On Sun, Oct 28, 2012 at 2:37 AM, H.J. Lu wrote:
>>> As suggested by Richard S. [1], after the patch that converts subreg:M
>>> (op:N (...)(...)) to op:M (subreg:M (...) subreg:M (...)), we can
>>> remove several peephole2 patterns that handle subregs of PLUS, MINUS
>>> and MULT operators. I have
Hi,
Right now aarch64-*-* is broken because the C front-end also defines
a __builtin_thread_pointer which causes an internal error. We don't
need to do the __builtin_thread_pointer support in the back-end. This
patch moves aarch64 to using the common code instead.
OK? Built for aarch64-thunde
Hi Tobi,
It looks straightforward to me but I am not convinced that a fortran
maintainer should be giving the green light on this :-)
Cheers
Paul
On 28 October 2012 16:28, Tobias Schlüter
wrote:
>
> Ping. This issue stands in the way of a very simple solution of PR
> fortran/51727. I've re-a
On Mon, Oct 29, 2012 at 9:07 AM, Uros Bizjak wrote:
As suggested by Richard S. [1], after the patch that converts subreg:M
(op:N (...)(...)) to op:M (subreg:M (...) subreg:M (...)), we can
remove several peephole2 patterns that handle subregs of PLUS, MINUS
and MULT operators.
On 29 Oct 2012, at 09:26, "Andrew Pinski"
wrote:
> Hi,
> Right now aarch64-*-* is broken because the C front-end also defines
> a __builtin_thread_pointer which causes an internal error. We don't
> need to do the __builtin_thread_pointer support in the back-end. This
> patch moves aarch64 to
Is_Signed_Integer_Arithmetic_Op is called to recognize an arithmetic operation
on signed integers, or on if expressions and case expressions whose dependent
expressions are arithmetic operation on signed integers. A special case was
included for N_Case_Expression_Alternative which should have been
This patch gives a warning when unary minus is used in a context that
requires parentheses, and no parentheses are present, as shown by the
following test:
1. function Test_Mod_B (N: in Integer) return Integer is
2. begin
3.return N mod -10;
|
>>>
Checks were suppressed to avoid undesired expansion in Alfa mode. It turns out
these checks are needed during semantic analysis for the frontend to properly
mark nodes as needing or not a run-time check, and the special expansion done
in Alfa mode prevents the undesired expansion of checks. So reen
If nested loops appear where the iteration ranges are over the
same range of a multi-dimensional range (where the range number
has been defaulted to 1), then a warning is issued as shown in
the following example:
1. procedure Warn2D is
2.type E is array (Integer range <>, Integer ran
This change modifies the source location assigned to expander generated nodes
produced in the context of the distributed systems annex. Previously, we
always assigned code generated for the conversion of an expression to
the intermediate Any representation the location of the expression. However
wh
ARM D.1.7 states that only one pragma priority or interrupt_priority can be
given for a given entity. This patch recognizes some illegal instances of these
two pragmas or aspects, that were previously undiagnosed.
Compiling tasking.ads in Ada 2012 mode must yield:
tasking.ads:17:07:
pr
A new warning flag -gnatw.k causes the compiler to emit a warning
if a declaration redefines an entity of package Standard. Such
redefinitions are usually not a good idea, since these entities
are directly visible, and this can lead to confusion. This warning
is off by default.
The following, comp
This patch remove a duplicated logic to check wether a package is in the
closure.
No functional change.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Tristan Gingold
* bindgen.adb (Check_File_In_Partition, Check_System_Restrictions_Used):
Removed.
(Check
If a binary digest is needed it is a waste of time to recompute
it from the string encoded digest.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Pascal Obry
* g-sechas.adb, g-sechas.ads: (Binary_Message_Digest): New subtype.
