Hi,
the attached patch uses the existing promote_function_mode hook. For
a libcall neither TYPE nor FNTYPE is available so I had to change a
few related function in order to deal with that.
The patch also fixes the s390 DFP problems.
Bye,
-Andreas-
2011-04-18 Andreas Krebbel
* ca
> The problem is quite real, it's just bitten us with some ARM changes.
> The fact that we don't have types for libcall function args is very
> unfortunate. Maybe that's something we need to change? It would be a
> much better fix, and it might not even be _that_ much work to add an
> optional type
On Mon, 18 Apr 2011, Jan Hubicka wrote:
> > > notice that OBJ_TYPE_REF is useless since foo got devitualized and it
> > > holds alive
> > > the useless s._vptr.S = &_ZTV1S[2]. We want to drop OBJ_TYPE_REF at time
> > > we fold
> > > first argument to constant.
> >
> > At some point we dropped
On Sun, Apr 17, 2011 at 11:27 PM, Easwaran Raman wrote:
> On Sun, Apr 17, 2011 at 1:39 PM, Steven Bosscher
> wrote:
>> On Sun, Apr 17, 2011 at 9:39 PM, Easwaran Raman wrote:
>>> @@ -372,8 +366,9 @@
>>> to elements will conflict. In case of unions we have
>>> to
Diego Novillo writes:
> On Thu, Apr 14, 2011 at 09:43, Richard Sandiford
> wrote:
>> +/* This file specifies a list of internal "functions". These functions
>> + differ from built-in functions in that they have no linkage and cannot
>> + be called directly by the user. They represent operat
On Mon, Apr 18, 2011 at 12:56 AM, Eric Botcazou wrote:
> Hi,
>
> this (partially) fixes the LTO bootstrap failure with Ada enabled on the
> mainline. The problem is an ICE whose origin is as follows:
> 1. during LTO stage, copy-prop eliminates a DECL_IN_CONSTANT_POOL variable
> from the GIMPLE I
Ping for this ARM PR fix:
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00200.html
which is a prerequisite for the ARM strided load/store stuff.
Richard
Ping for these two changes:
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00194.html
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00195.html
They just add a mode argument to TARGET_CANNOT_FORCE_CONST_MEM
and LEGITIMATE_CONSTANT_P (turning the latter into a hook).
This is directly needed fo
On Mon, 2011-04-04 at 11:34 +0100, Richard Sandiford wrote:
> This patch fixed PR target/46329, which is a problem that occurs
> when trying to reload:
>
> (set (reg:OI foo) (const_int 0))
>
> There is no move alternative for moving constants directly into large
> numbers of VFPs, and it pro
Richard Earnshaw writes:
> I'm uncomfortable about this. Generally the ARM port doesn't work well
> with the target-independent constant pool and it's better to assert that
> this is empty when it comes to final assembly generation. Can you
> clarify by way of example how this patch is working p
The following patches have remained unreviewed for more than a week:
[build, lto] Check plugin linker for level of plugin support
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00226.html
This one primarily needs a build maintainer and an LTO maintainer for
the backport question.
Denis Chertykov schrieb:
> 2011/4/17 Denis Chertykov :
>> 2011/4/15 Georg-Johann Lay :
>>> Finally, I exposed alternative #3 of the insns to the register
>>> allocator, because it is not possible to distinguish between
>>> overlapping or non-overlapping regs, and #3 does not need a scratch.
>>>
>>>
STRING_CST is now derived from tree_typed but we still clear a
tree_common sized chunk. Nathan, maybe grep for other sizeof()s
around the tree?
Installed as obvious.
Richard.
2011-04-18 Richard Guenther
PR middle-end/48650
* tree.c (build_string): STRING_CST is now derived
Hi,
attached is a couple of tweaks to the AVR back-end that we have in our tree:
1. libgcc cannot be built with RTL checking; the error is
/home/eric/svn/gcc/libgcc/../gcc/unwind-dw2-fde.c: In function 'search_object':
/home/eric/svn/gcc/libgcc/../gcc/unwind-dw2-fde.c:992:1: internal compiler
On Tue, Apr 12, 2011 at 3:59 PM, Richard Sandiford
wrote:
> This patch adds vec_load_lanes and vec_store_lanes optabs for instructions
> like NEON's vldN and vstN. The optabs are defined this way because the
> vectors must be allocated to a block of consecutive registers.
