On 10/18/13 07:41, Andrew MacLeod wrote:
Also straightforward. includes sbitmap in the 5 files that need it.
I also happened to notice that tree-switch-conversion.c was including
tree-ssa-operands.h directly, and doesnt need it, so removed it too
bootstraps on x86_64-unknown-linux-gnu with no ne
Sandra Loosemore wrote:
>On 10/18/2013 04:50 AM, Richard Biener wrote:
>> On Sat, Sep 28, 2013 at 4:19 AM, Sandra Loosemore
>> wrote:
>>> This patch fixes various -fstrict-volatile-bitfields wrong-code
>bugs,
>>> including instances where bitfield values were being quietly
>truncated as
>>> well
On 10/17/2013 05:50 PM, Joseph S. Myers wrote:
I've applied this patch to the C11-atomic branch to fix various
miscellaneous issues.
excellent!
I put a hack in the C++ front end to disable checks on atomic
qualifiers so that libstdc++ builds - I think that if you want other
people to help on
On Oct 18, 2013, at 4:24 AM, Richard Biener wrote:
> I agree. Btw, the "implementation defined" precision of PDI on SH-5
> definitely
> looks interesting, but I wonder how you can perform "implementation defined"
> arithmetic on that PDI mode then ;)
Easy, perform it in the maximal size support
Back in 2011 I wrote code to detect cases when traversing a particular
path could be proven to trigger undefined behaviour (such as a null
pointer dereference). That original patch would find the control
dependent edges leading to the dereference and eliminate those edges.
The result being
On Fri, Oct 18, 2013 at 3:03 AM, Dominique Dhumieres wrote:
> Sriraman,
>
> The tests gcc.target/i386/funcspec-5.c and gcc.target/i386/pr57756.c fail
> on targets for which -msse is the default (see
> http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01365.html or
> http://gcc.gnu.org/ml/gcc-testre
On Fri, 18 Oct 2013, Andrew MacLeod wrote:
> right. We may re-visit it when we finalize the ABI changes for c++... Use of
> the attribute in c++ will ensure that C and C++ both treat an atomic object
> the same... My original discussions with lawrence and jeffrey over atomics
> concerned the de
On Oct 18, 2013, at 6:36 AM, David Malcolm wrote:
> Thanks everyone; I'm finding this very helpful.
Thanks for the feedback and making enough noise such that we fixed the problem.
:-)
On Fri, Oct 18, 2013 at 10:27 AM, Sriraman Tallam wrote:
> On Fri, Oct 18, 2013 at 3:03 AM, Dominique Dhumieres
> wrote:
>> Sriraman,
>>
>> The tests gcc.target/i386/funcspec-5.c and gcc.target/i386/pr57756.c fail
>> on targets for which -msse is the default (see
>> http://gcc.gnu.org/ml/gcc-tes
OK.
Jason
On 10/11/2013 01:59 PM, Dehao Chen wrote:
It's hard to get a testcase without
http://gcc.gnu.org/viewcvs/gcc?view=revision&revision=201856 because
none of these *INTERNAL* symbols will be emitted in debug info.
Why does that change cause one of these symbols to be emitted? As Cary
says, that
On 10/18/2013 01:27 PM, Joseph S. Myers wrote:
On Fri, 18 Oct 2013, Andrew MacLeod wrote:
right. We may re-visit it when we finalize the ABI changes for c++... Use of
the attribute in c++ will ensure that C and C++ both treat an atomic object
the same... My original discussions with lawrence
On Fri, Oct 18, 2013 at 10:39 AM, Jason Merrill wrote:
> On 10/11/2013 01:59 PM, Dehao Chen wrote:
>>
>> It's hard to get a testcase without
>> http://gcc.gnu.org/viewcvs/gcc?view=revision&revision=201856 because
>> none of these *INTERNAL* symbols will be emitted in debug info.
>
>
> Why does tha
> We use the container mode where possible.
> It is always possible for well-formed bit-fields.
This is the only part I really need.
> If necessary the user has to add anonymous bit field members, or
> convert normal members to bit-fields.
IIRC I added code to support normal members as well, th
> What I would suggest is to have a -fgnu-strict-volatile-bit-fields
Why a new option? The -fstrict-volatile-bitfields option is already
GCC-specific, and I think it can do what you want anyway.