(Digest): New versions returning a Bi
On Mon, Jul 16, 2012 at 8:57 AM, Steven Bosscher wrote:
> Hello,
>
> There are comments in basic-block.h that advise to update certain
> parts of the compiler if a new edge flag or basic block flag is added:
>
> -/* Always update the table in cfg.c dump_edge_info. */
>
> and
>
> - Always update
The new pragma Assert_And_Cut behaves exactly like pragma Assert for
compilation, but it informs formal verification tools of program points where
the tool may simpllify it treatment by using the given assertion.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Yannick Moy
A declaration for Is_Elaborated was emitted but never referenced.
No functional change.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Tristan Gingold
* bindgen.adb (Gen_Output_File_Ada): Do not emit declaration for
Is_Elaborated if not referenced.
Index: bindge
When a call to a subprogram in another package uses named parameters,
gnatxref will sometimes output duplicate entries for the parameter,
depending on the order in which the ALI files are read.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Emmanuel Briot
* xr_tabls.adb,
Initial work to implement pragma Partition_Elaboration_Policy. Currently,
only consistency is checked, the runtime only implements one policy.
A following patch will add a pragma Partition_Elaboration_Policy in the
runtime to enforce the policy (when tasking is used).
Tested on x86_64-pc-linux-gnu
On Mon, Oct 29, 2012 at 10:06 AM, Uros Bizjak wrote:
> As suggested by Richard S. [1], after the patch that converts subreg:M
> (op:N (...)(...)) to op:M (subreg:M (...) subreg:M (...)), we can
> remove several peephole2 patterns that handle subregs of PLUS, MINUS
> and MULT opera
This patch reimplements the mechanism which creates string names for entries
and families. The names are now created statically during object initialization
and associated with the concurrent object's Protection_Entries or ATCB. This
approach eliminates the need for runtime support (allocation, dea
This change introduces a new implementation defined pragma
"Attribute_Definition", which allows an attribute definition clause to be
expressed in a backward-compatible manner: compilers not supporting the
pragma, or the specified attribute, will just ignore it.
Tested on x86_64-pc-linux-gnu, commi
This makes the new support for extended overflow checking work on big-endian
platforms as well by fixing a thinko in the allocation routine of bignums.
No functional changes.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Eric Botcazou
* s-bignum.adb (Allocate_Bignum):
This change fixes incorrect values in s-oscons.ads on x86_64-freebsd,
which were caused by those values being processed as signed, instead of
unsigned, integers.
Tested on x86_64-pc-linux-gnu, committed on trunk
2012-10-29 Thomas Quinot
* s-oscons-tmplt.c: Fix signedness of ioctl requ
Normally pragma Optimize_Alignment (Space) forces an alignment of 1
for any packed record, but we can't do that for a variable length
record, since we can't implement that case, so ignore the pragma
in this case with a warning.
The following test program shows the patch in action:
1. pragma
Hi,
This patch adds me to the Write After Approval section of MAINTAINERS.
Regards,
James Greenhalgh
--
2012-10-26 James Greenhalgh
* MAINTAINERS (Write After Approval): Add myself.
wap.patch
Description: wap.patch
Hi,
When indent == 0, we call alloca with -1 bytes. This patch changes
it to indent + 1. This is a trunk only regression. OK to install?
Thanks.
H.J.
---
2012-10-29 H.J. Lu
* gimple-pretty-print.c (dump_gimple_bb_header): Correct alloca
length.
diff --git a/gcc/gimple-pr
On Mon, Oct 29, 2012 at 05:06:01AM -0700, H.J. Lu wrote:
> When indent == 0, we call alloca with -1 bytes. This patch changes
> it to indent + 1. This is a trunk only regression. OK to install?
> 2012-10-29 H.J. Lu
>
> * gimple-pretty-print.c (dump_gimple_bb_header): Correct alloca
>
> Index: gcc/cgraph.c
> ===
> --- gcc/cgraph.c (revision 192623)
> +++ gcc/cgraph.c (working copy)
> @@ -132,6 +132,74 @@ static GTY(()) struct cgraph_edge *free_edges;
> /* Did procss_same_body_aliases run? */
> bool same
On Mon, Oct 29, 2012 at 5:18 AM, Jakub Jelinek wrote:
> On Mon, Oct 29, 2012 at 05:06:01AM -0700, H.J. Lu wrote:
>> When indent == 0, we call alloca with -1 bytes. This patch changes
>> it to indent + 1. This is a trunk only regression. OK to install?