>
> Tested on x86_64-lin
Richard Guenther writes:
> On Tue, Apr 12, 2011 at 3:59 PM, Richard Sandiford
> wrote:
>> This patch adds vec_load_lanes and vec_store_lanes optabs for instructions
>> like NEON's vldN and vstN. The optabs are defined this way because the
>> vectors must be allocated to a block of consecutive re
On Fri, 15 Apr 2011, Steve Ellcey wrote:
> Vladimir and Dmitry,
>
> The new test you added, gcc.dg/pr48235.c, is failing on IA64 because
> it gets this excess message:
>
> cc1: note: -freorder-blocks-and-partition does not work on this architecture
>
> Could you modify the test to either filt
On Mon, Apr 18, 2011 at 11:26 AM, Richard Sandiford
wrote:
> Diego Novillo writes:
>> On Thu, Apr 14, 2011 at 09:43, Richard Sandiford
>> wrote:
>>> +/* This file specifies a list of internal "functions". These functions
>>> + differ from built-in functions in that they have no linkage and ca
We can fold OBJ_TYPE_REFs to direct calls if the callee is known. Even
easier now that we keep fntypes separately.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2011-04-18 Richard Guenther
* gimple.h (gimple_call_addr_fndecl): New function.
On Mon, Apr 18, 2011 at 1:24 PM, Richard Sandiford
wrote:
> Richard Guenther writes:
>> On Tue, Apr 12, 2011 at 3:59 PM, Richard Sandiford
>> wrote:
>>> This patch adds vec_load_lanes and vec_store_lanes optabs for instructions
>>> like NEON's vldN and vstN. The optabs are defined this way beca
Richard Guenther writes:
> On Mon, Apr 18, 2011 at 1:24 PM, Richard Sandiford
> wrote:
>> Richard Guenther writes:
>>> On Tue, Apr 12, 2011 at 3:59 PM, Richard Sandiford
>>> wrote:
Index: gcc/doc/md.texi
===
--- gcc/
This patchs makes us run the main C vect.exp tests twice, once with the
normal optimisation options and once with -flto added. It's a simple
way of covering the LTO streaming of the new "internal" functions,
but it should be useful more generally.
The testsuite runs quickly (even cross), so an ex
Richard,
> +foreach flags {"" "-flto"} {
> +dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/pr*.\[cS\]]] \
> + $flags $DEFAULT_VECTCFLAGS
> +dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/vect-*.\[cS\]]] \
> + $flags $DEFAULT_VECTCFLAGS
> +dg-runtest [lsort [glob -no
Hi,
[FWIW I can't approve patches, but some feedback nevertheless]
On Sun, 17 Apr 2011, Easwaran Raman wrote:
> This patch impoves the heuristic used in assigning stack location to
> stack variables. Currently, if there are 3 variables A, B and C with
> their sizes in increasing order and A an
On Mon, Apr 18, 2011 at 2:19 PM, Richard Sandiford
wrote:
> Richard Guenther writes:
>> On Mon, Apr 18, 2011 at 1:24 PM, Richard Sandiford
>> wrote:
>>> Richard Guenther writes:
On Tue, Apr 12, 2011 at 3:59 PM, Richard Sandiford
wrote:
> Index: gcc/doc/md.texi
> =
> "Rainer" == Rainer Orth writes:
Rainer> 2011-04-14 Rainer Orth
Rainer> * testsuite/libjava.jni/jni.exp
(gcj_jni_get_cxxflags_invocation):
Rainer> Add $libiconv to cxxflags for alpha*-dec-osf*.
This is ok, thanks.