On 10/18/2013 11:59 AM, Jeff Law wrote:
On 10/18/13 07:39, Andrew MacLeod wrote:
degenerate_phi_result was defined in tree-ssa-dom.c, I moved it to
tree-phinodes since all it does is determine whether the arguements of
the phi which are not the same as the result are all the same. This
reduced b
On Fri, 18 Oct 2013, Jeff Law wrote:
Back in 2011 I wrote code to detect cases when traversing a particular path
could be proven to trigger undefined behaviour (such as a null pointer
dereference). That original patch would find the control dependent edges
leading to the dereference and elimi
Hi,
this 4.9 Regression is just and ICE on invalid, but I think it points to
a (minor) matter or principle: the body of a constexpr constructor may
have, possibly nested, BIND_EXPRs, but in any case there should be no
BIND_EXPR_VARS (for this specific testcase the VAR_DECL has the
TREE_TYPE =
On 10/18/2013 12:10 PM, Jeff Law wrote:
On 10/18/13 07:37, Andrew MacLeod wrote:
gimple_check_call_matching_types() was being called from 3 or 4
different files,and seemed more appropriate as a cgraph routine (which
called it 3 times). So I moved that and its helper to cgraph.c.
After that, I
> I see why pr57756.c could fail, if -msse4.2 is turned on by default. I
> think this test needs {dg-options "-mno-sse4.2"}.
This change allows the test to pass. The failure of
gcc.target/i386/funcspec-5.c is
/opt/gcc/work/gcc/testsuite/gcc.target/i386/funcspec-5.c:34:1: warning: SSE
instructio
On 10/18/13 12:47, Marc Glisse wrote:
Maybe a new -fretroactive-undefined-behavior? (for later, obviously)
Something like that -- haven't thought much about the name.
* should cfg_altered be static (or a member of the pass class)?
At the minimum it should be static. Into the pass class woul
On 10/18/2013 12:06 PM, Jeff Law wrote:
On 10/18/13 07:45, Andrew MacLeod wrote:
bootstraps on x86_64-unknown-linux-gnu with no new regressions. OK?
OK.
jeff
All 8 patches were committed at once, revision 203833.
I'll submit another patch if any recommended followups surface.
Andrew
Jeff Law writes:
> On 09/28/13 09:30, Mikael Pettersson wrote:
> > This patch fixes PR58369, an ICE in subreg_get_info when compiling
> > boost for m68k-linux.
> >
> > choose_reload_regs attempts to reload a DFmode (8-byte) reg, finds
> > an XFmode (12-byte) reg in "last_reg", and calls subre
On Fri, 18 Oct 2013, Jeff Law wrote:
On 10/18/13 12:47, Marc Glisse wrote:
* tree-vrp has a function infer_nonnull_range, do you think we could
share it? We now store the VRP ranges for integers, but not for
pointers. If we did (or maybe just a non-null bit), the pass could just
test that bit o
Here is the patch updated to use the new parameter from r203830.
Passed bootstrap and regression tests.
2013-10-18 Jan Hubicka
Teresa Johnson
* predict.c (handle_missing_profiles): New function.
(counts_to_freqs): Don't overwrite estimated frequencies
when
On Fri, 18 Oct 2013, Andrew MacLeod wrote:
> I had thought about making the type and value of atomic_flag hooks , but never
> got to it. There is currently one for the TRUE value which is exposed to the
> C++ templates as" __GCC_ATOMIC_TEST_AND_SET_TRUEVAL"
I'm tending to think of this, and alig
On Oct 18, 2013, at 6:11 AM, Kenneth Zadeck wrote:
>>> Does this look ok? Kenny, can you double check the cases, think I have
>>> them right, but? a double check would be good.
>> That works for me.
>>
> i talked to mike about this last night, but did not follow up with an email
> until now.
On Fri, Oct 18, 2013 at 01:52:54PM -0700, Mike Stump wrote:
> On Oct 18, 2013, at 6:11 AM, Kenneth Zadeck wrote:
> >>> Does this look ok? Kenny, can you double check the cases, think I have
> >>> them right, but? a double check would be good.
> >> That works for me.
> >>
> > i talked to mike a
On 10/18/13 14:31, Marc Glisse wrote:
But once you have that pointer defined by a PHI containing a zero, you
look at all its uses, trying to find one that proves the pointer is
non-zero
What are you going to do with that information? The only use I can see
for this pass would be discovering mo
Hi Jeff,
Please see my comments below. Also, I am adding all these changes to
the files as you requested in my local directory. Should I send you an updated
patch along the way?
Thanks,
Balaji V. Iyer.
