On Wed, Dec 06, 2017 at 04:32:38PM -0600, Kelvin Nilsen wrote:
>
> I would like to backport the following patch to the GCC 7 branch.
>
> PR80101: Fix ICE in store_data_bypass_p
> https://gcc.gnu.org/ml/gcc-patches/2017-04/msg00953.html
>
>
> This patch has been bootstrapped and regression tes
On Tue, Dec 05, 2017 at 05:57:37PM +, Richard Sandiford wrote:
> Three related regression fixes:
>
> - We can't use asserts like:
>
> gcc_assert (GET_MODE_SIZE (mode) == 16);
>
> in aarch64_print_operand because it could trigger for invalid user input.
>
> - The output_operand_lossage
On Fri, Sep 29, 2017 at 09:29:37PM +0100, Steve Ellcey wrote:
> On Thu, 2017-09-28 at 12:31 +0100, Szabolcs Nagy wrote:
> >
> > i think this should be improved, see below.
>
> diff --git a/libatomic/Makefile.am b/libatomic/Makefile.am
> index d731406..92d19c6 100644
> --- a/libatomic/Makefile.am
Hi,
This is the overall loop interchange patch on gimple-linterchange branch. Note
the new pass
is enabled at -O3 level by default. Bootstrap and regtest on x86_64 and
AArch64(ongoing).
NOte after cost model change it is now far more conservative than original
version. It only
interchanges 11
On Thu, 7 Dec 2017, Martin Jambor wrote:
> Hi,
>
> On Wed, Dec 06 2017, Richard Biener wrote:
> > On Wed, 6 Dec 2017, Martin Jambor wrote:
>
> >> ...
>
> >> Second is the testcase I described in my previous email. When I saw
> >>
> >> FAIL: gcc.dg/guality/pr54970.c -O1 line 31 a[0] == 4
On Wed, Dec 6, 2017 at 4:11 PM, Richard Sandiford
wrote:
> Richard Biener writes:
>> On Thu, Nov 30, 2017 at 2:18 PM, Richard Sandiford
>> wrote:
>>> Richard Sandiford writes:
Richard Biener writes:
> On Wed, Nov 29, 2017 at 12:57 PM, Richard Sandiford
> wrote:
>> It was clea
On Wed, Dec 6, 2017 at 4:22 PM, Richard Sandiford
wrote:
> This patch switches most build_vector calls over to tree_vector_builder,
> using explicit encodings where appropriate. Later patches handle
> the remaining uses of build_vector.
>
> Tested on aarch64-linux-gnu, x86_64-linux-gnu and powerp
On Wed, Dec 6, 2017 at 4:23 PM, Richard Sandiford
wrote:
> This patch makes fold-const.c operate directly on the VECTOR_CST
> encoding when folding an operation that has a single VECTOR_CST input.
>
> Tested on aarch64-linux-gnu, x86_64-linux-gnu and powerpc64le-linux-gnu.
> Also spot-checked on s
On Wed, Dec 6, 2017 at 4:24 PM, Richard Sandiford
wrote:
> This patch makes fold-const.c operate directly on the VECTOR_CST
> encoding when folding an operation that has two VECTOR_CST inputs.
>
> Tested on aarch64-linux-gnu, x86_64-linux-gnu and powerpc64le-linux-gnu.
> Also spot-checked on sparc
On Wed, Dec 6, 2017 at 4:26 PM, Richard Sandiford
wrote:
> After the previous patches, the only remaining uses of build_vector are
> in the selftests in tree.c. This patch makes it static and moves it to
> the selftest part of the file.
>
> Tested on aarch64-linux-gnu, x86_64-linux-gnu and powerp
On Wed, Dec 6, 2017 at 4:25 PM, Richard Sandiford
wrote:
> This patch changes gimple_build_vector so that it takes a
> tree_vector_builder instead of a size and a vector of trees.
>
> Tested on aarch64-linux-gnu, x86_64-linux-gnu and powerpc64le-linux-gnu.
