Hi Arnaud, this is a great feature. How does it handle unchecked unions?
Will
it warn if you write to a field but only read the value via a different field?
It's a simple minded check, nothing fancy. It will warn when there are no
reference to the the object at all after the assignment, and
Pursuant to Arno's comments on PR ada/49444, here is another patch that
allows GCC to compile with Ada support on Debian GNU/kFreeBSD.
2011-08-04 Ludovic Brenta ludo...@ludovic-brenta.org
* gcc/ada/gcc-interface/Makefile.in (kfreebsd%):
- use s-taprop-posix.ad[bs] instead of
On 08/04/2011 12:21 PM, Romain Geissler wrote:
On Mon, Aug 01, 2011 at 02:27:49PM +0200, Romain Geissler wrote:
ping
I went ahead and bootstrapped/regtested/and make install tested your patch
(note your mailer wrapped it up so that it didn't apply cleanly), but now
that I think about it, it
2011/8/3 Richard Guenther richard.guent...@gmail.com:
On Wed, Aug 3, 2011 at 3:32 PM, Kai Tietz ktiet...@googlemail.com wrote:
2011/8/3 Michael Matz m...@suse.de:
Hi,
On Tue, 2 Aug 2011, Kai Tietz wrote:
this patch improves the ability of reassociation pass to simplifiy
more complex
On Thu, 4 Aug 2011, Richard Guenther wrote:
On Wed, 3 Aug 2011, Mikael Morin wrote:
Hello,
On Wednesday 03 August 2011 15:47:37 Richard Guenther wrote:
Comments? Any idea why reversing the loop would break?
Yes, the list of scalarized expressions has to be created in the same
This patch adjusts the dump output from modulo-sched.c:get_schdedule_window.
Dump output is very much down to personal preference, so please feel free
just to reject the change.
Anyway, the current output for a typical scheduling window looks like this:
Hello Richard,
This patch adjusts the dump output from
modulo-sched.c:get_schdedule_window.
Dump output is very much down to personal preference, so please feel free
just to reject the change.
The output format looks great to me although I cannot approve it.
Thanks,
Revital
This patch adds the missing support to the frontend to save and restore
consistently the settings of pragma Normalize_Scalars. Required to have
consistent setting of its value when the compilation of a unit causes
implicit analysis of runtime units. Before applying this patch the
compilation of
This adjusts the finalization code emitted in the binder-generated file so as
to make it symmetrical to the initialization code with regard to the handling
of the elaboration counter. Both now increment or decrement the counter only
once per library unit, even though the elaboration or
An allocator for a local task in a package declaration will raise Program_Error
because the elaboration check on the body will fail. THis patch adds a warning
for
this case.
Compiling the following must yield:
tasking.ads:6:18: warning: cannot activate task before body seen
This change fixes a defect whereby instantiating a Remote_Types unit
specifying user-defined stream attributes for one of its types in the
private part of an RCI unit would cause undefined symbols in calling stubs.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Thomas Quinot
On Thursday 04 August 2011 13:12:04 Richard Guenther wrote:
On Thu, 4 Aug 2011, Richard Guenther wrote:
On Wed, 3 Aug 2011, Mikael Morin wrote:
Hello,
On Wednesday 03 August 2011 15:47:37 Richard Guenther wrote:
Comments? Any idea why reversing the loop would break?
Yes,
In this patch, we enable calls to the binder and linker in gnatmake in
CodePeer mode, to help support e.g. detection of global uninitialized variables
or e.g. environment task race conditions.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Arnaud Charlet char...@adacore.com
To prevent anomalies with enumeration types with holes and scalar types with
complex predicates, the attributes First, Last, and Range cannot be applied to
subtypes with predicates. This rule applies only to scalar types.
The following must compile quietly in Ada2012 mode:
package Pred is
This change ensures that no statement SCO is emitted for a disabled pragma.
For the following compilation, the CS line shall contain only one entry
(for the assignment statement), and no entry for the (disabled) pragma Assert.
$ gcc -c -gnateS plop.adb
$ grep ^C plop.ali
C 1 plop.adb
CS 8:4-8:15
On 08/04/11 13:16, Richard Sandiford wrote:
Tested in the same way as the previous patch. OK to install?
You don't actually need to ask, and based on Revital's reaction, yes.
Bernd
GNAT was issuing a wrong message about a missing operand in some cases of
ill-parenthesized code in Ada 2012 mode. Now it issues a correct message about
the missing parentheses.
