On Tue, Oct 16, 2012 at 04:19:09PM -0700, Xinliang David Li wrote:
> I am not sure -- fasan is an error detecting feature -- the goal is to
> find bugs -- missing handling of commons etc. are not desirable.
> Besides if ABI changes consistently for all objects, why does it
> matter?
>
> Or making
There is a function defined in builtins.def as
__builtin_extract_return_addr. The function is used by that name in
libgcc/unwind-dw2.c. For some reason the documentation for this
function uses the name __builtin_extract_return_address. This patch
corrects the documentation to match the definitio
Differences from v1:
1) Fix configure.ac test to use it's own variable name, thanks
Rainer.
2) Get rid of __VIS__ cpp stuff.
3) The new compiler option is now called -mcbcond
4) Actually document the new option.
Thanks.
gcc/
2012-10-12 David S. Miller
* configure.ac: Add chec
On 10/16/2012 07:51 PM, Richard Henderson wrote:
On 2012-10-17 09:53, Aldy Hernandez wrote:
+/* Like memory_modified_in_insn_p, but return TRUE if INSN will
+ *SURELY* modify the memory contents of MEM. */
+bool
+memory_surely_modified_in_insn_p (const_rtx mem, const_rtx insn)
I don't like
On 2012-10-17 09:53, Aldy Hernandez wrote:
> +/* Like memory_modified_in_insn_p, but return TRUE if INSN will
> + *SURELY* modify the memory contents of MEM. */
> +bool
> +memory_surely_modified_in_insn_p (const_rtx mem, const_rtx insn)
I don't like the word "surely". Are we certain or not?
I
On 2012-10-12 20:42, Aldy Hernandez wrote:
> PR middle-end/54893
> * trans-mem.c (diagnose_tm_1_op): Allow volatiles inside relaxed
> transactions.
Ok.
r~
On 12-10-16 5:21 PM, Richard Sandiford wrote:
Vladimir Makarov writes:
On 10/16/2012 02:44 AM, Richard Sandiford wrote:
Vladimir Makarov writes:
On 12-10-15 4:25 PM, Richard Sandiford wrote:
Richard Sandiford writes:
Vladimir Makarov writes:
After committing a patch yesterday to i
On 12-10-12 10:29 AM, Richard Sandiford wrote:
Hi Vlad,
Comments for the rest of ira-constraints.c.
Vladimir Makarov writes:
+ saved_base_reg = saved_base_reg2 = saved_index_reg = NULL_RTX;
+ change_p = equiv_address_substitution (&ad, addr_loc, mode, as, code);
+ if (ad.base_reg_loc != NU
The following patch implements most proposals of Richard's review of
the 2nd part of lra-constraints.
The patch was successfully bootstrapped and tested on x86/x86-64.
Committed as rev. 192524.
2012-10-16 Vladimir Makarov
* lra-int.h (lra_split_pseudos): Rename to lra_split_regs.
On Tue, Oct 16, 2012 at 4:54 PM, Aldy Hernandez wrote:
> On 10/16/12 13:28, Jakub Jelinek wrote:
>
>> I see another problem with noce_can_store_speculate_p, the return false;
>> checks. The search for a store to exactly that location can be done
>> just on post dominator bbs, but I wonder how you
On Tue, Oct 16, 2012 at 4:53 PM, Aldy Hernandez wrote:
>
>> Clearly we could consider the possibility of a PARALLEL of SET insns,
>> but of course most the compiler won't handle that anyhow. I suppose
>> that would be a reason to use memory_surely_modified_in_insn_p, but in
>> that case you might
On 10/16/12 13:28, Jakub Jelinek wrote:
I see another problem with noce_can_store_speculate_p, the return false;
checks. The search for a store to exactly that location can be done
just on post dominator bbs, but I wonder how you can avoid looking for
volatile insns or non-const/pure calls (or
I don't see a reason to add this new function. I would just inline it
into noce_can_store_speculate_p, replacing the call to
memory_modified_in_p. And I'm not sure I see a reason to change the
comment for memory_modified_in_p, it seems to already be accurate.
