Hi!
As discussed in the PR, e.g. on x86_64 (both -m32 and -m64) there is no
double-word modulo and so we expand it to a __{,u}mod[dt]i3 call.
For certain constant divisors we can do better. E.g. consider
32-bit word-size, 0x1ULL % 3 == 1, so we can use partly the Hacker's
delight modulo b
On 11/26/20 10:06 AM, Richard Sandiford wrote:
Martin Sebor writes:
I do have one concern: the tendency to prioritize efficiency
over safety (this can be said about most GCC code). Specifically
in this class, the address bit twiddling makes me uneasy. I don't
think the object model in either l
On Fri, Nov 20, 2020 at 10:38 PM Kito Cheng wrote:
> On Sat, Nov 21, 2020 at 6:12 AM Jim Wilson wrote:
> > On Fri, Nov 20, 2020 at 12:34 AM Kito Cheng
> wrote:
> >
> > > - Define MULTILIB_DEFAULTS can reduce the total number of multilib if
> > >the default arch and ABI are listed in the mu
On Thu, Nov 26, 2020 at 1:04 AM Kito Cheng wrote:
> * gcc.c (print_multilib_info): Check default arguments not
> appeared in multi-lib option list with '!'
>
OK.
Jim
On 11/26/20 8:42 AM, H.J. Lu wrote:
On Tue, Nov 24, 2020 at 5:13 PM Martin Sebor via Gcc-patches
wrote:
The attached patch cleans up most remaining -Wformat-diag instances
in an x86_64-build. I tried to minimize using #pragma diagnostic
so I tweaked the code instead. A preferable solution mi
The preprocessor check for overflow (of linenum_type = unsigned int)
when reading the line number in a #line directive is incomplete; it
checks "reg < reg_prev" which doesn't cover all cases where
multiplying by 10 overflowed. Fix this by checking for overflow
before rather than after it occurs (u
On 27.11.20 18:31, Jakub Jelinek via Fortran wrote:
Depends on what does the OpenACC standard say.
If it has similar wording to OpenMP that '!$ ' stands ...
It only has '!$acc' (free) and !$acc + c$acc + *$acc (fixed).
cf.
https://www.openacc.org/sites/default/files/inline-images/Specification
Hello
This patch still needs review.
Thanks
Kwok
On 19/11/2020 6:07 pm, Kwok Cheung Yeung wrote:
On 29/10/2020 10:03 am, Jakub Jelinek wrote:
I'm actually not sure how this can work correctly.
Let's say we have
int foo () { return 1; }
int bar () { return 2; }
int baz () { return 3; }
int qu
Hello
This is an updated version of the WIP patch for task detach support. Any
comments would be welcome!
On 11/11/2020 7:06 pm, Kwok Cheung Yeung wrote:
- No error checking at the front-end.
The detach clause is now parsed properly in C, C++ and Fortran, and will raise
an error if the syn
Hi Folks,
this patch needs C++ review
thanks
Iain
Iain Sandoe wrote:
resending - the first & second attempt didn’t seem to make it to
gcc-patches.
Hi
This option allows the user to specify alternate C++ runtime libraries,
for example when a platform uses libc++ as the installed C++ runti
Hi
I believe this version addressed Joseph’s and Richard’s comment, but it
C++ review.
thanks
Iain
Iain Sandoe wrote:
Joseph Myers wrote:
This patch seems to be missing documentation for the new attribute in
extend.texi.
Apologies, for that omission, revised patch includes the documen
From: Matt Thomas
Fix an ICE with the handling of RTL expressions like:
(subreg:QI (mem/c:SI (plus:SI (plus:SI (mult:SI (reg/v:SI 0 %r0 [orig:67 i ]
[67])
(const_int 4 [0x4]))
(reg/v/f:SI 7 %r7 [orig:59 doacross ] [59]))
(const_int 40 [0x28])) [1
On Fri, 27 Nov 2020, Maciej W. Rozycki wrote:
> > That said, if it fixes the test suite errors you're seeing, it would
> > probably be OK to go with just this minimal change -- unless we can
> > just move to LRA as mentioned above.
