On Sat, Mar 28, 2015 at 8:21 PM, Alexandre Oliva wrote:
> Regstrapped on x86_64-linux-gnu native and on i686-pc-linux-gnu native
> on x86_64, so without lto plugin. The only regression is in
> gcc.dg/guality/pr54200.c, that explicitly disables VTA.
What about memory footprint? IIRC this pass was
Jakub Jelinek writes:
>> @@ -424,3 +424,8 @@ __cpu_indicator_init (void)
>>
>>return 0;
>> }
>> +
>> +#if defined SHARED && !defined _WIN32
>> +__asm__ (".symver __cpu_indicator_init, __cpu_indicator_init@GCC_4.8.0");
>> +__asm__ (".symver __cpu_model, __cpu_model@GCC_4.8.0");
>> +#endif
>
On Mon, Mar 30, 2015 at 07:08:00PM -0700, H.J. Lu wrote:
> --- a/gcc/gcc.c
> +++ b/gcc/gcc.c
> @@ -1566,11 +1566,13 @@ init_spec (void)
> if (in_sep && *p == '-' && strncmp (p, "-lgcc", 5) == 0)
> {
> init_gcc_specs (&obstack,
> + "-lgcc_nonshared "
>
On 03/26/2015 07:40 AM, Jakub Jelinek wrote:
Hi!
Rather than adding 130 or so missing headers to PLUGIN_HEADERS,
this just adds all *.h and *.def files from gcc/ directory.
For files from different directories, one has to still add them
into PLUGIN_HEADERS.
Tested on x86_64-linux and i686-linux
On 03/27/2015 10:44 AM, Caroline Tice wrote:
It took me a while to get a test case I'm happy with, so I'm
re-submitting the whole patch for approval.
2015-03-27 Caroline Tice
* final.c (final_scan_insn): Change 'cold_function_name' to
'cold_partition_name' and make it a gl
On 03/28/2015 01:21 PM, Alexandre Oliva wrote:
On Mar 27, 2015, Alexandre Oliva wrote:
This patch reworks the out-of-ssa expander to enable coalescing of SSA
partitions that don't share the same base name. This is done only when
optimizing.
The test we use to tell whether two partitions ca
-Wall -Wno-narrowing -Wwrite-strings -Wcast-qual
> -Wno-format -Wstrict-prototypes -Wmissing-prototypes
> -Wold-style-definition -isystem ./include -pipe -fno-common -g
> -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector -pipe
> -fno-common -I. -I. -I../../.././gcc
> -I../../..
;> -g -O2 -DIN_GCC-W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual
>>> -Wno-format -Wstrict-prototypes -Wmissing-prototypes
>>> -Wold-style-definition -isystem ./include -pipe -fno-common -g
>>> -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector -pipe
>&g
./include -pipe -fno-common -g
-DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector -pipe
-fno-common -I. -I. -I../../.././gcc
-I../../../../gcc-5-20150330/libgcc
-I../../../../gcc-5-20150330/libgcc/.
-I../../../../gcc-5-20150330/libgcc/../gcc
-I../../../../gcc-5-20150330/libgcc/../include
s -Wmissing-prototypes
>> -Wold-style-definition -isystem ./include -pipe -fno-common -g
>> -DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector -pipe
>> -fno-common -I. -I. -I../../.././gcc
>> -I../../../../gcc-5-20150330/libgcc
>> -I../../../../gcc-5-20150330/libgcc/
g-libgcc -fno-stack-protector -pipe
> -fno-common -I. -I. -I../../.././gcc
> -I../../../../gcc-5-20150330/libgcc
> -I../../../../gcc-5-20150330/libgcc/.
> -I../../../../gcc-5-20150330/libgcc/../gcc
> -I../../../../gcc-5-20150330/libgcc/../include -DHAVE_CC_TLS
> -DUSE_EMUTLS -fv
-Wwrite-strings -Wcast-qual
-Wno-format -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition -isystem ./include -pipe -fno-common -g
-DIN_LIBGCC2 -fbuilding-libgcc -fno-stack-protector -pipe
-fno-common -I. -I. -I../../.././gcc
-I../../../../gcc-5-20150330/libgcc
-I../../../.
