Hi,
This patch tries to reduce over-promotion of vector operations that
could be done with narrower elements, e.g., for
char a;
int b, c;
short d;
b = (int) a;
c = b << 2;
d = (short) c;
we currently produce six vec_unpack_lo/hi_expr statements for
char->int conversion and then two vec_pack_tru
Hi,
The vectorizer performs the following alias checks for data-refs with
unknown dependence:
((store_ptr_0 + store_segment_length_0) <= load_ptr_0)
|| (load_ptr_0 + load_segment_length_0) <= store_ptr_0))
where segment_length is data-ref's step in the loop multiplied by the
loop's number of i
I've been experimenting with a patch that adds a new define_peephole2
for MIPS. It was blowing up with a segfault in insn-recog.c while
building libstdc++, and looking at the generated code for my new pattern
there was clearly a bad offset being passed to peep2_next_insn. I
tracked it down to
Hi,
This patch adds the missing Pmode check and conversion. OK for trunk?
Thanks.
H.J.
---
2011-07-18 H.J. Lu
* config/i386/i386.c (ix86_legitimize_address): Convert to
Pmode if needed.
(ix86_expand_move): Likewise.
(ix86_expand_call): Likewise.
(ix8
Hi,
Address in x32 is 32bit and x32 doesn't support indirec branch via
memory. This patch properly handles address output and calls patterns
for x32. OK for trunk?
Thanks.
H.J.
---
2011-07-18 H.J. Lu
* config/i386/i386.c (ix86_output_addr_vec_elt): Check
TARGET_LP64 instea
Hi
This patch back ports r176293 from gcc-4_6-branch to google/gcc-4_6 branch.
BB reorder hides the optimization opportunity, so disable it for the test case.
Tested with
make check-gcc RUNTESTFLAGS="--target_board=arm-sim/thumb/arch=armv7-a
arm.exp=pr42093.c"
thanks
Carrot
2011-07-19 Guozhi
On Mon, Jul 18, 2011 at 19:36, Gabriel Charette wrote:
> 2011-07-18 Gabriel Charette
>
> * lib/dg-pph.exp (dg-pph-pos): Output actualSum on unexpected diff.
> Use native TCL functions instead of exec calls to tr and cut.
OK.
Diego.
On Mon, Jul 18, 2011 at 20:54, Gabriel Charette wrote:
> We potentially want to store which flags were on at the pph compile time and
> make sure the same flags
> are on when doing the real compile (otherwise recompile the pph or some other
> soltution...??)
That's up to the build system. Dif
The issue was that we were compiling the pph with some flags and weren't doing
so in the rest of the compilation (pph and non-pph), resulting in different
outputs.
The fact that pph would let this happen is potentially a problem and I opened a
separate thread "Flags and pph..." to discuss this.
LGTM
http://codereview.appspot.com/4752052/
On 07/18/2011 04:14 PM, Richard Henderson wrote:
On 07/18/2011 12:15 PM, Georg-Johann Lay wrote:
However, what you've done is try very hard to work around reload
doing the Right Thing with constant spilling, namely re-generate
the constant rather than spill and restore it. I cannot believe
that
On Thu, 14 Jul 2011, Ulrich Weigand wrote:
> Jeff Law wrote:
> > On 06/15/11 21:46, Hans-Peter Nilsson wrote:
> > > PR rtl-optimization/48542
> > > * reload.c (find_equiv_reg): Stop looking when finding a
> > > setjmp-type call.
> > > * reload1.c (reload_as_needed): Invalidate all reload
>
On Mon, Jul 18, 2011 at 3:05 PM, Andreas Schwab wrote:
> Gabriel Charette writes:
>
>> On Sat, Jul 16, 2011 at 12:59 AM, Andreas Schwab
>> wrote:
>>>
>>> gch...@google.com (Gabriel Charette) writes:
>>>
>>> > + set expectedSum [exec tr -d \} << [exec cut -f 4 -d\ <<
>>> > $xdiff_entry
Output the diff's sum for unexpected diffs as well (used to only output it when
it had changed, but outputting it for new diffs will make it easier to generate
the comment required to mark this diff as expected when needed).