My comment was just clearer to m
On 2012-10-17 08:21, Michael Meissner wrote:
> Now, the x86 actually maps the OPTION_ISA_xxx switches to TARGET_xxx switches,
> so it is easy to change all of the defines from OPTION_ISA_ to
> TARGET_ISA_. For Android, it was simple to change the one line of
> reference.
>
> 2012-10-16 Michael M
On Tue, Oct 16, 2012 at 4:02 PM, Jakub Jelinek wrote:
> On Tue, Oct 16, 2012 at 03:50:27PM -0700, Xinliang David Li wrote:
>> Does that mean that all globals defined in shared libraries can not be
>> protected as long as they are not protected or hidden? This sounds
>> like a big limitation. We n
On Tue, Oct 16, 2012 at 03:50:27PM -0700, Xinliang David Li wrote:
> Does that mean that all globals defined in shared libraries can not be
> protected as long as they are not protected or hidden? This sounds
> like a big limitation. We need to answer the following two questions:
For !DECL_ONE_ON
On Wed, 2012-10-17 at 00:45 +0200, Gerald Pfeifer wrote:
> On Tue, 9 Oct 2012, Oleg Endo wrote:
> > This documents the new thread pointer built-ins in the SH www changes
> > for 4.8.
>
> Thanks, Oleg.
>
> I've got one change and one question:
>
> +Added support for the built-in functions
> +
On Tue, Oct 16, 2012 at 2:52 PM, Jakub Jelinek wrote:
> On Tue, Oct 16, 2012 at 01:56:46PM -0700, Xinliang David Li wrote:
>> Looks good except for the following:
>>
>> 1) I am not sure if the stack slot sharing is handled correctly. If I
>> read the code correctly, the redzone var will be only cr
On Tue, Oct 16, 2012 at 1:38 PM, Paolo Bonzini wrote:
>>> Bottom line: This is a bug in df_kill_notes.
>
> Yep, and it could cause wrong code generation, couldn't it? Because the
> new (reg:DI 72) is obviously not equal to (plus:DI (reg:DI 72)
> (const_int 24)), but it could be propagated to subse
On 2012-10-16 18:50 , Ian Lance Taylor wrote:
That is fine with me as long as we acknowledge that the upstream
sources don't care about GCC and will think it is absurd that they
should modify their code to carry untested and unmaintained
GCC-specific annotations. It would be one thing if the GC
The problem is that the macro unwinder code is changing the original
diagnostic type and not restoring it, so the code detecting that we
ICE fails to abort, which triggers another ICE, and so on. But there
is no point in modifying the original diagnostic, we can simply create
a temporary copy and u
On Tue, Oct 16, 2012 at 3:03 PM, Jakub Jelinek wrote:
> On Tue, Oct 16, 2012 at 02:41:42PM -0700, Xinliang David Li wrote:
>> > +bool
>> > +asan_protect_global (tree decl)
>> > +{
>> > + rtx rtl, symbol;
>> > + section *sect;
>> > +
>> > + if (TREE_CODE (decl) != VAR_DECL
>> > + || DECL_TH
On Tue, Oct 16, 2012 at 2:33 PM, Eric Botcazou wrote:
>> Yes, that is true. But when you say "if this trend continues" you are
>> making a slippery slope argument that I don't think applies. To date,
>> the trend consists of a single example. We are discussing adding a
>> second example, and we
On Tue, 9 Oct 2012, Oleg Endo wrote:
> This documents the new thread pointer built-ins in the SH www changes
> for 4.8.
Thanks, Oleg.