>
> I've looked through the test results and indeed these suspic
On Thu, 26 Nov 2020, Roger Sayle wrote:
> NaNs have a sign-bit, so copyexpr (and negate and ABS_EXPR) are
> well-defined, and it's reasonable for nonnegative_p to reflect this.
> IMHO, the true bug is that we can't fold away (any) comparisons against
> NaN when flag_trapping_math, irrespective
On Thu, 26 Nov 2020, Richard Biener wrote:
> Is copysign (x, NaN) supposed to be well-defined? We'd stop folding
copysign with NaN arguments (including sNaN) is well-defined and copies
the sign bit without raising any exceptions.
> this then, no? I think the ABS_EXPR < 0 to false folding is
>
Hi,
When libphobos is configured with --enable-cet, this adds extra fields
to the Fiber class to support the ucontext_t fallback implementation.
These fields get omitted when compiling user code unless they also used
`-fversion=CET' to build their project, which resulted in data being
overwritten
Hi,
This patch merges the D front-end with upstream DMD db0df3f7e, removing
all support code and tests for the extern(Pascal) calling convention.
Bootstrapped and regression tested on x86_64-linux-gnu with -mx32/-m32,
committed to mainline.
Regards
Iain.
---
gcc/d/ChangeLog:
* dmd/MERG
Hi,
This patch merges the D runtime library with upstream druntime
d37ef985, adding support for FreeBSD/x86 53-bit precision reals,
updating bindings for FreeBSD 12.x, and removing all support code and
tests for the extern(Pascal) calling convention.
Bootstrapped and regression tested on x86_64-l
Hi,
This patch merges the D standard library with upstream phobos 38873fe6e,
adding support for FreeBSD/x86 53-bit precision reals, and removing all
support code and tests for the extern(Pascal) calling convention.
Bootstrapped and regression tested on x86_64-linux-gnu with -mx32/-m32,
committed
Hi,
For the math intrinsics: cos, fabs, ldexp, rint, rndtol, and sin, new
overloads have been added to the core.math module for matching float and
double types. These have been implemented in the compiler.
A recent change to dump_function_to_file started triggering some
scan-tree-dump tests to F
Excerpts from H.J. Lu's message of November 27, 2020 8:28 pm:
> On Fri, Nov 27, 2020 at 11:02 AM Matthias Klose wrote:
>>
>> On 11/27/20 3:54 PM, H.J. Lu via Gcc-patches wrote:
>> > On Fri, Nov 27, 2020 at 6:24 AM Richard Biener wrote:
>> >>
>> >> OK.
>> >>
>> >> On Fri, 27 Nov 2020, H.J. Lu wrot
On Fri, 27 Nov 2020 at 17:13, Jonathan Wakely via Gcc-patches
wrote:
>
> The default for the GCC testsuite is 300, i.e. 5 minutes, which is the
> same as the DejaGnu default.
>
> Libstdc++ overrides this to 600, i.e. 10 minutes.
>
> This seems ridiculously long. If any test takes that long on mode
On Fri, Nov 27, 2020 at 11:02 AM Matthias Klose wrote:
>
> On 11/27/20 3:54 PM, H.J. Lu via Gcc-patches wrote:
> > On Fri, Nov 27, 2020 at 6:24 AM Richard Biener wrote:
> >>
> >> OK.
> >>
> >> On Fri, 27 Nov 2020, H.J. Lu wrote:
> >>
> >>> PR other/98027
> >>> * doc/install: Default t
On Fri, 27 Nov 2020, Ulrich Weigand wrote:
> > NB I find the reindentation resulting in `push_reload' awful, just as I
> > do either version of the massive logical expression involved. Perhaps we
> > could factor these out into `static inline' functions sometime, and then
> > have them split
On 11/27/20 3:54 PM, H.J. Lu via Gcc-patches wrote:
> On Fri, Nov 27, 2020 at 6:24 AM Richard Biener wrote:
>>
>> OK.