> Also spot-checked on sparc64-linux-gnu.
On Wed, Dec 6, 2017 at 4:28 PM, Richard Sandiford
wrote:
> This patch makes various bits of code operate directly on the new
> VECTOR_CST encoding, instead of using VECTOR_CST_ELT on all elements
> of the vector.
>
> Previous patches handled operations that produce a new VECTOR_CST,
> while this p
On Wed, Dec 6, 2017 at 5:36 PM, Will Schmidt wrote:
> Hi,
> Add support for gimple folding of splat_u{8,16,32}.
> Testcase coverage is primarily handled by existing tests
> testsuite/gcc.target/powerpc/fold-vec-splat_*.c
>
> One new test added to verify we continue to receive
> an 'invalid argumen
On Thu, Dec 07, 2017 at 12:21:03PM +0100, Richard Biener wrote:
> > +if ( TREE_CODE (arg0) != INTEGER_CST)
Also watch formatting, the space after ( doesn't belong there.
> > + return false;
>
> Is there a reason to not do this for non-constants? (even not for
> float constants?
On Thu, Dec 7, 2017 at 12:26 PM, Jakub Jelinek wrote:
> On Thu, Dec 07, 2017 at 12:21:03PM +0100, Richard Biener wrote:
>> > +if ( TREE_CODE (arg0) != INTEGER_CST)
>
> Also watch formatting, the space after ( doesn't belong there.
>
>> > + return false;
>>
>> Is there a reason to
On Thu, Dec 7, 2017 at 11:28 AM, Bin Cheng wrote:
> Hi,
> This is the overall loop interchange patch on gimple-linterchange branch.
> Note the new pass
> is enabled at -O3 level by default. Bootstrap and regtest on x86_64 and
> AArch64(ongoing).
> NOte after cost model change it is now far mor
Hi!
On Wed, Dec 06, 2017 at 10:36:58AM -0600, Will Schmidt wrote:
> Add support for gimple folding of splat_u{8,16,32}.
> Testcase coverage is primarily handled by existing tests
> testsuite/gcc.target/powerpc/fold-vec-splat_*.c
>
> One new test added to verify we continue to receive
> an 'in
On Thu, Dec 7, 2017 at 11:39 AM, Richard Biener
wrote:
> On Thu, Dec 7, 2017 at 11:28 AM, Bin Cheng wrote:
>> Hi,
>> This is the overall loop interchange patch on gimple-linterchange branch.
>> Note the new pass
>> is enabled at -O3 level by default. Bootstrap and regtest on x86_64 and
>> AAr
On Thu, Dec 7, 2017 at 12:55 PM, Bin.Cheng wrote:
> On Thu, Dec 7, 2017 at 11:39 AM, Richard Biener
> wrote:
>> On Thu, Dec 7, 2017 at 11:28 AM, Bin Cheng wrote:
>>> Hi,
>>> This is the overall loop interchange patch on gimple-linterchange branch.
>>> Note the new pass
>>> is enabled at -O3 le
We limit the amount of copying for jump threading based on counting
stmts. This counting is overly pessimistic, because we will very
often delete stmts as a consequence of jump threading: when the final
conditional jump of a block is removed, earlier SSA names computed
exclusively for use in that
There's an unfortunate interaction between -save-temps=obj and
-fdump-tree- when subdirectories are in play.
Consider:
g++ -fdump-tree-all -save-temps=obj -c -o tgt/x.o sub/x.cc
we end up with a bunch of errors of the form:
sub/x.cc:1:0: error: could not open dump file
‘tgt/tgt/x.046t.profil
On Thu, Dec 7, 2017 at 1:04 PM, Alexandre Oliva wrote:
> We limit the amount of copying for jump threading based on counting
> stmts. This counting is overly pessimistic, because we will very
> often delete stmts as a consequence of jump threading: when the final
> conditional jump of a block is
Thanks for the review. Second revision of the patch attached (also
rebased on the current trunk). Issues that are not commented on
below have been resolved as suggested.
