On Wed, Aug 31, 2011 at 6:53 PM, Aldy Hernandez wrote:
>
Sure. I assume in this case, *bit_offset would be 0, right?
>>>
>>> It would be DECL_FIELD_BIT_OFFSET of that field. Oh, and
>>> *byte_offset would be
>>>
>>> *byte_offset = size_binop (MULT_EXPR, TREE_OPERAND (exp, 2),
>>>
On 31 Aug 2011, at 20:07, Iain Sandoe wrote:
On 31 Aug 2011, at 17:34, Arnaud Charlet wrote:
In particular I'd be curious to know if revision 178376 has the
failure or not.
different failure;
built with BOOT_CFLAGS="-O0 -g" ..
.. it fails debug-compare (ada/exp_ch6.o).
There are a lot
On Thu, Sep 01, 2011 at 02:32:51PM +0800, Terry Guo wrote:
> FAIL: gcc.dg/compat/struct-layout-1 c_compat_x_tst.o compile
> FAIL: gcc.dg/compat/struct-layout-1 c_compat_y_tst.o compile
> UNRESOLVED: gcc.dg/compat/struct-layout-1 c_compat_x_tst.o-c_compat_y_tst.o
> link
> UNRESOLVED: gcc.dg/compat/
Hello Jakub,
>
> On Thu, Sep 01, 2011 at 09:34:20AM +0800, Terry Guo wrote:
> > Here is the patch, is it OK to commit?
>
> Definitely not.
> struct-layout-1* already has support for short enums, see
> http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=110247
> So, if that doesn't work for you, you
On Thu, Sep 01, 2011 at 09:34:20AM +0800, Terry Guo wrote:
> Here is the patch, is it OK to commit?
Definitely not.
struct-layout-1* already has support for short enums, see
http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=110247
So, if that doesn't work for you, you need to debug why.
> 2011-08-
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 08/31/11 15:41, Bernd Schmidt wrote:
> On 08/29/11 18:02, Jeff Law wrote:
>> On 08/26/11 08:47, Bernd Schmidt wrote:
>>> In rest_of_reorder_blocks, we avoid reordering if
>>> !optimize_function_for_speed_p. However, we still call
>>> insert_sectio
Hello,
There are some bitfield definitions in this case, such as "enum E2 a:31".
For most targets, the ENUM type size equals INT type so this definition is
OK. But for some targets like arm_eabi, the default ENUM type size could be
the smallest integer type that can contain all of its enumerated v
On 08/29/11 18:02, Jeff Law wrote:
> On 08/26/11 08:47, Bernd Schmidt wrote:
>> In rest_of_reorder_blocks, we avoid reordering if
>> !optimize_function_for_speed_p. However, we still call
>> insert_section_bounary_note, which can cause problems because now, if
>> we have a sequence of HOT-COLD-HO
On 08/31/11 19:43, Jeff Law wrote:
> Presumably the jump & call flags are copied as part of the
> shallow_copy_rtx call, thus removing the explicit copy is safe?
That's the conclusion I came to.
Bernd
On 08/31/11 20:43, Richard Sandiford wrote:
> Bernd Schmidt writes:
>> This is necessary when adding shrink-wrapping; otherwise dwarf2cfi sees
>> inconsistent information and aborts.
>>
>> Tested on mips64-elf together with the rest of the shrink-wrapping
>> patches. Ok?
>
> It looks like the cur
Uros Bizjak writes:
> On Mon, Aug 15, 2011 at 11:57 AM, Richard Sandiford
> wrote:
>
Following on from the two patches I've just posted, this one makes
config/i386/*.md use match_test for .md attributes. Tested as
described here:
>>>
http://gcc.gnu.org/ml/gcc-patches/2011
On Wed, 31 Aug 2011, Artem Shinkarov wrote:
> On Wed, Aug 31, 2011 at 4:38 PM, Joseph S. Myers
> wrote:
> > On Wed, 31 Aug 2011, Artem Shinkarov wrote:
> >
> >> 1) Helper function for the pseudo-builtins.
