On 08/14/2014 09:46 AM, Yury Gribov wrote:
Something like this?
Forgot to mention: full tested the new patch on x64 (bootstrap, regtest,
Asan-bootstrap).
On 08/13/2014 02:13 PM, Jakub Jelinek wrote:
I wonder if we just shouldn't handle COMPONENT_REFs with
DECL_BIT_FIELD_REPRESENTATIVE that way unconditionally.
So, do the if (TREE_CODE (t) == COMPONENT_REF ... before the other checks,
and
else if (bitpos % BITS_PER_UNIT || bitsize != size_in_byt
PR60484 is marked as 4.7/4.8 regression and it is reported against 4.8
recently by an user.
OK backporting to 4.7/4.8?
- Joey
On Sat, Mar 15, 2014 at 1:43 AM, Joseph S. Myers
wrote:
> On Fri, 14 Mar 2014, Marek Polacek wrote:
>
>> This patch makes sure that we set the directory prefix of
>> dum
On 13 August 2014 20:29, Richard Earnshaw wrote:
> On 25/02/14 09:34, Zhenqiang Chen wrote:
>> Hi,
>>
>> Current value for max_insns_skipped is 6. For THUMB2, it needs 2 (IF-THEN)
>> or 3 (IF-THEN-ELSE) IT blocks to hold all the instructions. The overhead of
>> IT is 4 or 6 BYTES.
>>
>> If we do n
> From: Marek Polacek [mailto:pola...@redhat.com]
> Sent: Tuesday, August 12, 2014 5:43 PM
> On Tue, Aug 12, 2014 at 11:34:35AM +0200, Jakub Jelinek wrote:
> >
> > This looks wrong. c_token_starts_typename is just a function which tells
> > you if certain token can start a typename, issuing diagno
This bug is caused by my last patch, which did not differentiate
between explicit section names (via attributes) and implicit section
names (via -ffunction-section).
This patch fixes that.
--
diff --git gcc/bb-reorder.c gcc/bb-reorder.c
index 8f8c420..2115b01 100644
--- gcc/bb-reorder.c
+++ gcc/
On 08/12/2014 09:35 PM, Jakub Jelinek wrote:
As detailed in the PR, find_inc ignored any possible clobbers on
inc_insn (typically %cc/flags/etc. register) and thus we could ignore
all register dependencies between mem_insn and inc_insn even when
we could only safely ignore the mem_reg0 register d
On 08/06/14 11:21, David Malcolm wrote:
/
* rtx-classes-status.txt: Update
---
rtx-classes-status.txt | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/rtx-classes-status.txt b/rtx-classes-status.txt
index e350eaf..b22cb1e 100644
--- a/rtx-classes-status.txt
+++ b
On 08/13/14 18:11, David Malcolm wrote:
On Wed, 2014-08-13 at 14:34 -0600, Jeff Law wrote:
On 08/13/14 14:28, David Malcolm wrote:
Thanks. Although this function gets converted back to a macro in patch
191, I just realized that in the meantime that it's not inlined, as is
the case for some of
Hi,
Function split_live_ranges_for_shrink_wrap has code
if (!flag_shrink_wrap)
return false;
But flag_shrink_wrap is TRUE by default when optimize > 0 even if the
port does not support shrink-wrap. To make sure shrink-wrap is
enabled, "HAVE_simple_return" must be defined and "HAVE_simple_r
On 08/06/14 11:21, David Malcolm wrote:
gcc/
* resource.c (next_insn_no_annul): Strengthen local "next" from
rtx to rtx_insn *.
(mark_referenced_resources): Likewise for local "insn".
---
gcc/resource.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
OK.
As
On 08/06/14 11:21, David Malcolm wrote:
gcc/
* reload.h (struct insn_chain): Strengthen field "insn" from rtx
to rtx_insn *.
(find_reloads): Likewise for param 1.
(subst_reloads): Likewise for sole param.
(find_equiv_reg): Likwise for param 2.
(regn
On 8/14/14 10:14, Michael Eager wrote:
>> OK, thanks, and I shall analyze it, but excuse me, I have to do other
>> things this week, so hope I can finish it within next week (2014-08-24).
>> If this time point is too long to bare, please let me know.
>
> Take your time and let me know when you hav
On 08/13/14 18:35, Chen Gang wrote:
Firstly, thank you very much for spending your time resource on the
related 2 patches.