>
>> 2012-10-29 H.J. Lu
>>
>> * gi
Fix the mail subject line.
On 10/29/2012 06:56, JonY wrote:
> Hi,
>
> Workaround now exists for trunk mingw-w64 headers.
> Kai approved over IRC, so anybody with commit rights please push.
>
> ChangeLog
> 2012-10-29 Jonathan Yong
>
> * config/os/mingw32-w64/os_defines.h: Remove
> _GLI
On Thu, Oct 25, 2012 at 6:39 PM, Lawrence Crowl wrote:
> The sbitmap non-bool returning bitwise operations have been merged with
> the bool versions. Sometimes this merge involved modifying the non-bool
> version to compute the bool value, and sometimes modifying bool version to
> add additional
Hi,
This patch changes get_elimination to check register number instead of
RTX. Tested on Linux/x32 with -maddress-mode=long. OK to install?
Thanks.
H.J.
---
gcc/
2012-10-29 H.J. Lu
PR rtl-optimization/55093
* lra-eliminations.c (get_elimination): Check register number
Hi,
On Fri, 26 Oct 2012, Dehao Chen wrote:
> 1. abandon the changes in cfgexpand.c
Well, you merely moved the bogus code to gimple-low.c. It is bogus
because you unconditionally overwrite TREE_BLOCK of all operands (and all
operands operands) with the statements block. I assume the unit-test
On Mon, Oct 29, 2012 at 1:03 AM, Jan Hubicka wrote:
> Hi,
> this patch makes optimizer to predict &this->field to be optimized out.
> The main motivation for this is to make destructors that only calls destructor
> of inner type to be inlined, even in the cold regions.
>
> Bootstrapped/regtested x
On Mon, Oct 29, 2012 at 2:06 PM, Diego Novillo wrote:
> On Thu, Oct 25, 2012 at 6:39 PM, Lawrence Crowl wrote:
>
>> The sbitmap non-bool returning bitwise operations have been merged with
>> the bool versions. Sometimes this merge involved modifying the non-bool
>> version to compute the bool va
On Sun, Oct 28, 2012 at 4:14 PM, Marc Glisse wrote:
> Hello,
>
> this patch lets some predicates on floating point constants answer true for
> vectors, so optimizations are applied.
>
> Tested bootstrap + testsuite (default languages).
Ok. Bonus points if you quickly eyed users of whether they m
On Sun, Oct 28, 2012 at 7:33 PM, Steven Bosscher wrote:
> On Mon, Oct 22, 2012 at 11:09 PM, Jakub Jelinek wrote:
>> On Mon, Oct 22, 2012 at 10:51:43PM +0200, Steven Bosscher wrote:
>> Wouldn't it be way cheaper to just export dfs_find_deadend from cfganal.c
>> and call it in calc_dfs_tree on each
On Mon, Oct 29, 2012 at 2:51 PM, Michael Matz wrote:
> Hi,
>
> On Fri, 26 Oct 2012, Dehao Chen wrote:
>
>> 1. abandon the changes in cfgexpand.c
>
> Well, you merely moved the bogus code to gimple-low.c. It is bogus
> because you unconditionally overwrite TREE_BLOCK of all operands (and all
> ope
Hi,
On Mon, 29 Oct 2012, Richard Biener wrote:
> > Well, you merely moved the bogus code to gimple-low.c. It is bogus
> > because you unconditionally overwrite TREE_BLOCK of all operands (and all
> > operands operands) with the statements block. I assume the unit-test you
> > added shows the pr
Arnaud,
After updating my tree to rev 192942 this morning, I'm getting a
bootstrap failure with ada enabled:
raised CONSTRAINT_ERROR : a-comlin.adb:65 explicit raise
make[6]: *** [rts/s-oscons.ads] Error 1
make[6]: Leaving directory `bld-gcc/gcc/ada'
make[5]: *** [gnatlib-shared-default] Error 2
> After updating my tree to rev 192942 this morning, I'm getting a
> bootstrap failure with ada enabled:
Looks like make did not rebuild (or too late?) xoscons.