David Malcolm writes:
> To my naive eyes this seems like a useful addition, but I'm hoping
> someone with more knowledge of t
This reverts the fix for PR67769 and installs an alternate fix
which is more targeted. This helps preserving range info where
important (int this case for emitting a -Wstringop-overflow
warning).
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2017-12-07 Richard Biener
Hi Martin,
On 6 December 2017 at 00:51, Jeff Law wrote:
> On 12/05/2017 04:47 PM, Martin Sebor wrote:
>> PR middle-end/82646 - bogus -Wstringop-overflow with
>> -D_FORTIFY_SOURCE=2 on strncpy with range to a member array,
>>
>> The bug points out a false positive in a call to strncpy() when
>> _
Hi,
On Mon, 20 Nov 2017, Richard Biener wrote:
> > +static void
> > +fix_loop_structure (struct loop *loop, struct loop *old)
>
> We have a function with the same name in loop-init.c so please use a
> different name.
Renamed to merge_loop_tree.
> > + for (gsi = gsi_start_bb (bb); !gsi_end_p
On Thu, Dec 7, 2017 at 3:11 PM, Michael Matz wrote:
> Hi,
>
> On Mon, 20 Nov 2017, Richard Biener wrote:
>
>> > +static void
>> > +fix_loop_structure (struct loop *loop, struct loop *old)
>>
>> We have a function with the same name in loop-init.c so please use a
>> different name.
>
> Renamed to m
Hello,
The following patch improves both the speed and code size for 64 bit
unsigned max for RL78:
it emits a library function call instead of emitting code for the 64 bit
max for every single time.
The unsigned max function which was added in libgcc is hand written, so more
optimal than what GCC
On Wed, Dec 6, 2017 at 9:11 PM, Jeff Law wrote:
> On 11/13/2017 05:04 PM, Richard Sandiford wrote:
>> Richard Sandiford writes:
>>> Richard Sandiford writes:
This patch adds a new "poly_int" class to represent polynomial integers
of the form:
C0 + C1*X1 + C2*X2 ... + Cn*Xn
Hi,
I'm currently debugging a problem in lra, and got a bit lost in the 20k+
lines dump file.
I observed that:
- the lra dump file is one of the biggest ones
- lra itself consists of a looping of sub-passes.
So, I've:
- written a dump infrastructure addition that can be used within a pass
t
On 12/07/2017 07:46 AM, Richard Biener wrote:
> On Wed, Dec 6, 2017 at 9:11 PM, Jeff Law wrote:
>> On 11/13/2017 05:04 PM, Richard Sandiford wrote:
>>> Richard Sandiford writes:
Richard Sandiford writes:
> This patch adds a new "poly_int" class to represent polynomial integers
> of
On Thu, 2017-12-07 at 12:21 +0100, Richard Biener wrote:
> On Wed, Dec 6, 2017 at 5:36 PM, Will Schmidt
> wrote:
> > Hi,
> > Add support for gimple folding of splat_u{8,16,32}.
> > Testcase coverage is primarily handled by existing tests
> > testsuite/gcc.target/powerpc/fold-vec-splat_*.c
> >
> >
Hello,
The following patch improves both the speed and code size for 64 bit signed
max for RL78:
it emits a library function call instead of emitting code for the 64 bit
max for every single time.
The signed max function which was added in libgcc is hand written, so more
optimal than what GCC gen
On Thu, 2017-12-07 at 09:56 +, James Greenhalgh wrote:
>
> One obvious thing I missed in the review is that this change will break
> bootstrap on systems with older assemblers. Practically, that's those of
> us who are holding out on Ubuntu 14.04. -march=armv8-a+lse would go back
> a little fu
Hi!
save_template_attributes ignored flags, when ATTR_FLAG_TYPE_IN_PLACE
wasn't set on a type, it would happily attach the attributes to some
existing type (in this case to integer_type_node).