Tom
> "Diego" == Diego Novillo writes:
>> Unfortunately, five bits are not sufficient for the alternate use of
>> cpp_hashnode.directive_index as a named operator index. So, I have reverted
>> the number of bits from five back to seven. As a result, we now have 34 bits
>> in small fields, and t
This avoids changing -P output with the PR48248 fix which appearantly
breaks Chrome (with has bogus assumptions on gcc -E -P output).
For 4.7 I think we should instead go with the 2nd patch and make -P
output smaller (which is the whole reason for this code path).
Thus, first patch is for the 4.
On Mon, Apr 18, 2011 at 12:46:52PM +0200, Richard Guenther wrote:
> STRING_CST is now derived from tree_typed but we still clear a
> tree_common sized chunk. Nathan, maybe grep for other sizeof()s
> around the tree?
Ouch, thanks for fixing. I grepped for 'struct tree_common' and turned
up a few
On Sat, 2011-04-16 at 12:34 +0800, Carrot Wei wrote:
> Hi Richard
>
> Thank you for the detailed explanation. It sounds like an inherent
> difficulty of rtl passes. Then the only opportunity is ldrb/strb
> instructions because they never affect cc registers.
There are also some comparison operat
Richard Guenther writes:
>> +/* Expand a call to internal function FN. ARGS is an array of the
>> + function's arguments and LHS is where the result should be stored. */
>> +
>> +void
>> +expand_internal_call (enum internal_fn fn, tree lhs, tree *args)
>> +{
>> + internal_fn_expanders[(int) f
This fixes up upper_bound_in_type and lower_bound_in_type which
build non-canonical integer csts for sizetype.
Bootstrapped and tested on x86_64-unknown-linux-gnu, installed.
Richard.
2011-04-18 Richard Guenther
* tree.c (upper_bound_in_type): Build properly canonicalized
IN
An IRIX 6.5 bootstrap over the weekend broke in libgo:
/vol/gcc/src/hg/trunk/local/libgo/runtime/go-append.c:22:1: error: conflicting
types for '__go_append'
/vol/gcc/src/hg/trunk/local/libgo/runtime/go-append.c:18:1: note: previous
declaration of '__go_append' was here
make[4]: *** [go-append.l
On Mon, Apr 18, 2011 at 9:33 PM, Richard Earnshaw wrote:
>
> On Sat, 2011-04-16 at 12:34 +0800, Carrot Wei wrote:
>> Hi Richard
>>
>> Thank you for the detailed explanation. It sounds like an inherent
>> difficulty of rtl passes. Then the only opportunity is ldrb/strb
>> instructions because they
obj-c++.dg/dwarf-2.mm fails on Tru64 UNIX since the target doesn't
support DWARF-2:
FAIL: obj-c++.dg/dwarf-2.mm -fgnu-runtime (test for excess errors)
Excess errors:
/vol/gcc/src/hg/trunk/local/gcc/testsuite/obj-c++.dg/dwarf-2.mm:1:0: error:
target system does not support the "dwarf-2" debug form
Committed to trunk and the 4.6 branch.
Richard.
2011-04-18 Richard Guenther
PR middle-end/46364
* g++.dg/torture/pr46364.C: New testcase.
Index: gcc/testsuite/g++.dg/torture/pr46364.C
===
--- gcc/testsuite/g++.d
The new gcc.dg/ipa/pr48195.c testcase fails on Tru64 UNIX:
FAIL: gcc.dg/ipa/pr48195.c (test for excess errors)
Excess errors:
cc1: error: LTO support has not been enabled in this configuration
Fixed as follows. Tested with the appropriate runtest invocation,
installed on mainline.
Raine
Rainer Orth writes:
>> +foreach flags {"" "-flto"} {
>> +dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/pr*.\[cS\]]] \
>> +$flags $DEFAULT_VECTCFLAGS
>> +dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/vect-*.\[cS\]]] \
>> +$flags $DEFAULT_VECTCFLAGS
>> +dg-runtest [
Richard,
>> I think you need to make sure that the target supports lto,
>> cf. check_effective_target_lto.