Also switched from using exec calls to tr/cut, now using native TCL fun
On 07/14/2011 07:16 AM, Andrew Stubbs wrote:
> { dg-options "-O2 -march=armv7-a" }
The tests use "{ dg-options "-O2 -march=armv7-a" }" but -march will be
overridden for multilibs that specify -march, and might conflict with
other multilib options. If you really need that particular -march value
t
On Mon, Jul 18, 2011 at 1:42 PM, H.J. Lu wrote:
> On Mon, Jul 18, 2011 at 1:33 PM, Uros Bizjak wrote:
>> On Mon, Jul 18, 2011 at 10:25 PM, H.J. Lu wrote:
>>
>>> TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
>>> in x32 mode. OK for trunk?
>>
>> Do you p
On Tue, 2011-07-19 at 01:19 +0200, Torvald Riegel wrote:
> On Sat, 2011-07-09 at 16:30 +0200, Torvald Riegel wrote:
> > The attached patch makes flat nesting the default, while still
> > supporting closed nesting on demand (for user-controlled aborts via
> > __transaction_cancel).
> > Previously, a
On Sat, 2011-07-09 at 16:30 +0200, Torvald Riegel wrote:
> The attached patch makes flat nesting the default, while still
> supporting closed nesting on demand (for user-controlled aborts via
> __transaction_cancel).
> Previously, a new transaction object was created for each nested
> transaction.
On 07/19/2011 12:33 AM, Sebastian Pop wrote:
On Mon, Jul 18, 2011 at 16:40, Tobias Grosser wrote:
You are right. In case we want to downcast b we would also need to prove
that b itself fits into the smaller type. (This may happen, if the larger
type was introduced earlier to remove casts, but i
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 07/14/11 10:16, Ulrich Weigand wrote:
> Jeff Law wrote:
>> On 06/15/11 21:46, Hans-Peter Nilsson wrote:
>>> PR rtl-optimization/48542
>>> * reload.c (find_equiv_reg): Stop looking when finding a
>>> setjmp-type call.
>>> * reload1.c
On Mon, Jul 18, 2011 at 16:40, Tobias Grosser wrote:
> You are right. In case we want to downcast b we would also need to prove
> that b itself fits into the smaller type. (This may happen, if the larger
> type was introduced earlier to remove casts, but is not needed to store the
> result).
[...
Gabriel Charette writes:
> On Sat, Jul 16, 2011 at 12:59 AM, Andreas Schwab
> wrote:
>>
>> gch...@google.com (Gabriel Charette) writes:
>>
>> > + set expectedSum [exec tr -d \} << [exec cut -f 4 -d\ <<
>> > $xdiff_entry]]
>
> $xdiffentry is set to something like {{pph asm diff 123456}
Hi!
Especially the FEs and gimplification are highly recursive on more complex
(especially badly generated) testcases, the following patch attempts to
increase stack size to 64MB if possible. It is done in the driver
(where it can affect the children of the driver early) and also in
toplev_main t
On Sat, Jul 16, 2011 at 12:59 AM, Andreas Schwab wrote:
>
> gch...@google.com (Gabriel Charette) writes:
>
> > + set expectedSum [exec tr -d \} << [exec cut -f 4 -d\ <<
> > $xdiff_entry]]
$xdiffentry is set to something like {{pph asm diff 123456}}, so i
need tr and cut to extract 12345
On 07/18/2011 07:07 PM, Sebastian Pop wrote:
On Mon, Jul 18, 2011 at 11:11, Tobias Grosser wrote:
We introduce casts when we generate code for an expression with a type
smaller than the types of its constituents.
