Ok for google branches
It is better to have a higher level compiler option to be introduced
for this purpose instead of asking user to specify the plugin path.
The option should enable 1) ffunction-sections; 2) cgraph note section
genration; 3) enable the plugin. One possibility is to enhance th
"H.J. Lu" writes:
> 2011-08-02 H.J. Lu
>
> * config/i386/linux-unwind.h (RT_SIGRETURN_SYSCALL): New.
> (x86_64_fallback_frame_state): Use RT_SIGRETURN_SYSCALL and
> long long to check rt_sigreturn syscall.
This is OK.
Thanks.
Ian
Hi,
X32 uses 0x4206 for system call number __NR_rt_sigreturn and long
long is needed for 64bit integer. OK for trunk?
Thanks.
H.J.
---
2011-08-02 H.J. Lu
* config/i386/linux-unwind.h (RT_SIGRETURN_SYSCALL): New.
(x86_64_fallback_frame_state): Use RT_SIGRETURN_SYSCALL an
Hi,
This patch adds a new linker plugin to re-order functions. The
plugin constructs an annotated callgraph with edge profile information
and then repeatedly groups sections that are connected by hot edges
and passes the new function layout to the linker. The grouping is
done similar to the P
This patch fixes PR49437 with a single line change in ARM backend
and a regression test case for ARM target
ChangeLog:
2011-08-02 Matthew Gretton-Dann
PR target/49437
* config/arm/arm.c (arm_output_epilogue): Properly handle epilogue
when stack was realigned in interru
I got Jeff Law to review the reload change on IRC
and committed the composite patch.
Tested on x86_64, i586, avr, and h8300. Most other
tier1 targets ought not be affected, as this patch
only applies to ACCUMULATE_OUTGOING_ARGS == 0 targets.
r~
On Tue, Aug 2, 2011 at 2:02 PM, Jason Merrill wrote:
> On 06/30/2011 04:47 PM, H.J. Lu wrote:
>>
>> +@defmac REG_VALUE_IN_UNWIND_CONTEXT
>> +
>> +Define this macro if the target stores register values as
>> +@code{_Unwind_Word} type in unwind context. The default is to
>> +store register values a
On 08/02/2011 02:15 PM, Andreas Schwab wrote:
> Richard Henderson writes:
>
>> diff --git a/gcc/dwarf2cfi.c b/gcc/dwarf2cfi.c
>> index 99b37ab..69218cb2d 100644
>> --- a/gcc/dwarf2cfi.c
>> +++ b/gcc/dwarf2cfi.c
>
> Did you mean to commit this?
>
> Andreas.
>
Err, yes. But clearly I somehow m
Richard Henderson writes:
> diff --git a/gcc/dwarf2cfi.c b/gcc/dwarf2cfi.c
> index 99b37ab..69218cb2d 100644
> --- a/gcc/dwarf2cfi.c
> +++ b/gcc/dwarf2cfi.c
Did you mean to commit this?
Andreas.
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5
We need to look for packs in the type of a CONSTRUCTOR to handle
list-initialized temporaries.
Tested x86_64-pc-linux-gnu, applied to trunk.
commit 642c479a11613a80ea8eea986178be0cc60e5a39
Author: jason
Date: Tue Aug 2 21:09:08 2011 +
PR c++/49593
* pt.c (find_parameter_packs_r
The compiler was wrongly clearing uses_unions_p when recursing into the
anonymous struct, breaking the union logic.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit e7830050a2bd41b5109b9e09512b0f285c47001c
Author: jason
Date: Tue Aug 2 21:08:57 2011 +
PR c++/49803
* init.c
A couple of narrowing errors we were missing: we need to deal with
reshape_init discarding the {} and still complain, and we need to handle
unsignedness mismatch.
Tested x86_64-pc-linux-gnu, applied to trunk.
commit 2d486097a6240eb2a4d0caaf2cdf1a5f12faa249
Author: Jason Merrill
Date: Tue Aug
With lambdas, we can have a function body inside a default argument, so
we need to clear parser->local_variables_forbidden_p.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit bdbb09d8c20e71b3e4a023fd1bf49005a59b03a7
Author: Jason Merrill
Date: Tue Aug 2 16:22:24 2011 -0400
PR c++/4
On 06/30/2011 04:47 PM, H.J. Lu wrote:
+@defmac REG_VALUE_IN_UNWIND_CONTEXT
+
+Define this macro if the target stores register values as
+@code{_Unwind_Word} type in unwind context. The default is to
+store register values as @code{void *} type.