> >> In my case the builtin can have 2 or 3 arguments, and I think that I
> >> expressed tha
gfc_ss structs of type GFC_SS_SCALAR (and GFC_SS_REFERENCE) are not uncommon.
Let's share their initialization.
There is one single case of GFC_SS_REFERENCE, everything else is GFC_SS_SCALAR,
so I have decided to always set GFC_SS_SCALAR in the initialisation, and reset
it to the right value afte
Now that array gfc_ss structs are initialized right after allocation,
especially their DIM and DIMEN fields are set properly, later initializations
of those fields are now redundant.
This patch removes such initializations and/or replaces them with asserts that
the value is already correct.
OK?
2
gfc_ss structs of type GFC_SS_TEMP are not uncommon. Let's share their
initialization.
OK?
2011-08-30 Mikael Morin
* trans-array.h (gfc_get_temp_ss): New prototype.
* trans-array.c (gfc_get_temp_ss): New function.
(gfc_conv_resolve_dependencies): Re-use gfc_get_temp_ss.
All the gfc_ss of type GFC_SS_FUNCTION, GFC_SS_ARRAY, GFC_SS_CONSTRUCTOR,
GFC_SS_VECTOR, ... have the same kind of initialization. Let's share it.
OK?
2011-08-30 Mikael Morin
* trans-array.h (gfc_get_array_ss): New prototype.
* trans-array.c (gfc_get_array_ss): New function.
Hello,
the 4 follow-up patches try to refactor some common code initializing gfc_ss
structs.
Regression-tested (the 4 patches together only) on x86_64-freebsd8.2.
OK for trunk?
Mikael
trans-array.c | 181 -
trans-array.h |6 ++
tr
On 31 Aug 2011, at 20:07, Iain Sandoe wrote:
Same for revision 178311
will set this going .. prob. tomorrow before a report.
fails with:
"exp_light.ali" not found "exp_light.adb" must be compiled
(that issue was already reported by Richi).
so .. not much progress ... will try bisec
On 31 Aug 2011, at 17:34, Arnaud Charlet wrote:
Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0x
0x001c4fa0 in lib__writ__write_ali ()
(gdb) bt
#0 0x001c4fa0 in lib__writ__write_ali ()
#1 0x0036799a in _ada_gnat1drv ()
#2
Bernd Schmidt writes:
> This is necessary when adding shrink-wrapping; otherwise dwarf2cfi sees
> inconsistent information and aborts.
>
> Tested on mips64-elf together with the rest of the shrink-wrapping
> patches. Ok?
It looks like the current code doesn't handle the RESTORE instruction.
Could
Did you test Ada and enable the C++ memory model? ;)
See my earlier comment on Ada. Who would ever use the C++ memory model
on Ada?
Btw, even if the bitfield we access (and thus the whole region) is at a
constant offset, the field _following_ the bitregion (the one you query
above with ge
Sure. I assume in this case, *bit_offset would be 0, right?
It would be DECL_FIELD_BIT_OFFSET of that field. Oh, and
*byte_offset would be
*byte_offset = size_binop (MULT_EXPR, TREE_OPERAND (exp, 2),
size_int (DECL_OFFSET_ALIGN
(field) / BITS_PER_UNIT)
This is a new version of the original 4/6 shrink wrapping patch, minus
the preliminary bits that were already approved, and with some extra bug
fixes that were discovered in the meantime. This is now mostly contained
to just function.c. I'll resubmit the other parts (i.e. exposing more
shrink-wrapp
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 08/31/11 11:20, Bernd Schmidt wrote:
> With the final shrink-wrapping patch applied, I see failures in
> dwarf2cfi on mips64-elf. The problem is that reorg.c uses copy_rtx
> to copy instructions, and for some reason that clears the
> frame_related
On Aug 31, 2011, at 1:27 AM, Artem Shinkarov wrote:
>> If you're going to add vector shuffling builtins, you might consider adding
>> the same builtin that clang has for compatibility:
>> http://clang.llvm.org/docs/LanguageExtensions.html#__builtin_shufflevector
>>
>> It should be straight-forw
This is necessary when adding shrink-wrapping; otherwise dwarf2cfi sees
inconsistent information and aborts.