So the next question is: What is the type of the expression. As far as I
underst
The create_basic_block function will do the right thing
with ordering the block label and the block note. No
need for us to work hard at that within bb-reorder.
r~
* bb-reorder.c (fix_crossing_conditional_branches): Emit all insns
before calling create_basic_block.
diff --git
On Mon, Jul 18, 2011 at 1:33 PM, Uros Bizjak wrote:
> On Mon, Jul 18, 2011 at 10:25 PM, H.J. Lu wrote:
>
>> TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
>> in x32 mode. OK for trunk?
>
> Do you perhaps have a testcase to help in analyzing the problem?
On 07/16/2011 10:37 AM, Dodji Seketeli wrote:
+ /* This array of location is actually an array of pairs of
+ locations. The elements inside it thus look like:
+
+ x0,y0, x1,y1, x2,y2, , xn,yn.
Pairs of locations still seems limited, given the flexibility of macro
expansion;
On Mon, Jul 18, 2011 at 10:25 PM, H.J. Lu wrote:
> TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
> in x32 mode. OK for trunk?
Do you perhaps have a testcase to help in analyzing the problem?
>>>
>>> See:
>>>
>>> http://gcc.gnu.org/bugzilla/show_b
On Mon, Jul 18, 2011 at 1:05 PM, Uros Bizjak wrote:
> On Mon, Jul 18, 2011 at 8:48 PM, H.J. Lu wrote:
>
TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
in x32 mode. OK for trunk?
>>>
>>> Do you perhaps have a testcase to help in analyzing the problem?
>>>
>>
>>
Tobias Burnus wrote:
I would be happy if someone could carefully read the function in
expr.c - and confirm that the second test case ("dg-error" line) is
indeed invalid. I think it is invalid because one has an array element
of a coarray, which is a coarray but which is not simply contiguous.
On 07/18/2011 12:15 PM, Georg-Johann Lay wrote:
>> However, what you've done is try very hard to work around reload
>> doing the Right Thing with constant spilling, namely re-generate
>> the constant rather than spill and restore it. I cannot believe
>> that's the right way to proceed.
>
> You me
On Mon, Jul 18, 2011 at 8:48 PM, H.J. Lu wrote:
>>> TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
>>> in x32 mode. OK for trunk?
>>
>> Do you perhaps have a testcase to help in analyzing the problem?
>>
>
> See:
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49780
I
On 07/18/2011 12:09 PM, Steven Bosscher wrote:
> On Mon, Jul 18, 2011 at 8:48 PM, Richard Henderson wrote:
>> Split out a helper subroutine to insert a barrier,
>> after we've modified a basic block to not fallthru.
>
> Huh, aren't barriers emitted automatically when going out of cfglayout mode??
On 07/18/2011 12:09 PM, Steven Bosscher wrote:
> On Mon, Jul 18, 2011 at 8:48 PM, Richard Henderson wrote:
>> Split out a helper subroutine to insert a barrier,
>> after we've modified a basic block to not fallthru.
>
> Huh, aren't barriers emitted automatically when going out of cfglayout mode??
For the Fortran standard, quite a lot of objects are coarrays.
Essentially every object which as an attr.codimension - including array
sections, elements or components of it - unless the components are
allocatable or pointer. Note that this excludes coindexed objects. See
F2008, "2.4.7 Coarray"
Richard Henderson schrieb:
On 07/18/2011 11:05 AM, Georg-Johann Lay wrote:
What's bad with pre-reload splits?
The only weak point is in target-independent code because there
is nothing like split1_completed and other missing information
for better pass-awareness.
Nothing's wrong with pre-relo
On Mon, Jul 18, 2011 at 8:48 PM, Richard Henderson wrote:
> Split out a helper subroutine to insert a barrier,
> after we've modified a basic block to not fallthru.
Huh, aren't barriers emitted automatically when going out of cfglayout mode??
Ciao!
Steven
Split out a helper subroutine to insert a barrier,
after we've modified a basic block to not fallthru.
Tested on x86_64-linux and committed.
r~
* bb-reorder.c (emit_barrier_after_bb): Split out of ...
