On Sat, 24 Oct 2015, Jan Hubicka wrote:
> Hi, as discussed earlier, the types_compatible_p in operand_equal_p
> seems redundant. (it is callers work to figure out how much of type
> matching it wants. If not, we probably want to treat most of other
> references and casts simlar way).
>
> Boot
Hi,
this patch extends the simplification formerly done in fold_widened_comparison
and now in match.pd to all integral types instead of just integer types, and
comes with an Ada testcase for the enumeral type case.
The patch introduces a failure in the C testsuite:
FAIL: gcc.dg/atomic-noinline.
On Mon, Oct 26, 2015 at 4:39 PM, Eric Botcazou wrote:
> Hi,
>
> this patch extends the simplification formerly done in fold_widened_comparison
> and now in match.pd to all integral types instead of just integer types, and
> comes with an Ada testcase for the enumeral type case.
>
> The patch intro
> 2015-10-25 Steven G. Kargl
>
> PR fortran/36192
> * array.c (gfc_ref_dimen_size): Check for BT_INTEGER before calling
> mpz_set.
>
>
> 2015-10-25 Steven G. Kargl
>
> PR fortran/36192
> * gfortran.dg/pr36192.f90: New test.
OK. But I don’t understand why the
> The problem: Statement labels within a type declaration are put in the
> statement label tree belonging to the type declaration's namespace's (instead
> of the current namespace). When the line is otherwise empty and an error is
> issued, gfc_free_st_label tries to delete the label from the
On Fri, Oct 23, 2015 at 5:03 PM, Richard Sandiford
wrote:
> This required reinstating support for captures in the result
> of a simplification. That part (genmatch.c) is by Richard B.
>
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
>
> Thanks,
> Richard
>
On Fri, Oct 23, 2015 at 5:15 PM, Alan Lawrence wrote:
> On 19/10/15 12:49, Richard Biener wrote:
>
>> Err, you should always do the shift in the type of rhs1. You should also
>> avoid the chrec_convert of rhs2 above for shifts.
>
> Err, yes, indeed. Needed to keep the chrec_convert before the
>
Hi all,
After r228136, flag_reorder_blocks_and_partition is canceled when
-fprofile-use is not specified.
In this case check_effective_target_freorder() is not able to check the
proper target support.
This is a simple patch to add "-fprofile-use" option that effective
target check.
Okay to
On Sat, Oct 24, 2015 at 12:29 AM, Ilya Enkovich wrote:
> 2015-10-24 0:32 GMT+03:00 Jeff Law :
>> On 10/23/2015 09:26 AM, Ilya Enkovich wrote:
>>>
>>> Hi,
>>>
>>> This patch checks optab exists before using it vector vector statement
>>> lowering. It fixes compilation of test from PR68062 with -fu
Hi Nathan!
On Sun, 25 Oct 2015 10:01:55 -0400, Nathan Sidwell wrote:
> I've applied this patch to gomp4 branch. It's the reworking of IFN_UNIQUE
> suggested by Richard & Jakub.
>
> 1) IFN_UNIQUE is a ctrl-altering call, and thus ends up at the end of a BB.
> 2) tracer only needs to check that
On Sat, Oct 24, 2015 at 5:14 AM, Jason Merrill wrote:
> On 10/21/2015 12:16 AM, Richard Biener wrote:
>>
>> On Tue, Oct 20, 2015 at 9:10 PM, Jason Merrill wrote:
>>>
>>> I made this change on the delayed folding branch and then noticed that it
>>> broke pointer-arith-10.c, which you added to the
A patch with ChangeLog diffs is attached.