Tested on mips64-elf together with the rest of the shrink-wrapping
patches. Ok?
Bernd
* config/mips/mips.c (cfa_restores): New static variable.
(mips_restore_reg): Add to
On Wed, Aug 31, 2011 at 4:38 PM, Joseph S. Myers
wrote:
> On Wed, 31 Aug 2011, Artem Shinkarov wrote:
>
>> 1) Helper function for the pseudo-builtins.
>> In my case the builtin can have 2 or 3 arguments, and I think that I
>> expressed that in a pretty much short way without any helper function.
>
With the final shrink-wrapping patch applied, I see failures in
dwarf2cfi on mips64-elf. The problem is that reorg.c uses copy_rtx to
copy instructions, and for some reason that clears the frame_related
bit. We end up with a prologue insn in a delay slot, and dwarf2cfi
disregards its effects.
I se
> Program received signal EXC_BAD_ACCESS, Could not access memory.
> Reason: KERN_PROTECTION_FAILURE at address: 0x
> 0x001c4fa0 in lib__writ__write_ali ()
> (gdb) bt
> #0 0x001c4fa0 in lib__writ__write_ali ()
> #1 0x0036799a in _ada_gnat1drv ()
> #2 0x000305f5 in gnat_parse_file ()
I j
Hello,
I'd like to have tighter control over the individual situations that
-mrecip handles, and I think the user might appreciate this too. Hence
I've introduced four new target options -mrecip-div, -mrecip-sqrt,
-mrecip-vec-div and -mrecip-vec-sqrt. I've redefined -mrecip to be
equivalent
On 31 Aug 2011, at 16:57, Arnaud Charlet wrote:
(x86_64-unknown-linux-gnu) |
| Storage_Error stack overflow or erroneous memory
access |
| Error detected at system.ads:
175:5 |
+
=
=
=
=
=
=
===
On 31 Aug 2011, at 16:53, Joseph S. Myers wrote:
On Wed, 31 Aug 2011, Arnaud Charlet wrote:
+===GNAT BUG
DETECTED==+
| 4.7.0 20110831 (experimental) [trunk revision 161655]
(x86_64-unknown-linux-gnu) |
| Storage_Error stack overflow or
> (x86_64-unknown-linux-gnu) |
> | Storage_Error stack overflow or erroneous memory access |
> | Error detected at system.ads:175:5 |
> +==+
>
>
>
> Please include these
On Wed, 31 Aug 2011, Arnaud Charlet wrote:
> > >> +===GNAT BUG
> > >> DETECTED======+
> > >> | 4.7.0 20110831 (experimental) [trunk revision 161655]
> > >> (x86_64-unknown-linux-gnu) |
> > >&g
On Wed, 31 Aug 2011, Artem Shinkarov wrote:
> 1) Helper function for the pseudo-builtins.
> In my case the builtin can have 2 or 3 arguments, and I think that I
> expressed that in a pretty much short way without any helper function.
> Am I missing something?
The point is to refactor what's commo
*bit_offset = (TREE_INT_CST_LOW (DECL_FIELD_OFFSET (fld))
* BITS_PER_UNIT
+ TREE_INT_CST_LOW (DECL_FIELD_BIT_OFFSET (fld)))
- (TREE_INT_CST_LOW (DECL_FIELD_OFFSET (bitregion_start))
* BITS_PER_UNIT
+ TREE_INT_CST_LOW (DECL_FIELD_BIT_OFFSET (bitregion_sta
> >> +===GNAT BUG
> >> DETECTED======+
> >> | 4.7.0 20110831 (experimental) [trunk revision 161655]
> >> (x86_64-unknown-linux-gnu) |
> >> | Storage_Error stack overflow or erroneous memory access
orry about the pain. Yes, I'm trying hard to not break bootstrap, and when
> I do, I fix it as soon as possible.
>
>> +===GNAT BUG DETECTED==+
>> | 4.7.0 20110831 (experimental) [trunk revision 161655]
I am friendly pinging this patch.