(add_labels_and_missing_jumps): ... here.
(fix_up_fall_thru_edges, fix_
On Mon, Jul 18, 2011 at 11:44 AM, Uros Bizjak wrote:
> On Mon, Jul 18, 2011 at 8:39 PM, H.J. Lu wrote:
>
>> TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
>> in x32 mode. OK for trunk?
>
> Do you perhaps have a testcase to help in analyzing the problem?
>
See:
http://
On Mon, Jul 18, 2011 at 8:39 PM, H.J. Lu wrote:
> TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
> in x32 mode. OK for trunk?
Do you perhaps have a testcase to help in analyzing the problem?
Uros.
Hi,
TARGET_MEM_REF only works on ptr_mode. This patch allows 32bit address
in x32 mode. OK for trunk?
Thanks.
H.J.
---
2011-07-18 H.J. Lu
* config/i386/i386.c (ix86_legitimate_address_p): Support 32bit
address in x32 mode.
diff --git a/gcc/config/i386/i386.c b/gcc/config
On 07/18/2011 11:05 AM, Georg-Johann Lay wrote:
> What's bad with pre-reload splits?
> The only weak point is in target-independent code because there
> is nothing like split1_completed and other missing information
> for better pass-awareness.
Nothing's wrong with pre-reload splits.
However, wha
Richard Henderson wrote:
> On 07/18/2011 08:41 AM, Georg-Johann Lay wrote:
>> +(define_insn_and_split "*muluqihi3.uconst"
>> + [(set (match_operand:HI 0 "register_operand" "=r")
>> +(mult:HI (zero_extend:HI (match_operand:QI 1 "register_operand"
>> "r"))
>> +
I have started working on supporting external symbol references from
one PPH file into another. I don't yet have anything working, but I
will definitely not be using the decl referencing used by LTO, so this
patch cleans up the interfaces not to use it.
No functional changes since we were not us
Hello!
This alternative patch fixes the problem in ix86_decompose_address,
uncovered by x32 branch. Since x32 branch generates lots of SImode
subreg of DImode values to handle Pmode vs. ptr_mode restrictions, a
latent bug in x86_decompose_address allowed addresses in the (invalid,
see SImode subre
On 18/07/11 14:10, Chung-Lin Tang wrote:
> On 2011/7/18 04:46 PM, Richard Earnshaw wrote:
>> The patch to arm.c is ok, but the change to the test is not as it will cause
>> problems with multilib testing. A better fix is to skip the test if the
>> target is thumb1.
>>
>> The other test needs a s
Hi!
As the testcase below shows, fold_nonarray_ctor_reference doesn't have a
correct check whether a field might overlap with the given offset, size
access (in this case a BIT_FIELD_REF). The bitfield ref wants to read
8 bits starting at offset 0, the bitfield defined there goes from bit 1
1 bit
On 07/18/2011 10:44 AM, Jakub Jelinek wrote:
> PR middle-end/49675
> * tree.c (build_common_builtin_nodes): Register
> __builtin_return_address, __cyg_profile_func_enter
> and __cyg_profile_func_exit.
>
> * gfortran.dg/pr49675.f90: New test.
Ok.
r~
Hi,
I am checking the following patches to revert 32bit address changes in
x32 mode.
H.J.
commit 2087dd1362a4c2095a03ea92a06a9b67ab516a02
Author: H.J. Lu
Date: Sun Jul 17 16:16:16 2011 -0700
Remove ix86_simplify_base_index_disp.
diff --git a/gcc/ChangeLog.addr32 b/gcc/ChangeLog.addr32
n
On 07/18/2011 08:41 AM, Georg-Johann Lay wrote:
> +(define_insn_and_split "*muluqihi3.uconst"
> + [(set (match_operand:HI 0 "register_operand" "=r")
> +(mult:HI (zero_extend:HI (match_operand:QI 1 "register_operand" "r"))
> + (match_operand:HI 2 "u8_
Hi!