You're welcome.
On 8/13/14 23:10, Michael Eager wrote:
On 07/06/14 03:26, Chen Gang wrote:
* microblaze/mocroblaze.md (call_value_intern): Use 'SI' instead of
On Tue, Aug 12, 2014 at 2:07 AM, Matthew Fortune
wrote:
> Eric Christopher writes:
>> On Sat, Aug 9, 2014 at 12:00 PM, Matthew Fortune
>> wrote:
>> > Moore, Catherine writes:
>> >> > -Original Message-
>> >> > From: Steve Ellcey [mailto:sell...@mips.com]
>> >> > Sent: Friday, August 08,
Ah... sorry. Leftovers. I didn't have time to run a full bootstrap
build before heading out for a few days. I'll try to get those out
tomorrow afternoon-ish.
Andrew
On Wed, Aug 13, 2014 at 9:13 PM, Ed Smith-Rowland <3dw...@verizon.net> wrote:
> I get build fail:
>
> ../../gcc_concepts/gcc/cp/cal
Firstly, thank you very much for spending your time resource on the
related 2 patches.
On 8/13/14 23:10, Michael Eager wrote:
> On 07/06/14 03:26, Chen Gang wrote:
>>
>>* microblaze/mocroblaze.md (call_value_intern): Use 'SI' instead of
>>'VOID' for operand 1, just like 'call_internal1' h
I get build fail:
../../gcc_concepts/gcc/cp/call.c:8793:8: error: unused variable ‘m1’
[-Werror=unused-variable]
tree m1 = get_temploid (cand1);
^
../../gcc_concepts/gcc/cp/call.c:8794:8: error: unused variable ‘m2’
[-Werror=unused-variable]
tree m2 = get_temploid (cand2);
On 14 August 2014 03:41, Jason Merrill wrote:
> On 08/13/2014 07:06 PM, Ville Voutilainen wrote:
>
> Looks good, but
>
>> + int friendp = 0;
>
>
> Let's declare friendp when it's initialized.
>
> And I still need a ChangeLog entry.
Ok, modified patch attached. The changelog entry was attached to
On 08/13/2014 07:06 PM, Ville Voutilainen wrote:
Looks good, but
+ int friendp = 0;
Let's declare friendp when it's initialized.
And I still need a ChangeLog entry.
Jason
On Wed, 2014-08-06 at 13:19 -0400, David Malcolm wrote:
> This is the patch series I spoke about at Cauldron in the talk
> "A proposal for typesafe RTL"; slides here:
> http://dmalcolm.fedorapeople.org/presentations/cauldron-2014/rtl
>
> They can also be seen at:
> https://dmalcolm.fedorapeople.or
On Wed, 2014-08-13 at 14:34 -0600, Jeff Law wrote:
> On 08/13/14 14:28, David Malcolm wrote:
> > Thanks. Although this function gets converted back to a macro in patch
> > 191, I just realized that in the meantime that it's not inlined, as is
> > the case for some of the other macro->function conv
On 2014-08-13, 6:30 PM, Jakub Jelinek wrote:
On Tue, Aug 12, 2014 at 05:12:41PM -0400, Vladimir Makarov wrote:
On 08/12/2014 03:35 PM, Jakub Jelinek wrote:
Hi!
As detailed in the PR, find_inc ignored any possible clobbers on
inc_insn (typically %cc/flags/etc. register) and thus we could ignore
Hi,
This patch adds a PowerPC-specific pass just prior to the first cse RTL
pass. The pass runs only when generating little-endian code for Power8
with VSX enabled, and for -O1 and up. For this particular subtarget,
the use of the big-endian-biased vector load and store instructions
requires per
We didn't allow deleted friend definitions, because that was diagnosed
in grokdeclarator, which doesn't have knowledge of the kind of a function
declaration "initializer". Move the test to grokfield and adjust where
necessary. Tested on Linux x86-64.
diff --git a/gcc/cp/decl.c b/gcc/cp/decl.c
index
Hi,
I want to reuse some of the infrastructure in web.c (and df.h) for a
target-specific RTL pass, particularly the union-find stuff. I could
just copy it, I suppose, but it seems better to make the struct
web_entry into a class so that I can inherit what I need. That's what
this patch does.
