro-expansion by default.
>
> The idea is to address each issue I notice in the course of trying to
> bootstrap the compiler
On 04/30/2012 11:50 AM, Richard Earnshaw wrote:
> On 30/04/12 09:51, Andrew Haley wrote:
>> On 04/27/2012 10:30 PM, Richard Henderson wrote:
>>> We can perform a single-copy atomic load with an ldrexd insn.
>>> If the load is all we care about, we need not pair this with
>>> a strexd.
>>
>> Can we?
On 30/04/12 09:51, Andrew Haley wrote:
> On 04/27/2012 10:30 PM, Richard Henderson wrote:
>> We can perform a single-copy atomic load with an ldrexd insn.
>> If the load is all we care about, we need not pair this with
>> a strexd.
>
> Can we? It's good to know. I have had a long email exchange
Ping?
Also added Richard in Cc: as the author of the C front-end code I am
sharing with the C++ front-end.
Looks like I forgot to say it here (was in bugzilla), but the patch was
tested in a c,c++ bootstrap + make -k check on linux x86_64.
Reattaching the patch (now using the recommended d
On 04/27/2012 10:30 PM, Richard Henderson wrote:
> We can perform a single-copy atomic load with an ldrexd insn.
> If the load is all we care about, we need not pair this with
> a strexd.
Can we? It's good to know. I have had a long email exchange with
engineers at ARM, and they would not say th
On 27/04/12 22:30, Richard Henderson wrote:
> We can perform a single-copy atomic load with an ldrexd insn.
> If the load is all we care about, we need not pair this with
> a strexd.
>
> Ok?
>
>
> r~
>
>
> d-arm-ldi
>
>
> * config/arm/arm.md (UNSPEC_LL): New.
> * config/arm/sync.
In -gnatct mode, the compiler doesn't generate code and gigi is only invoked to
lay out and back-annotate types. There were old issues when representation
clauses are present: fields are shifted, sizes are wrong, component clauses
are overridden.
Tested on i586-suse-linux, applied on the mainl
The MS bitfields ABI and its stringent rules totally break packing in Ada, so
we cannot use the C layout in this case. The i386.c hunk fixes an oversight
which would result in gazillions of warnings for QUAL_UNION_TYPE.
Tested on i586-suse-linux and i686-pc-mingw32, applied on the mainline and
On Fri, Apr 27, 2012 at 3:30 PM, Paolo Bonzini wrote:
> tzcnt is encoded as "rep;bsf" and unlike lzcnt is a drop-in replacement
> if we don't care about the flags (it has the same semantics for non-zero
> values).
>
> Since bsf is usually slower, just emit tzcnt unconditionally. However,
> write
Although the Ada type system provides strong enough guarantees to make a safe
and effective usage of -fstrict-aliasing, there is an explicit "unsafe"
construct (Unchecked_Conversion) for which these guarantees aren't valid any
more. That's why GNAT also provides the No_Strict_Aliasing pragma (m
It is still OK :-)
On Mon, Apr 30, 2012 at 1:22 AM, Dodji Seketeli wrote:
> Dodji Seketeli writes:
>
>> Gabriel Dos Reis writes:
>>
>>> OK.
>>
>> Thank you.
>>
>> While bootstrapping the tree again, it appeared that an output
>> regression of the objc test objc.dg/foreach-7.m flew below my rada
75 matches
Mail list logo