>
> Good point. How's this? Tested as before.
looks good, thanks.
Rainer
>
> Richard
>
>
> gcc/testsuite/
> * gcc.dg/vect/vect.exp: Run the main tests twice, one with -flt
On Mon, 2011-04-18 at 22:17 +0800, Carrot Wei wrote:
> On Mon, Apr 18, 2011 at 9:33 PM, Richard Earnshaw wrote:
> >
> > On Sat, 2011-04-16 at 12:34 +0800, Carrot Wei wrote:
> >> Hi Richard
> >>
> >> Thank you for the detailed explanation. It sounds like an inherent
> >> difficulty of rtl passes.
The new gfortran.dg/fmt_g0_5.f08 test fails on Tru64 UNIX at runtime
with
Floating point exception
but passes if compiled with -mieee.
The following patch implements this, tested with the appropriate runtest
invocation.
Ok for mainline?
Rainer
2011-04-18 Rainer Orth
* gfo
Janne Blomqvist wrote:
> On Thu, Mar 24, 2011 at 18:51, Jim Meyering wrote:
>> Janne Blomqvist wrote:
>>> On Tue, Mar 8, 2011 at 19:53, Jim Meyering wrote:
Relative to v2, I've added libgo/ to the list of exempt directories and
added
this recently discussed gfc_free patch, at the
On Mon, Apr 18, 2011 at 3:37 PM, Richard Sandiford
wrote:
> Richard Guenther writes:
>>> +/* Expand a call to internal function FN. ARGS is an array of the
>>> + function's arguments and LHS is where the result should be stored. */
>>> +
>>> +void
>>> +expand_internal_call (enum internal_fn f
On Mon, Apr 18, 2011 at 17:59, Rainer Orth
wrote:
> The new gfortran.dg/fmt_g0_5.f08 test fails on Tru64 UNIX at runtime
> with
>
> Floating point exception
>
> but passes if compiled with -mieee.
>
> The following patch implements this, tested with the appropriate runtest
> invocation.
>
> Ok fo
> Ah, no. We _did_ that in CCP but now we only adjust the OBJ_TYPE_REF
> expr in CCP and defer to fold_stmt to eventually "devirtualize" it.
> See PR45878. Then rev.165435 was necessary, as we dropped OBJ_TYPE_REF
> for the non-devirtualized call as well. As both cases were because
Hmm, sounds
On Mon, 18 Apr 2011, Jan Hubicka wrote:
> > Ah, no. We _did_ that in CCP but now we only adjust the OBJ_TYPE_REF
> > expr in CCP and defer to fold_stmt to eventually "devirtualize" it.
> > See PR45878. Then rev.165435 was necessary, as we dropped OBJ_TYPE_REF
> > for the non-devirtualized call a
The gcc.dg/vect/pr48377.c execution test is currently failing on Solaris
8/x86 and Solaris/SPARC. On Solaris 8/x86, this happens because the
test is compiled with -msse2, but the OS cannot execute SSE2 insns.
This is usually taken care of by vect.exp, which only runs the tests as
compile tests if
Mike Stump wrote:
> On Apr 16, 2011, at 4:04 AM, Gerald Pfeifer wrote:
>> On Fri, 15 Apr 2011, Mike Stump wrote:
>>> I think these are obvious.
>>
>> Which means that you can commit them without getting explicit approval
>
> Well, technically, it means nothing... It only means something if the
>
Hi,
On Fri, Apr 15, 2011 at 05:26:44PM +0200, Richard Guenther wrote:
> On Fri, 15 Apr 2011, Martin Jambor wrote:
>
> > Hi,
> >
> > early inlining can create virtual calls based on the part of an object
> > that represents an ancestor. This patch makes ipa-prop analysis able
> > to recognize su
Hi,
On Fri, Apr 15, 2011 at 05:28:49PM +0200, Richard Guenther wrote:
> On Fri, 15 Apr 2011, Martin Jambor wrote:
>
> > Hi,
> >
> > in order to speed up astar, I had to persuade the function that
> > decides whether a statement potentially modifies the dynamic type of
> > an object by storing a
As already done for the g++.dg/guality driver, I'm also disabling the
remaining guality testsuites on Tru64 UNIX: the gcc.dg one doesn't even
start:
gdb: took too long to attach
and the gfortran tests either fail or are unsupported. It seems gdb on
this platform is in a miserable state right now
> On Mon, 18 Apr 2011, Jan Hubicka wrote:
>
> > > Ah, no. We _did_ that in CCP but now we only adjust the OBJ_TYPE_REF
> > > expr in CCP and defer to fold_stmt to eventually "devirtualize" it.