I've got one change and one question:
+Added support for the built-in functions
+__builtin_thread_pointer and
+__builtin_set_thread_pointer. This ass
The powerpc port is running out of switches that set bits in target_flags, and
I am in the process of changing all of the switches to use a secondary flag
word that is HOST_WIDE_INT (at least 64-bits on powerpc). One of the changes
is that the options machinery changes all of the TARGET_ names to
On Tue, Oct 16, 2012 at 02:41:42PM -0700, Xinliang David Li wrote:
> > +bool
> > +asan_protect_global (tree decl)
> > +{
> > + rtx rtl, symbol;
> > + section *sect;
> > +
> > + if (TREE_CODE (decl) != VAR_DECL
> > + || DECL_THREAD_LOCAL_P (decl)
> > + || DECL_EXTERNAL (decl)
> > +
On Tue, Oct 16, 2012 at 01:56:46PM -0700, Xinliang David Li wrote:
> Looks good except for the following:
>
> 1) I am not sure if the stack slot sharing is handled correctly. If I
> read the code correctly, the redzone var will be only created for the
> representative variable in a partition -- wi
On Tue, Oct 16, 2012 at 7:58 AM, Jakub Jelinek wrote:
> Hi!
>
> This is a WIP patch for globals protection.
> I'm not filling names yet and has_dynamic_init is always
> false (wonder how to figure it has_dynamic_init out, especially
> with LTO, TYPE_ADDRESSABLE (TREE_TYPE (decl)) probably isn't it
> Yes, that is true. But when you say "if this trend continues" you are
> making a slippery slope argument that I don't think applies. To date,
> the trend consists of a single example. We are discussing adding a
> second example, and we may decide against it. There are no current
> prospects o
Vladimir Makarov writes:
> On 10/16/2012 02:44 AM, Richard Sandiford wrote:
>> Vladimir Makarov writes:
>>> On 12-10-15 4:25 PM, Richard Sandiford wrote:
Richard Sandiford writes:
> Vladimir Makarov writes:
>> After committing a patch yesterday to implement proposals from a
>>
Looks good except for the following:
1) I am not sure if the stack slot sharing is handled correctly. If I
read the code correctly, the redzone var will be only created for the
representative variable in a partition -- will this lead to false
negatives? As I asked before, should stack slot sharing
Janus,
Your patch works as advertised without disturbing my pet bugs.
Just a nit pick: the double parentheses in
+ if ((CLASS_DATA (select_type_stack->selector)->attr.dimension
|| CLASS_DATA (select_type_stack->selector)->attr.codimension))
do not seem necessary.
Note for Paul: I
ping!
2012/10/11 Janus Weil :
> Hi all,
>
> here is an OOP patch for the above PR, which has two disconnected parts:
>
> 1) It fixes a problem with ASSOCIATED, when it is fed a CLASS-valued
> function as argument (i.e. the ICE in the bug title). This is the
> trans-intrinsic part of the patch. In
On Tue, Oct 16, 2012 at 04:58:48PM +0200, Jakub Jelinek wrote:
> @@ -2022,6 +2036,12 @@ assemble_variable (tree decl, int top_le
>if (DECL_ALIGN (decl) > BITS_PER_UNIT)
> ASM_OUTPUT_ALIGN (asm_out_file, floor_log2 (DECL_ALIGN_UNIT (decl)));
>assemble_variable_contents (decl, n
Committed.
google/gcc-4_6: r192510.
google/gcc-4_7: r192511.
google/main: r192513.
On Tue, Oct 16, 2012 at 12:44 PM, Delesley Hutchins wrote:
> Yes, but that won't happen for a while yet.
>
> On Tue, Oct 16, 2012 at 12:43 PM, Diego Novillo wrote:
>> On Tue, Oct 16, 2012 at 3:19 PM, Delesley Hutc
Attached patch applied.
2012-10-16 François Dumont
* include/debug/formatter.h (_Debug_msg_id): Add
__msg_bucket_index_oob.
* include/debug/macros.h (__glibcxx_check_bucket_index): New.
* include/debug/unordered_set (unordered_set<>::begin(size_type)):
Add check on bucket
Yes, but that won't happen for a while yet.