I'd suggest doing a rm -rf gcc/ada/bldtools from your obj directory and
see if this fixes the error.
If not, you might need to do a mak
On Mon, 29 Oct 2012, Richard Biener wrote:
On Sun, Oct 28, 2012 at 4:14 PM, Marc Glisse wrote:
Hello,
this patch lets some predicates on floating point constants answer true for
vectors, so optimizations are applied.
Tested bootstrap + testsuite (default languages).
Ok. Bonus points if yo
On Mon, Oct 29, 2012 at 10:36 AM, Arnaud Charlet wrote:
>> After updating my tree to rev 192942 this morning, I'm getting a
>> bootstrap failure with ada enabled:
>
> Looks like make did not rebuild (or too late?) xoscons.
>
> I'd suggest doing a rm -rf gcc/ada/bldtools from your obj directory and
> Well, this was a build from scratch. Maybe there is a parallel make
> bug in the new ada changes? Some undeclared dependency?
There is not in the recent changes. Maybe there's a latent bug, although I
have never experienced it.
At which revision are you?
Arno
On Mon, Oct 29, 2012 at 10:39 AM, Arnaud Charlet wrote:
>> Well, this was a build from scratch. Maybe there is a parallel make
>> bug in the new ada changes? Some undeclared dependency?
>
> There is not in the recent changes. Maybe there's a latent bug, although I
> have never experienced it.
>
> >> Well, this was a build from scratch. Maybe there is a parallel make
> >> bug in the new ada changes? Some undeclared dependency?
> >
> > There is not in the recent changes. Maybe there's a latent bug, although I
> > have never experienced it.
> >
> > At which revision are you?
>
> rev 19294
On 12-10-28 8:49 PM, H.J. Lu wrote:
On Sun, Oct 28, 2012 at 5:43 PM, Vladimir Makarov wrote:
The following patch fixes PR55106. A value in GENERAL_REGS is inherited
into a move with destination pseudo of SSE_REGS. It results into secondary
move for which inheritance is tried again an again
On 12-10-29 9:06 AM, H.J. Lu wrote:
Hi,
This patch changes get_elimination to check register number instead of
RTX. Tested on Linux/x32 with -maddress-mode=long. OK to install?
Yes. Thanks, H.J.
On 12-10-28 11:03 PM, H.J. Lu wrote:
Hi Vladimir,
There are many trialling white spaces in LRA codes. This patch removes
them.
Thank you for pointing this out. I should be more careful with emacs
macros using TAB. It would be nice if you install the patch.
On Mon, Oct 29, 2012 at 10:44 AM, Arnaud Charlet wrote:
>> >> Well, this was a build from scratch. Maybe there is a parallel make
>> >> bug in the new ada changes? Some undeclared dependency?
>> >
>> > There is not in the recent changes. Maybe there's a latent bug, although I
>> > have never exp
On Mon, Oct 29, 2012 at 10:48 AM, Diego Novillo wrote:
> On Mon, Oct 29, 2012 at 10:44 AM, Arnaud Charlet wrote:
>>> >> Well, this was a build from scratch. Maybe there is a parallel make
>>> >> bug in the new ada changes? Some undeclared dependency?
>>> >
>>> > There is not in the recent chang
> Yup, it does:
OK then can you check the timestamp of ./gcc/ada/bldtools/oscons/xoscons
vs e.g. ./gcc/ada/bldtools/oscons/xoscons/xoscons.adb and
$(srcdir)/gcc/ada/xoscons.adb?