>>
>> On Fri, 27 Nov 2020, H.J. Lu wrote:
>>
>>> PR other/98027
>>> * doc/install: Default to --enable-cet=auto.
>>> ---
>>> gcc/doc/install.texi | 9 -
>>> 1
Hello,
-
This test should ensure that we can compile with hwasan, that such a compiled
binary runs as expected, *and* that we're running on a kernel which implements
the capability to ignore the top bytes of pointers in syscalls.
It was expected that a basic test of `int main(void) { return
On Fri, Nov 27, 2020 at 8:22 AM Gerald Pfeifer wrote:
>
> On Fri, 27 Nov 2020, H.J. Lu via Gcc-patches wrote:
> > +@code{--enable-cet=auto} is default. CET is enabled on Linux/x86 if
> > +target binutils supports @code{Intel CET} instructions and disabled
> > +otherwise. In this case, the target
Richard Sandiford via Gcc-patches writes:
> One thing I'd deliberately tried to avoid was converting integers
> “back” to pointers, because that seemed like a more dangerous thing.
> That's why:
>
>>> +template
>>> +inline T2 *
>>> +pointer_mux::second_or_null () const
>>> +{
>>> + // Micro optim
On Thu, Nov 26, 2020 at 6:02 PM Jonathan Wakely wrote:
>
> On 26/11/20 22:33 +, Jonathan Wakely wrote:
> >On 26/11/20 15:33 -0500, David Edelsohn via Libstdc++ wrote:
> >>Hi, Jonathan
> >>
> >>Thanks for the recent fixes. I still see a few errors that have crept
> >>in on AIX. All the same:
On 20/11/20 16:30 -0800, Thomas Rodgers wrote:
From: Thomas Rodgers
Should include all discussion on and off list to date.
Most of the comments in
https://gcc.gnu.org/pipermail/gcc-patches/2020-November/558090.html
still apply to this version.
Adds
libstdc++/ChangeLog:
* include/
Add missing returns and remove unnecessary postfix increments.
2020-11-27 Uroš Bizjak
PR testsuite/98036
gcc/testsuite/
* gcc.target/i386/fma4-256-maccXX.c (check_maccps):
Remove unnecessary postfix increment on a returned variable.
(check_maccpd): Ditto.
* gcc.target/i386
On Fri, Nov 27, 2020 at 06:18:46PM +0100, Tobias Burnus wrote:
> '!' starts in Fortran a comment (+ fixed-form variants)
> OpenACC defines as sentinel "!$acc" (likewise)
> But OpenMP has two: Besides "!$omp" there is additionally
> "!$ " (with space) to permit conditional compilation.
>
> Currentl
'!' starts in Fortran a comment (+ fixed-form variants)
OpenACC defines as sentinel "!$acc" (likewise)
But OpenMP has two: Besides "!$omp" there is additionally
"!$ " (with space) to permit conditional compilation.
Currently, -fopenacc also executes the '!$ ' code, which I
think does not make sen
On 25/11/20 15:05 +0100, Michael Weghorn via Libstdc++ wrote:
I'd like to ping for this patch:
https://gcc.gnu.org/pipermail/gcc-patches/2020-September/553870.html
Thanks, I'll take another look next week.
Michael
On 11/10/2020 19.22, Michael Weghorn via Gcc-patches wrote:
On 22/09/2020 12
Michael Matz writes:
> Hello,
>
> On Thu, 26 Nov 2020, Richard Sandiford via Gcc-patches wrote:
>
>> >> The aim is only to provide a different view of existing RTL instructions.
>> >> Unlike gimple, and unlike (IIRC) the old RTL SSA project from way back,
>> >> the new framework isn't a “native” S
This allows using "testsuite/*" in libstdc++-v3/ChangeLog entries, which
was one of the original motivations for adding wildcard support in the
first place:
https://gcc.gnu.org/pipermail/gcc/2020-June/232719.html
contrib/ChangeLog:
* gcc-changelog/git_commit.py (wildcard_prefixes): Add li
On Fri, 27 Nov 2020, H.J. Lu via Gcc-patches wrote:
> +@code{--enable-cet=auto} is default. CET is enabled on Linux/x86 if
> +target binutils supports @code{Intel CET} instructions and disabled
> +otherwise. In this case, the target libraries are configured to get
> +additional @option{-fcf-prote
The default for the GCC testsuite is 300, i.e. 5 minutes, which is the
same as the DejaGnu default.