-finstrument-functions with any function written in Fortran, Go (and
presumably also Java and Ada, haven't tried) fails because
non-C/C++ FEs don't register __builtin_return_address (),
__cyg_profile_func_enter () and __cyg_profile_func_exit () builtins.
This worked until 4.5 because gimplific
On Mon, 2011-07-18 at 13:31 +0200, Rainer Orth wrote:
> Steve,
> I had a fresh look and it seems this isn't as bad as it might have been:
> the only SHLIB_ macros used are SHLIB_(DIR|OBJS|SOLINK|SLIBDIR_QUAL),
> which are all present in libgcc/config/t-slibgcc and are substituted by
> configure.
On Mon, Jul 18, 2011 at 11:11, Tobias Grosser wrote:
>> We introduce casts when we generate code for an expression with a type
>> smaller than the types of its constituents.
>
> So the next question is: What is the type of the expression. As far as I
> understand we decide ourselves which type the
On 07/18/2011 11:27 AM, Bernd Schmidt wrote:
On 07/18/11 17:09, Vladimir Makarov wrote:
On 07/11/2011 05:09 PM, Bernd Schmidt wrote:
I came up with the notion of adding a new transition to the NDFA, one
which collapses a nondeterministic state (which is composed of multiple
possible determinist
A small tweak to avoid inserting the same decl more than once in the
symbol table for a pph image.
Tested on x86_64. Committed to branch.
Diego.
* pph-streamer-in.c (pph_in_struct_function): Revert previous
change. Do not allow shared struct functions.
* pph-streamer-
On 07/15/2011 06:56 AM, Rainer Orth wrote:
> 2011-07-14 Rainer Orth
>
> gcc:
> * config.gcc (alpha*-*-linux*): Add crtfastmath.o to extra_parts.
> (alpha*-*-freebsd*): Likewise.
> (i[34567]86-*-linux*, i[34567]86-*-kfreebsd*-gnu,
> i[34567]86-*-knetbsd*-gnu, i[3456
On Mon, 2011-07-18 at 13:16 +0200, Rainer Orth wrote:
> Unfortunately, nobody reviewed the patch. There's now also PR
> bootstrap/49769 for Linux/SPARC.
> Rainer
I tested the patch on IA64 Linux and verified that it fixed that
bootstrap. I can approve the IA64 part of the patch. Perhap
On 9 July 2011 14:46, Jonathan Wakely wrote:
> This adds the new header. Currently only
> std::vector has the necessary support for C++0x allocators to be
> usable with std::scoped_allocator_adaptor.
>
> * include/Makefile.am: Add new header.
> * include/Makefile.in: Regenerate.
>
On 07/18/2011 05:03 PM, Sebastian Pop wrote:
On Sun, Jul 17, 2011 at 19:41, Tobias Grosser wrote:
Accessing a[100] is undefined in C, so i<100, and so n<= 100.
Sure. I am just surprised we already include this information in graphite.
Do we add this to the context?
Yes, that's added to
Hi,
committed to mainline.
Paolo.
//
2011-07-18 Paolo Carlini
PR libstdc++/48430
* include/bits/shared_ptr.h (struct hash>): Use
__hash_base.
* include/bits/unique_ptr.h (struct hash>): Likewise.
Index: include/bits/shared_ptr.h
==
This patch improves 16=8*8 and 16*16*8 multiplication.
mulhi3 is extended to care for small, signed 9-bit constants
that can be handled more efficiently than loading the constants
and performing a 16-bit multiplication.
Some combine patterns are added to match if one operand is a small
constant w
On 07/18/11 17:09, Vladimir Makarov wrote:
> On 07/11/2011 05:09 PM, Bernd Schmidt wrote:
>> I came up with the notion of adding a new transition to the NDFA, one
>> which collapses a nondeterministic state (which is composed of multiple
>> possible deterministic ones) to just one of its component
On 07/11/2011 05:09 PM, Bernd Schmidt wrote:
On C6X, we want to use the "ndfa" option to give the scheduler maximum
freedom when assigning units to instructions. After scheduling is
complete, we process the insns again in c6x_reorg, looking at each cycle
and assigning a unit specifier to each ins
On Sun, Jul 17, 2011 at 19:41, Tobias Grosser wrote:
>> Accessing a[100] is undefined in C, so i< 100, and so n<= 100.