+
+@end defmac
This ought to suggest why a port
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 08/02/11 22:04, Tobias Burnus wrote:
> Daniel Kraft wrote:
>> after a lng while, here's another patch from me. It is a
>> one-line fix for PR 49885, which was introduced by me last year.
>> While the patch should be simple, it wasn't that easy
Hello,
this patch improves the ability of reassociation pass to simplifiy
more complex bitwise-binary
operations with comparisons. We break-up for this patch statements
like (X | Y) != 0 to X != 0 | Y != 0,
and (X | Y) == 0 to expanded X == 0 & Y == 0.
Additionally we expand logical-not expressio
Daniel Kraft wrote:
after a lng while, here's another patch from me. It is a one-line
fix for PR 49885, which was introduced by me last year. While the patch
should be simple, it wasn't that easy to find the fix (for me at least)
... but I think it should be fairly easy to see now. For som
The h8sx multilib fails to build because h8300_swap_into_er6
used the push function, which set RTX_FRAME_RELATED_P. This
resulted in mismatched (and incorrect) unwind info across two
different execution paths.
The trivial solution is to unset RTX_FRAME_RELATED_P from
the push, but it's now possi
On Tue, 2 Aug 2011 19:13:04 +0200
Romain Geissler wrote:
> > You should build MELT in an empty build tree. Please give us the log file
> > of your make
> > (don't use make -j, only a sequential make).
>
> Ok, i'll send you that tomorrow. Tell me if you want that i explore this
> issue on my con
The Java frontend currently produces an incorrectly typed tree if
a method call is rewritten. Specifically, the callee operand is a
NOP_EXPR casting the address of the target method to the original
method type. This patch causes the NOP_EXPR cast to use the type of
the target method instead.
The
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi,
after a lng while, here's another patch from me. It is a one-line
fix for PR 49885, which was introduced by me last year. While the patch
should be simple, it wasn't that easy to find the fix (for me at least)
... but I think it should be fa
On 08/02/2011 12:01 PM, Georg-Johann Lay wrote:
> Didn't try it yet. Is that capable of fixing the runtime FAILs?
Apparently. I installed your simulator and get comparable results to
http://gcc.gnu.org/ml/gcc-testresults/2011-07/msg03632.html
r~
The following patch decreases ARM code size degradation for LRA. It
permits achieve practically the same SPECINT2000 code size on ARM as for
reload (there is still small ARM code size degradation on SPECFP2000).
The patch was successfully bootstrapped on x86-64.
2011-08-02 Vladimir Makarov
Sorry, had a pasto in testcase.
Fixed at rev. 166205
2011-08-02 Kai Tietz
PR middle-end/49947
* gcc.dg/tree-ssa/forwprop-15.c
Tested on x86_64-pc-linux-gnu. Applied as obvious fix.
Regards,
Kai
Index: gcc.dg/tree-ssa/forwprop-15.c
On 08/02/2011 12:01 PM, Georg-Johann Lay wrote:
> == avrtest ==
>
> There is a text
> http://lists.gnu.org/archive/html/avr-gcc-list/2011-06/msg00015.html
> and a README:
> http://winavr.cvs.sourceforge.net/viewvc/winavr/avrtest/README?view=markup
>
> In the case there are questions: Ask.
Ah, a
In CCing Jörg.
Richard Henderson schrieb:
On 08/02/2011 12:52 AM, Georg-Johann Lay wrote:
There are still unrecognizables:
gcc.c-torture/execute/complex-7.c:56:1: error: unrecognizable insn:
(insn 17 14 18 3 (set (mem:SF (post_dec:HI (reg/f:HI 32 __SP_L__)) [0 S4 A8])
(reg:SF 43 [ f5.0
The problem here is that reload finds a push of arg_pointer_rtx
and disables register elimination of AP->SP because of it. This
leads to a register elimination (and assertion) failure when it
comes time to output debug info.
Preventing eliminable registers from being pushed fixes this.