Libstdc++ overrides this to 600, i.e. 10 minutes.
This seems ridiculously long. If any test takes that long on modern
hardware, something is wrong. We've seen this a lot recently with
buggy tests,
OpenMP-related patch pings:
Kwok's:
* Re: [PATCH] openmp: Implicit 'declare target' for C++ static initializers
https://gcc.gnu.org/pipermail/gcc-patches/2020-November/559624.html
* RFC:
Nested declare target support
https://gcc.gnu.org/pipermail/gcc-patches/2020-November/559758.html
Chung
Hi David!
On 2020-11-27T10:47:17-0500, David Edelsohn wrote:
> Actually, yes, AIX does allow dereference of a NULL pointer.
Oh. That's not what I expected! Heh.
Anyway, that then indeed completely explains what was going on here --
which is good. :-)
Grüße
Thomas
> On Fri, Nov 27, 2020
Hello,
On Thu, 26 Nov 2020, Richard Sandiford via Gcc-patches wrote:
> >> The aim is only to provide a different view of existing RTL instructions.
> >> Unlike gimple, and unlike (IIRC) the old RTL SSA project from way back,
> >> the new framework isn't a “native” SSA representation. This means
This introduces two new procs to replace boilerplate in the
effective-target checks.
libstdc++-v3/ChangeLog:
* testsuite/lib/libstdc++.exp (v3_try_preprocess): Define
new proc to preprocess a chunk of code.
(v3_check_preprocessor_condition): Define new proc to test
Hi, Thomas
Actually, yes, AIX does allow dereference of a NULL pointer.
Thanks, David
On Fri, Nov 27, 2020 at 9:15 AM Thomas Schwinge wrote:
>
> Hi David!
>
> On 2020-11-24T15:29:17-0500, David Edelsohn via Gcc-patches
> wrote:
> > On Tue, Nov 24, 2020 at 5:19 AM Thomas Schwinge
> > wrote:
Hi Thomas,
On 25.11.20 12:58, Tobias Burnus wrote:
On 15.11.20 18:52, Thomas Koenig via Fortran wrote:
+#define ADD_CHAR(c) do { *fp++ = c; *fp++ = ' '; } while(0)
...
+ ADD_CHAR ('.'); /* Function return. */
Shouldn't this be ".c" instead of ". " as neither global memory is
read nor writte
On Fri, 27 Nov 2020 at 15:13, Andre Vieira (lists)
wrote:
>
> Hi Christophe,
>
> On 26/11/2020 15:31, Christophe Lyon wrote:
> > Hi Andre,
> >
> > Thanks for the quick feedback.
> >
> > On Wed, 25 Nov 2020 at 18:17, Andre Simoes Dias Vieira
> > wrote:
> >> Hi Christophe,
> >>
> >> Thanks for thes
On 10/29/20 8:11 PM, H.J. Lu wrote:
Add the --enable-pgo-build[=lto] configure option. When binutils+gdb
is not built together with GCC, --enable-pgo-build enables the PGO build:
I would like to support the patch. It significantly speeds up gas for free
(based on my measurements).
Thanks,
Mar
On 11/20/20 3:37 PM, Richard Biener wrote:
On Fri, Nov 20, 2020 at 9:57 AM Martin Liška wrote:
On 11/19/20 3:46 PM, Richard Biener wrote:
OK, so can you send an updated patch?
+ tree pos_one = build_int_cst (type, 1);
+ if (!left->m_has_forward_bb
+ && !right->m_has_forwar
On Fri, Nov 27, 2020 at 6:24 AM Richard Biener wrote:
>
> OK.