>
> Sure. I am just surprised we already include this information in graphite.
> Do we add this to the context?
Yes, that's added to the context as a bound on the loop iteration
On Mon, Jul 18, 2011 at 3:14 PM, Andrew Stubbs wrote:
> As far as I can tell, the patch series so far works great as long as the
> input type of the accumulate value is the same as the output type.
> Unfortunately you get an ICE otherwise doh!
>
> This patch should fix the problem.
>
> I coul
> "Jakub" == Jakub Jelinek writes:
Tom> I don't think I really understood DW_MACINFO_GNU_define_opcode, so the
Tom> implementation here is probably wrong.
Jakub> Well, I think you've skipped it correctly and furthermore even patched
Jakub> GCC doesn't emit it. The point of it was to allow s
On Mon, Jul 18, 2011 at 04:29:29PM +0200, Romain Geissler wrote:
> Hi,
>
> This patch allows a third party plugin to define new GTY-ed types
> by compiling and installing gengtype for host and installing gtype.state
> in plugin_includedir.
>
> Builds and installs fine on a x86_64 red hat.
I am
> "Jakub" == Jakub Jelinek writes:
Jakub> Ok, so how about this way (as DWARF4 modifications, of course for
Jakub> DWARF5 proposal GNU_ would be gone and the ops would have different
Jakub> codes):
Thanks very much for writing it up this way. I think it is very
important that all our DWARF
The 4.5 branch lacked the fix from
[build] Link libgcjgc with -lrt on Solaris > 8
http://gcc.gnu.org/ml/gcc-patches/2011-01/msg02288.html
Since this issue has now been reported as PR boehm-gc/49381, I've
backported the fix to the 4.5 branch after reconfiguring and
successfully bui
This rewrites TRUTH_NOT_EXPRs during gimplification
as BIT_NOT_EXPR respective BIT_XOR_EXPR as RTL expansion does. This
completes the series lowering the TRUTH_*_EXPRs to operations that
are also available on HW and allows us to unify code handling
TRUTH_*_EXPRs and BIT_*_EXPRs.
The patch has se
Hi,
This patch allows a third party plugin to define new GTY-ed types
by compiling and installing gengtype for host and installing gtype.state
in plugin_includedir.
Builds and installs fine on a x86_64 red hat.
Ok for the trunk ?
Romain Geissler
gcc/
2011-07-18 Romain Geissler
* g
Recent boolifications have introduced exprs with saved_location,
that's bogus.
Committed as obvious.
Richard.
2011-07-18 Richard Guenther
* gimplify.c (gimplify_expr): Use input_location, not saved_location
when building new trees.
Index: gcc/gimplify.c
On Thursday 14 July 2011 11:01:29 Tobias Burnus wrote:
> As the title says: Mark "token" of static coarrays "restrict"
>
> Bootstrapped and regtested on x86-64-linux.
> OK for the trunk?
>
> Tobias
OK. (and obvious)
Mikael
*ping*
http://gcc.gnu.org/ml/fortran/2011-07/msg00142.html
On 07/14/2011 11:01 AM, Tobias Burnus wrote:
As the title says: Mark "token" of static coarrays "restrict"
Bootstrapped and regtested on x86-64-linux.
OK for the trunk?
Tobias
As far as I can tell, the patch series so far works great as long as the
input type of the accumulate value is the same as the output type.
Unfortunately you get an ICE otherwise doh!
This patch should fix the problem.
I could have inserted this fix into the correct spot in the existing
On 2011/7/18 04:46 PM, Richard Earnshaw wrote:
> The patch to arm.c is ok, but the change to the test is not as it will cause
> problems with multilib testing. A better fix is to skip the test if the
> target is thumb1.