My first approach was to just call build_type_attribute_variant, but
that ICEs on g++.dg/cpp0x/alias-dec
Hi!
As mentioned in the PR, the code emitted by ix86_expand_rint
doesn't work with rounding to +/- infinity.
This patch adjusts it if flag_rounding_math to do something that works
well even for that case (should be just one insn longer).
Bootstrapped/regtested on x86_64-linux and i686-linux, ok f
On Wed, Dec 06, 2017 at 05:30:53PM +0100, Jakub Jelinek wrote:
> On Wed, Dec 06, 2017 at 09:20:15AM -0700, Martin Sebor wrote:
> > Attached is a patch with the comment updated/simplified.
> > The tests do the job they need to do today so I just removed
> > the useless attribute but otherwise left t
GCC Maintainers:
The following patch adds support for missing builtin instances
documented in the ABI specification. The patch includes test cases for
missing builtins.
The patch has been run on:
powerpc64le-unknown-linux-gnu (Power 8 LE)
powerpc64le-unknown-linux-gnu (Power 8 BE)
power
Hi!
When committing the previous PR81281 patch, I've removed all the @@0 cases
on plus:c, used @0 instead, to make sure we don't regress.
This patch readds those where possible. For the cases where there is
just P and A, it was mostly a matter of @@0 and convert? instead of convert
plus using ty
Hi!
As mentioned in the PR, we need to treat POINTER_DIFF_EXPR similarly
how we treat comparisons with pointer operands (i.e. that either
there is useless conversion from one type to the other, or from the
other to the one, or the pointer type modes are the same), because
we have exceptions for po
On Thu, Dec 7, 2017 at 5:48 PM, Jakub Jelinek wrote:
> Hi!
>
> As mentioned in the PR, the code emitted by ix86_expand_rint
> doesn't work with rounding to +/- infinity.
> This patch adjusts it if flag_rounding_math to do something that works
> well even for that case (should be just one insn long
On 12/07/2017 05:29 AM, Boris Kolpackov wrote:
Thanks for the review. Second revision of the patch attached (also
rebased on the current trunk). Issues that are not commented on
below have been resolved as suggested.
David Malcolm writes:
To my naive eyes this seems like a useful addition, b
The following patch fixes PR83252 and PR80818.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83252
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80818
The patch was successfully tested and bootstrapped on ppc64 and x86_64.
Committed as rev. 255471.
Jakub, the patch does not include a test cas
On Wed, Dec 6, 2017 at 1:29 PM, Richard Biener wrote:
>
> The following fixes a vectorization issue that appears when trying
> to vectorize the bwaves mat_times_vec kernel after interchange was
> performed by the interchange pass. That interchange inserts the
> following code for the former reduc
Hello,
The following patch improves both the speed and code size for 64 bit signed
min for RL78:
it emits a library function call instead of emitting code for the 64 bit
min for every single time.
The signed min function which was added in libgcc is hand written, so more
optimal than what GCC gen
On December 7, 2017 5:55:47 PM GMT+01:00, Jakub Jelinek
wrote:
>On Wed, Dec 06, 2017 at 05:30:53PM +0100, Jakub Jelinek wrote:
>> On Wed, Dec 06, 2017 at 09:20:15AM -0700, Martin Sebor wrote:
>> > Attached is a patch with the comment updated/simplified.
>> > The tests do the job they need to do t
On 12/07/2017 09:55 AM, Jakub Jelinek wrote:
On Wed, Dec 06, 2017 at 05:30:53PM +0100, Jakub Jelinek wrote:
On Wed, Dec 06, 2017 at 09:20:15AM -0700, Martin Sebor wrote:
Attached is a patch with the comment updated/simplified.
The tests do the job they need to do today so I just removed
the use
When the testcases were regenerated due to PR libstdc++/PR66689,
-ffp-contract=off was omitted from (at least) the hyperg
check_value.cc testcases, causing runtime regression. This patch
restores the dg-options directive.
This returns the libstdc++ testsuite to all passing on AIX.