> > > See PR45878. Then rev.165435 was necessary, as we dropped OBJ_TYPE_REF
> > > for the non-devirtua
Hi,
On Fri, Apr 15, 2011 at 05:38:50PM +0200, Richard Guenther wrote:
> On Fri, 15 Apr 2011, Martin Jambor wrote:
>
> > Hi,
> >
> > this is the patch that actually speeds up astar (together with the
> > previous one). It implements devirtualization based on global
> > objects. In the past we h
On Mon, Apr 18, 2011 at 18:08, Jim Meyering wrote:
> I've rebased and divided/reordered these changes as you suggested.
> Here are the fortran parts. I'll post the other parts separately.
> Parts 1 and 3 are manual. Part 2 is the big mechanical change with
> two manual adjustments:
>
> [PATCH 1
> -Original Message-
> From: Eric Botcazou [mailto:ebotca...@adacore.com]
> Sent: Monday, April 18, 2011 4:57 AM
> To: gcc-patches@gcc.gnu.org
> Subject: [AVR] Couple of tweaks
>
> Hi,
>
> attached is a couple of tweaks to the AVR back-end that we have in our
> tree:
>
> 1. libgcc can
2011/4/18 Georg-Johann Lay :
> Denis Chertykov schrieb:
>> 2011/4/17 Denis Chertykov :
>>> 2011/4/15 Georg-Johann Lay :
Finally, I exposed alternative #3 of the insns to the register
allocator, because it is not possible to distinguish between
overlapping or non-overlapping regs, and
Janne Blomqvist wrote:
> On Mon, Apr 18, 2011 at 18:08, Jim Meyering wrote:
>> I've rebased and divided/reordered these changes as you suggested.
>> Here are the fortran parts. I'll post the other parts separately.
>> Parts 1 and 3 are manual. Part 2 is the big mechanical change with
>> two man
Encountered them when testing IMAGE_INDEX.
Build and regtested on x86-64-linux.
OK for the trunk?
Tobias
2011-04-18 Tobias Burnus
PR fortran/18918
* array.c (gfc_match_array_ref): Check for too many codimensions.
* check.c (gfc_check_image_index): Check number of elements
in SUB argument
2011/4/18 Denis Chertykov :
> 2011/4/18 Georg-Johann Lay :
> Few years ago I have played with early splitting of anything possible
> (move,add,sub,and,...). The results was very bad.
> It's happened because flow of splitted insns (8bits insns) becomes
> unreadable for most of GCC optimisation passe
On Mon, 18 Apr 2011, Jakub Jelinek wrote:
> The GNU Compiler Collection version 4.4.6 has been released.
And the GCC 4.4 status adjusted as follows...
Gerald
Index: index.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/index.html,v
retri
On 04/18/11 18:34, Tobias Burnus wrote:
Encountered them when testing IMAGE_INDEX.
Build and regtested on x86-64-linux.
OK for the trunk?
Ok.
Thanks!
Daniel
--
http://www.pro-vegan.info/
--
Done: Arc-Bar-Cav-Kni-Ran-Rog-Sam-Tou-Val-Wiz
To go: Hea-Mon-Pri
Ping? Ping?
On Mon, Apr 11, 2011 at 11:00 AM, Paul Pluzhnikov
wrote:
> Ping?