--- Begin Message ---
Hello,
Consider this code snippet:
class C
{
struct Private { };
};
template
struct exploit1
{
C::Private type;
};
exploit1 x1; //#1
At the instantiation point of exploit1 in #1, the compiler
$subject says it all, COND_EXPR to follow. rhs1 will still be
a "single" tree, namely a is_gimple_condexpr () operand. That's
not easy to change (without forcing it to a separate statement,
thus making it a is_gimple_val) because it embeds another tree code.
Bootstrapped and tested on x86_64-un
On Sat, Jul 30, 2011 at 12:21 AM, Tom de Vries wrote:
>
> This is an updated version of the patch. I have 2 new patches and an updated
> testcase which I will sent out individually.
>
> Patch set was bootstrapped and reg-tested on x86_64.
>
> Ok for trunk?
>
> Thanks,
> - Tom
>
> 2011-07-30 Tom d
On 9 August 2011 11:30, David Gilbert wrote:
> Other than Ramana's comment about my comment on 4/4, has anyone else got any
> other input? Otherwise I'd like to fix that comment and then get it in.
We now have comments from Ramana on patches 3/4 and 4/4 - anyone for
anything on the
main pair ?
Hello Richard,
Thanks for the review!
> The fact that you have to adjust gcc.dg/tree-ssa/pr38533.c looks problematic
> to me. Can you investigate the problem report, look at the geneated
> code with the atom default of the param and see whether it's still
> reasonably "fixed" (maybe you'd alread
On Wed, Aug 31, 2011 at 1:02 PM, Artem Shinkarov
wrote:
> Here is a newer version of the patch, which transforms the builtin to
> the VEC_SHUFFLE_EXPR in the front-end.
>
> Several comments:
> 1) Helper function for the pseudo-builtins.
> In my case the builtin can have 2 or 3 arguments, and I thi
and when
I do, I fix it as soon as possible.
> +===GNAT BUG DETECTED======+
> | 4.7.0 20110831 (experimental) [trunk revision 161655]
> (x86_64-unknown-linux-gnu) |
> | Storage_Error stack overflow or erroneous memory access
On Wed, Aug 31, 2011 at 9:45 AM, Richard Guenther
wrote:
> On Tue, Aug 30, 2011 at 5:01 PM, Aldy Hernandez wrote:
>> [I'm going to respond to this piece-meal, to make sure I don't drop
>> anything. My apologies for the long thread, but I'm pretty sure it's in
>> everybody's kill file by now.]
>>
Hi,
tested x86_64-linux, committed to mainline.
Paolo.
///
2011-08-31 Paolo Carlini
* include/std/chrono (operator*(const _Rep1&, const duration<>&)):
Fix order of template parameters per LWG 2004.
Index: include/std/chrono
===
don't break bootstrap all the time? I'm
now simply testing patches w/o Ada for the time being (having wasted
another hour for verifying it wasn't my patch causing the last bootstrap
error I ran into, loads of
+===GNAT BUG DETECTED=
If a declaration within a generic unit has aspects, the capture of references
within the aspect expressions has to be done in a separate step because the
aspect specificatios are not directly attached to the tree for the declaration.
The following must compile quietly:
gcc -c -gnat12 -gnata ml
This change ensures that partitions in a DSA application always use a
64 bit representation when exchanging data of type Stream_Element_Offset or
Storage_Offset, to avoid interoperability issues between 32- and 64-bit hosts.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Thomas Qui
On Windows the directory separator can be / or \, in some cases this
was not properly handled. This patch fixes this issue.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Pascal Obry
* a-direct.adb: Use Dir_Seps everywhere to properly handle all
directory spepara
When an indexed reference to a packed array is expanded, it is rewritten as
a boolean operation on the prefix that denotes the packed array. If the prefix
is a function call it must not be reanalyzed, because it will have been expanded
already. and may carry extra actuals for accessibility checks.
In an instance, errors that report that an aggregate (sub)component is limited
must be ignaored because they originate in view conflicts. If the original
aggregate is legal and the actuals are legal, the aggregate in the instance
itself is legal.