Wh
> A while ago I've removed a couple of those 'typedef struct' things, as
> they are not required in C++ anymore. Is there any particular reason
> why this couldn't be simply 'struct int_n_data_t' ?
No particular reason.
The Go frontend passes closures through to functions using the functions
__go_set_closure and __go_get_closure. The expectation is that there
are no function calls between set_closure and get_closure. However, it
turns out that there can be function calls if some of the function
arguments require
On Tue, Aug 12, 2014 at 05:12:41PM -0400, Vladimir Makarov wrote:
> On 08/12/2014 03:35 PM, Jakub Jelinek wrote:
> > Hi!
> >
> > As detailed in the PR, find_inc ignored any possible clobbers on
> > inc_insn (typically %cc/flags/etc. register) and thus we could ignore
> > all register dependencies b
On Wed, 2014-08-13 at 18:11 -0400, DJ Delorie wrote:
> This is the main __int128 -> __intN conversion patch. We support up
> to four __int types per target, which may include __int128. I
> include the minimal change to the msp430 backend to illustrate how to
> add a target-specific __int type.
>
2014-08-13 22:51 GMT+02:00 Steve Ellcey :
> On Wed, 2014-08-13 at 20:42 +, Joseph S. Myers wrote:
>> On Wed, 13 Aug 2014, Steve Ellcey wrote:
>>
>> > This is a ping on a patch I sent out a while ago to fix the GCC build
>> > when building with the mingw toolset.
>>
>> This is not a review.
>>
Changes to the testsuite to make tests more portable to targets with
unusual address spaces.
* testsuite/
* lib/target-supports.exp (check_effective_target_size32plus):
Increase size to avoid false positives on 24-bit address spaces.
* gcc.c-torture/compile/limits-stringli
This is the MSP430-specific use of the new intN framework to enable
true 20-bit pointers. Since I'm one of the MSP430 maintainers, this
patch is being posted for reference, not for approval.
gcc/config/msp430
* config/msp430/msp430-modes.def (PSI): Add.
* config/msp430/msp430-pr
The purpose of this set of changes is to remove assumptions in GCC
about type sizes. Previous to this patch, GCC assumed that all types
were powers-of-two in size, and used naive math accordingly.
Old:
POINTER_SIZE / BITS_PER_UNIT
TYPE_SIZE
GET_MODE_BITSIZE
New:
This patch is part of the __intN series, but is independent. It
provides an additional optimization opportunity, since the MSP430 does
a lot of conversions between HImode and PSImode.
* expr.c (convert_move): If the target has an explicit converter,
use it.
Index: gcc/expr.c
===
The following five patches are the latest in my ongoing work to
replace the hard-coded __int128 type with a more flexible __int
system that allows target-specific types that correspond to
partial-int modes. Specifically, this will allow targets to have
pointers that aren't powers-of-two in size.
Jeff Law wrote:
> I'm pretty sure jump threading *could* handle it, but after looking
> at the full testcase when it was posted, I'm not sure it's *wise* to
> handle this in jump threading.
Thanks for clearing my doubts.
Sebastian
Hi Richard,
sorry for the belated reply due to spending lazily my time at the sea ...
The patch is okay - and I would even claim that it is obvious.
Regarding the data type, I have no idea why it uses a hard-coded 32bit
integer instead of a size-type one. I have added it as item to
https://gc
Hi,
This patch makes gcov-tool build configurable. A new configure option
--disable-gcov-tool is added to disable the build.
It also fixes some build issues for mingw.
Tested with regression and bootstrap. mingw test is ongoing.
OK to check in if mingw build passes?
Thanks,
-Rong
gcov_tool_
On 08/13/14 14:55, Sebastian Pop wrote:
Steve Ellcey wrote:
+/* This file implements an optimization where, when a variable is set
+ to a constant value and there is a path that leads from that definition
+ to a switch statement that uses that variable as its controlling expression
+ we du
Steve Ellcey wrote:
> +/* This file implements an optimization where, when a variable is set
> + to a constant value and there is a path that leads from that definition
> + to a switch statement that uses that variable as its controlling
> expression
> + we duplicate the blocks on this path
On Wed, 2014-08-13 at 20:42 +, Joseph S. Myers wrote:
> On Wed, 13 Aug 2014, Steve Ellcey wrote:
>
> > This is a ping on a patch I sent out a while ago to fix the GCC build
> > when building with the mingw toolset.