This patch to the Go frontend marks some builtin calls with erroneous
as erroneous. This avoids a compiler crash on invalid code
(http://golang.org/issue/10285). Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Committed to mainline.
Ian
diff -r 6780bf5a7069 go/expressions.cc
---
On Sun, Mar 29, 2015 at 7:40 PM, H.J. Lu wrote:
> On Sun, Mar 29, 2015 at 7:34 PM, H.J. Lu wrote:
>> On Sun, Mar 29, 2015 at 7:25 PM, H.J. Lu wrote:
>>> We shouldn't call external function, __cpu_indicator_init, while an object
>>> is being relocated since its .got.plt section hasn't been update
On 03/30/2015 07:03 PM, Jan Hubicka wrote:
Hello.
Following patch fixed the issue, where we have to check if function summary
for IPA CP has been already created.
Tested on x86_64-linux-pc w/o any new regression introduced.
Ready for trunk?
Thanks,
Martin
>From c33680093e67328863836
The Samsung Exynos M1 implements the ARMv8 ISA and this patch adds support
for it through the -mcpu command-line option.
The patch was checked on arm-unknown-linux-gnueabihf without new failures.
OK for trunk?
--
Evandro Menezes Austin, TX
0001-ARM-Add-option-fo
The Samsung Exynos M1 implements the ARMv8 ISA and this patch adds support
for it through the -mcpu command-line option.
The patch was checked on aarch64-unknown-linux-gnu without new failures.
OK for trunk?
--
Evandro Menezes Austin, TX
0001-AArch64-Add-option-
On Mon, 30 Mar 2015 18:42:02 +0200
Jakub Jelinek wrote:
> On Thu, Mar 26, 2015 at 11:41:30PM +0300, Ilya Verbin wrote:
> > Here is the latest patch for libgomp and mic plugin.
> > make check-target-libgomp using intelmic emul passed.
> > Also I used a testcase from the attachment.
>
> This appli
>
> 2015-03-30 Jakub Jelinek
>
> PR ipa/65610
> * ipa-utils.h (inlined_polymorphic_ctor_dtor_block_p): Declare.
> * ipa-polymorphic-call.c (inlined_polymorphic_ctor_dtor_block_p): New
> function.
> (decl_maybe_in_construction_p, noncall_stmt_may_be_vtbl_ptr_store)
Jeff Law writes:
> On 03/30/2015 08:00 AM, Ilya Enkovich wrote:
>> Hi,
>>
>> Currently check_effective_target_mpx doesn't check libs availability for
>> non Linux targets but tests require them. This patch modifies test to
>> requre wrappers library to link. I suppose it should fix failing test
This is a simple patch that ensures that a .size directive is emitted
when space is allocated for a static variable in the BSS on bare-metal
ARM targets. This allows other tools such as GDB to look up the size of
the object correctly.
Before:
$ readelf -s pr26702.o
Symbol table '.symtab' con
On March 30, 2015 6:45:34 PM GMT+02:00, Alan Lawrence
wrote:
>-O2 was what I first used; it also occurs at -O1. -fno-tree-sra fixes
>it.
>
>The problem appears to be in laying out arguments, specifically
>varargs. From
>the "good" -fdump-rtl-expand:
>
>(insn 18 17 19 2 (set (mem:SI (reg/f:SI 107
On 03/30/2015 08:00 AM, Ilya Enkovich wrote:
Hi,
Currently check_effective_target_mpx doesn't check libs availability for non
Linux targets but tests require them. This patch modifies test to requre
wrappers library to link. I suppose it should fix failing tests on Solaris and
other non-Lin
Hi,
On Mon, 30 Mar 2015 12:33:47, Richard Biener wrote:
>
> So - shouldn't the check be
>
> if (MEM_ALIGN (op0) < GET_MODE_ALIGNMENT (fieldmode))
> return false;
>
No. Because this example would access memory beyond the end of structure
on m66k:
volatile struct s
{
unsigned x:15;
} g;
int x =
On Mon, Mar 30, 2015 at 08:45:51PM +0200, Jan Hubicka wrote:
> We do another round of polymorphic call analysis in PRE pass. (it is not that
> important to do so as it is too late to inline the call, but we do that)
Ok.