>
> The other test needs a similar check as it seems to expect a movs instr
The following two libgcc patches have seen almost no comments, and
certainly neither testing or review in a week:
CFT: [build] Move fp-bit support to toplevel libgcc
http://gcc.gnu.org/ml/gcc-patches/2011-07/msg00927.html
CFT: [build] Move soft-fp support to toplevel libgc
On 18/07/11 12:09, Tom de Vries wrote:
> Hi,
>
> PR43597 was fixed by http://gcc.gnu.org/viewcvs?view=revision&revision=172032.
>
> This patch adds a testcase.
>
> OK for trunk?
>
> Thanks,
> - Tom
>
> 2011-07-18 Tom de Vries
>
> PR target/43597
> * gcc.target/arm/pr43597.c: New te
Jakub Jelinek writes:
> On Mon, Jul 18, 2011 at 01:53:40PM +0200, Rainer Orth wrote:
>> As described in the PR, the fact that the libgomp testsuite always links
>> with -lgomp, even with -fno-openmp, causes testsuite failures on targets
>> (like Solaris) that need special options when linking lib
on x86_64-unknown-linux-gnu.
Richard.
2011-07-18 Richard Guenther
* expr.c (expand_expr_real_2): Properly truncate the BIT_NOT_EXPR
expansion result to bitfield precision if required.
* gcc.dg/torture/20110718-1.c: New testcase.
Index: gcc/expr.c
On 05/27/11 14:18, Jason Merrill wrote:
On 05/26/2011 02:37 PM, Aldy Hernandez wrote:
The narrowest integer mode containing the bit field is still 32, so we
access the bitfield with an SI instruction as expected.
OK, then:
struct A
{
int i: 4;
int j: 4;
int k: 8;
int l: 8;
int m: 8;
};
now t
On Mon, Jul 18, 2011 at 01:53:40PM +0200, Rainer Orth wrote:
> As described in the PR, the fact that the libgomp testsuite always links
> with -lgomp, even with -fno-openmp, causes testsuite failures on targets
> (like Solaris) that need special options when linking libgomp. With
> -fgomp, this is
As described in the PR, the fact that the libgomp testsuite always links
with -lgomp, even with -fno-openmp, causes testsuite failures on targets
(like Solaris) that need special options when linking libgomp. With
-fgomp, this is taken care of by the relevant specs (like adding
-pthread), but not
Steve,
>> It looks like a prefix is missing somewhere since it is trying to access
>> /libunwind.so. This
>> may be something messed up in my build area again but I did run autoconf in
>> libgcc so I am
>> not sure what is going on. I'll dig around some more but I thought I would
>> see if th
On Mon, Jul 18, 2011 at 12:51:34AM +0200, Gerald Pfeifer wrote:
> I am happy to announce that the steering committee is appointing our
> three release managers Jakub Jelinek, Joseph S. Myers, and Richard
> Guenther to the additional roles of global reviewers.
Thanks.
> PS: I assume you guys know
Jakub Jelinek writes:
>> There's a far simpler fix, inspired by gcc.dg/dfp/pr35739.c. I think it
>> is better since it gives the testcase wider coverage. Tested with the
>> appropriate runtest invocation on i386-pc-solaris2.11 and
>> alpha-dec-osf5.1b.
>
> libmudflap testsuite also iterates ove
Rainer Orth writes:
> I think the following patch should fix all broken targets. For those
> that previously had crtfastmath.o in EXTRA_PARTS, I'm adding it to
> config.gcc (extra_parts) explicitly.
>
> The non-Linux x86 glibc targets were missed before. I move the
> extra_parts addition down s
On Mon, Jul 18, 2011 at 12:22:01PM +0200, Rainer Orth wrote:
> >> The problem here is that the type of the POINTER_PLUS_EXPR is
> >> incorrect and also the non folded version leaks to the IR. This patch
> >> fixes those two problems and fixes the ICE.