> -Original Message-
> From: Jeff Law [mailto:l...@redhat.com]
> Sent: Thurs
On Fri, 18 Oct 2013, Jeff Law wrote:
On 10/18/13 14:31, Marc Glisse wrote:
But once you have that pointer defined by a PHI containing a zero, you
look at all its uses, trying to find one that proves the pointer is
non-zero
What are you going to do with that information?
Uh? I must have been
> Here is the patch updated to use the new parameter from r203830.
> Passed bootstrap and regression tests.
>
> 2013-10-18 Jan Hubicka
> Teresa Johnson
>
> * predict.c (handle_missing_profiles): New function.
> (counts_to_freqs): Don't overwrite estimated frequenc
On 10/18/2013 04:40 PM, Joseph S. Myers wrote:
On Fri, 18 Oct 2013, Andrew MacLeod wrote:
I had thought about making the type and value of atomic_flag hooks , but never
got to it. There is currently one for the TRUE value which is exposed to the
C++ templates as" __GCC_ATOMIC_TEST_AND_SET_TRUE
On 10/18/13 15:15, Marc Glisse wrote:
On Fri, 18 Oct 2013, Jeff Law wrote:
On 10/18/13 14:31, Marc Glisse wrote:
But once you have that pointer defined by a PHI containing a zero, you
look at all its uses, trying to find one that proves the pointer is
non-zero
What are you going to do with t
DJ,
Is the below patch your purview? If not, do you know whose it is?
Sterling
-- Forwarded message --
From: Sterling Augustine
Date: Thu, Oct 17, 2013 at 11:52 AM
Subject: [PATCH] Add gdb/gdb-index.h to gcc tree.
To: gcc-patches , d...@redhat.com
The enclosed patch to "merge
On Thu, Oct 17, 2013 at 7:47 PM, Sriraman Tallam wrote:
> On Thu, Oct 17, 2013 at 4:51 PM, Sriraman Tallam wrote:
>> On Thu, Oct 17, 2013 at 3:08 PM, Sriraman Tallam wrote:
>>> On Thu, Oct 17, 2013 at 1:23 PM, Mike Stump wrote:
On Oct 17, 2013, at 10:23 AM, Sriraman Tallam wrote:
>> Y
I'm not sure either, but if it's been approved in gdb and you're
willing to cede control of it to gcc's policies, I'm OK with it.
Note that this will be a new directory in gcc, and I think the
automerge scripts will automatically pick it up. Which means, after
committing it, any future changes m
I've committed the following to dmalcolm/jit:
Add unary ops.
Add more binary ops.
Add more comparison modes.
Convert field access API from:
extern gcc_jit_lvalue *
gcc_jit_context_new_field_access (gcc_jit_context *ctxt,
gcc_jit_location *loc,
As the comment in this patch said, DFS approach is faster in simple
regex, but exponentially slower in complex(many quantifier) cases.
DFA optimization could be added, but I'm afraid it needs rewriting
regex_automaton.*.
Tested under -m32 and -m64.
Thanks!
--
Tim Shen
a.patch
Description: B
I've applied this patch to C11-atomic branch with more miscellaneous
fixes. stdatomic.h had more typo fixes, and a substantive definition
of kill_dependency that I hope is right (it seems to match libstdc++,
and I think it matches the standard). Atomic increment and decrement
are made to go throu
On Thu, Oct 17, 2013 at 3:10 PM, Jakub Jelinek wrote:
> On Thu, Oct 17, 2013 at 12:30:46PM -0700, Sriraman Tallam wrote:
>> I checked my build again for these tests and they all pass.
>
> Even on x86_64-linux I can reproduce all of those with
> -m32 -mno-sse.
Figured out why this happens in -m32
On Fri, Oct 18, 2013 at 9:13 PM, Tim Shen wrote:
> As the comment in this patch said, DFS approach is faster in simple
> regex, but exponentially slower in complex(many quantifier) cases.
Actually I suggest to use DFS where number of quantifiers < 2, to make
this 'optimization' more conservative.
On 10/18/13 03:56, Richard Biener wrote:
On Thu, 17 Oct 2013, Cong Hou wrote:
I tested this case with -fno-tree-loop-im and -fno-tree-pre, and it
seems that GCC could hoist j+1 outside of the i loop:
t3.c:5:5: note: hoisting out of the vectorized loop: _10 = (sizetype) j_25;
t3.c:5:5: note: ho
101 - 143 of 143 matches
Mail list logo