>
> This is not a review.
>
> When pinging, please give *the gcc.gnu.org URL
On Thu, 14 Aug 2014, Alexander Ivchenko wrote:
> Hi,
>
> This patch disables a bunch of tests that fail when using Bionic libc.
> But this is expected; three reasons:
> - Bionic does not support complex functions.
> - Bionic does not have tgmath.h and error.h headers.
> - Bionic does not have mem
On Wed, 13 Aug 2014, Steve Ellcey wrote:
> This is a ping on a patch I sent out a while ago to fix the GCC build
> when building with the mingw toolset.
This is not a review.
When pinging, please give *the gcc.gnu.org URL of the patch being pinged*.
--
Joseph S. Myers
jos...@codesourcery.com
Cesar Philippidis wrote:
According to section 2.6.1 in the openacc spec, fortran loop variables
should be implicitly private like in openmp. This patch does just so.
Makes sense. Looking at the patch, I wonder whether the context is
properly handled when mixing OpenMP and OpenACC. I have the f
On 08/13/14 14:28, David Malcolm wrote:
Thanks. Although this function gets converted back to a macro in patch
191, I just realized that in the meantime that it's not inlined, as is
the case for some of the other macro->function conversions in patches
13-16.
Do I need to convert them to inline
On Tue, 2014-08-12 at 15:20 -0600, Jeff Law wrote:
> On 08/06/14 11:19, David Malcolm wrote:
> > DF_REF_INSN looks up the "insn" field of the referenced df_insn_info.
> > This will eventually be an rtx_insn *, but for now is just an rtx.
> >
> > As further scaffolding: for now, convert DF_REF_INSN
On 08/11/14 07:49, Ilya Tocar wrote:
Hi,
I've observed SPEC2006 failure on avx512-vlbwdq branch.
It was caused by hardreg_cprop. In maybe_mode_change it was
assumed, that all values of the same register class and same mode.
are ok. This is not the case for i386/avx512. We need to honor
HARD_REG
On 08/13/14 13:32, David Malcolm wrote:
On Wed, 2014-08-06 at 13:22 -0400, David Malcolm wrote:
---
gcc/insn-addr.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/gcc/insn-addr.h b/gcc/insn-addr.h
index e255ac4..aec09fd 100644
--- a/gcc/insn-addr.h
+++ b/gcc/insn-addr.h
@
On 08/06/14 11:21, David Malcolm wrote:
gcc/
* lra-int.h (struct lra_insn_recog_data): Strengthen field "insn"
from rtx to rtx_insn *.
(lra_push_insn): Likewise for 1st param.
(lra_push_insn_and_update_insn_regno_info): Likewise.
(lra_pop_insn): Likewise fo
Hi,
This patch disables a bunch of tests that fail when using Bionic libc.
But this is expected; three reasons:
- Bionic does not support complex functions.
- Bionic does not have tgmath.h and error.h headers.
- Bionic does not have mempcpy and stpcpy.
diff --git a/gcc/testsuite/ChangeLog b/gcc/
On Wed, 2014-08-13 at 12:07 +0200, Richard Biener wrote:
> On Wed, Aug 13, 2014 at 12:01 PM, Martin Jambor wrote:
> > Hi,
> >
> > On Wed, Aug 06, 2014 at 01:19:44PM -0400, David Malcolm wrote:
> >> In many circumstances, is_a_helper ::test assumes that the pointer is
> >> non-NULL, but sometimes y
This patch re-implements explicit instantiation and specialization.
The latter isn't fully tested just yet, but it's close.
Constraints for explicit instantiations and specializations are
deduced from their candidates, not explicitly specified as part of the
declaration.
2014-08-13 Andrew Sutton
On 08/06/14 11:20, David Malcolm wrote:
In particular, after this patch, the first param passed to the
FINAL_PRESCAN_INSN macro is strengthened from rtx to rtx_insn *.
gcc/
* output.h (final_scan_insn): Strengthen return type from rtx to
rtx_insn *.