>
> On Fri, 27 Nov 2020, H.J. Lu wrote:
>
> > PR other/98027
> > * doc/install: Default to --enable-cet=auto.
> > ---
> > gcc/doc/install.texi | 9 -
> > 1 file changed, 4 insertions(+), 5 deletions(-)
> >
> > diff
OK.
On Fri, 27 Nov 2020, H.J. Lu wrote:
> PR other/98027
> * doc/install: Default to --enable-cet=auto.
> ---
> gcc/doc/install.texi | 9 -
> 1 file changed, 4 insertions(+), 5 deletions(-)
>
> diff --git a/gcc/doc/install.texi b/gcc/doc/install.texi
> index 5f879ca4cea..021
Hi David!
On 2020-11-24T15:29:17-0500, David Edelsohn via Gcc-patches
wrote:
> On Tue, Nov 24, 2020 at 5:19 AM Thomas Schwinge
> wrote:
>> On 2020-11-21T10:50:10-0500, David Edelsohn wrote:
>> > I see
>> >
>> > "during GIMPLE pass: omplower"
>> >
>> > message for kernels-decompose-ice-2.c. k
Hi Christophe,
On 26/11/2020 15:31, Christophe Lyon wrote:
Hi Andre,
Thanks for the quick feedback.
On Wed, 25 Nov 2020 at 18:17, Andre Simoes Dias Vieira
wrote:
Hi Christophe,
Thanks for these! See some inline comments.
On 25/11/2020 13:54, Christophe Lyon via Gcc-patches wrote:
This pat
Hi!
On 2020-11-13T23:22:30+0100, I wrote:
> [...] I've pushed to master branch [...] commit
> e898ce7997733c29dcab9c3c62ca102c7f9fa6eb "Decompose OpenACC 'kernels'
> constructs into parts, a sequence of compute constructs", see attached.
>
>> There's more work to be done there, and we're aware of
PR other/98027
* doc/install: Default to --enable-cet=auto.
---
gcc/doc/install.texi | 9 -
1 file changed, 4 insertions(+), 5 deletions(-)
diff --git a/gcc/doc/install.texi b/gcc/doc/install.texi
index 5f879ca4cea..021c347cc09 100644
--- a/gcc/doc/install.texi
+++ b/gcc/d
Many tests do `for (const path& p : test_paths)` where test_paths is an
array of strings. To avoid -Wrange-loop-construct warnings the loop
variable should be an object, not a reference bound to a temporary.
Tested powerpc64le-linux and x86_64-w64-mingw32, pushed to trunk.
commit 4a7c7999086756b
> > On Wed, Nov 25, 2020 at 3:11 PM Jan Hubicka wrote:
> > >
> > > > On Tue, 24 Nov 2020, Jan Hubicka wrote:
> > > >
> > > > > Hi,
> > > > > at the end of processing function body we loop over basic blocks and
> > > > > free all edges while we do not free the rest. I think this is
> > > > > left
Kyrylo Tkachov writes:
>> -Original Message-
>> From: Richard Sandiford
>> Sent: 27 November 2020 08:29
>> To: Richard Biener
>> Cc: Kyrylo Tkachov ; gcc-patches@gcc.gnu.org
>> Subject: Re: [PATCH] aarch64: Introduce --param=aarch64-autovec-
>> preference to select autovec preference in
It handles the following:
(cherry picked from commit c0c7270cc4efd896fe99f8ad5409dbef089a407f (testsuite
changes only))
contrib/ChangeLog:
* gcc-changelog/git_commit.py: Use regex for cherry pick prefix.
* gcc-changelog/test_email.py: Test it.