I ran "make && make install && make check-fortran" at the top level on
x86_64-pc-linux-gnu. (I always do that before I post a patch and again before
I actually check anything in. I sometimes forget to include files when I
finally commit, but I'm alway
On Sat, Oct 24, 2015 at 7:31 PM, Jan Hubicka wrote:
> Hello,
> this patch adds OBJ_TYPE_REF that is the only code handled by ipa-icf-gimple
> but not bu operand_equal_p. I tried to make the following testcase:
>
> /* { dg-do compile } */
> /* { dg-options "-O2 -fdump-tree-pre" } */
> struct foo {
On Sat, Oct 24, 2015 at 11:15 PM, Marc Glisse wrote:
> On Fri, 23 Oct 2015, Hurugalawadi, Naveen wrote:
>
> + (minus (bit_and:cs @0 (bit_not @1)) (bit_and:s @0 @1))
>
> I am not sure why we have :c on one bit_and but not the other.
Clearly an omission.
> + (bit_ior:c (bit_and:c @0 (bit_not @1))
On Mon, Oct 26, 2015 at 10:26 AM, Richard Biener
wrote:
> On Sat, Oct 24, 2015 at 11:15 PM, Marc Glisse wrote:
>> On Fri, 23 Oct 2015, Hurugalawadi, Naveen wrote:
>>
>> + (minus (bit_and:cs @0 (bit_not @1)) (bit_and:s @0 @1))
>>
>> I am not sure why we have :c on one bit_and but not the other.
>
> I ran "make && make install && make check-fortran" at the top level on
> x86_64-pc-linux-gnu. (I always do that before I post a patch and again
> before I actually check anything in. I sometimes forget to include files
> when I finally commit, but I'm always confident that the ones I remembe
On 26 Oct 10:09, Richard Biener wrote:
> On Sat, Oct 24, 2015 at 12:29 AM, Ilya Enkovich
> wrote:
> > 2015-10-24 0:32 GMT+03:00 Jeff Law :
> >> On 10/23/2015 09:26 AM, Ilya Enkovich wrote:
> >>>
> >>> Hi,
> >>>
> >>> This patch checks optab exists before using it vector vector statement
> >>> low
An upcoming patch adds a fold from hypot(x,x) to fabs(x)*sqrt(2).
This is unusual in that it could trigger in the gimplifier but would
require new SSA names to be created. This patch makes sure that we
don't try to create new SSA names when we're not yet in SSA form.
Tested on x86_64-linux-gnu, a
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_hypot): Delete.
(fold_builtin_2): Handle constant hypot arguments here.
* match.pd: Fold hypot(x, 0) and hypot(0, x) to x. Canonicalize
An upcoming patch adds a match.pd rule that folds pow(pow(x,y),z)
to pow(x,y*z). This fold can reuse the existing pow gimple statement
and simply replace the operands with x and y*z. However, the y*z
itself requires a separate gimple statement and the code wasn't
prepared to handle that.
Tested
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_pow): Delete in favor of...
(fold_const_builtin_pow): ...this new function. Only handle constant
arguments.
(fold_builtin_2): Updat
On Sat, Oct 24, 2015 at 02:10:26PM -0700, Cesar Philippidis wrote:
> +static tree
> +c_parser_oacc_shape_clause (c_parser *parser, omp_clause_code kind,
> + const char *str, tree list)
> +{
> + const char *id = "num";
> + tree ops[2] = { NULL_TREE, NULL_TREE }, c;
> + loc
On Mon, Oct 26, 2015 at 2:18 AM, Mikhail Maltsev wrote:
> On 10/21/2015 01:57 PM, Richard Biener wrote:
>> Ugh (stupid templates).
>>
>> @@ -387,10 +389,10 @@ base_pool_allocator ::allocate ()
>>block = m_virgin_free_list;
>>header = (allocation_pool_list*) allocation_object::get_d
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_powi): Delete.
(fold_builtin_2): Handle constant powi arguments here.
* match.pd: Add rules previously handled by fold_builtin_powi.
gcc/te
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_exponent): Delete.
(fold_builtin_2): Handle constant expN arguments here.
* match.pd: Fold expN(logN(x)) -> x.
diff --git a/gcc/builtins.c
On Mon, Oct 26, 2015 at 9:39 AM, Eric Botcazou wrote:
> Hi,
>
> this patch extends the simplification formerly done in fold_widened_comparison
> and now in match.pd to all integral types instead of just integer types, and
> comes with an Ada testcase for the enumeral type case.