(final_forward_branch_
On Wed, 2014-08-06 at 13:22 -0400, David Malcolm wrote:
> ---
> gcc/insn-addr.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/gcc/insn-addr.h b/gcc/insn-addr.h
> index e255ac4..aec09fd 100644
> --- a/gcc/insn-addr.h
> +++ b/gcc/insn-addr.h
> @@ -38,7 +38,7 @@ extern int
On Wed, Aug 13, 2014 at 08:02:22PM +0100, Jonathan Wakely wrote:
> On 13/08/14 20:47 +0200, Jakub Jelinek wrote:
> >On Wed, Aug 13, 2014 at 07:40:57PM +0100, Jonathan Wakely wrote:
> >>This should fix the Hurd bug in PR61841 as well as the problem
> >>described in https://gcc.gnu.org/ml/gcc/2013-09
> This should fix an undefined behavior in diagnostics.c.
> Under certain circumstances, max_width is (INT_MAX - 1),
> and right_margin is -4 -> the subtraction overflows.
> Changing the types to unsigned would involve changing
> much more code than just one cast.
>
> BTW, the diagnostics we output
On 13/08/14 20:47 +0200, Jakub Jelinek wrote:
On Wed, Aug 13, 2014 at 07:40:57PM +0100, Jonathan Wakely wrote:
This should fix the Hurd bug in PR61841 as well as the problem
described in https://gcc.gnu.org/ml/gcc/2013-09/msg00196.html
Tested x86_64-linux, committed to trunk.
This is not suita
> 2014-08-13 Ilya Enkovich
>
> * cgraph.c (cgraph_node::function_symbol): Fix wrong
> cgraph_function_node to cgraph_node::function_symbol
> refactoring.
OK, thanks1
Honza
>
> diff --git a/gcc/cgraph.c b/gcc/cgraph.c
> index 5a0b903..370a96a 100644
> --- a/gcc/cgraph.c
> +++
On 08/07/14 03:43, Bin Cheng wrote:
Hi,
The case depends on GCC inlining of global function, but the callee won't be
inlined because it's global function and considered over-writable during
run-time.
This patch fixes the failure by binding it to pic locally. Is it OK?
Thanks,
bin
gcc/testsuit
On Wed, Aug 13, 2014 at 07:40:57PM +0100, Jonathan Wakely wrote:
> This should fix the Hurd bug in PR61841 as well as the problem
> described in https://gcc.gnu.org/ml/gcc/2013-09/msg00196.html
>
> Tested x86_64-linux, committed to trunk.
>
> This is not suitable for the branches as it requires a
On 08/13/14 12:39, Jeff Law wrote:
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* combine.c (i2mod): Strengthen this variable from rtx to rtx_insn *.
(struct reg_stat_struct): Likewise for fields "last_death",
"last_set".
(subst_insn): Likewise for this variable.
(added_links_ins
This patch causes the std::thread constructor that launches a new
thread to odr-use pthread_create on glibc-based targets, so that the
program depends directly on pthread_create (and must be linked to
libpthread) not just on the weak symbol in gthr-posix.h
This should fix the Hurd bug in PR61841
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* combine.c (i2mod): Strengthen this variable from rtx to rtx_insn *.
(struct reg_stat_struct): Likewise for fields "last_death", "last_set".
(subst_insn): Likewise for this variable.
(added_links_insn): Likewise.
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* cfgrtl.c (can_delete_note_p): Require a const rtx_note * rather
than a const_rtx.
(can_delete_label_p): Require a const rtx_code_label * rather than
a const_rtx.
(delete_insn): Add checked cast to rtx_code_labe
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* basic-block.h (flow_find_cross_jump): Strengthen params 3 and 4
"f1" and "f2" from rtx * to rtx_insn **.
(flow_find_head_matching_sequence): Likewise.
* cfgcleanup.c (try_simplify_condjump): Strengthen local
"
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* bb-reorder.c (copy_bb_p): Strengthen local "insn" from rtx to
rtx_insn *.
(get_uncond_jump_length): Likewise for locals "label", "jump".
(fix_up_crossing_landing_pad): Likewise for locals "new_label",
"jump", "
On 08/06/14 11:20, David Malcolm wrote:
Note to self: verified the compile on pdp11-aout
gcc/
* auto-inc-dec.c (struct inc_insn): Strengthen field "insn" from
rtx to rtx_insn *.
(struct mem_insn): Likewise for field "insn".
(reg_next_use): Strengthen from rtx * to
On 08/06/14 11:20, David Malcolm wrote:
This is an example of strengthening rtx. For example, we
now have strong enough types provided by the existing scaffolding to
turn "insn" and "insns" in this:
for (insn = insns; insn; insn = NEXT_INSN (insn))
from plain rtx into rtx_insn *.
gcc/
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* alias.c (init_alias_analysis): Strengthen local "insn" from rtx
to rtx_insn *.