* gcc-changelog/test_patche
On 27/11/20 11:15 +, Jonathan Wakely wrote:
On 27/11/20 09:43 +0100, Stephan Bergmann via Libstdc++ wrote:
On 24/11/2020 15:59, Jonathan Wakely via Gcc-patches wrote:
Most initialization of locales and facets happens before main() during
startup, when the program is likely to only have one
Hi Uros,
> On Fri, Nov 27, 2020 at 11:08 AM Rainer Orth
> wrote:
>>
>> When using the Solaris/x86 assembler with gcc, a couple of testcases
>> currently FAIL. Those failures follow two patterns:
>>
>> FAIL: gcc.target/i386/avx512bw-vpmovb2m-2.c (test for excess errors)
>> Excess errors:
>> Assem
On Fri, Nov 27, 2020 at 11:08 AM Rainer Orth
wrote:
>
> When using the Solaris/x86 assembler with gcc, a couple of testcases
> currently FAIL. Those failures follow two patterns:
>
> FAIL: gcc.target/i386/avx512bw-vpmovb2m-2.c (test for excess errors)
> Excess errors:
> Assembler: avx512bw-vpmovb
> -Original Message-
> From: Richard Sandiford
> Sent: 27 November 2020 08:29
> To: Richard Biener
> Cc: Kyrylo Tkachov ; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] aarch64: Introduce --param=aarch64-autovec-
> preference to select autovec preference in backend
>
> Richard Biener
The change to make PRE insertion iterate less had a typo in checking
successors RPO state. Fixing this exposes that regular PRE insertion
when facing a value that is the same on all edges inserts an
assignment in place of a PHI node but fails to set up things so that
this insertion is not repeated
On 27/11/20 09:43 +0100, Stephan Bergmann via Libstdc++ wrote:
On 24/11/2020 15:59, Jonathan Wakely via Gcc-patches wrote:
Most initialization of locales and facets happens before main() during
startup, when the program is likely to only have one thread. By using
the new __gnu_cxx::__is_single_t
Doh! Wrong patch2.txt file. Sorry.
-Original Message-
From: Roger Sayle
Sent: 27 November 2020 10:52
To: 'Jakub Jelinek' ; 'Joseph S. Myers'
Cc: 'Richard Biener' ; 'gcc-patches@gcc.gnu.org'
Subject: RE: [PATCH] fold-const: Don't consider NaN non-negative [PR97965]
Hi Jakub,
Technic
On Tue, Nov 24, 2020 at 06:19:41AM +, Maciej W. Rozycki wrote:
> NB I find the reindentation resulting in `push_reload' awful, just as I
> do either version of the massive logical expression involved. Perhaps we
> could factor these out into `static inline' functions sometime, and then
>
Hi Jakub,
Technically, PR97965 doesn't explicitly mention equality/inequality, but
you're
right, it makes sense to tackle this missed optimization at the same time as
we
fix the wrong-code.
On Thu, Nov 26, 2020, Jakub Jelinek wrote:
>On Thu, Nov 26, 2020 at 01:56:03PM -, Roger Sayle wrote:
>>
Found when grepping & post commit. There were two !$omp issues in
the testcase – which (currently) do not show up due to the previous
(and intended) errors.
Tobias
-
Mentor Graphics (Deutschland) GmbH, Arnulfstraße 201, 80634 München / Germany
Registergericht München HRB 106955,
On Fri, Nov 27, 2020 at 10:29:53AM +, Iain Sandoe wrote:
> --- a/gcc/configure.ac
> +++ b/gcc/configure.ac
> @@ -3114,34 +3114,38 @@ AC_MSG_RESULT($gcc_cv_ld_ro_rw_mix)
> gcc_AC_INITFINI_ARRAY
>
> # Check if we have .[us]leb128, and support symbol arithmetic with it.
> +# Older versions of G
Andrea Corallo via Gcc-patches writes:
> "Richard Earnshaw (lists)" writes:
>
>> On 26/11/2020 13:53, Andrea Corallo via Gcc-patches wrote:
>>> Hi all,
>>>
>>> I'd like to submit the following simple patch to clean some Low Loop
>>> Overhead test failing on hard float configurations.
>>>
>>> l
On Fri, 27 Nov 2020, Jakub Jelinek wrote:
> Hi!
