Oleg Endo wrote:
> This improves some cases where the T bit is stored as all ones or all
> zeros to a GP reg.
> Tested on rev 189916 with
> make -k check RUNTESTFLAGS="--target_board=sh-sim
> \{-m2/-ml,-m2/-mb,-m2a/-mb,-m2a-single/-mb,-m4/-ml,-m4/-mb,
> -m4-single/-ml,-m4-single/-mb,-m4a-single/-m
Oleg Endo wrote:
> In cases where dynamic shifts are available, the ashlsi3_d pattern is
> picked first and never converted back to a short sequence of constant
> shifts. Preferring short constant shift sequences over dynamic shifts
> can potentially reduce the need for an extra register to hold
Oleg Endo wrote:
> The attached patch adds the combine patterns as discussed in the PR.
> Tested on rev 189916 with
> make -k check RUNTESTFLAGS="--target_board=sh-sim
> \{-m2/-ml,-m2/-mb,-m2a/-mb,-m2a-single/-mb,-m4/-ml,-m4/-mb,
> -m4-single/-ml,-m4-single/-mb,-m4a-single/-ml,-m4a-single/-mb}"
>
> From: Ulrich Weigand
> Date: Fri, 27 Jul 2012 17:24:08 +0200
> Richard (Earnshaw) has asked me to take over working on this patch now.
>
> I've now made the change requested above and removed the size argument.
> The target is now simply asked to return the required alignment for the
> given v
Hello,
In cases where dynamic shifts are available, the ashlsi3_d pattern is
picked first and never converted back to a short sequence of constant
shifts. Preferring short constant shift sequences over dynamic shifts
can potentially reduce the need for an extra register to hold the shift
count.
Hello,
The attached patch adds the combine patterns as discussed in the PR.
Tested on rev 189916 with
make -k check RUNTESTFLAGS="--target_board=sh-sim
\{-m2/-ml,-m2/-mb,-m2a/-mb,-m2a-single/-mb,-m4/-ml,-m4/-mb,
-m4-single/-ml,-m4-single/-mb,-m4a-single/-ml,-m4a-single/-mb}"
and no new failures.
Hello,
In cases where dynamic shifts are available, the ashlsi3_d pattern is
picked first and never converted back to a short sequence of constant
shifts. Preferring short constant shift sequences over dynamic shifts
can potentially reduce the need for an extra register to hold the shift
count.
Hello,
This improves some cases where the T bit is stored as all ones or all
zeros to a GP reg.
Tested on rev 189916 with
make -k check RUNTESTFLAGS="--target_board=sh-sim
\{-m2/-ml,-m2/-mb,-m2a/-mb,-m2a-single/-mb,-m4/-ml,-m4/-mb,
-m4-single/-ml,-m4-single/-mb,-m4a-single/-ml,-m4a-single/-mb}"
a
I will add the comment. However, my hope is that will be short lived
knowledge, but i am a couple of patches away from that.
kenny
On 07/29/2012 05:31 PM, Steven Bosscher wrote:
On Sun, Jul 29, 2012 at 6:11 PM, Kenneth Zadeck
wrote:
* rtl.h (CONST_DOUBLE_AS_INT_P, CONST_DOUBLE_AS_FLOAT
On Sun, Jul 29, 2012 at 6:11 PM, Kenneth Zadeck
wrote:
> * rtl.h (CONST_DOUBLE_AS_INT_P, CONST_DOUBLE_AS_FLOAT_P): New
> macros.
Hello Kenny,
Thanks for doing this!
Can you please also update the comment before CONST_DOUBLE in rtl.def?
According to rtl.def CONST_DOUBLEs are only for flo
Tested only as far as cross-compile. I had a browse through
objdump of libatomic for a brief sanity check.
Can you please test on real hw and report back?
r~
---
gcc/config/s390/s390-protos.h |3 +-
gcc/config/s390/s390.c| 90 +-
gcc/config/s390/s390.md | 373
On 07/29/2012 07:38 PM, Jonathan Wakely wrote:
Yes please. The point is to allow people to review and comment before
the patch is applied, and some people only subscribe to gcc-patches
not libstdc++.
I don't have a strong opinion, but I must say that I don't understand
why those people don't su
On 29 July 2012 18:15, François Dumont wrote:
> Patch applied. I usually CC to gcc-patches when I signal that it has been
> applied. Should I send it all my patch proposals ?
Yes please. The point is to allow people to review and comment before
the patch is applied, and some people only subscribe
Patch applied. I usually CC to gcc-patches when I signal that it has
been applied. Should I send it all my patch proposals ?
François
On 07/28/2012 11:18 PM, Jonathan Wakely wrote:
Please remember to CC gcc-patches too.
On 28 July 2012 21:49, François Dumont wrote:
Hi
Here is the patch
sorry, will fix all of this. thanks.
On 07/29/2012 12:33 PM, Richard Sandiford wrote:
Kenneth Zadeck writes:
Given that Richard Sandiford advised on all of the non trivial changes,
I am going to check this patch in in the next few days unless i hear
some comments otherwise.
TBH I'd only look
> Given that Richard Sandiford advised on all of the non trivial changes,
> I am going to check this patch in in the next few days unless i hear
> some comments otherwise. This patch has been fully tested on the x86-64.
Pasto in the rtl.h change. Please also avoid the long lines in there.
--
Kenneth Zadeck writes:
> Given that Richard Sandiford advised on all of the non trivial changes,
> I am going to check this patch in in the next few days unless i hear
> some comments otherwise.
TBH I'd only looked at the ones you flagged. This time...
> diff -puNr '--exclude=.git' '--exclude
This patch is a minor, mostly mechanical cleanup of CONST_DOUBLE. It
wraps all of the checks of CONST_DOUBLE in one of three macros:
CONST_DOUBLE_AS_INT_P, CONST_DOUBLE_AS_FLOAT_P, or CONST_DOUBLE_P is it
used for both. There were some non obvious changes that Richard
Sandiford told me the ho
Hi all,
here is a second patch for PR 51081, which fixes the rejection of the
valid test case in comment 1.
The problem was that the checking for intrinsics came too early (in
gfc_match_rvalue), so that INT2 was marked as intrinsic, before it was
known that there is a contained procedure which sh
Hello world,
here is an updated patch for PR 54033, this time with test cases.
Thanks to Janis for pointing me in the right direction with these.
Regression-tested. OK for trunk?
Thomas
2012-07-29 Thomas König
PR fortran/54033
* scanner.c (add_path_to_list): Emit warning if
Andrew Pinski writes:
> On Fri, Jul 27, 2012 at 12:15 PM, Sandra Loosemore
> wrote:
>> Richard,
>>
>> This ancient patch to tweak mips_legitimize_address
>>
>> http://gcc.gnu.org/ml/gcc/2008-11/msg00294.html
>>
>> seems to never have been applied. Do you have any idea whether this is
>> still a
21 matches
Mail list logo