OK.
jeff
On 08/06/14 11:20, David Malcolm wrote:
This commit is a placeholder for me when rebasing, to help organize the
patch kit.
/
* rtx-classes-status.txt: Update
OK.
jeff
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* rtl.h (delete_related_insns): Strengthen return type from rtx to
rtx_insn *.
* jump.c (delete_related_insns): Likewise, also for locals "next"
and "prev".
OK.
Jeff
This is a ping on a patch I sent out a while ago to fix the GCC build
when building with the mingw toolset.
The problem was that crtstuff.c would not compile because it saw two
different (incompatible) definitions of caddr_t, one coming from
auto-host.h (set by the configure script) and one coming
On 08/06/14 11:20, David Malcolm wrote:
(define_bypass) clauses in .md files can specify the name of a guard
function as their final operand. Currently these functions are called
with a pair of rtx. This patch strengthens insn-automata.c so that such
guard functions are passed a pair of rtx_ins
On 08/06/14 11:20, David Malcolm wrote:
Strengthen "insn" from rtx to rtx_insn * within the generated get_attr_
functions in insn-attrtab.c, without imposing a strengthening from rtx
to rtx_insn * on the param itself and thus the callers.
gcc/
* genattrtab.c (write_attr_get): Within the
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* output.h (peephole): Strengthen return type from rtx to rtx_insn *.
* rtl.h (delete_for_peephole): Likewise for both params.
* genpeep.c (main): In generated "peephole" function, strengthen
return type and local "insn"
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* rtl.h (try_split): Strengthen return type from rtx to rtx_insn *.
* emit-rtl.c (try_split): Likewise, also for locals "before" and
"after". For now, don't strengthen param "trial", which requires
adding checked casts
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* debug.h (struct gcc_debug_hooks): Strengthen param 1 of hook
"label" from rtx to rtx_code_label *. Strengthen param 1 o
"var_location" hook from rtx to rtx_insn *.
(debug_nothing_rtx): Delete in favor of...
(d
On 08/06/14 11:20, David Malcolm wrote:
With this patch, "insn" and "curr_insn" as used from C++ fragments in .md
files are strengthened from rtx to rtx_insn *, allowing numerous
target-specific functions to have their params be similiar strengthened.
The top-level interfaces ("recog", "split",
[ dup sorry ]
On Aug 13, 2014, at 5:12 AM, Felix Yang wrote:
> The qsort library function may have different behavior on
> different hosts (say Linux vs MinGW).
> OK for trunk?
I just have one question. Why if we want to use a stable sort, and we program
in C++, would we not want to just us
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* sel-sched-ir.h (create_insn_rtx_from_pattern): Strengthen return
type from rtx to rtx_insn *.
* sel-sched-ir.h (create_copy_of_insn_rtx): Likewise.
* sel-sched-ir.c (create_insn_rtx_from_pattern): Likewise.
* s
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* rtl.h (find_first_parameter_load): Strengthen return type from
rtx to rtx_insn *.
* rtlanal.c (find_first_parameter_load): Strengthen return type
from rtx to rtx_insn *. Add checked cast for now, to postpone
s
On Aug 13, 2014, at 5:12 AM, Felix Yang wrote:
>The qsort library function may have different behavior on
> different hosts (say Linux vs MinGW).
> OK for trunk?
I just have one question. Why if we want to use a stable sort, and we program
in C++, would we not want to just use C++:
http
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* sel-sched-ir.h (exit_insn): Strengthen from rtx to rtx_insn *.
(sel_bb_head): Strengthen return type insn_t (currently just an
rtx) to rtx_insn *.
(sel_bb_end): Likewise.
* sel-sched-ir.c (exit_insn): Strength
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* basic-block.h (get_last_bb_insn): Strengthen return type from
rtx to rtx_insn *.
* cfgrtl.c (get_last_bb_insn): Likewise, and for locals "tmp" and
end".
OK.
jeff
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* rtl.h (unlink_insn_chain): Strengthen return type from rtx to
rtx_insn *.
(duplicate_insn_chain): Likewise.