The chang
ping:
On Tue, 28 Jun 2011 22:15:04 +0200, Jan Kratochvil wrote:
Hi,
a mechanical patch which fixes during
#define CP_DEMANGLE_DEBUG
make check
->
/bin/sh: line 1: 9179 Segmentation fault ./test-demangle <
./demangle-expected
which also fixes confusing output for _Z1hI1AIiEdEDTcldtfp_1gIT
Even in a template, if the range-init is not type-dependent, we can
deduce 'auto' in the for-range-declaration.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit 5384cac2db4875bc2f34bce11c6d1a3c360cd66d
Author: Jason Merrill
Date: Tue Aug 2 13:42:21 2011 -0400
PR c++/49834
* pa
On 08/02/2011 12:52 AM, Georg-Johann Lay wrote:
> There are still unrecognizables:
>
> gcc.c-torture/execute/complex-7.c:56:1: error: unrecognizable insn:
> (insn 17 14 18 3 (set (mem:SF (post_dec:HI (reg/f:HI 32 __SP_L__)) [0 S4 A8])
> (reg:SF 43 [ f5.0+4 ]))
> /mnt/nfs/home/georg/gnu/gcc
I think the following patch fixes this problem; it consists only in
copying a few lines from s-osinte-linux.ads to
s-osinte-kfreebsd-gnu.ads:
Index: b/src/gcc/ada/s-osinte-kfreebsd-gnu.ads
===
--- a/src/gcc/ada/s-osinte-kfreebsd-gnu.a
Hello
Here is my next try to put the stack check into rtl at prologue stage.
To me, it was not as easy as I hoped.
I've had little problems to get push/pop and the compare/jump working.
Hoping the way i choose is acceptable.
With rtl no extra pool to hold pointer or size values is required any mo
Le 2 août 2011 à 18:38, Basile Starynkevitch a écrit :
> On Tue, 2 Aug 2011 17:02:05 +0200
> Romain Geissler wrote:
>
>> Hi,
>>
>> I added a few primitives about trees, plus a string_to_long
>> conversion primitive and a foreach_pair_between iterator
>> that allows to iterate between two pairs
Hi!
--- index.html 15 Jul 2011 09:48:15 - 1.808
+++ index.html 2 Aug 2011 17:01:52 -
@@ -53,6 +53,13 @@ mission statement.
+August 2, 2011
+An implementation of the http://www.openmp.org/mp-documents/OpenMP3.1.pdf";>OpenMP v3.1
+parallel programming interface for C, C++ and Fo
On Tue, Aug 2, 2011 at 6:14 AM, Kai Tietz wrote:
> 2011/8/2 Richard Guenther :
>> On Tue, Aug 2, 2011 at 12:17 PM, Kai Tietz wrote:
>>> Hello,
>>>
>>> this patch removes in forward-propagation useless comparisons X != 0
>>> and X != ~0 for boolean-typed X. For one-bit precision typed X we
>>> si
Richard Guenther wrote on 02/08/2011 04:25:58 PM:
>
> Thinking about it it probably makes sense to keep a variant of this
> in the vectorizer - after all it has quite specific requirements on
> operand sizes while VRP would probably demote as far as possible
> (maybe taking PROMOTE_MODE into a
On Tue, 2 Aug 2011 17:02:05 +0200
Romain Geissler wrote:
> Hi,
>
> I added a few primitives about trees, plus a string_to_long
> conversion primitive and a foreach_pair_between iterator
> that allows to iterate between two pairs of the same list (and
> thus iterate on sublist).
>
> Note: again,
On Tuesday 02 August 2011 17:47:37 Tobias Burnus wrote:
> Simple patch: Coarrays are identified by a token; this patch passes the
> token (which is stored in the descriptor allocatable coarrays) to
> libcaf's registering function.
>
> In terms of token and allocatable coarrays: The next step is to
On 08/02/2011 06:01 PM, Mikael Morin wrote:
It's a pity that I've made the same fixes myself. I was planing to commit
soon. :-(
Indeed it would have been more useful to avoid replicating the work. :-(
On Tuesday 02 August 2011 17:34:46 Tobias Burnus wrote:
* Wrong case: All module files are
This patch fixes two issues:
a) LOCK(coarray%lock_type_comp) is also a coarray.
b) The following constraint was incompletely checked for: C1302. For
reference, I also list C1303/C1304.