No simple example available.
Tested on x86_64-pc-l
This patch fixes an obscure bug, in which an error "prefix of dereference must
be an access type" is given incorrectly. The spurious error would occur in an
instance of a generic if an implicit dereference of an expression of a private
type is used.
No simple test is available.
Tested on x86_64-pc
.. I applied the patch after running make info and make dvi, seems
obvious to me.
Thanks,
Paolo.
Priority may be specified for protected and task types and objects using the
standard aspect notation. This patch makes the analysis of the aspect
argument earlier, so it is taken into account when the task or protected
object is been created.
The following must compile quietly in Ada2012 mode and
Replaced a comment about lock status with a proper assertion, and added a
comment to explain the representation invariant for the lock status.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Matthew Heaney
* a-rbtgbo.adb (Clear_Tree): Assert representation invariant for l
When generating scil files (-gnatcC), we want to remove any old scil files
corresponding to the main unit being compiled in case of errors, to avoid
re-analyzing these old scil files.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Arnaud Charlet
* comperr.adb, comperr.a
When the SPARK restriction was set, GNAT was not issuing violations on some
derived types. Now corrected.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Marc Sango
* restrict.adb (Check_SPARK_Restriction): Change Comes_From_Source (N)
by Comes_From_Source (Origi
This is a small refactoring that remove some duplicate code.
No functional change.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Tristan Gingold
* exp_ch7.ads, exp_ch7.adb (Finalization_Exception_Data): New type to
hold variables between these following subprogr
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Yannick Moy
* exp_alfa.adb (Expand_Alfa_N_Package_Declaration,
Expand_Alfa_N_Subprogram_Body): Remove useless procedures which simply
call Qualify_Entity_Names.
(Expand_Alfa): call Qualify_Entity_Names
On Wed, Aug 31, 2011 at 10:35 AM, Duncan Sands wrote:
> Hi Artem,
>
> On 31/08/11 10:27, Artem Shinkarov wrote:
>>
>> On Wed, Aug 31, 2011 at 12:51 AM, Chris Lattner
>> wrote:
>>>
>>> On Aug 30, 2011, at 10:01 AM, Artem Shinkarov wrote:
>>
>> The patch at the moment lacks of some examples
When a node is allocated from the free store, its pointer components (the links
to other nodes in the tree) must also be initialized (to 0, the equivalent of
null). This simplifies the post-allocation handling of nodes inserted into
terminal positions.
Tested on x86_64-pc-linux-gnu, committed on t
A selected_component whose selector_name denotes an entity of a concurrent
tagged type may be ambiguous because the target entity may be covered by
a class-wide subprogram. This patch adds this missing test to the frontend
to report the ambiguity. The following test must now compile with errors:
p
When manipulating bounded tree-based container objects, that do not otherwise
have any explicit initialization expression, the compiler would emit a warning
about the object not being initialized, because the Nodes component of the tree
type had not been given an initialization expression.
This wa
The compiler incorrectly rejects a generic instantiation that passes an
undiscriminated private type when the full type has discriminants and
the generic does an assignment to a dereferenced access value denoting
an object of the formal type. The error message complains that the type
has no discrim
When making sizetypes no longer sign-extended (they are unsigned)
we run into extract_muldiv_1 miscompiling the Ada RTS during
secondary stack initialization while folding sizes for an allocation.
From
((sizetype) (_GLOBAL.SZ4_system.secondary_stack (.last, .first) /[cl] 8) + 15 &
0x0ff
Hash values must be folded by the length of the buckets array. The length of
the buckets array is a discriminant of the container type, which means that it
is possible for the user to specify 0 as the length value. Insert must
therefore check for this case specifically, in order to prevent divide-b
When the frontend resolves a dispatching call through the object
operation notation it must also check if there is a class-wide
subprogram covering the target primitive. This check was missing
in the frontend. After this patch the following test must
compile with errors:
package Pkg1 is
type If
Alfa cross references should only contain 'call' references for operators.
