Paul Hua 于2018年11月30日周五 下午8:46写道：
>
> In some older Loongson3 processors there is a LL/SC errata that can
> cause the CPU to deadlock occasionally. The details are very
> complicated. We find a way to work around this errata by a) adding a
> sync before ll/lld instruction, b) adding a sync
>
On Mon, Dec 03, 2018 at 01:26:48AM +, Iain Sandoe wrote:
> Hi,
>
> Recent changes to the lowerings for call handling in rs6000 have broken
> bootstrap for powerpc-darwin.
>
> I’m proposing two patches;
> 1. the minimal fix to bootstrap - which I believe also fixes some possible
> typos in
On Wed, 31 Oct 2018 at 04:26, Joseph Myers wrote:
>
> This patch (diffs to generated files omitted below) updates GCC to use
> autoconf 2.69 and automake 1.15.1. (That's not the latest automake
> version, but it's the one used by binutils-gdb, with which consistency
> is desirable, and in any
Hi,
The second patch normalises Darwin’s call handling to follow the same pattern
that Alan introduced recently.
Hopefully it makes it clearer where there are differences (actually quite small
in this case, but enough to make it better to separate the lowering). It isn’t
needed to fix
Hi,
Recent changes to the lowerings for call handling in rs6000 have broken
bootstrap for powerpc-darwin.
I’m proposing two patches;
1. the minimal fix to bootstrap - which I believe also fixes some possible
typos in the patterns for sysv.
2. a second patch that brings Darwin into line with
Hi,
GCC does not export construction vtable symbols from shared libraries**.
The symbols are marked hidden in the objects; for Darwin that makes them also
external (“private_extern” is Darwin’s hidden) which means that they show up in
the list of possible symbols for export from libstdc++, and
TFmode is moved differently than XFmode from a while ago.
2018-12-02 Uros Bizjak
* config/i386/i386.c (inline_memory_move_cost): Correctly
compute number of TFmode GPR moves.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
Committed to mainline SVN.
Uros.
Index:
On Sun, 2 Dec 2018, Iain Buclaw wrote:
> I assume that the notes on the about.html page are all that are
> needed to commit and push?
Yes, they *should*.
If you run into anything that's missing or have any questions,
just let me know and I'll (a) help directly and (b) see how to
improve the
On Sun, Dec 02, 2018 at 05:26:32PM +0100, Dominique d'Humières wrote:
> Hi Andi,
>
> % gcc9 -S -pg -mfentry -minstrument-return=call -mrecord-return
> /opt/gcc/work/gcc/testsuite/gcc.target/i386/returninst1.c -m32
> cc1: sorry, unimplemented: -mfentry isn't supported for 32-bit in combination
>
On Sun, 2 Dec 2018 at 18:23, Gerald Pfeifer wrote:
>
> Hi Iain,
>
> On Mon, 12 Nov 2018, Iain Buclaw wrote:
> > As suggested, this adds an announcement of the D front end addition
> > to the news items on the GCC home page, and from what I can tell, the
> > relevant pages where the language
On Sun, Dec 2, 2018 at 8:48 PM Jakub Jelinek wrote:
>
> On Sun, Dec 02, 2018 at 08:23:21PM +0100, Uros Bizjak wrote:
> > OK with a constraint adjustment below.
> >
> > > +(define_insn "*vec_concatv4sf_0"
> > > + [(set (match_operand:V4SF 0 "register_operand" "=v")
> > > +
On Sun, Dec 02, 2018 at 08:23:21PM +0100, Uros Bizjak wrote:
> OK with a constraint adjustment below.
>
> > +(define_insn "*vec_concatv4sf_0"
> > + [(set (match_operand:V4SF 0 "register_operand" "=v")
> > + (vec_concat:V4SF
> > + (match_operand:V2SF 1 "nonimmediate_operand"
On Fri, Nov 30, 2018 at 10:14 PM Jakub Jelinek wrote:
>
> Hi!
>
> As mentioned in the PR, while in vec_concatv2df or vec_concatv2di we have
> alternatives where the lower half is low 64-bit part of a xmm reg or 64-bit
> memory and upper half is zero using movq/vmovq (or movq2dq), for
>
On Sun, 2 Dec 2018, Segher Boessenkool wrote:
what is the point of !!count when we take the max with 1 on the very
next line? Is it in prevision of a time when we may remove the MAX? (sorry
if this was covered in previous iterations)
By the way, not related to the patch, but I wonder why we
When a large number of bytes are pushed in a function call sequence on
the H8, the call patterns to fail to match because they expect the
operand to be a valid HImode operand -- even when pointers are SImode.
The fact that the operand is expected to be in HImode is a bit of a
historical wart
Hi!