> > Also, looking for better name of the function if you have ideas.
>
> I t
> On 03/30/2015 02:36 PM, Jan Hubicka wrote:
> >Will there be only one copy even on targets that do not support aliases?
>
> Yes, on those targets we use thunks.
I see, and only if thinks fails on variadic arguments we lose, right?
>
> >Still, it would be great if we can make context to be the
> Hi!
>
> As discussed in the PR, we have -fcompare-debug failures, because
> remove_unused_scope_block_p can sometimes remove blocks relevant
> to ipa-polymorphic-call.c analysis (where it is checking if
> there is some ctor or dtor in BLOCKs from current tree block upward in
> BLOCK_SUPERCONTEXT
On 03/30/2015 02:36 PM, Jan Hubicka wrote:
Will there be only one copy even on targets that do not support aliases?
Yes, on those targets we use thunks.
Still, it would be great if we can make context to be the actual function.
That sounds messy.
Otherwise I will need to figure out some w
> On 03/30/2015 01:19 PM, Jan Hubicka wrote:
> >I think the problem is that C++ FE does not ask about copy_forbidden and
> >clone
> >ctor anyway?
>
> maybe_clone_body checks tree_versionable_function_p and doesn't
> clone if it's false.
>
> ...ok, now I've actually checked the testcase. In this
On 03/30/2015 01:19 PM, Jan Hubicka wrote:
I think the problem is that C++ FE does not ask about copy_forbidden and clone
ctor anyway?
maybe_clone_body checks tree_versionable_function_p and doesn't clone if
it's false.
...ok, now I've actually checked the testcase. In this case there is a
On 30/03/15 18:59 +0100, Jonathan Wakely wrote:
In the library we instantiate these functions:
template S operator+(const C*, const S&);
template S operator+(C, const S&);
template S operator+(const S&, const S&);
Hmm, that's a bit cryptic, sorry :-)
It would have been useful to explain that
Hi!
As discussed in the PR, we have -fcompare-debug failures, because
remove_unused_scope_block_p can sometimes remove blocks relevant
to ipa-polymorphic-call.c analysis (where it is checking if
there is some ctor or dtor in BLOCKs from current tree block upward in
BLOCK_SUPERCONTEXT hierarchy) -
In the library we instantiate these functions:
template S operator+(const C*, const S&);
template S operator+(C, const S&);
template S operator+(const S&, const S&);
We have a Fedora bug report for a package which uses
-fno-implicit-templates and then expects to find these instantations
in th
Hi!
This patch makes sure we have OMP_CLAUSE_LINEAR_NO_COPYIN set on
iterator vars, both with implicit and explicit linear clause, so that
we don't ICE on those - the iterator vars don't need original list item
value. The patch also fixes up the step value for the artificial count
iterator var, w
OK, thanks.
Jason
On 03/30/2015 01:23 PM, Jan Hubicka wrote:
Jason probably knows better, but I think only real C++ types comply the One
Defintion
Type and should be merged. Anything we create artifically in compiler is
probably
not covered by this.
Agreed, compiler internals are outside the scope of the langu
Dear all,
please find attach a patch fixing pr44672:
integer, dimension(:) :: arr
allocate(arr, source = [1,2,3])
as for F2008:C633 now is no longer flagged, beside when you insist on
-std=f2003 or lower. Furthermore does the patch implement the F2008 feature of
obsoleting the explicit array spe
On Thu, Mar 26, 2015 at 03:28:53PM +0300, Marat Zakirov wrote:
> 2015-02-25 Marat Zakirov
>
> * asan.c (asan_emit_stack_protection): Support for misalign accesses.
> (asan_expand_check_ifn): Likewise.
> * params.def: New option asan-catch-misaligned.