>
> The patch intro
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_signbit): Delete.
(fold_builtin_2): Handle constant signbit arguments here.
* match.pd: Add rules previously handled by fold_builtin_signbit
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_copysign): Delete.
(fold_builtin_2): Handle constant copysign arguments here.
* match.pd: Add rules previously handled by fold_builtin_copys
On Mon, Oct 26, 2015 at 10:35 AM, Ilya Enkovich wrote:
> On 26 Oct 10:09, Richard Biener wrote:
>> On Sat, Oct 24, 2015 at 12:29 AM, Ilya Enkovich
>> wrote:
>> > 2015-10-24 0:32 GMT+03:00 Jeff Law :
>> >> On 10/23/2015 09:26 AM, Ilya Enkovich wrote:
>> >>>
>> >>> Hi,
>> >>>
>> >>> This patch che
fold_builtin_fabs and fold_builtin_abs had code to handle constant
arguments, but this simply duplicated what the following fold_build1_loc
would do for ABS_EXPR.
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_
Just makes an upcoming patch a bit clearer.
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_fma): Remove constant handling.
(fold_builtin_3): Handle constant fma arguments here.
diff --git a/gcc
On Mon, Oct 26, 2015 at 10:36 AM, Richard Sandiford
wrote:
> An upcoming patch adds a fold from hypot(x,x) to fabs(x)*sqrt(2).
> This is unusual in that it could trigger in the gimplifier but would
> require new SSA names to be created. This patch makes sure that we
> don't try to create new SSA
On Mon, Oct 26, 2015 at 10:38 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_hypot): Delete.
> (fold_builtin_2): Handle constant
This handles both integer and floating-point arguments. It's needed
for the follow-on patch to move fmin and fmax to match.pd.
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* fold-const.c (fold_minmax): Delete.
(fold_bi
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_fmin_fmax): Delete.
(fold_builtin_2): Handle constant fmin and fmax arguments here.
* match.pd: Add rules previously handled by fold_builtin
Hi all,
unfortunately did my last patch create a segfault on some 32-bit
system. This happens because in the scalarizer the lower bound of the
deferred length array of the source= expression was taken to be
constant zero instead of taking that information from the array
descriptor. This patch fixe
fold_builtin_logb and fold_builtin_significand now only handle
constant arguments, so this patch renames them to fold_const...,
to match fold_const_builtin_pow. The idea is to differentiate
constant-only folds so that they can be moved to a const_binop-like
function in future.
The functions also
On Mon, Oct 26, 2015 at 09:59:49AM +0100, Jakub Jelinek wrote:
> Ok for trunk with those changes fixed.
Oops, I've missed that there is no checking of the type (that the
expressions have INTEGRAL_TYPE_P); in the C FE, this is sometimes done
already during the parsing, sometimes during c_finish_omp
Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
OK to install?
Thanks,
Richard
gcc/
* builtins.c (fold_builtin_load_exponent): Rename to...
(fold_const_builtin_load_exponent): ...this and only handle
constant arguments.
(fold_builtin_2): Updat
On 23/10/15 04:56, Mike Frysinger wrote:
> 2015-10-22 Mike Frysinger
>
> * scripts/check_jni_methods.sh.in: Run sort with LC_ALL=C, and
> combine `sort|uniq` into `sort -u`.
Looks OK to me.
Andrew.
On Mon, Oct 26, 2015 at 10:41 AM, Richard Sandiford
wrote:
> An upcoming patch adds a match.pd rule that folds pow(pow(x,y),z)
> to pow(x,y*z). This fold can reuse the existing pow gimple statement
> and simply replace the operands with x and y*z. However, the y*z
> itself requires a separate gi
On Fri, Sep 25, 2015 at 02:59:41PM +0100, Dominik Vogt wrote:
> The following set of two patches implements the function
> __attribute__ ((target("..."))) and the corresponding #pragma GCC
> target("...") on S/390. It comes with certain limitations:
>
> * It is not possible to change any options
On Mon, Oct 26, 2015 at 10:44 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_pow): Delete in favor of...