C1302 A named variable of type LOCK TYPE shall be a coarray. A named
variable with a noncoarray subcomponen
It's a pity that I've made the same fixes myself. I was planing to commit
soon. :-(
On Tuesday 02 August 2011 17:34:46 Tobias Burnus wrote:
> This patch fixes dg-final cleanup-modules issues:
> * Missing cleanups
> * Wrong case: All module files are lower case
This is not actually a problem as cl
PING.
On Thu, Jul 28, 2011 at 3:01 PM, H.J. Lu wrote:
> Hi Richard, Jason,
>
> Is this patch
>
> http://gcc.gnu.org/ml/gcc-patches/2011-06/msg02401.html
>
> OK for trunk?
>
> Thanks.
>
>
> H.J.
> On Mon, Jul 11, 2011 at 3:21 PM, H.J. Lu wrote:
>> Ping.
>>
>> On Wed, Jul 6, 2011 at 2:20 PM, H.J.
Ping.
Could one of the configure maintainers review these changes?
Thanks,
Sebastian
On Thu, Jul 21, 2011 at 18:00, Tobias Grosser wrote:
> 2011-07-21 Tobias Grosser
>
> * configure: Regenerated.
> * config/cloog.m4: Do not define CLOOG_ORG
>
> and in gcc/
>
> 2011-07-21 Tobia
Simple patch: Coarrays are identified by a token; this patch passes the
token (which is stored in the descriptor allocatable coarrays) to
libcaf's registering function.
In terms of token and allocatable coarrays: The next step is to fix
passing actual arguments to assumed-shape coarrays dummie
Ping.
Could one of the configure maintainers review these changes?
Thanks,
Sebastian
On Thu, Jul 21, 2011 at 18:00, Tobias Grosser wrote:
> 2011-07-21 Tobias Grosser
>
> * configure: Regenerated.
> * configure.ac: Require cloog isl 0.16.3
> ---
> ChangeLog | 5 +
>
Ping.
Could one of the configure maintainers review these changes?
Thanks,
Sebastian
On Thu, Jul 21, 2011 at 18:00, Tobias Grosser wrote:
> 2011-07-21 Tobias Grosser
>
> * configure: Regenerated.
> * config/cloog.m4: Remove support for CLooG-ppl and CLooG-parma,
> both c
2011/8/2 Richard Guenther :
> On Tue, Aug 2, 2011 at 12:39 PM, Kai Tietz wrote:
Thanks, yes, I noticed that. Patch adjusted for cfg_tree.
ChangeLog
2011-08-02 Kai Tietz
* tree-ssa-forwprop.c (simplify_bitwise_binary):
Remove possible unused statement after optimization
On 06/22/2011 11:50 AM, Jason Merrill wrote:
This failure was happening on all targets that don't use either CFI
assembler directives or asynchronous unwind tables: we were mistakenly
deciding that the _FUN thunk for returning from the lambda function
pointer conversion operator couldn't throw be
When searching for an executable to run, Unix shells check every directory
on PATH and stop at the first matching executable found.
GNAT.OS_Lib.Locate_Exec_On_Path, however, returned the first matching
regular file, even if it wasn't executable. As a result, when an executable
tries to find where i
In SPARK, some parts of code that are ignored in the analysis can be "hidden",
through the use of a HIDE directive formatted as a special comment. There is no
benefit in detecting violations of SPARK restriction in such hidden parts, so
we now recognize the HIDE directive from SPARK and do not repo
A tree node must never be analyzed if it is not attached to the tree for the
current compilation, because the parent link is used for numerous semantic
checks and code insertions. This patch fixes a violation of this rule in the
analysis of array aggregates, where expressions in component associati
Follow-up of changes to define which entities are in the ALFA subset for formal
verification. Here, we define a flag Body_Is_In_ALFA that applies to entities
for subprograms, which is set to True when the subprogram body can be analyzed
formally. This is the initial definition, to be refined.
Test
On Tue, Aug 2, 2011 at 04:50, Richard Guenther wrote:
> On Tue, 2 Aug 2011, Sebastian Pop wrote:
>
>> --- a/gcc/graphite-scop-detection.c
>> +++ b/gcc/graphite-scop-detection.c
>> @@ -196,6 +196,12 @@ graphite_can_represent_scev (tree scev)
>> if (chrec_contains_undetermined (scev))
>> ret
The ALFA subset corresponds to those entities which can be formally analyzed
through the SPARK or Why tool-sets. This is the initial work to identify these
entities, for integer/enumeration types and subtypes, objects of such types
and subprogram specifications.