> * params.h: New
The Go language permits converting any type whose underlying type is
uintptr to unsafe.Pointer. The Go frontend did not implement this
correctly. This is http://golang.org/issue/10284. This patch from
Minux fixes the problem. Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Commi
> On Mon, Mar 30, 2015 at 07:06:39PM +0200, Jan Hubicka wrote:
> > > On Mon, Mar 30, 2015 at 05:02:57 +0200, Jan Hubicka wrote:
> > > > * lto-symtab.c (lto_symtab_merge_decls_2): Silence warnings on
> > > > artificial decls.
> > >
> > > Shouldn't this patch fix libgomp.c++/target-3
On Mon, Mar 30, 2015 at 07:06:39PM +0200, Jan Hubicka wrote:
> > On Mon, Mar 30, 2015 at 05:02:57 +0200, Jan Hubicka wrote:
> > > * lto-symtab.c (lto_symtab_merge_decls_2): Silence warnings on
> > > artificial decls.
> >
> > Shouldn't this patch fix libgomp.c++/target-3.C in an offloading-enab
> On 30 Mar 14:05, Ilya Enkovich wrote:
> > 2015-03-27 18:23 GMT+03:00 Jan Hubicka :
> > > Index: symtab.c
> > > ===
> > > --- symtab.c(revision 221734)
> > > +++ symtab.c(working copy)
> > > @@ -1130,15 +1130,20 @@ symtab_node
> On 03/30/2015 05:32 AM, Richard Biener wrote:
> >IMHO that's bogus. But I thought we decided that functions with
> >non-local labels shouldn't be cloned?
>
> Yes, why isn't copy_forbidden flagging this function? It has a
> check for exactly this situation.
I think the problem is that C++ FE d
> On Mon, Mar 30, 2015 at 19:06:39 +0200, Jan Hubicka wrote:
> > > On Mon, Mar 30, 2015 at 05:02:57 +0200, Jan Hubicka wrote:
> > > > * lto-symtab.c (lto_symtab_merge_decls_2): Silence warnings on
> > > > artificial decls.
> > >
> > > Shouldn't this patch fix libgomp.c++/target-3.C
On Mon, Mar 30, 2015 at 19:06:39 +0200, Jan Hubicka wrote:
> > On Mon, Mar 30, 2015 at 05:02:57 +0200, Jan Hubicka wrote:
> > > * lto-symtab.c (lto_symtab_merge_decls_2): Silence warnings on
> > > artificial decls.
> >
> > Shouldn't this patch fix libgomp.c++/target-3.C in an offloading-enable
> On Mon, Mar 30, 2015 at 05:02:57 +0200, Jan Hubicka wrote:
> > * lto-symtab.c (lto_symtab_merge_decls_2): Silence warnings on
> > artificial decls.
>
> Shouldn't this patch fix libgomp.c++/target-3.C in an offloading-enabled
> configuration? It still fails...
>
> libgomp/testsuite/libg
> > Index: ipa-cp.c
> > ===
> > --- ipa-cp.c(revision 221757)
> > +++ ipa-cp.c(working copy)
> > @@ -811,6 +811,41 @@ set_all_contains_variable (struct ipcp_p
> >return ret;
> > }
> >
> > +/* Worker of call_for_s
> Hello.
>
> Following patch fixed the issue, where we have to check if function summary
> for IPA CP has been already created.
>
> Tested on x86_64-linux-pc w/o any new regression introduced.
> Ready for trunk?
>
> Thanks,
> Martin
> >From c33680093e67328863836a845e847bf1b1b23d0e Mon Sep
-O2 was what I first used; it also occurs at -O1. -fno-tree-sra fixes it.
The problem appears to be in laying out arguments, specifically varargs. From
the "good" -fdump-rtl-expand:
(insn 18 17 19 2 (set (mem:SI (reg/f:SI 107 virtual-outgoing-args) [0 S4 A32])
(reg:SI 111 [ b1$16 ]))
On Thu, Mar 26, 2015 at 11:41:30PM +0300, Ilya Verbin wrote:
> Here is the latest patch for libgomp and mic plugin.
> make check-target-libgomp using intelmic emul passed.
> Also I used a testcase from the attachment.
This applies cleanly.