Arno
On Wed, 24 Oct 2012, Dodji Seketeli wrote:
> Jakub Jelinek writes:
> > On Tue, Oct 23, 2012 at 03:11:29PM +0200, Dodji Seketeli wrote:
> > > * asan.c (insert_if_then_before_iter)
> > > (instrument_mem_region_access)
> > > (maybe_instrument_builtin_call, maybe_instrument_call): New static
> >
> > {
> > tree rhs = gimple_assign_rhs1 (stmt);
> > @@ -1584,6 +1585,20 @@ eliminated_by_inlining_prob (gimple stmt
> > /* Reads of parameter are expected to be free. */
> > if (unmodified_parm (stmt, inner_rhs))
> > rhs_free = true;
> >
"H.J. Lu" writes:
> Hi,
>
> This patch changes get_elimination to check register number instead of
> RTX. Tested on Linux/x32 with -maddress-mode=long. OK to install?
FWIW, this doesn't sound right to me, at least not without more justification.
The idea is that things like frame_pointer_rtx ar
Uros Bizjak writes:
> On Sun, Oct 28, 2012 at 2:37 AM, H.J. Lu wrote:
>
As suggested by Richard S. [1], after the patch that converts subreg:M
(op:N (...)(...)) to op:M (subreg:M (...) subreg:M (...)), we can
remove several peephole2 patterns that handle subregs of PLUS, MINUS
On Mon, Oct 29, 2012 at 10:51 AM, Arnaud Charlet wrote:
>> Yup, it does:
>
> OK then can you check the timestamp of ./gcc/ada/bldtools/oscons/xoscons
> vs e.g. ./gcc/ada/bldtools/oscons/xoscons/xoscons.adb and
> $(srcdir)/gcc/ada/xoscons.adb?
I have no directory called gcc/ada/bldtools/oscons/xos
On Mon, Oct 29, 2012 at 7:17 AM, Michael Matz wrote:
> Hi,
>
> On Mon, 29 Oct 2012, Richard Biener wrote:
>
>> > Well, you merely moved the bogus code to gimple-low.c. It is bogus
>> > because you unconditionally overwrite TREE_BLOCK of all operands (and all
Emm, then in gimple-low.c, we should
> I have no directory called gcc/ada/bldtools/oscons/xoscons. That's an
> executable.
Yes, that's as expected, I asked you to check the timestamp of this file.
> Still happens with -j1 (after starting from scratch), so it's probably
> not a parallel make issue.
Right, no surprise here.
> We ch
On Oct 26, 2012, Jakub Jelinek wrote:
> On Fri, Oct 26, 2012 at 04:30:41AM -0200, Alexandre Oliva wrote:
>> From: Alexandre Oliva
Fixed .git/config and the patches to use my @redhat.com address for
these patches.
>> +++ b/gcc/testsuite/gcc.dg/guality/pr54693.c
> The #include looks useless t
On Sat, Oct 27, 2012 at 10:12 PM, David Miller wrote:
> From: David Miller
> Date: Sun, 28 Oct 2012 00:31:27 -0400 (EDT)
>
>> The size is 24, and my patch definitely makes the crashes go away.
>>
>> It seems like a vector is being used for a mixed set of objects.
>> I'll try to figure out how tha
Hi,
this patch updates inlining functions called once into inlining functions
called multiple times when the overall unit size for inlining everywhere
shriks. This is now quite possible in the cases where inline predicates catch
some DCE as seen in the silly testcase attached. We already caught mo
Ping ^ 3.
On 10/18/2012 10:30 AM, Meador Inge wrote:
> Ping ^ 2.
>
> On 10/09/2012 09:44 PM, Meador Inge wrote:
>> Ping.