Bootstrapped o
On 12/07/2017 11:03 AM, Martin Sebor wrote:
> On 12/07/2017 09:55 AM, Jakub Jelinek wrote:
>> On Wed, Dec 06, 2017 at 05:30:53PM +0100, Jakub Jelinek wrote:
>>> On Wed, Dec 06, 2017 at 09:20:15AM -0700, Martin Sebor wrote:
Attached is a patch with the comment updated/simplified.
The tests
C11 DR#444 dealt with how C11 intended to allow alignment specifiers
on struct and union members, but failed to include that in the syntax.
The final resolution of that DR also allows alignment specifiers in
type names in compound literals (in order to apply an increased
alignment to the unnamed ob
On 12/07/2017 06:46 AM, Christophe Lyon wrote:
Hi Martin,
On 6 December 2017 at 00:51, Jeff Law wrote:
On 12/05/2017 04:47 PM, Martin Sebor wrote:
PR middle-end/82646 - bogus -Wstringop-overflow with
-D_FORTIFY_SOURCE=2 on strncpy with range to a member array,
The bug points out a false pos
On 07/12/17 13:13 -0500, David Edelsohn wrote:
When the testcases were regenerated due to PR libstdc++/PR66689,
-ffp-contract=off was omitted from (at least) the hyperg
check_value.cc testcases, causing runtime regression. This patch
restores the dg-options directive.
Ah yes, good catch.
We m
GCC Maintainers:
The following patch adds missing test cases for various builtins. The
patch does not make any functional changes to GCC.
The patch has been run on:
powerpc64le-unknown-linux-gnu (Power 8 LE)
powerpc64le-unknown-linux-gnu (Power 8 BE)
powerpc64le-unknown-linux-gnu (Power
On 11/25/2017 05:53 PM, Martin Sebor wrote:
> On 11/22/2017 04:50 PM, Jeff Law wrote:
>> On 11/16/2017 02:29 PM, Martin Sebor wrote:
On 10/23/2017 08:42 PM, Martin Sebor wrote:
> Attached is a reworked solution to enhance -Wrestrict while
> avoiding changing tree-vrp.c or any other VRP
On 11/27/2017 05:44 AM, Richard Biener wrote:
>
> +
> + if (const strinfo *chksi = olddsi ? olddsi : dsi)
> +if (si
> + && !check_bounds_or_overlap (stmt, chksi->ptr, si->ptr, NULL_TREE,
> len))
> + /* Avoid transforming strcpy when out-of-bounds offsets or
> +overlapping
On Thu, 7 Dec 2017, Jakub Jelinek wrote:
When committing the previous PR81281 patch, I've removed all the @@0 cases
on plus:c, used @0 instead, to make sure we don't regress.
This patch readds those where possible. For the cases where there is
just P and A, it was mostly a matter of @@0 and co
On 12/06/17 23:35, Jason Merrill wrote:
> On Fri, Dec 1, 2017 at 7:42 AM, Bernd Edlinger
> wrote:
>> this version of the patch improves the heuristic check to take the
>> target hook into account, to handle cases correctly when both or only
>> one parameter is _not_ promoted to int.
>
> In lookin
On Jul 7, 2017, Segher Boessenkool wrote:
> I meant, just double check if your new
> code does the correct thing for the set count.
Sorry this took me so long to get back to. Even this was difficult for
me to answer for sure, then and now.