On the following code we get:
$ gcc -c -gnat2012 -gnaty3 pred.ads
pred.ads:5:30: conditional expression must be
GNAT was evaluating the right part of an AND-THEN pre- or postcondition before
the left part, which could cause a wrong exception to be raised in case of a
failing pre- or postcondition.
On the code attached, compiling and executing now raises a precondition
failure instead of a division by zero:
-state.c cannot parse
(!version 4.7.0 20110804 (experimental) [trunk revision 161655])
please fix.
Jakub
On Thu, Aug 4, 2011 at 2:07 AM, Kirill Yukhin kirill.yuk...@gmail.com wrote:
HJ, are you sure your Changlog entry is OK?
It was fixed.
--
H.J.
On Wednesday 03 August 2011 17:55:00 Tobias Burnus wrote:
Though variables in the general case can be components, I don't think it
is the case here as only named variables are involved here.
Does that sound right?
The first part of the sentence sounds wrong: A component itself is not a
On Wed, Aug 3, 2011 at 1:14 PM, Jason Merrill ja...@redhat.com wrote:
On 08/03/2011 08:46 AM, Richard Guenther wrote:
If that's reasonable then adding the malloc attribute should be, too.
Finally. Please. Doesn't C++0x maybe fix the issue we were
discussing to death?
Nope, as far as I can
, committed.
This breaks bootstrap with a ./contrib/gcc_update updated tree because
gengtype-state.c cannot parse
(!version 4.7.0 20110804 (experimental) [trunk revision 161655])
please fix.
False alarm - I was confused by seeing libstdc++ build. The above is
caused by a local VRP patch of mine
On Thu, Aug 4, 2011 at 2:58 PM, Gabriel Dos Reis
g...@integrable-solutions.net wrote:
On Wed, Aug 3, 2011 at 1:14 PM, Jason Merrill ja...@redhat.com wrote:
On 08/03/2011 08:46 AM, Richard Guenther wrote:
If that's reasonable then adding the malloc attribute should be, too.
Finally. Please.
There was one use of floating point in the front end, in a function
to estimate the equivalent decimal exponent for a universal real.
While this use was probably OK, because a different estimate
shouldn't affect compilation results, in general we do not want
anything in the compiler that may
When resolving a prefixed call with class-wide actuals, we iterate over the
class-wide operations of all ancestors of the controlling type. If the context
is a function call any overloading is resolved by context. if the context is
a procedure call there must be only one candidate interpretation,
Hi,
On Wed, 3 Aug 2011, Kai Tietz wrote:
This machinery doen't work in this case
That's why you have to extend it.
The issue about this machinery is that it assumes that the statement
itself gets transformed, but for normalized form of invert of bitwise
operations it is essential to
If a library project extending another one has no sources of its own,
and there is an exception name with capital letters in the project
being extended, then on platform with case-insensitive file names
(such a Windows, Darwin or VMS), the invocation of gnatmake to build
the library will fail.
AI05-0025 specifies that a formal package is illegal if it includes a named box
initialization for an overloaded formal subprogram. This is an extension of an
existing rule for instantiations.
Compiling proc1.adb in Ada2005 mode must yield the following:
proc1.adb:10:05: instantiation
In formal verification mode, in order to properly detect which parameters may
be read/written by a subprogram, a new information is added to ALFA cross
references.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy m...@adacore.com
* alfa.ads (ALFA_Xref_Record):
On Thu, Aug 4, 2011 at 3:01 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Aug 4, 2011 at 2:58 PM, Gabriel Dos Reis
g...@integrable-solutions.net wrote:
On Wed, Aug 3, 2011 at 1:14 PM, Jason Merrill ja...@redhat.com wrote:
On 08/03/2011 08:46 AM, Richard Guenther wrote:
If
On 08/04/2011 02:44 PM, Mikael Morin wrote:
and I belief both var%comp and
var are named.
My reading is that a named variable is a variable that is an object-name.