Tested on x86_64-pc-linux-gnu, comm
This patch treats expression functions as functions with an implicit pragma
Inline_Always. This ensures that they are chained to the list of inlined
subprograms of the enclosing unit, and made available to the back end for
inlining. The front-end does not examine whether inlining is actually possib
Hi,
I added a few primitives about trees, plus a string_to_long
conversion primitive and a foreach_pair_between iterator
that allows to iterate between two pairs of the same list (and
thus iterate on sublist).
Note: again, it's a git patch for Pierre until Basil returns.
Romain
0001-Add-a-few-
The compiler improperly checks subtype conformance of null exclusions on
anonymous access parameters in Ada 2005 cases involving dispatching operations.
In Ada 2005, controlling access parameters are defined to implicitly exclude
null (more properly, their anonymous access type excludes null). The
This patch does not change the functionality of the compiler. It removes
one argument that can be evaluated by the called routine.
No test required.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Javier Miranda
* exp_atag.ads, exp_atag.adb
(Build_Common_Dispatch
>From the text of AI05-0071:
If a generic unit has a subprogram_default specified by a box, and
the corresponding actual parameter is omitted, then it is equivalent
to an explicit actual parameter that is a usage name identical to the
defining name of the formal. {If a subtype_mark in t
In ALFA mode, generate an additional section in ALI files for so-called 'local'
cross-references, which 1) group the cross-references in each subprogram or
package; 2) add references to object definitions ('D' or 'I' with
initialization). This new section should be used in specific back-ends which
If the actual for a formal type with unknown discriminants is class-wide, then
a call to a primitive operation of the formal that dispatches on result raises
program_error in the instance if the context cannot provide a tag for the call.
This is the case for a declaration of an object of the formal
When an error is reported for an excluded source that is listed in a file
include the file name and the number of the line of the excluded source.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Vincent Celier
* prj-nmsc.adb (File_Found): New components Excl_File and Excl
On Tue, Aug 2, 2011 at 4:34 PM, Kai Tietz wrote:
> 2011/8/2 Richard Guenther :
>> On Tue, Aug 2, 2011 at 3:14 PM, Kai Tietz wrote:
>>> 2011/8/2 Richard Guenther :
On Tue, Aug 2, 2011 at 12:17 PM, Kai Tietz wrote:
> Hello,
>
> this patch removes in forward-propagation useless com
The new concatenation circuitry installed some time ago, results in
missing debug information when a constant string is initialized with
the result of a concatenation. This is because the resulting generated
renaming did not generate required debug information. This patch fixes
the problem with con
This patch allows the run-time to be compiled with normalize scalars.
More specifically, it fixes a warning about initialiation overlays
when Ada.Strings.Unbounded.Aux is compiled with this pragma. Note
that the previous fix to pragma Suppress_Initialization is required.
The following test should
This patch enables detection of conflicting external tags when compiling
under Ada 2005 mode. Required since this AI is a binding interpretation.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Javier Miranda
* exp_disp.adb (Make_DT): Generate call to Check_TSD in Ada 200
2011/8/2 Richard Guenther :
> On Tue, Aug 2, 2011 at 3:14 PM, Kai Tietz wrote:
>> 2011/8/2 Richard Guenther :
>>> On Tue, Aug 2, 2011 at 12:17 PM, Kai Tietz wrote:
Hello,
this patch removes in forward-propagation useless comparisons X != 0
and X != ~0 for boolean-typed X. For
This patch adds a runtime check to the elaboration of tagged types to raise
Program_Error if a user-specified external tag is the same as the external
tag for some other declaration. The following test must raise Program_Error
if compiled with -gnat12
package Pkg1 is
type Typ is tagged null rec
Ada Issue AI05-0230 specifies that null procedures can only have inheritable
'Class preconditions. For compatibility with the earlier GNAT pragmas for
preconditions, this rule only applies to Ada2012 aspect specifications.
Compilation of p.ads must yield:
p.ads:5:06: aspect "Pre" requires 'Cl
Hi,
For now, plugins can't compare types. This patch allows
c-tree.h to be installed as a plugin header, allowing
plugins to see "comptypes" (among other things).