> (fold_const_builtin_p
With the patch compiling the original test still gives
…
pr36192.f90:39:10:
x_n, v_n, & ! Configuration at t+dt with step dt
1
Error: The module or main program array 'x_n' at (1) must have constant shape
f951: internal compiler error: Segmentation fault: 11
Domini
On Mon, Oct 26, 2015 at 10:48 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_powi): Delete.
> (fold_builtin_2): Handle constant
On Mon, Oct 26, 2015 at 10:50 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_exponent): Delete.
> (fold_builtin_2): Handle const
On 26 Oct 10:56, Richard Biener wrote:
> On Mon, Oct 26, 2015 at 10:35 AM, Ilya Enkovich
> wrote:
> > On 26 Oct 10:09, Richard Biener wrote:
> >> On Sat, Oct 24, 2015 at 12:29 AM, Ilya Enkovich
> >> wrote:
> >> > 2015-10-24 0:32 GMT+03:00 Jeff Law :
> >> >> On 10/23/2015 09:26 AM, Ilya Enkovich
> I think excluding enums had sth to do with C++ -f[no-]strict-enums
> (whatever is the default). Just tried to figure where the check
> came from ... both calls are keyed on INTEGER_TYPE...
Here's the relevant block of code from the C++ front-end:
/* The middle-end currently assumes that
On Mon, Oct 26, 2015 at 10:52 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_signbit): Delete.
> (fold_builtin_2): Handle consta
On Sat, Oct 24, 2015 at 02:11:41PM -0700, Cesar Philippidis wrote:
> @@ -29582,6 +29592,144 @@ cp_parser_oacc_data_clause_deviceptr (cp_parser
> *parser, tree list)
>return list;
> }
>
> +/* OpenACC 2.0:
> + auto
> + independent
> + nohost
> + seq */
> +
> +static tree
> +cp_parser_
Richard Biener writes:
> On Mon, Oct 26, 2015 at 10:41 AM, Richard Sandiford
> wrote:
>> An upcoming patch adds a match.pd rule that folds pow(pow(x,y),z)
>> to pow(x,y*z). This fold can reuse the existing pow gimple statement
>> and simply replace the operands with x and y*z. However, the y*z
On Mon, 26 Oct 2015, Richard Sandiford wrote:
+ /* expN(logN(x)) -> x. */
+ (simplify
+ (exps (logs @0))
+ @0))
We are not very consistent about it, but wasn't there an idea that we
should use non_lvalue in most such places?
--
Marc Glisse
Hi all,
The auto_inc_dec pass can handle 4 types of sequences, described in the comment
at the start of auto-inc-dec.c.
In two of those: FORM_PRE_ADD and FORM_POST_ADD the resulting sequence is a
move followed by a POST_INC or PRE_INC
memory operation.
In the FORM_POST_ADD case the pass transfo
On Mon, Oct 26, 2015 at 10:54 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_copysign): Delete.
> (fold_builtin_2): Handle const
On Mon, Oct 26, 2015 at 10:56 AM, Richard Sandiford
wrote:
> fold_builtin_fabs and fold_builtin_abs had code to handle constant
> arguments, but this simply duplicated what the following fold_build1_loc
> would do for ABS_EXPR.
>
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueab
On Mon, Oct 26, 2015 at 10:57 AM, Richard Sandiford
wrote:
> Just makes an upcoming patch a bit clearer.
>
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_fma): Remov
On Mon, Oct 26, 2015 at 11:00 AM, Richard Sandiford
wrote:
> This handles both integer and floating-point arguments. It's needed
> for the follow-on patch to move fmin and fmax to match.pd.
>
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Ric
This patch fixes a bug that caused the compiler to issue the error "default
iterator must be unique" when one of the alleged "duplicates" is overridden by
another. This can happen in cases involving types derived from types declared
in generic formal packages. The error message (when it is correct)
This patch ensures that only source package and subprogram bodies "freeze" the
contract of the nearest enclosing package body.