>>
>> On 10/04/2012 03:45 PM, Meador Inge wrote:
>>> Hi All,
>>>
>>> Currently the gcc-{ar,nm,ranlib} utilities assume that binutils is in
>>> path when invoking the wrapped bi
H.J. in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55116
reported an interesting address
(and:DI (subreg:DI (plus:SI (ashift:SI (reg:SI 96 [ glob_vol_int.22 ])
(const_int 2 [0x2]))
(symbol_ref:SI ("glob_vol_int_arr") 0x703c2720 glob_vol_int_arr>)) 0)
(c
On Mon, Oct 29, 2012 at 4:25 PM, Dehao Chen wrote:
> On Mon, Oct 29, 2012 at 7:17 AM, Michael Matz wrote:
>> Hi,
>>
>> On Mon, 29 Oct 2012, Richard Biener wrote:
>>
>>> > Well, you merely moved the bogus code to gimple-low.c. It is bogus
>>> > because you unconditionally overwrite TREE_BLOCK of
> > We chatted about this on IRC, and Richi is seeing the same problem on
> > his builds.
>
> Very strange. I'm looking at it.
OK, this comes from code duplicated between gcc-interface/Makefile.in and
Make-generated.in, pretty messy.
Testing the patch below, will commit if success.
2012-10-29
Vladimir Makarov writes:
>H.J. in
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55116
>
>reported an interesting address
>
> (and:DI (subreg:DI (plus:SI (ashift:SI (reg:SI 96 [ glob_vol_int.22 ])
> (const_int 2 [0x2]))
> (symbol_ref:SI ("glob_vol_int_arr")
> Testing the patch below, will commit if success.
>
> 2012-10-29 Arnaud Charlet
>
>* gcc-interface/Makefile.in (s-oscons.ads): Adjust call to
>xoscons.
Committed on trunk now. Sorry for the breakage, my build was incremental
and missed this missing part.
Arno
Sorry, forgot a line:
Richard Sandiford writes:
> Vladimir Makarov writes:
>> Index: rtlanal.c
>> ===
>> --- rtlanal.c (revision 192942)
>> +++ rtlanal.c (working copy)
>> @@ -5459,6 +5459,11 @@ strip_address_mutations (rtx *loc
> ICK ...
>
> Why not sth as simple as
>
> return num_ssa_operands (stmt, SSA_OP_USE);
>
> ? a[1][2] and b[2] really have the same cost, variable length
> objects have extra SSA operands in ARRAY_REF/COMPONENT_REF for
> the size. Thus, stmt cost somehow should reflect the number
> of depe
Hi,
On Mon, 29 Oct 2012, Dehao Chen wrote:
> On Mon, Oct 29, 2012 at 7:17 AM, Michael Matz wrote:
> > Hi,
> >
> > On Mon, 29 Oct 2012, Richard Biener wrote:
> >
> >> > Well, you merely moved the bogus code to gimple-low.c. It is bogus
> >> > because you unconditionally overwrite TREE_BLOCK of a
On 12-10-29 12:21 PM, Richard Sandiford wrote:
Vladimir Makarov writes:
H.J. in
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55116
reported an interesting address
(and:DI (subreg:DI (plus:SI (ashift:SI (reg:SI 96 [ glob_vol_int.22 ])
(const_int 2 [0x2]))
On 12-10-29 12:30 PM, Richard Sandiford wrote:
Sorry, forgot a line:
i.e. please use SUBREG_REG rather than XEXP.
I've just fixed it.
Hello!
Since we use simplify_gen_subreg on address part, we can use it for
DImode subreg of SImode address as well. simplify_gen_subreg knows how
to strip subreg.
2012-10-29 Uros Bizjak
* config/i386/i386.c (ix86_decompose_address): Use simplify_gen_subreg
for all addresses, z
On Mon, Oct 29, 2012 at 9:10 AM, Richard Biener
wrote:
> On Mon, Oct 29, 2012 at 4:25 PM, Dehao Chen wrote:
>> On Mon, Oct 29, 2012 at 7:17 AM, Michael Matz wrote:
>>> Hi,
>>>
>>> On Mon, 29 Oct 2012, Richard Biener wrote:
>>>
> Well, you merely moved the bogus code to gimple-low.c. It is
The first two patches I've applied. The remaining two are needed to fully
enable building the VxWorks flavor of GCC, but those bits affect parts
outside of fixincludes and there is some breakage somewhere.