That's actually my problem with the standard - it never quite tells what
a named variable exactly is. I think at least
There was a problem with generation of ALFA cross-references for formal
verification, which causes a reference to appear in two scopes. Now corrected.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy m...@adacore.com
* lib-xref-alfa.adb (Add_ALFA_Xrefs):
In ALFA mode, we modify expansion so that pragma check are kept in the code,
while pre- and postconditions are kept attached to entities rather than
being inserted in the code.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy m...@adacore.com
* exp_prag.adb
This AI specifies that a user-defined equality on an anonymous access type
whose designated type is private does not lead to an ambiguity with the
universal access equality operator in the body or child units of the defining
package. The same is true for a multiplication operator on a private type
This patch reverses the previous patch which saved/restore the values of
pragma Normalize_Scalars because the value of Normalize_Scalars must not
be saved/restored because once set to true its value never changes. That
is, if a compilation unit has pragma Normalize_Scalars then it forces
that
The formal verification backend relies on unique names for the variables named
in ALFA sections of ALI files. Thus, generate these unique names with a new
function.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy m...@adacore.com
* frontend.adb (Frontend):
The function Identifer_For, called from the code that generates the type
declarations for string types in the standard package, now also sets the
entity of the generated identifier.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Johannes Kanig ka...@adacore.com
*
Jan Hubicka hubi...@ucw.cz writes:
Cool, will this also help to handle alternate entry points, so we can move
ahead
on this area we got stuck years ago? (i.e. add support for alternate entry
points on
i386 skipping IP pointer load in PIC mode/using register passing
conventions)
Follow-up of changes for Test_Case pragma. Pragma is not allowed inside
subprogram body, only after separate declaration. No two test cases with same
name allowed on same entity. Correct error in checking procedure.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy
The dynamic stack usage engine has been rewritten to slightly simplify it.
It now provides accurate results on machine where the stack base is known.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Tristan Gingold ging...@adacore.com
* s-tassta.adb (Task_Wrapper): Rewrite
On 08/04/2011 08:58 AM, Gabriel Dos Reis wrote:
Do you intend to rule out garbage collectors?
No, I suppose the rule should be that interleaved access through the
returned pointer and other ways is undefined.
Should not access as raw memory (e.g. through char* or void*) be allowed?
No,
This patch improves the support for interface conversions in the .NET/JVM
compiler extending the current support for attribute 'tag and adding the
missing runtime checks required in interface conversions when the tag of
the source is unknown at compile time. After this patch the following test
On Thu, Aug 4, 2011 at 3:50 PM, Gabriel Dos Reis
g...@integrable-solutions.net wrote:
On Thu, Aug 4, 2011 at 8:43 AM, Jason Merrill ja...@redhat.com wrote:
On 08/04/2011 08:58 AM, Gabriel Dos Reis wrote:
Do you intend to rule out garbage collectors?
No, I suppose the rule should be that
Follow-up of implementation of Test_Case pragma/aspect.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy m...@adacore.com
* sem_prag.adb (Check_Arg_Is_String_Literal): remove useless procedure
(Analyze_Pragma): allow static string expression for name of
ALFA mode used for formal verification requires different cross references,
in which read of constants is absent, and formal is not referenced when used
as selector of parameter association.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Yannick Moy m...@adacore.com
*
When manipulating bounded hash-table 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 hash table
type had not been given an initialization
On Wed, 3 Aug 2011, Jakub Jelinek wrote:
2011-08-03 Jakub Jelinek ja...@redhat.com
* gcc.c (self_spec): New variable.
(static_specs): Add self_spec.
(main): Call do_self_spec on self_spec specs after reading
user specs files. Move compare_debug handling right
This patch fixes the result check for coarrays / variables with coarray
subcomponents. It was working with a separate RESULT() variable - but
not if the function name was the result variable.
Build and regtested on x86-64-linux.
OK for the trunk?
Tobias
2011-08-04 Tobias Burnus
When the SPARK restriction was set, GNAT was not issuing violations on generic.
Now corrected.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Marc Sango sa...@adacore.com
* sem_ch12.adb (Analyze_Generic_Package_Declaration,
Analyze_Generic_Subprogram_Declaration,
It is not possible anymore to generate the C version of
the binder file. Switches -A and -C of gnatbind are now
removed.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Tristan Gingold ging...@adacore.com
* bindgen.adb (Gen_Adainit_C): Remove.
(Gen_Adafinal_C):
This new restriction is intended to prevent the use of the predefined stream
attributes for elementary types. A consequence of this restriction is that
the default implementation of stream attributes for composite types cannot
be created if any of its elementary components lacks user-defined Read
GNAT compilation error when SPARK restriction mode was set with attribut
reference violation is now correct.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-08-04 Marc Sango sa...@adacore.com
* sem_attr.adb (Analyze_Attribute): Replace the message
invisible attribute
This change removes circuitry in gnatls that tried to approximate what
is done in Prj.Env.Initialize_Default_Project_Path to set the project
search path, and instead uses that routine directly. This fixes an
inconsistency between gnatls' output and the actual behaviour of
other project aware
This patch adds initial support for PPH images that include other PPH
images. Currently, we support this but it is inefficient. When a
parent image includes a child image, the parent embeds all the
contents of the child in its image. This is a waste of space.