-- Source --
-- expr_funcs.ads
package Expr_Funcs
with SPARK_Mode,
Abstract_State => State
is
Var_1 : Integer := 1;
function I
On Mon, Oct 26, 2015 at 11:02 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_fmin_fmax): Delete.
> (fold_builtin_2): Handle cons
On Mon, Oct 26, 2015 at 11:05 AM, Richard Sandiford
wrote:
> fold_builtin_logb and fold_builtin_significand now only handle
> constant arguments, so this patch renames them to fold_const...,
> to match fold_const_builtin_pow. The idea is to differentiate
> constant-only folds so that they can be
With This patch the compiler properly rejects a subprogram body whose
signature is type conformant with a previous expression function in
the same scope.
Compiling same_signature,adb must yield:
same_signature.adb:8:04: body conflicts with expression function at line 5
---
with Ada.Text_IO; us
Hi,
this no-functional-changes patch inlines get_vi_for_tree into
intra_create_variable_infos.
It exposes the interaction with create_variable_info_for_1, and makes it
clear that we're using none of the constraint handling from
create_variable_info_for.
Bootstrapped and reg-tested on x86_6
Hi,
this no-functional-changes patch adds some vars in
intra_create_variables_info.
Bootstrapped and reg-tested on x86_64.
Committed as trivial.
Thanks,
- Tom
Add vars in intra_create_variables_info
2015-10-26 Tom de Vries
* tree-ssa-structalias.c (intra_create_variable_infos): Add
re
On 10/26/2015 12:47 PM, Richard Biener wrote:
> I committed the attached to fix build with the valgrind annotations active.
>
> Richard.
>
Doh! Sorry for breakage.
--
Regards,
Mikhail Maltsev
Hi,
this no-functional-changes patch factors a new function
make_restrict_var_constraints out of intra_create_variable_infos.
Bootstrapped and reg-tested on x86_64.
Committed as trivial.
Thanks,
- Tom
Add make_restrict_var_constraints
2015-10-26 Tom de Vries
* tree-ssa-structalias.c (m
This patch implements the following semantic rules:
The SPARK_Mode aspect can be used in the following places in the code:
* on a library-level subprogram spec or body (which now includes entry
declarations)
* on a library-level task spec or body
* on a library-level protected
VEC_COND_EXPRs have been changed to use boolean vectors for the first
argument.
This changes the vectorizing conditional expressions to use this new
format,
fixing the compiler failures for the 65947-*.c tests.
2015-10-26 Alan Hayward
gcc/
PR tree-optimization/65947
* tree-vect
On 10/26/2015 11:40 AM, Kyrill Tkachov wrote:
In the FORM_POST_ADD case the pass transforms:
*a
...
b <- a + c
into
b <- a
...
*(b += c) post
However, the code in attempt_change that compares the costs of the
before and
This patch modifies the analysis of pragmas Depends, Global, Refined_Depends
and Refined_Global when they appear in a subprogram body. The pragmas are now
analyzed immediately and in pairs. This preserves the inherent dependency of
[Refined_]Depends on [Refined_]Global and prevents a visibility iss
This patch modifies the analysis of pragmas that appear on package specs and
and subprogram bodies. The new model of analysis allows for pragmas to appear
out of order, but they are still analyzed according to internal precedence. The
patch removes the restrictions concerning the declarations order
On Mon, Oct 26, 2015 at 11:08 AM, Richard Sandiford
wrote:
> Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabi.
> OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * builtins.c (fold_builtin_load_exponent): Rename to...
> (fold_const_builtin_
Hi,
this no-functional-changes patch copies the restrict var declaration
code from intra_create_variable_infos to create_variable_info_for_1.
The code was copied rather than moved, since in fipa-pta mode the
varinfo p for the parameter t may already exist due to
create_function_info_for, in
Hi,
this patch enables recursive restrict handling in
create_variable_info_for_1.
This allows us to interpret all restricts in a function parameter
"int *restrict *restrict *restrict a".
This patch is the first step towards implementing a generic fix for PR67742.