On Tue, Oct 16, 2012 at 12:43 PM, Diego Novillo wrote:
> On Tue, Oct 16, 2012 at 3:19 PM, Delesley Hutchins
> wrote:
>> This patch adds the clang thread safety warning flags as recognized
>> warning flags that do nothing, for command-line compatibilit
On Tue, Oct 16, 2012 at 3:19 PM, Delesley Hutchins wrote:
> This patch adds the clang thread safety warning flags as recognized
> warning flags that do nothing, for command-line compatibility with
> clang. Enclosed patch is for branches/google/gcc-4_6, I have
> identical patches for google/gcc-4_
Quoting Richard Sandiford :
Joern Rennecke writes:
2012-10-04 Joern Rennecke
* final.c (get_attr_length_1): Use direct recursion rather than
calling get_attr_length.
(get_attr_lock_length): New function.
(INSN_VARIABLE_LENGTH_P): Define.
(shorte
This patch adds the clang thread safety warning flags as recognized
warning flags that do nothing, for command-line compatibility with
clang. Enclosed patch is for branches/google/gcc-4_6, I have
identical patches for google/gcc-4_7 and google/main.
Tested on linux x86.
--
DeLesley Hutchins | S
> Can you confirm that the class_optional_1.f90 of the trunk works correctly?
Yes: I have regtested with
make -k check-gfortran RUNTESTFLAGS="dg.exp=class_optional*
--target_board=unix'{-m32,-m64}'"
without failures.
The profile I have posted was for the former version of class_optional_2.f90.
On Tue, Oct 16, 2012 at 03:02:47PM +, Joseph S. Myers wrote:
> That is:
>
> 1. Patch adding TARGET_FOO aliases for OPTION_FOO (small change to the awk
> scripts and associated documentation, I expect).
>
> 2. Large, mechanical, automatically generated patch to change existing
> OPTION_FOO us
Dominique,
Dominique Dhumieres wrote:
The previous class_optional_2.f90 runs after your commit, but it takes
~168s (so it may have run with the previous patch also, but I was not
patient enough to see it).
Here, it takes (current version) < 2s to compile and 0.150s to run the
program. If I lo
On 10/16/2012 08:17 AM, Jan Hubicka wrote:
> * builtins.c (define_builtin): Accept ECF flags and
> use set_call_expr_flags.
> (initialize_builtins): Update; add BULIT_IN_UNREACHALE.
>
> * calls.c (set_call_expr_flags): New.
> * tree.h (set_call_expr_flags): Declare.
On 2012-10-16 10:43 , Richard Biener wrote:
This patch shows work-in-progress (read: implementation uglyness
hopefully to vanish ...) regarding to moving LTO type merging
work from WPA to compile stage.
You mean to LTRANS, the stage after WPA, right?
The patch re-organizes lto_output_tree (t
On Tue, Oct 16, 2012 at 10:40 AM, Jakub Jelinek wrote:
> On Tue, Oct 16, 2012 at 09:48:28AM -0700, Xinliang David Li wrote:
>> The way to do this is not to use the shadow memory, but to clobber
>> directly the variable itself with specified byte pattern -- though
>> error diagnostics won't be as n
On Tue, Oct 16, 2012 at 09:48:28AM -0700, Xinliang David Li wrote:
> The way to do this is not to use the shadow memory, but to clobber
> directly the variable itself with specified byte pattern -- though
> error diagnostics won't be as nice.
Clobbering the memory directly is definitely easy, but
Hi,
here is third revised version of the complette unroling changes. While working
on the RTL variant I noticed PR54937 and the fact that I was overly aggressive
on forcing single exit of the last iteration to be taken, because loop may
terminate
otherwise (by EH or by exitting the program). Sam
Tobias,
The previous class_optional_2.f90 runs after your commit, but it takes
~168s (so it may have run with the previous patch also, but I was not
patient enough to see it). The culprits are given by the following
profile:
+ 100.0%, start, a.out
| + 100.0%, main, a.out
| | + 100.0%, MAIN__, a.o
On Tue, Oct 16, 2012 at 10:19:54AM -0700, Ian Lance Taylor wrote:
> On Mon, Oct 15, 2012 at 5:28 AM, Aldy Hernandez wrote:
> I don't see a reason to add this new function. I would just inline it
> into noce_can_store_speculate_p, replacing the call to
> memory_modified_in_p. And I'm not sure I s
On Mon, Oct 15, 2012 at 5:28 AM, Aldy Hernandez wrote:
>
> Here we have a store data race because noce_can_store_speculate_p()
> incorrectly returns true. The problem is that memory_modified_in_insn_p()
> returns true if an instruction *MAY* modify a memory location, but the store
> can only be s
> On Tue, Oct 16, 2012 at 9:44 AM, Jan Hubicka wrote:
> >
> > Hmm, I was not aware of these bits (and yes, I agree it is non-sence this
> > being
> > duplicated everywhere). I will add UNREACHABLE there. What about rest of
> > the
> > change (i.e. adding the proper bits)?