Filter out unwanted references which are otherwise generated on intrinsic
operator.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-31 Yannick Moy
* lib-xref-alfa.adb (Add_Alfa_Xrefs): Do not take int
The attribute reference 'Old ppears within a postcondition, but refers to
an entity in the enclosing subprogram. If the prefix is a component of a formal
its expansion might generate actual subtypes that may be referenced in an inner
context, and which must be elaborated within the subprogram itse
When compiling with -gnatc (or any mode that turns off the expansion phase),
the compiler gets an Assertion_Failure when attempting to retrieve the body
discriminal of a discriminant of a protected type. In the absence of expansion,
a concurrent record type is not created, so this possibility has t
Symmetric_Difference iterates over each hash table, populating the result set
as it identifies items in one set not in the other set. The iterator works by
passing a node index value back to the caller, which it in turn uses to
dereference the node array. Both the Left and Right sets are visible in
Hi Artem,
On 31/08/11 10:27, Artem Shinkarov wrote:
On Wed, Aug 31, 2011 at 12:51 AM, Chris Lattner wrote:
On Aug 30, 2011, at 10:01 AM, Artem Shinkarov wrote:
The patch at the moment lacks of some examples, but mainly it works
fine for me. It would be nice if i386 gurus could look into the w
On Wed, Aug 31, 2011 at 12:51 AM, Chris Lattner wrote:
> On Aug 30, 2011, at 10:01 AM, Artem Shinkarov wrote:
The patch at the moment lacks of some examples, but mainly it works
fine for me. It would be nice if i386 gurus could look into the way I
am doing the expansion.
M
On Wed, Aug 31, 2011 at 1:51 AM, Chris Lattner wrote:
> On Aug 30, 2011, at 10:01 AM, Artem Shinkarov wrote:
The patch at the moment lacks of some examples, but mainly it works
fine for me. It would be nice if i386 gurus could look into the way I
am doing the expansion.
Mi
On Tue, Aug 30, 2011 at 7:01 PM, Artem Shinkarov
wrote:
> On Tue, Aug 30, 2011 at 2:03 PM, Richard Guenther
> wrote:
>> On Tue, Aug 30, 2011 at 4:31 AM, Artem Shinkarov
>> wrote:
>>> Hi
>>>
>>> This is a patch for the explicit vector shuffling we have discussed a
>>> long time ago here:
>>> http
On Tue, Aug 30, 2011 at 8:13 PM, Aldy Hernandez wrote:
>
>> Btw, *byte_offset is still not relative to the containing object as
>> documented, but relative to the base object of the exp reference
>> tree (thus, to a in a.i.j.k.l instead of to a.i.j.k). If it were supposed
>> to be relative to a.i
On Tue, Aug 30, 2011 at 6:15 PM, Aldy Hernandez wrote:
>
>> *bit_offset is supposed to be relative to *byte_offset then it should
>> be easy to calculate it without another get_inner_reference.
>
> Since, as you suggested, we will terminate early on variable length offsets,
> we can assume both DE
On Tue, Aug 30, 2011 at 5:01 PM, Aldy Hernandez wrote:
> [I'm going to respond to this piece-meal, to make sure I don't drop
> anything. My apologies for the long thread, but I'm pretty sure it's in
> everybody's kill file by now.]
>
>> + /* Be as conservative as possible on variable offsets. *
On Tue, Aug 30, 2011 at 6:50 PM, Martin Jambor wrote:
> Ping. Re-bootstrapped and re-tested yesterday on x86_64-linux.
Ok. Does this also apply (maybe in modifed form) to the 4.6 branch?
Thanks,
Richard.
> THanks,
>
> Martin
>
>
> On Fri, Jul 29, 2011 at 10:55:31PM +0200, Martin Jambor wrote:
On Tue, Aug 30, 2011 at 6:38 PM, Martin Jambor wrote:
>
> Hi,
>
> this is a ping of a patch that I have originally posted almost two
> months ago.
>
> Since (same body) aliases have their own cgraph_nodes, the check for
> them in cgraph_redirect_edge_call_stmt_to_callee is now unnecessary
> becaus
82 matches
Mail list logo