> Latest ptx part is here, I guess:
> https://gcc.gnu.org
On Mon, Mar 30, 2015 at 05:02:57 +0200, Jan Hubicka wrote:
> * lto-symtab.c (lto_symtab_merge_decls_2): Silence warnings on
> artificial decls.
Shouldn't this patch fix libgomp.c++/target-3.C in an offloading-enabled
configuration? It still fails...
libgomp/testsuite/libgomp.c++/../l
Turned out that my earlier fix for this PR didn't handle cases where we aren't
dealing with an INDIRECT_REF expression, which means we fail to compile valid
code. Fixed by moving the folding to cxx_eval_constant_expression, where we
ought to be able to reduce all POINTER_PLUS_EXPRs in a constexpr
Apparently 41-bit bit-fields with type long can't be used on archs with 32-bit
long.
Applying to trunk.
2015-03-30 Marek Polacek
* c-c++-common/pr65556.c: Change the width of bit-fields.
diff --git gcc/testsuite/c-c++-common/pr65556.c
gcc/testsuite/c-c++-common/pr65556.c
index c672
On Mon, Mar 30, 2015 at 04:06:56PM +0200, Marek Polacek wrote:
> On Mon, Mar 30, 2015 at 04:02:36PM +0200, Jakub Jelinek wrote:
> > On Mon, Mar 30, 2015 at 03:00:54PM +0100, James Greenhalgh wrote:
> > > Likewise on ARM. The testcase is fairly obviously not going to work
> > > for any target with 3
On Mon, Mar 30, 2015 at 04:02:36PM +0200, Jakub Jelinek wrote:
> On Mon, Mar 30, 2015 at 03:00:54PM +0100, James Greenhalgh wrote:
> > Likewise on ARM. The testcase is fairly obviously not going to work
> > for any target with 32-bit long:
> >
> > struct S
> > {
> > long l: 1;
> > long
On 03/30/2015 05:32 AM, Richard Biener wrote:
IMHO that's bogus. But I thought we decided that functions with
non-local labels shouldn't be cloned?
Yes, why isn't copy_forbidden flagging this function? It has a check
for exactly this situation.
And cloning static vars looks bogus
as well
On Mon, Mar 30, 2015 at 03:00:54PM +0100, James Greenhalgh wrote:
> Likewise on ARM. The testcase is fairly obviously not going to work
> for any target with 32-bit long:
>
> struct S
> {
> long l: 1;
> long l2: 41;
> unsigned long ul: 1;
> unsigned long ul2: 41;
> } s;
>
>
Hi,
Currently check_effective_target_mpx doesn't check libs availability for non
Linux targets but tests require them. This patch modifies test to requre
wrappers library to link. I suppose it should fix failing tests on Solaris and
other non-Linux systems. Patch also replaces alloca calls w
Hi,
This patch fixes the warning:
...
terminals.c:1266:21: warning: argument to ‘sizeof’ in ‘bzero’ call is the same
expression as the destination; did you mean to remove the addressof?
[-Wsizeof-pointer-memaccess]
...
bootstrapped and reg-tested on x86_64.
OK for stage 4/stage1?
Thanks,
-
On Fri, Mar 27, 2015 at 11:14:31PM +, H.J. Lu wrote:
> On Fri, Mar 27, 2015 at 7:38 AM, Marek Polacek wrote:
> > In this testcase we were crashing while trying to gimplify a switch, because
> > the types of the switch condition and case constants didn't match. This ICE
> > started with my -Ws
Ping.
https://gcc.gnu.org/ml/gcc-patches/2015-03/msg01196.html
Thanks,
Kyrill
On 23/03/15 16:15, Kyrill Tkachov wrote:
Hi all,
The ICE in the PR happens on arm during the hoist pass when the code
generates a SET rtx of the form:
(set (reg:OI) (const_int 0)). It checks whether const_int 0 is a
g
Hi,
On Mon, 30 Mar 2015 12:33:47, Richard Biener wrote:
>
> On Mon, Mar 16, 2015 at 11:53 AM, Bernd Edlinger
> wrote:
>>
>> Hi,
>>
>>
>> when looking at the m68k I realized the following, which is
>> a general problem...