>
> My recent match.pd change required quite a lot of code due to the separate
> need to handle INTEGER_CSTs and SSA_NAMEs, and after all, I didn't even
> handle one case there, when in x * y / y the x is INTEGER_CST and y is
> SSA_NAME.
> The follo
Hi Folks.
Iain Sandoe via Gcc-patches wrote:
Rainer Orth wrote:
maybe this is enough to cover all bases without having to do any
version or
target checks. (untested)
objdump is not available on quite a few Darwin versions with perfectly
functional
uleb128 - so I don’t want to punt on thos
Hi!
My recent match.pd change required quite a lot of code due to the separate
need to handle INTEGER_CSTs and SSA_NAMEs, and after all, I didn't even
handle one case there, when in x * y / y the x is INTEGER_CST and y is
SSA_NAME.
The following patch allows to simplify it, by allowing non-SSA_NAM
Now committed as r11-5489-gf324479caf0ac326534f4fcf72cb12991ccddb3d
Tobias
On 25.11.20 12:00, Tobias Burnus wrote:
Here, the problem was that we were mixing types when calculating
the product of loop-steps, each type was taken from the loop var.
Solution: take the largest type as diff_type and
Two fixes – simple, see patch + commit text.
Longer description:
* options:
Background:
- OpenMP, OpenACC and imply that a function is called
concurrently and -frecursive implies recusive. In all those
cases, the code may fail if a local variable is in static memory
instead of stack or heap. – I
When using the Solaris/x86 assembler with gcc, a couple of testcases
currently FAIL. Those failures follow two patterns:
FAIL: gcc.target/i386/avx512bw-vpmovb2m-2.c (test for excess errors)
Excess errors:
Assembler: avx512bw-vpmovb2m-2.c
"/var/tmp//ccPh3IRc.s", line 58 : Invalid instructi
On Fri, 27 Nov 2020 at 02:03, Stam Markianos-Wright
wrote:
>
> On 26/11/2020 09:01, Christophe Lyon wrote:
> > On Wed, 25 Nov 2020 at 14:24, Stam Markianos-Wright via Gcc-patches
> > wrote:
> >>
> >> Hi all,
> >>
> >> A while back I submitted GCC10 commit:
> >>
> >>44f77a6dea2f312ee1743f3dde4
This fixes a minor oversight in Integer_To_Fixed and restricts the
previous change to System.Value_R.Scan_Raw_Real for the sake of
backward compatibility with earlier conversions of the compiler.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/s-valuef.adb (Integer_T
When using the Shift_Right operator on negative values (e.g. -1 on a
signed type), the wrong value is incorrectly folded after recent changes
in sem_eval.adb to add compile time evaluation of the shift operators.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_eval.adb (
When working on LLVM/CCG, we found that the generated tree had a wrong
reference before declaration usage. Fixed for LLVM/CCG. The issue might
still be latent with Modify_Tree_For_C, but this expansion will
disappear in the future, so not worth bothering.
Tested on x86_64-pc-linux-gnu, committed o
Replace an access type parameter, which is unusual in the GNAT sources
and more difficult to understand, with a parameter of mode out. Code
cleanup only; behaviour is unaffected.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_eval.ads (Compile_Time_Compare): Restore par
This makes System.Value_R.Scan_Raw_Real round the extra digit it returns
to the caller, so that it is precise to the last 'Aft digits for strings
that are the exact representation of a number, i.e. not already rounded.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/
Similar to what is done for compilation, range checks should not be
applied inside generics during GNATprove analysis. This fixes an
assertion failure in GNATprove.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* checks.adb (Selected_Range_Checks): Adapt the condition for
This restricts the use of 128-bit division for fixed-point types to
the cases where there is a 128-bit type in the source code, that is
to say this prevents the compiler from using it for 64-bit types.
This is done mainly for the sake of backward compatibility with earlier
compilers not supporting
This is only an initial implementation, subject to many limitations.