There is one other missing piece
On 08/03/2011 04:57 AM, Torvald Riegel wrote:
Test static constructors inside of transactional code.
* testsuite/libitm.c++/static_ctor.C: New file.
Ok.
+static void *thread (void *dummy __attribute__((unused)))
This is C++. Unnamed parameters work.
r~
2011/8/4 Richard Henderson r...@redhat.com:
On 08/03/2011 11:09 PM, Denis Chertykov wrote:
2011/8/4 Richard Henderson r...@redhat.com:
When a frame pointer is in use, we can optimize popping all
queued parameters via a simple move from the frame pointer
instead of an addition to the stack
On 08/04/2011 09:04 AM, Denis Chertykov wrote:
Thank you for explanation.
I have a very clean understanding of whole picture.
May be better to use define_peephole2 with 3 insns as input and 1 as
output for easy understanding.
*shrug* Maybe. Then you also have to check for whether TMP
is
2011/8/4 Richard Henderson r...@redhat.com:
On 08/04/2011 09:04 AM, Denis Chertykov wrote:
Thank you for explanation.
I have a very clean understanding of whole picture.
May be better to use define_peephole2 with 3 insns as input and 1 as
output for easy understanding.
*shrug* Maybe. Then
Richard Henderson wrote:
On 08/04/2011 09:04 AM, Denis Chertykov wrote:
Thank you for explanation.
I have a very clean understanding of whole picture.
May be better to use define_peephole2 with 3 insns as input and 1 as
output for easy understanding.
*shrug* Maybe. Then you also have to
In fixing the string table memory problem, I found that this code
is not really used. Removed.
Tested on x86_64. Committed to branch.
Diego.
* pph-streamer-in.c (pph_get_section_data): Remove unused function.
(pph_free_section_data): Remove unused function.
This patch separates the string tables from input streams to allow the
string tables to remain allocated throughout the compilation process.
Gab, I think this addresses the problem you were seeing with your
line number changes. Let me know if it doesn't.
Tested on x86_64. Committed to branch.
On Wed, Aug 3, 2011 at 14:15, Gabriel Charette gch...@google.com wrote:
We will probably need the strings in here until the end of the
compilation, so we could either just not free it (i.e. memory leak,
but allow important when program is over...)
Or we could strcpy the strings pph'ed in,
On Thu, Aug 04, 2011 at 03:18:46PM +0200, Tobias Burnus wrote:
On 08/04/2011 02:44 PM, Mikael Morin wrote:
and I belief both var%comp and
var are named.
My reading is that a named variable is a variable that is an object-name.
That's actually my problem with the standard - it never quite
Rainer Orth r...@cebitec.uni-bielefeld.de writes:
Another easy part in the toplevel libgcc move was sync.c and related
stuff. While doing this, it turned out to be easier to move the rest of
gcc/config/mips/t-libgcc-mips16 rather than leave it behind.
The patch is untested except for
On 08/04/2011 09:22 AM, Torvald Riegel wrote:
On Thu, 2011-08-04 at 08:43 -0700, Richard Henderson wrote:
On 08/03/2011 04:04 AM, Torvald Riegel wrote:
Move local_tid from gtm_thread to gtm_transaction.
* config/generic/tls.h (gtm_thread): Move local_tid from here
...
On 08/04/2011 11:50 AM, Kai Tietz wrote:
* config/i386/i386.c (setup_incoming_varargs_ms_64): Set
ix86_varargs_gpr_size and ix86_varargs_fpr_size to zero.
Ok.
r~
On Thu, 2011-08-04 at 11:55 -0700, Richard Henderson wrote:
On 08/04/2011 09:22 AM, Torvald Riegel wrote:
On Thu, 2011-08-04 at 08:43 -0700, Richard Henderson wrote:
On 08/03/2011 04:04 AM, Torvald Riegel wrote:
Move local_tid from gtm_thread to gtm_transaction.
*
On 08/04/2011 12:24 PM, Torvald Riegel wrote:
On Thu, 2011-08-04 at 11:55 -0700, Richard Henderson wrote:
On 08/04/2011 09:22 AM, Torvald Riegel wrote:
On Thu, 2011-08-04 at 08:43 -0700, Richard Henderson wrote:
On 08/03/2011 04:04 AM, Torvald Riegel wrote:
Move local_tid from gtm_thread
Hello,
recently we got a bootstrap failure in libfortran's transfer.c:sset
function. This is caused by optimizing alloca into prologue. This
can cause for x64 with SEH to too large prologue-frame-size for SEH
information.