On Sun, Dec 02, 2018 at 06:23:23PM +0100, Marc Glisse wrote:
> On Sun, 2 Dec 2018, Segher Boessenkool wrote:
>
> >diff --git a/gcc/tree-inline.c b/gcc/tree-inline.c
> >index 5aa782b..7e9ed99 100644
> >--- a/gcc/tree-inline.c
> >+++ b/gcc/tree-inline.c
> >@@ -4109,6 +4109,9 @@
On Sun, 2 Dec 2018, Segher Boessenkool wrote:
diff --git a/gcc/tree-inline.c b/gcc/tree-inline.c
index 5aa782b..7e9ed99 100644
--- a/gcc/tree-inline.c
+++ b/gcc/tree-inline.c
@@ -4109,6 +4109,9 @@ estimate_num_insns (gimple *stmt, eni_weights *weights)
with very long asm statements.
Hi Iain,
On Mon, 12 Nov 2018, Iain Buclaw wrote:
> As suggested, this adds an announcement of the D front end addition
> to the news items on the GCC home page, and from what I can tell, the
> relevant pages where the language should get a mention.
I noticed this hasn't gone in; apologies for
The Linux kernel people want a feature that makes GCC pretend some
inline assembler code is tiny (while it would think it is huge), so
that such code will be inlined essentially always instead of
essentially never.
This patch lets you say "asm inline" instead of just "asm", with the
result that
PR55681 observes that currently only one qualifier is allowed for
inline asm, so that e.g. "volatile asm" is allowed, "const asm" is also
okay (with a warning), but "const volatile asm" gives an error. Also
"goto" has to be last.
This patch changes things so that only "asm-qualifiers" are
Hi!
v2, with the input from Joseph taken into account.
This is the same "asm inline" patch as before, but now preceded by a
patch that makes all orderings of volatile/goto/inline valid, all other
type qualifiers invalid, all repetitions of qualifiers invalid.
Tested on powerpc64-linux
Hi Andi,
% gcc9 -S -pg -mfentry -minstrument-return=call -mrecord-return
/opt/gcc/work/gcc/testsuite/gcc.target/i386/returninst1.c -m32
cc1: sorry, unimplemented: -mfentry isn't supported for 32-bit in combination
with -fpic
The tests should be protected.
TIA
Dominique
Hi Wilco,
On Thu, Nov 08, 2018 at 01:33:19PM +, Wilco Dijkstra wrote:
> > But the max. error in sinh/cosh/atanh is less than 2 ULP, with some math
> > libraries. It could be < 1 ULP, in theory, so sinh(atanh(x)) less than
> > 2 ULP even.
>
> You can't add ULP errors in general - a tiny
On Sat, Dec 01, 2018 at 07:11:08PM -0500, Jason Merrill wrote:
> > On the following testcase, build_conditional_expr_1 tries hard to make sure
> > that if both arguments are xvalue_p (or one is and the other throw) the
> > result is still xvalue_p. But, later on we call unary_complex_lvalue,
> >
Hi!
As promised, this patch adds the missing clauses handling to tree-nested.c
and also handling of clauses for taskgroup construct.
Bootstrapped/regtested on powerpc64{,-le}-linux, committed to trunk.
2018-12-02 Jakub Jelinek
* tree-nested.c (convert_nonlocal_omp_clauses,
Hi!
Taskloop cancellation which OpenMP 5.0 explicitly allows worked fine on the
library side and on the compiler side if the cancel taskgroup or
cancellation point taskgroup is inside of explicit task and there is no
outer parallel/teams etc. in that function, taskloop is in another function
that
Hi!
On Sat, Dec 01, 2018 at 08:36:41AM +0100, Jakub Jelinek wrote:
> I'm still seeing some regressions, ICEs like:
> FAIL: c-c++-common/asan/clone-test-1.c -O1 (internal compiler error)
> FAIL: c-c++-common/asan/clone-test-1.c -O1 (test for excess errors)
And here is a fix for that. It
On Sat, 1 Dec 2018 at 19:28, Iain Buclaw wrote:
>
> On Thu, 29 Nov 2018 at 15:12, Rainer Orth
> wrote:
> >
> > Hi Iain,
> >
> > > On Tue, 27 Nov 2018 at 20:32, Rainer Orth
> > > wrote:
> > >>
> > >> Hi Mike,
> > >>
> > >> > On Nov 27, 2018, at 2:18 AM, Rainer Orth
> > >> >
> > >> > wrote:
>
Hi,
This patch merges the D front-end implementation with dmd upstream 5220ad51e.
Backports Ddoc fix that is present in upstream, but did not make its
way into the C++ port of the D front-end implementation.
The old special types for C long, unsigned long, and long double have
also been removed
On Wed, 14 Nov 2018, claz...@gmail.com wrote:
> I've just committed the attached patch containing the news for the ARC
> backend.
Thanks!
> LRA is now on by default for the ARC target. This can be
> controlled by -mlra.
May that make it sound a bit like one still needs to specify -mlra
to
Applied.
Gerald
Index: readings.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/readings.html,v
retrieving revision 1.307
diff -u -r1.307 readings.html
--- readings.html 1 Dec 2018 19:17:56 - 1.307
+++ readings.html
31 matches
Mail list logo