>
> I suppose it's b
On Tue, Oct 16, 2012 at 9:44 AM, Jan Hubicka wrote:
>
> Hmm, I was not aware of these bits (and yes, I agree it is non-sence this
> being
> duplicated everywhere). I will add UNREACHABLE there. What about rest of the
> change (i.e. adding the proper bits)?
I suppose it's basically fine but add
On 10/16/2012 01:44 AM, Bin Cheng wrote:
Hi Steven, Jeff,
I found a flaw in original patch, which results in inaccurate register
pressure.
Here comes the updated patches, improving code size a little bit on
Thumb2/powerpc comparing to original patches.
Please review.
Thanks
2012-10-16 Bin Chen
I have overhauled the interface to VEC over the last few
weeks. I implemented most of the plan I outlined in
http://gcc.gnu.org/ml/gcc/2012-08/msg00268.html.
I have implemented the embedded and space-efficient vectors. The
diff for vec.[ch] is sufficiently confusing that I'm just
attaching both
> > "Jan" == Jan Hubicka writes:
>
> Jan> this patch udpates Java frontend to declare __bulitin_unreachable
> Jan> and also fixes flags of synchronize bulitins to match ones from C
> Jan> FE.
>
> Jan> Regtested x86_64-linux, OK?
>
> The java bits are ok.
>
> Jan> + /* Looping const or pur
The way to do this is not to use the shadow memory, but to clobber
directly the variable itself with specified byte pattern -- though
error diagnostics won't be as nice.
To use shadow memory, the scope start marker is also needed.
David
On Tue, Oct 16, 2012 at 9:12 AM, Jakub Jelinek wrote:
> On
> On Tue, Oct 16, 2012 at 8:14 AM, Jan Hubicka wrote:
>
> > this patch udpates go-frontend to deifine unreachable bultin I need for loop
> > and LTO optimizations. I also noticed that GO ignores existence of all
> > flags
> > except for CONST and thus I synchronized the flags with C FE variants
> "Jan" == Jan Hubicka writes:
Jan> this patch udpates Java frontend to declare __bulitin_unreachable
Jan> and also fixes flags of synchronize bulitins to match ones from C
Jan> FE.
Jan> Regtested x86_64-linux, OK?
The java bits are ok.
Jan> + /* Looping const or pure is imlpies by noretu
On 16/10/12 17:34, Ian Bolton wrote:
Subject: [PATCH][AArch64] Restrict usage of SBFIZ to valid range only
This fixes an issue where we were generating an SBFIZ with
operand 3 outside of the valid range (as determined by the
size of the destination register and the amount of shift).
My patch ch
On Tue, Oct 16, 2012 at 8:14 AM, Jan Hubicka wrote:
> this patch udpates go-frontend to deifine unreachable bultin I need for loop
> and LTO optimizations. I also noticed that GO ignores existence of all flags
> except for CONST and thus I synchronized the flags with C FE variants.
I can't see
> Subject: [PATCH][AArch64] Restrict usage of SBFIZ to valid range only
>
> This fixes an issue where we were generating an SBFIZ with
> operand 3 outside of the valid range (as determined by the
> size of the destination register and the amount of shift).
>
> My patch checks that the range is va
Hi,
I seem to have uncovered what seems to be a bug with handling SUBREG (MEM) with
the MEM having side-effects while testing aarch64-4.7. This bug seems to be
latent on trunk.
Here is a test case reduced from gcc.c-torture/execute/scal-to-vec1.c.