* cfgrtl.c (unlink_insn_chain): Strengthen return type from rtx to
rtx_insn *, also for locals "prevfi
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* rtl.h (next_cc0_user): Strengthen return type from rtx to
rtx_insn *.
(prev_cc0_setter): Likewise.
* emit-rtl.c (next_cc0_user): Strengthen return type from rtx to
rtx_insn *, adding checked casts for now as n
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* expr.h (emit_move_insn): Strengthen return type from rtx to
rtx_insn *.
(emit_move_insn_1): Likewise.
(emit_move_complex_push): Likewise.
(emit_move_complex_parts): Likewise.
* expr.c (emit_move_via_in
On 08/06/14 11:20, David Malcolm wrote:
More scaffolding: strengthen the return types from the various emit_
functions from rtx to rtx_insn * (or to the rtx_barrier * subclass in a
few cases).
These will ultimately have their params strengthened also, but we
postpone that until much later in the
On 08/13/2014 10:28 AM, Richard Biener wrote:
Sofar the patch survived building stage2 in a LTO bootstrap on the
4.9 branch, full testing is scheduled for trunk.
The patch breaks a lot of C++ testcases, such as
g++.old-deja/g++.other/cvt1.C; I think you need to share the "set the
canonical typ
On Wed, Aug 13, 2014 at 4:23 PM, Kirill Yukhin wrote:
>> >> In a couple of places, checks are changed to GET_MODE_SIZE
>> >> (GET_MODE (operands[0])) and in a similar way, mode checks were
>> >> changed to GET_MODE (operands[0]). The previous checks are more
>> >> efficient and are preferred. Is
On 08/13/14 11:08, David Malcolm wrote:
On Wed, 2014-08-13 at 07:44 -0600, Jeff Law wrote:
On 08/06/14 11:20, David Malcolm wrote:
gcc/
* function.h (struct rtl_data): Strengthen field
"x_parm_birth_insn" from rtx to rtx_insn *.
* function.c (struct assign_parm_data_all)
On Wed, 2014-08-13 at 07:44 -0600, Jeff Law wrote:
> On 08/06/14 11:20, David Malcolm wrote:
> > gcc/
> > * function.h (struct rtl_data): Strengthen field
> > "x_parm_birth_insn" from rtx to rtx_insn *.
> > * function.c (struct assign_parm_data_all): Strengthen fields
> > "first_con
OK.
Jason
On 08/06/14 10:21, David Malcolm wrote:
gcc/
* config/microblaze/microblaze.c (microblaze_call_tls_get_addr):
Strengthen return type and local "insns" from rtx to rtx_insn *.
(microblaze_legitimize_tls_address): Likewise for local "insns".
(microblaze_block_move_lo
Hi,
Here is the updated patch. offload_funcs/vars are now declared in omp-low.h,
the functions have a comment. Also it fixes the issue of offload_funcs/vars
corruption by the garbage collector. OK for gomp-4_0-branch?
-- Ilya
---
gcc/Makefile.in|1 +
gcc/gengtype.c |
On 2014-08-13, 4:36 AM, James Greenhalgh wrote:
On Tue, Aug 12, 2014 at 04:53:38PM +0100, pins...@gmail.com wrote:
On Aug 12, 2014, at 7:40 AM, Alan Lawrence wrote:
(It is no more expensive.)
Yes on some processors it could be.
Haven't we been here before [1]?
This disparaging mechanis
Ian Lance Taylor wrote:
> On Tue, Aug 12, 2014 at 10:11 AM, Gary Benson wrote:
> > Ian Lance Taylor wrote:
> > > I think that by default the program should stop. That will make
> > > it possible to eventually run as part of "make check". Give it
> > > some number of iterations that stops it in a
On 07/05/14 20:59, Chen Gang wrote:
* microblaze/microblaze.md: Remove redundant '@' to avoid compiling
warning.
The related warning:
../../gcc/gcc/config/microblaze/microblaze.md:516: '@' is redundant for
output template with single alternative
Signed-off-by: Chen Gang
---
gcc/
On 07/08/14 20:32, Richard Henderson wrote:
On 08/07/2014 02:57 AM, Kyrill Tkachov wrote:
+ if (one_match > zero_match)
+{
+ /* Set either first three quarters or all but the third. */
+ mask = 0xll << (16 - first_not__match);
+ emit_insn (gen_rtx_SET (VOIDmode, des
1 - 100 of 178 matches
Mail list logo