Bootstrapped and reg-tested
On Mon, Oct 26, 2015 at 11:30 AM, Richard Sandiford
wrote:
> Richard Biener writes:
>> On Mon, Oct 26, 2015 at 10:41 AM, Richard Sandiford
>> wrote:
>>> An upcoming patch adds a match.pd rule that folds pow(pow(x,y),z)
>>> to pow(x,y*z). This fold can reuse the existing pow gimple statement
>>>
On Mon, Oct 26, 2015 at 11:32 AM, Marc Glisse wrote:
> On Mon, 26 Oct 2015, Richard Sandiford wrote:
>
>> + /* expN(logN(x)) -> x. */
>> + (simplify
>> + (exps (logs @0))
>> + @0))
>
>
> We are not very consistent about it, but wasn't there an idea that we should
> use non_lvalue in most su
The front-end cannot inline subprograms that contain certain declarations,
such as nested subprograms. If the the subprogram to inline includes a
subtype declaration with predicates, it cannot be inlined because the
analysis of the predicate will generate (nested) predicate functions.
Source:
--
This patch implements pragmas [Refined_]Depends and [Refined_]Global on
protected entries and task units. As a prerequisite, the patch implements
aspect specifications on entry bodies as specified in Ada 2012 (AI12-0169):
ENTRY_BODY ::=
entry DEFINING_IDENTIFIER ENTRY_BODY_FORMAL_PART
On Mon, Oct 26, 2015 at 12:12 PM, Alan Hayward wrote:
> VEC_COND_EXPRs have been changed to use boolean vectors for the first
> argument.
> This changes the vectorizing conditional expressions to use this new
> format,
> fixing the compiler failures for the 65947-*.c tests.
Ok.
Richard.
>
> 201
On 10/26/2015 12:12 PM, Bernd Schmidt wrote:
But isn't that balanced by the fact that it doesn't seem to take into
account the gain of removing the inc_insn either? So I think this can't
be right.
Argh, misread the code. The patch is OK with the change I suggested.
Bernd
I've applied the attached one liner as a quick fix for PR target/68091
which is a 6 regression. sh_vector_mode_supported_p returns true even
if the target has no mov insns for some vector modes and causes ICEs
for several vector tests. Tested on sh4-unknown-linux-gnu.
Regards,
kaz
--
201
Hi,
the patch below fixes PR68087, an ICE caused by referring to a negative
constexpr array element.
Bootstrapped and tested on ppc64le.
OK for trunk?
gcc-5 also needs a slightly different fix. I'll post that as a
follow-up, once this one gets approved.
PR c++/68087
* constexpr.
This patch fixes a bug in which GNAT.Expect.Non_Blocking_Spawn fails to find an
executable when the PATH environment variable is not set. The executable
should be found if it contains a directory name.
The following test should execute quietly.
gnatmake -q do_nothing.adb
gnatmake -q spawn.adb
uns
This patch removes spurious errors triggered by the switches -gnatE (Dynamic
elaboration) and -gnatwl (list elaboration warnings) when the elaboration
call is appears in the prefix of an access attribute.
Compiling
gcc -c dds-translator.adb -gnatE -gnatwl
must yield:
dds-translator-genera
On 10/23/2015 10:40 PM, Martin Sebor wrote:
The original code deliberately avoids diagnosing the case of last
array members with bounds greater than 1 (see the comment about
"a poor man's flexible array member" added with a fix for bug
41935) and I didn't want to change that.
Jakub added that,
This patch cleans the way getpagesize() is imported and removes an
inconsistency wrt the underlying C routine. No change in functionality.
Tested on x86_64-pc-linux-gnu, committed on trunk
2015-10-26 Arnaud Charlet
* s-osinte-hpux.ads, s-osinte-aix.ads, s-osinte-solaris-posix.ads,
On Mon, Oct 26, 2015 at 12:40:18PM +0100, Bernd Schmidt wrote:
> On 10/23/2015 10:40 PM, Martin Sebor wrote:
> >
> >The original code deliberately avoids diagnosing the case of last
> >array members with bounds greater than 1 (see the comment about
> >"a poor man's flexible array member" added with
This patch fixes the handling of overloaded indexing operations that are
inherited by a type derived from one that carries an indexing aspect.