#define vector(elcount, type) \
__attribute__(
From: Eric Botcazou
Date: Tue, 16 Oct 2012 12:10:51 +0200
>> I've scanned the documentation and there is no indication of any
>> preprocessor predefines or anything like that.
>>
>> And keep in mind that __VIS__ is our very own invention.
>>
>> Sun's compilers never predefined this.
>>
>> Thei
On Tue, Oct 16, 2012 at 09:06:22AM -0700, Xinliang David Li wrote:
> I don't get it. Clobber marks the end of lifetime of a variable so it
> is safe to emit code to really clobber its value -- otherwise how
> would clobber based slot sharing work?
If you at CLOBBER protect the var in the shadow m
On Mon, Oct 15, 2012 at 11:12 PM, Jakub Jelinek wrote:
> On Mon, Oct 15, 2012 at 10:48:13PM -0700, Xinliang David Li wrote:
>> Another error checking feature is to poison stack vars on entry and
>> exit of the lexical scope to catch uninit variable reference and out
>> of scope references:
>>
>> S
On 16/10/12 16:10, Marcus Shawcroft wrote:
I've just committed this patch to aarch64-trunk to resolve an ICE in
aarch64_split_doubleword_move when attempting to split v->v moves.
/Marcus
2012-10-16 Marcus Shawcroft
* config/aarch64/aarch64-protos.h (aarch64_split_doubleword_move):
On 15/10/12 11:03, Marcus Shawcroft wrote:
On 11/09/12 15:02, Chris Schlumberger-Socha wrote:
This patch adds predefines for AArch64 code models. These code models are
added as an effective target for the AArch64 platform.
I've committed this patch to aarch64-trunk.
/Marcus
.. and back p
Hi,
s/imlpies/implied/
Sent with AquaMail for Android
http://www.aqua-mail.com
Hi,
while looking into cases where loop-iv.c still deduce useful bounds that are not
recorded by tree level I noticed that we do not duplicate the bounds when
copying
the loop.
Fixed thus.
Bootstrapped/regtested x86_64-linux, OK?
Honza
* cfgloopmanip.c (copy_loop_info): New function.
Hi!
This is a WIP patch for globals protection.
I'm not filling names yet and has_dynamic_init is always
false (wonder how to figure it has_dynamic_init out, especially
with LTO, TYPE_ADDRESSABLE (TREE_TYPE (decl)) probably isn't it,
and for more I'm afraid we need a langhook).
--- gcc/varasm.c.j
Hi!
Here is an updated patch, which emits also the shadow clearing sequence
at the end of function in the right spot.
2012-10-16 Jakub Jelinek
* Makefile.in (asan.o): Depend on $(EXPR_H) $(OPTABS_H).
(cfgexpand.o): Depend on asan.h.
* asan.c: Include expr.h and optabs.
On Tue, Oct 16, 2012 at 08:40:21AM +0200, Tom de Vries wrote:
> 2012-10-16 Tom de Vries
>
> * expr.c (move_by_pieces, move_by_pieces_ninsns, can_store_by_pieces)
> (store_by_pieces_1): Don't enter loop when no more data is left.
Okay.
Jakub
On Mon, Oct 15, 2012 at 11:39:52PM -0700, Wei Mi wrote:
> --- gcc/gcc.c (revision 192487)
> +++ gcc/gcc.c (working copy)
> @@ -679,6 +679,7 @@ proper position among the other output f
> %{fgnu-tm:%:include(libitm.spec)%(link_itm)}\
> %(mflib) " STACK_SPLIT_SPEC "\
> %{fprofile-arcs|f
On Mon, Oct 15, 2012 at 10:48:13PM -0700, Xinliang David Li wrote:
> Another error checking feature is to poison stack vars on entry and
> exit of the lexical scope to catch uninit variable reference and out
> of scope references:
>
> S* sp;
> {
> S s;
> sp = &s;
> }
> .. *sp ...
>
Hi,
this patch add __buliltin_unreachable to Fortran FE and also cleans up the code
a bit
Bootstrapped/regtested x86_64-linux, OK?