We don't (and can't?) know whether the REG_UNUSED not
On Apr 7, 2017, Alexandre Oliva wrote:
> On Mar 21, 2017, Alexandre Oliva wrote:
>> On Jan 27, 2017, Alexandre Oliva wrote:
>>> On Oct 19, 2016, Alexandre Oliva wrote:
On Sep 23, 2016, Alexandre Oliva wrote:
> On Aug 30, 2016, Alexandre Oliva wrote:
>> Handling non-template fri
On 12/01/2017 11:06 AM, Martin Sebor wrote:
> On 12/01/2017 01:26 AM, Jeff Law wrote:
>> On 11/30/2017 01:30 PM, Martin Sebor wrote:
>>> On 11/22/2017 05:03 PM, Jeff Law wrote:
On 11/21/2017 12:07 PM, Martin Sebor wrote:
> On 11/21/2017 09:55 AM, Jeff Law wrote:
>> On 11/19/2017 04:28
On 11/30/2017 09:40 AM, Thomas Preudhomme wrote:
> Hi,
>
> Effective target fstack_protector fails to return an error for
> newlib-based target (such as arm-none-eabi targets) which does not
> support stack protector. This is due to the test being too simplist for
> stack protection code to be gen
On 11/29/2017 04:36 PM, Martin Sebor wrote:
> I've finished reimplementing the patch as a standalone pass.
> In the attached revision I also addressed your comments below
> as well as Richard's to allowing the strlen optimizations even
> for overlapping accesses.
>
> While beefing up the tests I f
On 12/07/2017 02:14 PM, Jeff Law wrote:
On 11/29/2017 04:36 PM, Martin Sebor wrote:
I've finished reimplementing the patch as a standalone pass.
In the attached revision I also addressed your comments below
as well as Richard's to allowing the strlen optimizations even
for overlapping accesses.
On 12/07/2017 02:28 PM, Martin Sebor wrote:
> On 12/07/2017 02:14 PM, Jeff Law wrote:
>> On 11/29/2017 04:36 PM, Martin Sebor wrote:
>>> I've finished reimplementing the patch as a standalone pass.
>>> In the attached revision I also addressed your comments below
>>> as well as Richard's to allowin
On 11/29/2017 04:36 PM, Martin Sebor wrote:
> I've finished reimplementing the patch as a standalone pass.
> In the attached revision I also addressed your comments below
> as well as Richard's to allowing the strlen optimizations even
> for overlapping accesses.
>
> While beefing up the tests I f
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> This patch removes unused RTL functions, introduces alternate ones for
> use in a later SFN patch, and regroups other related functions so that
> they appear in a more consistent order.
>
> for gcc/ChangeLog
>
> * emit-rtl.c (next_nondebug_i
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> This patch introduces a number of new macros and functions that will
> be used to distinguish between different kinds of debug stmts, insns
> and notes, namely, preexisting debug bind ones and to-be-introduced
> nonbind markers.
>
> In a seemingly m
Jeff Law writes:
> On 12/07/2017 07:46 AM, Richard Biener wrote:
>> On Wed, Dec 6, 2017 at 9:11 PM, Jeff Law wrote:
>>> On 11/13/2017 05:04 PM, Richard Sandiford wrote:
Richard Sandiford writes:
> Richard Sandiford writes:
>> This patch adds a new "poly_int" class to represent poly
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> This patch adjusts numerous parts of the compiler that would
> malfunction should they find debug markers at points where they may be
> introduced. The changes purport to allow the compiler to pass
> bootstrap-debug-lean (-fcompare-debug in stage3)
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> If find_bb_boundaries is given a block with zero or one nondebug insn
> beside debug insns, it shouldn't purge dead edges, because without
> debug insns we wouldn't purge them at that point. Doing so may change
> the order in which edges are process
On 12/07/2017 03:38 PM, Richard Sandiford wrote:
>> So I think that's the final ack on this series.
>
> Thanks to both of you, really appreciate it!
Sorry it took so long.
>
>> Richard S. can you confirm? I fully expect the trunk has moved some
>> and the patches will need adjustments -- consi
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> Introduce a command line option to enable statement frontiers, enabled
> by default in optimized builds with DWARF2+ debug information.
>
> This patch depends on an earlier patch that completed the
> infrastructure for debug markers, and on another
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> The inline_entry hook will be given a definition in a later patch.
>
> for gcc/ChangeLog
>
> * debug.h (gcc_debug_hooks): Add inline_entry.
> * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
> * debug.c (do_nothing_debu
On 11/30/2017 02:18 AM, Olivier Hainque wrote:
> Hello,
>
> Bootstrap is currently broken with Ada on x86_64-windows using SEH exceptions,
> with several stage2/stage3 object comparison failures like
>
> Bootstrap comparison failure!
> gcc/ada/ali-util.o differs
> gcc/ada/ali.o differs
>
* Claudiu Zissulescu [2017-11-27 12:09:55
+0100]:
> From: Claudiu Zissulescu
>
> Make sure we check the constants in all cases.
>
> gcc/
> 2017-10-14 Claudiu Zissulescu
>
> * config/arc/arc.c (arc_legitimate_constant_p): Always check all
> constants.
>
> testsuite/
> 2017-10-
* Claudiu Zissulescu [2017-11-27 12:09:56
+0100]:
> From: claziss
>
> Track which regs are set fixed/call saved/call used from commnad line.
> Do not try to override their properties if user says otherwise.
>
> gcc/
> 2017-06-08 Claudiu Zissulescu
>
> * config/arc/arc.c (overridereg
* Claudiu Zissulescu [2017-11-27 12:09:58
+0100]:
> From: claziss
>
> The accumulator registers are freely used by the compiler. However,
> there are a number of instructions which are having an intrinsic use
> of these registers. Update patterns to inform the compiler which ones.
>
> gcc/
>
James,
Here is a patch that will turn off the use of IFUNC and the LSE
instructions in libatomic if the compiler/assembler toolchain do not
understand the '-march=armv8-a+lse' option (changed from
-march=armv8.1-a). Rather than check the assembler directly, I used
the existing ACX_PROG_CC_WARNING
On 11/09/2017 07:34 PM, Alexandre Oliva wrote:
> This patch completes the infrastructure for the introduction of
> statement frontiers in C-family languages.
>
> It brings in all the code remaining code needed to introduce and
> transform begin stmt trees, gimple stmts, insns and notes, and
> ulti
So the underlying issue here is quite simple. Given something like
x = (cond) ? res1 : res2;
EVRP analysis will compute the resultant range using vrp_meet of the
ranges for res1 and res2. Seems pretty natural.
vrp_meet makes optimistic assumptions if either range is VR_UNDEFINED
and will set
On 12/07/2017 03:23 PM, Jeff Law wrote:
On 11/29/2017 04:36 PM, Martin Sebor wrote:
I've finished reimplementing the patch as a standalone pass.
In the attached revision I also addressed your comments below
as well as Richard's to allowing the strlen optimizations even
for overlapping accesses.
This adds srodata section support to the RISC-V port, putting small read-only
data in the .srodata section instead of the .sdata section. There is already
code to put small read-only rtx in .srodata* instead of .rodata*. This
does the same for small read-only trees to be consistent.
Tested with
My series of patches from earlier this fall
https://gcc.gnu.org/ml/gcc-patches/2017-10/msg01309.html
caused code size regressions in cases where constant integers are used
as addresses; specifically, LRA was getting confused and trying to
generate a stack spill because it couldn't figure out h
I previously neglected to add address splitters for the io-variant
memory read and write insns, as I did for all the other insns that take
memory operands. In the examples I looked at, I found there was still
valid code coming out, but it seemed like an accident that it was
falling through to
I've checked in these new constant-integer address test cases to go with
the previous two patches in this series.
-Sandra
2017-12-07 Sandra Loosemore
gcc/testsuite/
* gcc.target/nios2/const-addr-1.c: New.
* gcc.target/nios2/const-addr-2.c: New.
* gcc.target/nios2/const-addr-3.c: New.
Ind
Hello Julia,
On 08 Nov 12:32, Koval, Julia wrote:
> Hi, this patch enables VAESDEC instruction from VAES isaset, defined here:
> https://software.intel.com/sites/default/files/managed/c5/15/architecture-instruction-set-extensions-programming-reference.pdf
>
> Ok for trunk?
Patch is OK. I've check
86 matches
Mail list logo