--
Paul Pluzhnikov
Hi,
this is a ping of a patch from the big cgraph_removal series that has
not got in yet
(http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00523.html). I'd like
some Fortran reviewers to have a look, the fewer the calls to
cgraph_get_create_node the better.
Thanks a lot,
Martin
---
David,
> Thanks Rainer, if you cons up a patch to add the ".align" directive I'll ACK
> that as well.
here's the patch that I've been testing over the weekend. It caused no
change for HAVE_AS_SPARC_GOTDATA_OP on anything but Solaris 10 with gas
and Sun ld, where HAVE_AS_SPARC_GOTDATA_OP is 1 now
This is a port of an old patch of mine that got integrated in some
avr-gcc distributions.
Linking against __do_copy_data resp. __do_clear_bss is only needed if
there is actually stuff in .[ro]data resp. .bss. This saves some space
on tiny targets.
Regression-tested for C
Johann
2011-04-18 Geor
As described in the PR, fmt FAILs on targets without split-stack support
(anything but Linux, it seems) since the 32-bit TestScanInts tests
overflows the default thread stack of 1 MB. To avoid this, I've reduced
the recursion depth from 1000 to 800 which lets the test pass.
Rainer
2011-
On Fri, Apr 8, 2011 at 10:51 PM, Alan Modra wrote:
> This patch enables sibling calls for powerpc in a few more cases, and
> fixes bugs exposed by that change. We now
> a) Allow sibling calls via function pointer. At the time
> rs6000_function_ok_for_sibcall was written, I don't think access t
On Sat, Apr 09, 2011 at 12:21:46PM +0930, Alan Modra wrote:
> a) Allow sibling calls via function pointer. At the time
>rs6000_function_ok_for_sibcall was written, I don't think access to
>arg types of function pointer calls was available in the target
>hook/macro.
>
> + /* Functi
Rainer Orth writes:
> An IRIX 6.5 bootstrap over the weekend broke in libgo:
>
> /vol/gcc/src/hg/trunk/local/libgo/runtime/go-append.c:22:1: error:
> conflicting types for '__go_append'
> /vol/gcc/src/hg/trunk/local/libgo/runtime/go-append.c:18:1: note: previous
> declaration of '__go_append' w
Hi.
+/* To track if code will use .bss and/or .data */
+static int avr_need_clear_bss_p = 0;
+static int avr_need_copy_data_p = 0;
Change type avr_need_clear_bss_p and avr_need_copy_data_p vars to bool.
-#define ASM_OUTPUT_COMMON(STREAM, NAME, SIZE, ROUNDED)\
-do {\
- fputs (
Georg-Johann Lay schrieb:
> This is a port of an old patch of mine that got integrated in some
> avr-gcc distributions.
>
> Linking against __do_copy_data resp. __do_clear_bss is only needed if
> there is actually stuff in .[ro]data resp. .bss. This saves some space
> on tiny targets.
>
> Regress
Compiling dwarf2out.c with older versions of GCC yields a warning because
is_redundant_typedef has a prototype without the 'inline' keyboard and a
declaration with it, and is called from another function in-between.
The attached patchlet adds 'inline' to the prototype, as is done for other
func
This is slightly updated from my previous patch proposal. The test now
correctly work if multiple tests are executed in parallell.
2011-04-17 Martin Thuresson
* gcc/doc/invoke.tex: Document support for relative profile paths.
* gcc/testsuite/gcc.dg/pr47793.c: New test.
On Sat, 2011-04-16 at 00:49 +0200, Michael Matz wrote:
> Callers of expand_expr are expected to deal with the situation that the
> returned object doesn't have the desired mode, hence I think it's okay for
> expand_expr to return a DImode code_label rtx. Meaning we have to deal
> with it. The
On Mon, 18 Apr 2011, Rainer Orth wrote:
> [build, doc] Cleanup --enable-threads support
> http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00735.html
>
> This one needs build and doc maintainers.
The doc side is fine once the build side is approved.