Honza
* f95-lang.c (gfc_define_builtin): Use set_call_expr_flags.
(ATTR_NOTHROW_LEAF_MALLOC_LIST,
ATTR_CONST_NORETURN_NOTHROW_LEAF_LIST): New
Hi!
Now that SIZEOF_EXPR isn't folded immediately, e.g. REAL_TYPE can
be an argument of it, and tsubst_copy* doesn't handle that, but tsubst does.
Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux,
ok for trunk?
2012-10-15 Jakub Jelinek
PR c++/54844
* pt.c (t
On Mon, Oct 15, 2012 at 10:38:02AM +0200, Richard Biener wrote:
> On Fri, Oct 12, 2012 at 8:16 PM, Jakub Jelinek wrote:
> > Apparently vectorizable_load is another spot that could create vector
> > CONSTRUCTORs that wouldn't pass the new CONSTRUCTOR verification.
> >
> > Fixed thusly, bootstrapped
Hi,
this patch udpates Java frontend to declare __bulitin_unreachable and also
fixes flags of synchronize bulitins to match ones from C FE.
Regtested x86_64-linux, OK?
Honza
* builtins.c (define_builtin): Accept ECF flags and
use set_call_expr_flags.
(initialize_builtins)
Hi,
this patch udpates go-frontend to deifine unreachable bultin I need for loop
and LTO optimizations. I also noticed that GO ignores existence of all flags
except for CONST and thus I synchronized the flags with C FE variants.
(I plan to use set_call_expr_flags in other FEs too)
Regtested x86_
On 10/16/2012 02:44 AM, Richard Sandiford wrote:
Vladimir Makarov writes:
On 12-10-15 4:25 PM, Richard Sandiford wrote:
Richard Sandiford writes:
Vladimir Makarov writes:
After committing a patch yesterday to implement proposals from a
review, I found that GCC crashes on SPEC2000 gap.
I've just committed this patch to aarch64-trunk to resolve an ICE in
aarch64_split_doubleword_move when attempting to split v->v moves.
/Marcus
2012-10-16 Marcus Shawcroft
* config/aarch64/aarch64-protos.h (aarch64_split_doubleword_move):
Rename to aarch64_split_128bit_move.
On Tue, 16 Oct 2012, Richard Biener wrote:
>
> This patch shows work-in-progress (read: implementation uglyness
> hopefully to vanish ...) regarding to moving LTO type merging
> work from WPA to compile stage.
>
> The patch re-organizes lto_output_tree (the write_tree streamer
> hook for LTO) in
On Mon, 15 Oct 2012, Michael Meissner wrote:
> > I think changing the names is safer - it's immediately obvious as a build
> > failure if you missed anything. If you have MASK_* names for bits in more
> > than one flags variable, there's a risk of accidentally testing a bit in
> > the wrong va
On Mon, 15 Oct 2012, Manuel L?pez-Ib??ez wrote:
> 2012-10-15 Manuel L?pez-Ib??ez
>
> PR c/53063
> PR c/40989
> * doc/options.texi (EnabledBy): Document new form.
> * optc-gen.awk: Handle new form of EnabledBy.
> * common.opt (Wunused-but-set-parameter): Use Enable
This patch shows work-in-progress (read: implementation uglyness
hopefully to vanish ...) regarding to moving LTO type merging
work from WPA to compile stage.
The patch re-organizes lto_output_tree (the write_tree streamer
hook for LTO) in a way so that we output all tree fields
in easy to discov
On Sun, 14 Oct 2012, Manuel L?pez-Ib??ez wrote:
> 2012-10-14 Manuel L?pez-Ib??ez
>
> PR c/53063
> PR c/40989
> gcc/
> * optc-gen.awk: Handle new form of LangEnabledBy.
> * opts.c (set_Wstrict_aliasing): Declare here. Make static.
> * common.opt (Wstrict-aliasing=,
On Tue, Oct 16, 2012 at 4:30 PM, Joern Rennecke
wrote:
> Quoting Zdenek Dvorak :
>
>> no -- you should also test that latch contains no active insns. I.e.,
>> factorize
>> out whatever forwarder_block_p does except for the test
>> "(dest->loop_father->header == dest)" test,
>
>
> Like this?