Romain Geissler
2011-08-02 Romain Geissler
* Makefile.in (PLUGIN_HEADERS): Add C_TREE_H.
Index: gcc/Makefile.in
==
This patch is a partial implementation of pre/postconditions that apply to
generic subprograms and are inherited by the corresponding instantiations.
This implementation does not defer the analysis of the corresponding aspects
to a later point, and therefore is restricted to conditions that only de
If the elements of the container are indefinite, the iterator must build the
actual subtypes of each element in the iteration. As this is done during
expansion, the analysis of the loop body must be delayed until the loop is
expanded as an iterator loop.
The following must compile quietly in Ada201
On Aug 2, 2011, at 12:51 AM, Paolo Bonzini wrote:
> On 08/01/2011 09:10 PM, Dimitrios Apostolou wrote:
>>
>> Keeping my patch exactly the same, just changing the
>> hook_void_hard_reg_set to receive a (HOST_WIDEST_FAST_INT *) arg and
>> doing the necessary typecasts, added an extra 3 M instructi
This patch adds code to ensure the timely finalization of a local element copy
when iterating over a container.
-- Source --
-- types.ads
with Ada.Containers.Doubly_Linked_Lists;
package Types is
package Lists is new Ada.Containers.Doubly_Linked_Lists (Integer);
On Tue, Aug 2, 2011 at 3:14 PM, Kai Tietz wrote:
> 2011/8/2 Richard Guenther :
>> On Tue, Aug 2, 2011 at 12:17 PM, Kai Tietz wrote:
>>> Hello,
>>>
>>> this patch removes in forward-propagation useless comparisons X != 0
>>> and X != ~0 for boolean-typed X. For one-bit precision typed X we
>>> si
This patch supresses the finalization of an intermediate copy produced when a
cursor iterates over a collection. The intermediate copy is still finalized
when the associated loop goes out of scope.
-- Source --
-- types.ads
with Ada.Containers.Doubly_Linked_Lists;
pack
On Tue, 2 Aug 2011, Ira Rosen wrote:
>
>
> Richard Guenther wrote on 02/08/2011 01:33:49 PM:
> >
> > On Tue, 2 Aug 2011, Ira Rosen wrote:
> >
> > >
> > > > + /* Now we have matched the statement pattern
> > > > +
> > > > +rhs1 = (T1)x;
> > > > +rhs2 = (T1)y;
> > > > +o
This patch completes the implementation of the Default_Value and
Default_Component_Value aspects as described in AI05-0228. Note
that there is no matching pragma or attribute definition clause
for these aspects (because it is difficult to get these working
because of freezing problems resulting fro
2011/8/2 Richard Guenther :
> On Tue, Aug 2, 2011 at 12:17 PM, Kai Tietz wrote:
>> Hello,
>>
>> this patch removes in forward-propagation useless comparisons X != 0
>> and X != ~0 for boolean-typed X. For one-bit precision typed X we
>> simplifiy X == 0 (and X != ~0) to ~X, and for X != 0 (and X
If the range of iteration in an Ada2012 iterator is a function call returning
a container, finalization actions will in general be created because the
predefined containers are controlled. The finalization actions must be taken
into account when rewriting the iteration as a while-loop.
Compiling a
Switch.C had needless dependencies on most of the front end, causing
a circular dependency on Back_End for back ends written in Ada.
This patch cleans up the dependencies through introduction of
a new Warnsw package specifically for warning switches, and by
moving variables related to the preproces
The compiler does not handle well the accessibility check of anonymous
access types that are formals of anonymous access to subprogram
components of record types. The execution of the program may
crash or have unexpected behavior since the check is performed
with an expected actual (the accessibili
This patch does not change the behavior of the compiler. It adds
a new attribute to facilitate some non gcc back-end work to locate
the protected subprogram type entity associated with an internally
generated access to subprogram type.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02
[1] Add d.b debug option for showing available back end types
This debug option prints out information on all types that the
back end indicates it supports.