Thanks,
Gerald
Hello Rainer,
* Rainer Orth wrote on Mon, Apr 04, 2011 at 06:15:28PM CEST:
> Here's a patch that does this. I'm not at all happy with the patch
> since it partially duplicates the logic to determine linker version
> numbers. While this could (and probably should) be generalized along
> the lines
Hi Rainer,
* Rainer Orth wrote on Mon, Mar 21, 2011 at 12:55:23PM CET:
> 2011-03-20 Rainer Orth
>
> gcc:
> * gthr-solaris.h: Remove.
> * gthr.h (_SOLARIS_THREADS): Don't include gthr-solaris.h, remove.
> * config/sol2.h (CPP_SUBTARGET_SPEC): Remove -threads support.
>
* Rainer Orth wrote on Mon, Apr 18, 2011 at 12:16:57PM CEST:
> The following patches have remained unreviewed for more than a week:
>
> [build, lto] Check plugin linker for level of plugin support
> http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00226.html
>
> This one primarily needs
Hello.
This patch change memory_move_cost function to stop using back end specific
type 'enum reg_class' in favor to reg_class_t. Also this allow do small
cleanup in ia64_register_move_cost function.
The patch has been bootstrapped on and regression tested on
x86_64-unknown-linux-gnu and ia
* Ralf Wildenhues wrote on Mon, Apr 18, 2011 at 08:23:37PM CEST:
> * Rainer Orth wrote on Mon, Apr 18, 2011 at 12:16:57PM CEST:
> > [build, doc] Cleanup --enable-threads support
> > http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00735.html
> >
> > This one needs build and doc maintainers
Hi,
this patch is result of another pass through ipa-inline.c cleaning up
confused and outdated comments.
I also fixed accounting of growth caused by the small function inlining.
We do not want to base it on program size before flattening, but afterwards.
Just like always inlines are not (and shou
Hi,
the test whether function contains computed goto is re-done by tree-inline (and
re-done
after optimization, so it is bit more sensitive). There is also nothing
language specific
in this.
Bootstrapped/regtested x86_64-linux, OK?
* cp/semantics.c (finish_goto_stmt): Do set UNINLINABL
Hi Jim,
* Jim Meyering wrote on Mon, Apr 18, 2011 at 05:40:03PM CEST:
> Mike Stump wrote:
> > On Apr 16, 2011, at 4:04 AM, Gerald Pfeifer wrote:
> >> On Fri, 15 Apr 2011, Mike Stump wrote:
> >>> I think these are obvious.
> >>
> >> Which means that you can commit them without getting explicit appr
On Mon, Apr 18, 2011 at 10:48 AM, Martin Thuresson wrote:
> This is slightly updated from my previous patch proposal. The test now
> correctly work if multiple tests are executed in parallell.
Forgot to mention that this patch has been bootstrapped without
regression on x86_64 and should be consi
On Apr 18, 2011, at 8:40 AM, Jim Meyering wrote:
> If you hadn't said anything, I would have committed those typo fixes
> by now, based on what I perceived as your review/approval and on my
> reading of this part of http://gcc.gnu.org/svnwrite.html:
>
>Free for all
>
>The following change
On 4/16/11, Diego Novillo wrote:
> On Apr 14, 2011 Lawrence Crowl wrote:
> > Unfortunately, five bits are not sufficient for the alternate
> > use of cpp_hashnode.directive_index as a named operator index.
> > So, I have reverted the number of bits from five back to seven.
> > As a result, we now
On Mon, Apr 18, 2011 at 14:58, Lawrence Crowl wrote:
> On 4/16/11, Diego Novillo wrote:
>> On Apr 14, 2011 Lawrence Crowl wrote:
>> > +/* Make sure the bitfield directive_index in include/cpplib.h is large
>> > + enough to index the entire table. */
>> > +
>> > +unsigned char too_many_directi
On 4/18/2011 11:47 AM, Jan Hubicka wrote:
> Bootstrapped/regtested x86_64-linux, OK?
>
> * cp/semantics.c (finish_goto_stmt): Do set UNINLINABLE flag on computed
> gotos.