>>
>> If the alignment of the structure is less than sizeof(field), the
>> s
The following patch fixes PR65626 - an ordering issue with removing
BBs and fixing up noreturn stmts. The fix is simple - delay
BB removal to the CFG cleanup run and only split the BB at
fixup_noreturn_stmt time so CFG cleanup can find it in O(1).
(similarly EH/abnormal cleanup should only remove
On Mon, 30 Mar 2015, Richard Biener wrote:
> On Mon, 30 Mar 2015, Alan Lawrence wrote:
>
> > ...actually attach the testcase...
>
> What compile options?
Just tried -O2. The GIMPLE IL assumes 64bit alignment of .LC0 but
I can't see anything not guaranteeing that:
.section.roda
On Mon, 30 Mar 2015, Alan Lawrence wrote:
> ...actually attach the testcase...
What compile options?
> Alan Lawrence wrote:
> > We've been seeing a bunch of new failures in the *libffi* testsuite on ARM
> > Linux (arm-none-linux-gnueabi, arm-none-linux-gnueabihf), following this
> > one-liner fi
...actually attach the testcase...
Alan Lawrence wrote:
We've been seeing a bunch of new failures in the *libffi* testsuite on ARM Linux
(arm-none-linux-gnueabi, arm-none-linux-gnueabihf), following this one-liner
fix. I've reduced the testcase down to the attached (including removing any
depe
We've been seeing a bunch of new failures in the *libffi* testsuite on ARM Linux
(arm-none-linux-gnueabi, arm-none-linux-gnueabihf), following this one-liner
fix. I've reduced the testcase down to the attached (including removing any
dependency on libffi); with gcc r221347, this prints the expec
Hi,
On Sun, Mar 29, 2015 at 05:43:20PM +0200, Jan Hubicka wrote:
> Hi,
thanks for committing the patch, I'm just wondering why...
> this patch improve crafty performance by avoiding ipa-cp clonning of
> Search function that specializes the first iteration of the recursion.
> The patch is by Mart
On 30 Mar 14:05, Ilya Enkovich wrote:
> 2015-03-27 18:23 GMT+03:00 Jan Hubicka :
> > Index: symtab.c
> > ===
> > --- symtab.c(revision 221734)
> > +++ symtab.c(working copy)
> > @@ -1130,15 +1130,20 @@ symtab_node::verify_symta
Hi!
Sorry for delay in sending this patch for review.
Please find attached updated patch.
In P5600, 2 consecutive loads/stores of same type which access contiguous memory locations are bonded together by instruction issue unit to dispatch
single load/store instruction which accesses both locati
Hello.
Following patch fixed the issue, where we have to check if function summary
for IPA CP has been already created.
Tested on x86_64-linux-pc w/o any new regression introduced.
Ready for trunk?
Thanks,
Martin
>From c33680093e67328863836a845e847bf1b1b23d0e Mon Sep 17 00:00:00 2001
Fr
2015-03-27 18:23 GMT+03:00 Jan Hubicka :
> Hi,
> this patch fixes bug in symtab_node::verify_symtab_nodes pointed out by Ilya.
> The loop checking that there all comdats are linked by same_comdat_group was
> completely bogus. In addition it checked also external symbols that are
> currently not ke
On Mon, Mar 16, 2015 at 11:53 AM, Bernd Edlinger
wrote:
>
> Hi,
>
>
> when looking at the m68k I realized the following, which is
> a general problem...
>
> If the alignment of the structure is less than sizeof(field), the
> strict volatile bitfields code may read beyond the end of the
> structure
2015-03-27 18:23 GMT+03:00 Jan Hubicka :
> Hi,
> this patch fixes bug in symtab_node::verify_symtab_nodes pointed out by Ilya.