IST
> Quoting Zdenek Dvorak :
>
> >no -- you should also test that latch contains no active insns.
> >I.e., factorize
> >out whatever forwarder_block_p does except for the test
> >"(dest->loop_father->header == dest)" test,
>
> Like this?
>
> * basic-block.h (forwarder_block_p_1): Declare.
>
Quoting Zdenek Dvorak :
no -- you should also test that latch contains no active insns.
I.e., factorize
out whatever forwarder_block_p does except for the test
"(dest->loop_father->header == dest)" test,
Like this?
* basic-block.h (forwarder_block_p_1): Declare.
* cfgrtl
Ulrich Weigand wrote:
Tejas Belagod wrote:
Ulrich Weigand wrote:
The following patch implements this idea; it passes a basic regression
test on arm-linux-gnueabi. (Obviously this would need a lot more
testing on various platforms before getting into mainline ...)
Can you have a look whether t
Hi,
> The loop appears to be entered at the top, yet both my original and your test
> fail.
>
> Looking what happens with your test in more detail, I find that
>
> loop->latch == desc->in_edge->dest
>
> is true, but forwarder_block_p (loop->latch) fails:
>
> 562 if (dest->loop_fathe
Tejas Belagod wrote:
> > Ulrich Weigand wrote:
> >> The following patch implements this idea; it passes a basic regression
> >> test on arm-linux-gnueabi. (Obviously this would need a lot more
> >> testing on various platforms before getting into mainline ...)
> >>
> >> Can you have a look whether
Richard Earnshaw wrote:
> On 20/09/12 16:49, Ulrich Weigand wrote:
> > Richard Earnshaw wrote:
> >
> >> Hmm, this is going to cause bottlenecks on Cortex-A15: writing a Neon
> >> single-precision register and then reading it back as a double-precision
> >> value will cause scheduling problems.
[sni
On Tue, Oct 16, 2012 at 4:05 AM, Rainer Orth
wrote:
>
> But importing different upstream testsuites with different annotation
> systems allows them to make GCC maintainer's lives miserable? If this
> trend continues, maintainers need to cope with several different
> annoation systems with differe
Quoting Zdenek Dvorak :
+ entered_at_top = loop_preheader_edge (loop)->dest == desc->in_edge->dest;
is equivalent to
+ entered_at_top = loop->header == desc->in_edge->dest;
But, I don't think it will do what you want. Loops are canonicalized so that
their latch blocks have single successor
On Tue, Oct 16, 2012 at 11:39 AM, Eric Botcazou wrote:
>> As I'm not sure how to best do that I suggest we do a more proper RTL
>> DSE hack by adding a 'libcall-call-escape'-set which we can add to
>> instead of calling mark_addressable this late. We need to add all
>> partitions of a decl here,
Hi Tobias,
I did not yet appied you latest patch to gfortran, but I ran the new tests
with gfortran patched with the previous patch and both pass (manual
testing without option, but -fcoarray=single). Note that for
class_optional_1.f90, valgrind --leak-check=full gives
==45665== 4 bytes in 1 bloc
Il 16/10/2012 12:35, Steven Bosscher ha scritto:
> On Tue, Oct 16, 2012 at 12:15 PM, Steven Bosscher wrote:
>> On Mon, Oct 15, 2012 at 3:26 PM, Steven Bosscher wrote:
>>> On Mon, Oct 15, 2012 at 3:21 PM, Paolo Bonzini wrote:
Il 15/10/2012 14:53, Steven Bosscher ha scritto:
> I think I've s
On 2012-10-16 07:05 , Rainer Orth wrote:
Diego Novillo writes:
On Mon, Oct 15, 2012 at 11:55 AM, Ian Lance Taylor wrote:
In my opinion, supporting the full range of GCC testsuite annotations
means imposing a lot of mechanism that the Go testsuite does not
require. It would complicate the G
1 - 100 of 119 matches
Mail list logo