Provided that we have this check elsewhere in the compiler and are
therefore confident we won't accidentally inli
On Mon, Apr 18, 2011 at 11:40, Jim Meyering wrote:
> Mike Stump wrote:
>> On Apr 16, 2011, at 4:04 AM, Gerald Pfeifer wrote:
>>> On Fri, 15 Apr 2011, Mike Stump wrote:
I think these are obvious.
>>>
>>> Which means that you can commit them without getting explicit approval
>>
>> Well, techni
Hello!
Attached patch macroizes movmsk and maskmov instructions. As an added
bonus, it also implements maskmovdqu tests, so "%%%" comment can be
removed.
2011-04-18 Uros Bizjak
* config/i386/i386.h (SSE_VEC_FLOAT_MODE_P): Remove.
(AVX_FLOAT_MODE_P): Ditto.
(AVX128_VEC
Since v3, I've rebased these and moved the fortran changes to precede
these (changing gfc_free to free introduced a few more instances
that this transformation deals with). As mentioned before, the
conditional-removing transformation was done mechanically with
two manual corrections. One to deal
LGTM
http://codereview.appspot.com/4446047/diff/1/gcc/libgcov.c
File gcc/libgcov.c (right):
http://codereview.appspot.com/4446047/diff/1/gcc/libgcov.c#newcode155
gcc/libgcov.c:155: filename? filename : "", e, v);
Better split it into two different printfs as the format is different --
otherwise
> "Jim" == Jim Meyering writes:
Jim> Since v3, I've rebased these and moved the fortran changes to precede
Jim> these (changing gfc_free to free introduced a few more instances
Jim> that this transformation deals with). As mentioned before, the
Jim> conditional-removing transformation was do
Tom Tromey wrote:
>> "Jim" == Jim Meyering writes:
>
> Jim> Since v3, I've rebased these and moved the fortran changes to precede
> Jim> these (changing gfc_free to free introduced a few more instances
> Jim> that this transformation deals with). As mentioned before, the
> Jim> conditional-re
Hi,
the attached patch replaces gfc_getmem with calls to xcalloc (from
libiberty). Apart from reducing duplicated code, calloc is better than
malloc + memset, as the allocator knows that the kernel always gives
out zeroed pages so in some cases it can avoid memset'in the area.
Regtested on x86_64
On Mon, Apr 18, 2011 at 11:44 AM, Jan Hubicka wrote:
> Hi,
> this patch is result of another pass through ipa-inline.c cleaning up
> confused and outdated comments.
>
> I also fixed accounting of growth caused by the small function inlining.
> We do not want to base it on program size before flatt
Diego Novillo wrote:
...
> Feel free to commit any of the patches in this series of typo fixes
> you just posted.
Thanks.
Done, with this:
From 7d5f43492809f2d96149853941bc8e0a9e7ecd10 Mon Sep 17 00:00:00 2001
From: Jim Meyering
Date: Mon, 18 Apr 2011 22:31:36 +0200
Subject: [PATCH] fix doubled-
On Mon, Apr 18, 2011 at 11:41:33PM +0300, Janne Blomqvist wrote:
> Hi,
>
> the attached patch replaces gfc_getmem with calls to xcalloc (from
> libiberty). Apart from reducing duplicated code, calloc is better than
> malloc + memset, as the allocator knows that the kernel always gives
> out zeroed
Martin Jambor wrote:
It seems to me that fortran can call cgraph_create_node directly
without checking for its existence first.
Bootstrapped and tested on x86_64-linux without any problems, tests on
i686 in progress.
The patch is OK.
Tobias
2011-03-18 Martin Jambor
* trans-decl.c
On Apr 18, 2011, at 10:24 AM, David Edelsohn wrote:
>> This patch does affect darwin, not only in removing what are now
>> redundant insn patterns, but also (a) and (b) above. Would someone
>> please run a bootstrap and regression test on darwin for me? I don't
>> have a machine handy to confirm
1 - 100 of 126 matches
Mail list logo