All evidence seems to me to show fixincludes still doing its thing correctly,
but somewhere
On Mon, Oct 29, 2012 at 12:29 PM, Arnaud Charlet wrote:
> Committed on trunk now. Sorry for the breakage, my build was incremental
> and missed this missing part.
Great. Thanks for the quick fix!
Diego.
Hi, Micheal,
Thanks for explaining the design principle of debug info with gimple,
now I can understand your concerns. And thanks for providing the
patch.
However, in some places after gimplification (e.g. tree-inline.c), we
still updates the block/location info. And EXPR_LOCATION is still used
w
On Oct 29, 2012, at 2:02 AM, Paul Richard Thomas
wrote:
> It looks straightforward to me but I am not convinced that a fortran
> maintainer should be giving the green light on this :-)
Well, seems reasonable to me as well… I'd let Diego or Jason approve it. :-)
On 12-10-29 12:21 PM, Richard Sandiford wrote:
> Vladimir Makarov writes:
>>H.J. in
>>
>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55116
>>
>>reported an interesting address
>>
>>> (and:DI (subreg:DI (plus:SI (ashift:SI (reg:SI 96 [ glob_vol_int.22 ])
>> (const_int 2 [0
On Mon, Oct 29, 2012 at 5:55 AM, Jan Hubicka wrote:
>> Index: gcc/cgraph.c
>> ===
>> --- gcc/cgraph.c (revision 192623)
>> +++ gcc/cgraph.c (working copy)
>> @@ -132,6 +132,74 @@ static GTY(()) struct cgraph_edge *free_edges
> > ICK ...
> >
> > Why not sth as simple as
> >
> > return num_ssa_operands (stmt, SSA_OP_USE);
> >
> > ? a[1][2] and b[2] really have the same cost, variable length
> > objects have extra SSA operands in ARRAY_REF/COMPONENT_REF for
> > the size. Thus, stmt cost somehow should reflect th
Status
==
I'd like to close the stage 1 phase of GCC 4.8 development
on Monday, November 5th. If you have still patches for new features you'd
like to see in GCC 4.8, please post them for review soon. Patches
posted before the freeze, but reviewed shortly after the freeze, may
still go in, f
From: Ian Lance Taylor
Date: Mon, 29 Oct 2012 08:43:42 -0700
> This patch should fix it. Bootstrapped and ran libbacktrace testsuite
> on x86_64-unknown-linux-gnu. Committed to mainline.
I can tell just by looking at your patch that it doesn't fix the
problem.
It's "vec->base + vec->size" tha
From: Jakub Jelinek
Date: Mon, 29 Oct 2012 18:56:42 +0100
> I'd like to close the stage 1 phase of GCC 4.8 development
> on Monday, November 5th. If you have still patches for new features you'd
> like to see in GCC 4.8, please post them for review soon. Patches
> posted before the freeze, but
> Bootstrapped/regtested x86_64-linux, will commit it shortly.
>
> Honza
>
> * gcc.dg/ipa/inline-6.c: New testcase.
> * ipa-inline.c (want_inline_function_called_once_p): Rename to ...
> (want_inline_function_to_all_callers_p): check also functions with
> multiple callers.
On Mon, Oct 29, 2012 at 6:07 PM, Dehao Chen wrote:
> Hi, Micheal,
>
> Thanks for explaining the design principle of debug info with gimple,
> now I can understand your concerns. And thanks for providing the
> patch.
>
> However, in some places after gimplification (e.g. tree-inline.c), we
> still
On Mon, Oct 29, 2012 at 11:03 AM, David Miller wrote:
> From: Ian Lance Taylor
> Date: Mon, 29 Oct 2012 08:43:42 -0700
>
>> This patch should fix it. Bootstrapped and ran libbacktrace testsuite
>> on x86_64-unknown-linux-gnu. Committed to mainline.