> The loop checking that there all comdats are linked by same_comdat_group was
> completely bogus. In addition it checked also external symbols that are
> currently not ke
On 30-03-15 10:15, Jan Hubicka wrote:
diff --git a/gcc/tree-cfg.c b/gcc/tree-cfg.c
index 64bdc92..c7a7c4d 100644
--- a/gcc/tree-cfg.c
+++ b/gcc/tree-cfg.c
@@ -6177,6 +6177,7 @@ gimple_duplicate_sese_tail (edge entry ATTRIBUTE_UNUSED,
edge exit ATTRIBUTE_UNU
gphi *phi;
tree def;
struc
On 30-03-15 10:15, Jan Hubicka wrote:
Also move_sese_region_to_fn seem to mis updating of counts. Can you, please,
add that and send updated patch?
Like this?
OK for stage1 if bootstrap and reg-test on x86_64 are ok?
Thanks,
- Tom
Preserve edge count in move_sese_region_to_fn
2015-03-30
On Mon, Mar 30, 2015 at 11:07 AM, Jan Hubicka wrote:
> Hi,
> unforutnately the patch ICE on:
> int i;
> struct C
> {
> C();
> };
>
> C::C()
> {
> static void *labelref = &&label;
> goto *labelref;
> label: i = 1;
> }
>
> int main()
> {
> C c;
> return (i != 1);
> }
>
> The problem is th
On Mon, Mar 30, 2015 at 3:45 AM, Jan Hubicka wrote:
> Hi,
> this patch makes inline metric to be
>estimate_speedup
> (over)
> edge_growth^2*overall_growth
>
> This is intended to make stronger push on inliner to preffer functions with
> smaller growth and d
On Fri, Mar 27, 2015 at 4:26 PM, Jan Hubicka wrote:
> Hi,
> I notieced that GGC hash tables are now accounted to
> hash-table.h:alloc_entries
> that is not very informative (we do not have any stats for heap hash tables
> that
> would be nice to have).
>
> This patch fixes the first problem by a
Hi,
unforutnately the patch ICE on:
int i;
struct C
{
C();
};
C::C()
{
static void *labelref = &&label;
goto *labelref;
label: i = 1;
}
int main()
{
C c;
return (i != 1);
}
The problem is that decl_function_context of labelref is not C::C
itself but an abstract function that is abst
> On Mon, 30 Mar 2015, Jan Hubicka wrote:
>
> > Hi,
> > when compiling C++ program that define different number of virtual method
> > in different classes, we output warnings about their virtual tables
> > being of different type. THese warnings looks ugly and we are able
> > to diagnose the situa
On Mon, 30 Mar 2015, Jan Hubicka wrote:
> Hi,
> when compiling C++ program that define different number of virtual method
> in different classes, we output warnings about their virtual tables
> being of different type. THese warnings looks ugly and we are able
> to diagnose the situation in more s
Hi,
also this code ignores counts and probabilities:
/* Rewire the entry and exit blocks. The successor to the entry
block turns into the successor of DEST_FN's ENTRY_BLOCK_PTR in
the child function. Similarly, the predecessor of DEST_FN's
EXIT_BLOCK_PTR turns into the predecesso
> diff --git a/gcc/tree-cfg.c b/gcc/tree-cfg.c
> index 64bdc92..c7a7c4d 100644
> --- a/gcc/tree-cfg.c
> +++ b/gcc/tree-cfg.c
> @@ -6177,6 +6177,7 @@ gimple_duplicate_sese_tail (edge entry
> ATTRIBUTE_UNUSED, edge exit ATTRIBUTE_UNU
>gphi *phi;
>tree def;
>struct loop *target, *aloop, *
Hi,
this patch fixes PR65511.
For the testcase, before transform_to_exit_first_loop the back-edge probability
is 99%:
...
;; basic block 5, loop depth 1, count 0, freq 7920, maybe hot
;; prev block 11, next block 6, flags: (NEW)
;; pred: 11 [100.0%] (FALLTHRU)
;; 7 [100.0
Hi,
this patch fixes problem with partitioning WRT named labels. This problem
reproduces with linux kernel LTO and also on firefox with -fno-toplevel-reorder
at least.
Bootstrapped/regtested x86_64-linux and tested it fixes the problem on Firefx,
intend to commit it tomorrow after some more testi
86 matches
Mail list logo