This patch simply assumes that iff the function is using an alloca
call,
On 08/04/2011 01:53 PM, Kai Tietz wrote:
diff = frame-stack_pointer_offset - frame-hard_frame_pointer_offset;
- if (diff 240 || (diff 15) != 0)
+ if (diff 240 || (diff 15) != 0 || cfun-calls_alloca != 0)
Hmm. Why didn't the diff 240 test trigger?
r~
On Wednesday 03 August 2011 22:55:34 Tobias Burnus wrote:
Mikael Morin wrote:
Patch updated.
Looks good! Thanks for the further clean up!
Committed as revision 177415.
I also added the gfortran.dg/ directory in the ChangeLog for you previous
patch (as revision 177416).
Mikael
Index:
On Thursday 04 August 2011 22:45:34 Steve Kargl wrote:
On Tue, Aug 02, 2011 at 05:34:46PM +0200, Tobias Burnus wrote:
There might be still some missing or wrong cleanups, but most remaining
.mod files are from gfortran.fortran-torture.
Yes, you missed a few.
And you missed my two patches
Hi all,
attached is a draft patch fixing the PR in the subject line and
extending the checks for overriding type-bound functions. It regtests
cleanly on x86_64-unknown-linux-gnu already, but I would like to have
some feedback.
The patch is rather large, but most of it is just mechanical, due to
On Thu, Aug 04, 2011 at 11:36:11PM +0200, Mikael Morin wrote:
On Thursday 04 August 2011 22:45:34 Steve Kargl wrote:
On Tue, Aug 02, 2011 at 05:34:46PM +0200, Tobias Burnus wrote:
There might be still some missing or wrong cleanups, but most remaining
.mod files are from
So if I understand correctly, the reference system doesn't work yet, you
just put out the design in this patch and you'll fill up functionality
in an upcoming patch so that's it's clearer what you're adding?
Cheers,
Gab
http://codereview.appspot.com/4847044/diff/1/gcc/cp/pph-streamer-in.c
File
The following patch fixes a bug introduced by a recent LRA patch.
The patch was successfully bootstrapped on x86-64 and ppc64.
2011-08-04 Vladimir Makarov vmaka...@redhat.com
* lra-constraints.c (uses_hard_regs_p): Use hard register
elimination.
Index: lra-constraints.c
On Wed, 20 Apr 2011, Steffen Dettmer wrote:
I'm trying to build gcc-4.6.0 with binutils-2.21 in the same
source directory which fails with
cc1: warnings being treated as errors
../../../gcc-4.6.0/binutils/bfd/compress.c: In function
`bfd_compress_section_contents':
This fixes the docbook html output with respect to images. Now,
filerefs for PNG data are relative to the html install directory, and an
images directory is copied and populated so that the links match PNG
files.
tested x86/linux
-benjamin2011-08-04 Benjamin Kosnik b...@redhat.com
*
IMPORTANT this patch goes on top of trunk patch issue4835047, which has yet to
be approved (I included the change as part of this patch for pph for now).
We now stream out the line_table from the pph'ed header and merge it back in
when compiling reading a pph.
All of the source_location's read
On 3 August 2011 08:48, Jiangning Liu jiangning@arm.com wrote:
This patch is to generate more conditional compare instructions in Thumb2
state. Given an example like below,
int f(int i, int j)
{
if ( (i == '+') || (j == '-') ) {
return i;
} else {
return j;
}
}
Without
This seems to do the trick for sim testing of sh-elf and cris-elf.
I'm interested in advice re debugging experiences with delay slots.
It seems like for calls there's no alternative but to have the unwind
info be incorrect when stopped at the call insn itself.
It also seems like we've similarly
-Original Message-
From: Georg-Johann Lay [mailto:a...@gjlay.de]
Sent: Wednesday, August 03, 2011 4:42 AM
To: Weddington, Eric
Cc: gcc-patches@gcc.gnu.org; Anatoly Sokolov; Denis Chertykov; Joerg
Wunsch
Subject: Re: [Patch,AVR]: Cleanup libgcc.S
It should be fairly easy to
On 29 July 2011 10:58, Dmitry Melnik d...@ispras.ru wrote:
This patch adds two define_insn patterns for NEON vabd instruction to make
combine pass recognize expressions matching (vabs (vsub ...)) patterns as
vabd.
Interesting but I would be a bit defensive and make sure that this
matches only
201 - 284 of 284 matches
Mail list logo