Some interactions with derived types and inheritance are not
implemented. Other limitations are idenitified in the code with "???"
comments. However, Stable_Properties and Stable_Properties'Class aspect
specifications are implem
Some code currently relies on To_GM_Time returning some reasonable value
in this case rather than e.g. raising an exception and it appears that
Windows returns 1970-01-01 except with second set to -1 (triggering an
exception), while linux returns 1970-01-01 minus 1 second (so 1969-12-31
at 23:59:59
Default_Initial_Condition was created for SPARK, but has been adopted by
Ada 202x, so now it is an RM defined aspect and an assertion policy.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* doc/gnat_rm/implementation_defined_pragmas.rst:
(Assertion_Policy): Move "Def
In the listing of assertion kinds accepted for pragma Assertion_Policy
some items were wrongly included while other were missing. Now this
listing is synchronized with Sem_Prag.Is_Valid_Assertion_Kind.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* doc/gnat_rm/implementatio
The following pragma is currently accepted:
pragma Assertion_Policy (Assertion_Policy => Ignore);
because of what appears to be a typo in a previous commit on this topic.
With this patch the above pragma is rejected as illegal.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
Enabling checks in the runtime made a latent bug with the sjlj runtime
visible. We used to handle abort deferral differently between GCC ZCX
and GNAT "Front-End" SJLJ, which is no longer the case with GCC SJLJ, so
remove differences to simplify the code and fix the inconsistency.
Tested on x86_64-
Replace an access type parameter, which is unusual in the GNAT sources
and more difficult to understand, with a parameter of mode out. Code
cleanup only; behaviour is unaffected.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* sem_eval.ads (Compile_Time_Compare): Change para
This reimplements the aforementioned generic package according to the
requirements of the Ada 2020 RM, namely that To_Big_Real be exact and
that From_Big_Real use the common floating-point conversion rules,
which are round to nearest, ties to even, in GNAT as per IEEE 754.
Tested on x86_64-pc-linu
This set of changes aimed at optimizing the generation of range and
overflow checks for fixed-point types contains two parts:
1. a cleanup to the generation of range checks for type conversions
involving fixed-point types, which is now more clearly deferred
entirely to after the expans
When reorganizing the mechanism which evaluates relational operators and
potentially folds them, the "General case" in routine Eval_Relational_Op
was put into an ELSE branch. This was subtly wrong, because it must be
executed unless the special case in the THEN branch exits with a RETURN
statement.
This aligns the behavior of gnatmake with the way the runtime is now
built.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* make.adb (GNAT_Flag): Change to "-gnatg".
(Compile): Adjust comments accordingly.diff --git a/gcc/ada/make.adb b/gcc/ada/make.adb
--- a/gcc/ada
None of these programs analyzes the resulting node on return any more.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* exp_fixd.adb (Build_Conversion): Adjust head comment.
(Build_Divide): Likewise.
(Build_Double_Divide): Likewise.
(Build_Multiply): L
It can block the transformation of the multiplication by a power of 2
into a shift operation when the context is Universal_Integer and checks
are disabled.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* exp_ch4.adb (Expand_N_Op_Multiply): Move down block calling
Nar
Add support for the recently added 128bits integer support in
Big_Integers.To_Big_Integer.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/s-genbig.ads, libgnat/s-genbig.adb (To_Bignum): New
variant taking an Unsigned_128.
* libgnat/a-nbnbin.adb (To_B
In most cases, the common operations for fixed-point types are implemented
by means of corresponding integer operations, in particular multiplication
and division. Because scaling is required to go back and forth between
the two representations, the expander needs to control the magnitude of
opera
This removes a use of the absolute value that could potentially fail
the overflow check and add assertions guarding a second use and the
manipulation of string buffers, the latter for the sake of CodePeer.
Tested on x86_64-pc-linux-gnu, committed on trunk
gcc/ada/
* libgnat/s-imagef.adb
This commit adds checks for rule 6.5.1 4/3 of the Ada RM which declares
nonreturning procedures and nonreturning generic procedure instances
illegal.
The reason this check is performed in sem_prag.adb rather than in
Check_Returns in sem_ch6.adb is that generic procedure instances won't
have their
1 - 100 of 105 matches
Mail list logo