> >>
> >> OK? Bootstrapped and tested on x8
Hi,
PR43597 was fixed by http://gcc.gnu.org/viewcvs?view=revision&revision=172032.
This patch adds a testcase.
OK for trunk?
Thanks,
- Tom
2011-07-18 Tom de Vries
PR target/43597
* gcc.target/arm/pr43597.c: New test.
Index: gcc.target/arm/pr43597.c
==
Here is a summary - probably, it doesn't cover every single piece in
the patch, but I tried to describe the major changes. I hope this will
help you a bit - and of course I'll answer your further questions if
they appear.
The changes could be logically divided into two parts (though, these
parts h
Ira Rosen writes:
>> This test fails on spu-elf, and presumably all other targets that
>> don't have mmap (or ) ...
>>
>> Can the test be restricted to those targets that actually support
>> that OS feature?
>
> I copied the list of targets from gcc.dg/20050826-1.c.
Sorry, but this is a bad idea
Jakub Jelinek writes:
> On Wed, Jul 13, 2011 at 01:37:22PM -0700, Andrew Pinski wrote:
>> Hi,
>> The problem here is that the type of the POINTER_PLUS_EXPR is
>> incorrect and also the non folded version leaks to the IR. This patch
>> fixes those two problems and fixes the ICE.
>>
>> OK? Boot
On Sun, Jul 17, 2011 at 4:34 PM, Richard Sandiford
wrote:
> This patch replaces the rtxes in mem_attrs with (bool, HOST_WIDE_INT) pairs.
> There's already room for the bools in the padding after "addrspace".
> (There's room for one more bool after this patch; we could move to
> bitfields if more a
On Sun, Jul 17, 2011 at 4:32 PM, Richard Sandiford
wrote:
> This patch makes the mechanical MEM_OFFSET interface change,
> along the same lines as the MEM_SIZE one.
Ok if there are no objections from target maintainers.
Thanks,
Richard.
> Richard
>
>
> gcc/
> * doc/rtl.texi (MEM_OFFSET_K
On Sun, Jul 17, 2011 at 4:31 PM, Richard Sandiford
wrote:
> This patch does the mechanical MEM_SIZE interface change. It also
> makes set_mem_size take a HOST_WIDE_INT, and adds clear_mem_size
> for resetting the size to "unknown".
>
> The i386.c part includes an obvious pasto fix for:
>
> if (d
On Sun, Jul 17, 2011 at 4:29 PM, Richard Sandiford
wrote:
> Internally, emit-rtl.c uses the idiom:
>
> MEM_ATTRS (mem) = get_mem_attrs (MEM_ALIAS_SET (mem), MEM_EXPR (mem),
> MEM_OFFSET (mem), size, MEM_ALIGN (mem),
> MEM_ADDR_SP
On Sun, Jul 17, 2011 at 4:18 PM, Richard Sandiford
wrote:
> This patch adds an array, mode_mem_attrs, that gives the default attributes
> for a given mode. The defaults depend on STRICT_ALIGNMENT, which on rs6000
> and v850 depends on command-line flags, and which could therefore in
> principle b
On Sun, Jul 17, 2011 at 7:57 AM, Ira Rosen wrote:
> On 15 July 2011 15:56, Ulrich Weigand wrote:
>> Ira Rosen wrote:
>>
>>> * gcc.dg/vect/pr49038.c: New test.
>>
>>> Index: testsuite/gcc.dg/vect/pr49038.c
>>> ===
>>> --- testsu
On 18 Jul 2011, at 07:15, "Chung-Lin Tang" wrote:
> On 2011/6/15 09:12 PM, Richard Earnshaw wrote:
>> On 22/04/11 16:21, Chung-Lin Tang wrote:
>>> Hi Richard, this part's for you.
>>>
>>> The ARM backend changes needed are very little after the prior patches,
>>> basically just a case in arm_can
1 - 100 of 104 matches
Mail list logo