Source:
---
with Ada.Text_Io; use Ada.Text_Io;
with References;
procedure Main is
A : aliased References.Iterated;
begin
A (1) := 42;
Put_Line ("A
On 26/10/15 11:28, Bernd Schmidt wrote:
On 10/26/2015 12:12 PM, Bernd Schmidt wrote:
But isn't that balanced by the fact that it doesn't seem to take into
account the gain of removing the inc_insn either? So I think this can't
be right.
Argh, misread the code. The patch is OK with the change
This patch allows First_Component and First_Component_Or_Discriminant to apply
to protected and task types. No change in behavior, no test needed. The patch
is aimed at clients of the compiler front-end.
Tested on x86_64-pc-linux-gnu, committed on trunk
2015-10-26 Hristian Kirtchev
*
This patch adds a legality check to element iterators over containers: the
element is a constant if the container object is a constant, even if the
container type has a Variable_Indexing aspect.
Compiling the following must be rejected with:
strange_behavior.adb:12:07: left hand side of assignme
On 10/26/2015 12:47 PM, Jakub Jelinek wrote:
Because the amount of code that uses this (including GCC itself) is just too
huge, so everywhere in the middle-end we also special case last array members of
structs. While C99+ has flexible array members, e.g. C++ does not, so users
are left with us
On Mon, Oct 26, 2015 at 12:57:53PM +0100, Bernd Schmidt wrote:
> On 10/26/2015 12:47 PM, Jakub Jelinek wrote:
>
> >Because the amount of code that uses this (including GCC itself) is just too
> >huge, so everywhere in the middle-end we also special case last array
> >members of
> >structs. While
Hi Andre,
Yes, that's good to fix the problem. OK to commit
Thanks for the rapid fix.
Paul
On 26 October 2015 at 11:03, Andre Vehreschild wrote:
> Hi all,
>
> unfortunately did my last patch create a segfault on some 32-bit
> system. This happens because in the scalarizer the lower bound of th
This patch implements a new restriction No_Dynamic_Sized_Objects, which is
intended to prevent the creation of composite objects of non-static size.
The following test should get an error.
gcc -c dynamic_string.adb -gnatws
dynamic_string.adb:4:18: violation of restriction "No_Dynamic_Sized_Object
On Fri, Sep 25, 2015 at 02:59:41PM +0100, Dominik Vogt wrote:
> The following set of two patches implements the function
> __attribute__ ((target("..."))) and the corresponding #pragma GCC
> target("...") on S/390. It comes with certain limitations:
>
> * It is not possible to change any options
Ping.
https://gcc.gnu.org/ml/gcc-patches/2015-10/msg01350.html
and the backport for 4.9/5
https://gcc.gnu.org/ml/gcc-patches/2015-10/msg01351.html
Thanks,
Kyrill
On 14/10/15 13:23, Kyrill Tkachov wrote:
Hi all,
This patch fixes the referenced PR by rewriting the vfp3_const_double_for_bits
f
On Mon, 2015-10-26 at 12:12 +0100, Bernd Schmidt wrote:
> On 10/26/2015 11:40 AM, Kyrill Tkachov wrote:
> > In the FORM_POST_ADD case the pass transforms:
> > *a
> > ...
> > b <- a + c
> >
> > into
> >
> > b <- a
> > ...
> >
On Mon, Oct 26, 2015 at 12:32 PM, Markus Trippelsdorf
wrote:
> Hi,
>
> the patch below fixes PR68087, an ICE caused by referring to a negative
> constexpr array element.
>
> Bootstrapped and tested on ppc64le.
> OK for trunk?
>
> gcc-5 also needs a slightly different fix. I'll post that as a
> fol
Hi,
After spending some time looking at ealias/pta dumps, I realized that
they're hard to understand because we use varinfo names to identify
varinfos, while those names are not necessarily unique.
F.i., for a function f:
...
void
f (int *__restrict__ a, int *__restrict__ b)
{
*a = 1;
*b
1 - 100 of 206 matches
Mail list logo