[2] Allow fpt types with more than Long_Long_Float'Digits digits
This patch takes advantage of the new infrastructure in Cstand to allow
de
This patch adds a new mechanism to have the Ada front end query what types are
supported by the back end. Types other than floating point types are currently
ignored. The information is needed because the list of floating point types
that may be present is open-ended, so it is impossible to add que
This fixes an oversight in the previous change: the check for library-level
inlined functions must be adjusted to the result of Get_Code_Unit_Entity.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Eric Botcazou
* inline.adb (Add_Inlined_Body): Adjust check for library-le
> -Original Message-
> From: Georg-Johann Lay [mailto:a...@gjlay.de]
> Sent: Tuesday, August 02, 2011 4:24 AM
> To: gcc-patches@gcc.gnu.org
> Cc: Anatoly Sokolov; Denis Chertykov; Weddington, Eric
> Subject: [Patch,AVR]: Cleanup libgcc.S
>
> This patch fixes RCALL/RJMP instructions to ot
This patch slightly reduces compilation time at -O0 in typical conditions by
streamlining the implementation of the Sem_Type.Covers predicate.
No functional changes.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Eric Botcazou
* sem_type.adb (Covers): Move trivial case
In Ada2012, the domain of iteration of a loop or quantified expression can be
a function call that yields a container. This patch implements the support for
default iterators over such expressions, that is to say iterators that use the
default indexing machinery present in all containers.
The follo
2011/8/2 Georg-Johann Lay :
> This patch fixes RCALL/RJMP instructions to other modules by replacing them
> with XCALL resp. XJMP.
>
> Moreover, now each function (except _cleanup) is enclosed in DEFUN/ENDF pairs
> so that size information is available for each function.
>
> Ok?
Approved.
Denis.
Richard Guenther wrote on 02/08/2011 01:33:49 PM:
>
> On Tue, 2 Aug 2011, Ira Rosen wrote:
>
> >
> > > + /* Now we have matched the statement pattern
> > > +
> > > +rhs1 = (T1)x;
> > > +rhs2 = (T1)y;
> > > +op_result = rhs1 OP rhs2;
> > > +lhs = (T2)op_result;
>
On Tue, Aug 2, 2011 at 12:39 PM, Kai Tietz wrote:
> Hello,
>
> this patch adds some statement-cleanup to forward-propagation.
>
> ChangeLog
>
> 2011-08-02 Kai Tietz
>
> * tree-ssa-forwprop.c (simplify_bitwise_binary):
> Remove possible unused statement after optimization.
>
> 2011
On Tue, Aug 2, 2011 at 12:17 PM, Kai Tietz wrote:
> Hello,
>
> this patch removes in forward-propagation useless comparisons X != 0
> and X != ~0 for boolean-typed X. For one-bit precision typed X we
> simplifiy X == 0 (and X != ~0) to ~X, and for X != 0 (and X == ~0) to
> X.
> For none one-bit p
The frontend does not expand record equalities found in Ada 2012 pre/post
conditions. After this patch the execution of the following test fails
the postcondition (as expected).
procedure Old_Bug is
type Val is new Integer;
type Index is new Integer range 1 .. 10;
type Point is record X,
This patch extends the functionality of routine Is_Variable to avoid
duplicating part of its code in routine Side_Effect_Free.
No functionality change.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Javier Miranda
* sem_util.ads, sem_util.adb (Is_Variable): Add a new fo
This patch turns off style checking in instances.
The following should compile quietly with -gnatyO:
gcc -c -gnatyO -gnatl p.ads
with Ada.Strings.Bounded;
package P is new
Ada.Strings.Bounded.Generic_Bounded_Length (100);
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Bob Duff
Hello,
this patch adds some statement-cleanup to forward-propagation.
ChangeLog
2011-08-02 Kai Tietz
* tree-ssa-forwprop.c (simplify_bitwise_binary):
Remove possible unused statement after optimization.
2011-08-02 Kai Tietz
* gcc.dg/tree-ssa/forwprop-9.c: Add tes
This patch significantly reduces compilation time at -O0 in typical conditions
by removing useless big local temporaries in Allocate_Initialize_Node.
No functional changes.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-02 Eric Botcazou
* atree.adb (Allocate_Initialize_Nod
When using a configurable run time, it is very likely that stream routines
for string types are not present (they require file system support). In this
case, the specific stream routines for strings are not used, relying on the
regular stream mechanism instead.
This patch changes only the behaviou
When a default config project file (for example default.cgpr in
.../share/gpr) exists, gnatmake was parsing it, and would fail if, for
example, Default_Language was not declared.
This patch ensures that an existing config project file is never taken
into account by gnatmake.
The test is to invoke g
1 - 100 of 142 matches
Mail list logo