>
> I can tell just by looking at your patch t
From: Ian Lance Taylor
Date: Mon, 29 Oct 2012 11:13:19 -0700
> Sure, if you can confirm that this patch fails, I will take another look.
It fails.
Same problem, SIGBUS.
From: Ian Lance Taylor
Date: Mon, 29 Oct 2012 11:13:19 -0700
> I changed it from vec->base + vec->size. Now it is vec->base
> + size, where size is the aligned version of vec->size.
It is not.
You don't modify 'size' at all.
On 10/26/2012 10:14 PM, Bin Cheng wrote:
Actually I am going to submit a patch monitoring the change of register
pressure accurately as we discussed before. With that patch, expressions
will be hoisted anyway if it can decrease register pressure thus this test
case won't fail anymore. So I am won
> I'd like to get the Sparc cbcond stuff in (3 revisions posted) which
> is waiting for Eric B. to do some Solaris specific work.
>
> I'd also like to enable LRA for at least 32-bit sparc, even if I can't
> find the time to work on auditing 64-bit completely.
End of stage #1 isn't a hard limit fo
On Mon, Oct 29, 2012 at 11:15 AM, David Miller wrote:
> From: Ian Lance Taylor
> Date: Mon, 29 Oct 2012 11:13:19 -0700
>
>> I changed it from vec->base + vec->size. Now it is vec->base
>> + size, where size is the aligned version of vec->size.
>
> It is not.
>
> You don't modify 'size' at all.
What was supposed to be a simple fix, turned out to be a nightmare...
I am working on adding an uninstrumented code path for TM, in
expectation of hardware TM. This means that we eventually want to
generate something like this:
prop = _ITM_beginTransaction(PR_multiwayCode);
From: Ian Lance Taylor
Date: Mon, 29 Oct 2012 11:31:33 -0700
> On Mon, Oct 29, 2012 at 11:15 AM, David Miller wrote:
>> From: Ian Lance Taylor
>> Date: Mon, 29 Oct 2012 11:13:19 -0700
>>
>>> I changed it from vec->base + vec->size. Now it is vec->base
>>> + size, where size is the aligned vers
From: Eric Botcazou
Date: Mon, 29 Oct 2012 20:25:15 +0100
>> I'd like to get the Sparc cbcond stuff in (3 revisions posted) which
>> is waiting for Eric B. to do some Solaris specific work.
>>
>> I'd also like to enable LRA for at least 32-bit sparc, even if I can't
>> find the time to work on a
From: David Miller
Date: Mon, 29 Oct 2012 14:33:33 -0400 (EDT)
> From: Ian Lance Taylor
> Date: Mon, 29 Oct 2012 11:31:33 -0700
>
>> On Mon, Oct 29, 2012 at 11:15 AM, David Miller wrote:
>>> From: Ian Lance Taylor
>>> Date: Mon, 29 Oct 2012 11:13:19 -0700
>>>
I changed it from vec->base
On Mon, Oct 29, 2012 at 11:40 AM, David Miller wrote:
> From: David Miller
> Date: Mon, 29 Oct 2012 14:33:33 -0400 (EDT)
>
>> From: Ian Lance Taylor
>> Date: Mon, 29 Oct 2012 11:31:33 -0700
>>
>>> On Mon, Oct 29, 2012 at 11:15 AM, David Miller wrote:
From: Ian Lance Taylor
Date: Mon,
On 10/29/12, Richard Biener wrote:
> On Oct 29, 2012 Diego Novillo wrote:
> > On Oct 25, 2012 Lawrence Crowl wrote:
> > > The sbitmap non-bool returning bitwise operations have been
> > > merged with the bool versions. Sometimes this merge involved
> > > modifying the non-bool version to comput
1 - 100 of 141 matches
Mail list logo
