I'm happy enough with the way that the gmp and mpfr libraries are building.
GMP has its own fallback (printf/repl_vsnprintf.c) to use when vsnprintf
is found to be unavailable or faulty in behaviour.
That fallback seems to work quite well for gmp, but mpfr does not adapt
perfectly to it.
However,
On 11/12/24 18:52, Kirill Makurin wrote:
Hi,
I have made some changes to the patch3, including using LN_S instead of `cp -f`
(but see below).
First of all, I split it into two different patches. One for Makefile.am and
configure.ac, and another for tests/Makefile.am since the changes are not
On Tue, 12 Nov 2024, Pali Rohár wrote:
- The commit message explanation of _swprintf_c seems weird. The commit
message says:
_(v)swprintf_c - clears the first wide character in buffer on error
(error means that there is no space for nul-term or
On Tuesday 12 November 2024 23:13:28 Martin Storsjö wrote:
> On Tue, 12 Nov 2024, Pali Rohár wrote:
>
> > > See https://github.com/mstorsjo/llvm-mingw/commits/c95-test for the added
> > > tests. At the end, I have two "WIP: test: Waive ..." commits that makes
> > > the
> > > tests pass, covering
On Tue, 12 Nov 2024, Pali Rohár wrote:
See https://github.com/mstorsjo/llvm-mingw/commits/c95-test for the added
tests. At the end, I have two "WIP: test: Waive ..." commits that makes the
tests pass, covering up the issues that I found.
These tests are in the form of one large file, test/crt-t
On Tuesday 12 November 2024 21:30:22 Pali Rohár wrote:
> On Tuesday 12 November 2024 12:30:08 Martin Storsjö wrote:
> > - The commit message explanation of _swprintf_c seems weird. The commit
> > message says:
> >
> > _(v)swprintf_c - clears the first wide character in buffer on error
> >
On Tuesday 12 November 2024 12:30:08 Martin Storsjö wrote:
> On Thu, 24 Oct 2024, Pali Rohár wrote:
>
> > > While debugging this, I also noted that libc++ fails to build (when using
> > > __USE_MINGW_ANSI_STDIO=1) after patch "crt: Add support for C95
> > > (v)swprintf() functions when __USE_MINGW
On 2024-11-12 Jeremy Drake wrote:
> On Tue, 12 Nov 2024, Martin Storsjö wrote:
>
> > Regarding the above, I wouldn't exactly think that those apps ignore
> > argv[] and get the command line some other way - but I'm not
> > entirely sure; ideally we'd take it up with them to understand
> > exactly
On Tue, 12 Nov 2024, Martin Storsjö wrote:
> Regarding the above, I wouldn't exactly think that those apps ignore
> argv[] and get the command line some other way - but I'm not entirely
> sure; ideally we'd take it up with them to understand exactly what their
> setup is and how it works, and how
Hi,
I have made some changes to the patch3, including using LN_S instead of `cp -f`
(but see below).
First of all, I split it into two different patches. One for Makefile.am and
configure.ac, and another for tests/Makefile.am since the changes are not
directly related.
A few days after I subm
On 2024-11-12 Pali Rohár wrote:
> On Monday 11 November 2024 22:10:01 Lasse Collin wrote:
> > If permissive mode can be enabled the same way as _dowildcard works,
> > then argv[] could be constructed with the same code as in strict
> > mode. The current crtexe.c in master has
> >
> > for (int i
在 2024-11-12 23:17, Lasse Collin 写道:
I'm suggesting it because the patch clearly is harmful for some types of
apps which don't need it for security reasons. I leave the final
decision to MinGW-w64 maintainers.
Done now: https://sourceforge.net/p/mingw-w64/mingw-w64/ci/c6857dc
Meanwhile, I see
On 2024-11-12 LIU Hao wrote:
> 在 2024-11-12 22:49, Lasse Collin 写道:
> > The _exit(255) trick is clearly not an acceptable default. My patch
> > should be reverted until a better one is made. It might be that all
> > apps that want safe argv[] for main() will need something special on
> > Windows (l
On Tue, 12 Nov 2024, Lasse Collin wrote:
On 2024-11-12 Martin Storsjö wrote:
FYI, see https://github.com/msys2/MINGW-packages/issues/22462 where
someone ran into issues caused by this.
Also see
https://ffmpeg.org/pipermail/ffmpeg-devel/2024-November/335936.html where
someone else observes othe
在 2024-11-12 22:49, Lasse Collin 写道:
I understand only now that many apps use main() but ignore argv[] and
instead get the args by some other method. I was too focused on porting
apps written for POSIX systems. I'm sorry.
It doesn't matter. That's not your fault.
The _exit(255) trick is clea
On 2024-11-12 Martin Storsjö wrote:
> FYI, see https://github.com/msys2/MINGW-packages/issues/22462 where
> someone ran into issues caused by this.
>
> Also see
> https://ffmpeg.org/pipermail/ffmpeg-devel/2024-November/335936.html where
> someone else observes other issues related to this. (The
GMP configure checks for the availability of %n format specifier.
Because of security reasons, Microsoft has disabled the %n format specifier by
default.
To use the %n format specifier, _set_printf_count_output(1) must be called.
---
Best regards,
Yuta
-Original Message-
From: Martin St
On Tue, 12 Nov 2024, sisyphus wrote:
(Otherwise the gmp configure step decides that vsnprintf does not work, and
an mpfr library built against the resultant gmp library exhibits some buggy
behaviour.)
The intent is that those stdio functions would be working correctly enough
in general on UCR
On Fri, 8 Nov 2024, Pali Rohár wrote:
Function _copysignf is in i386 natively available only in msvcr120 and
UCRT. mingw-w64 already provides emulation in crtdll and msvcrt i386 import
libraries. Include this emulation also into all other i386 CRT import
libraries.
Note that function _copysignf
On Fri, 8 Nov 2024, Pali Rohár wrote:
Source file purecall.c provides emulation of function _set_purecall_handler()
so rename this file to _set_purecall_handler.c. This file does not provide
function named purecall().
Function _set_purecall_handler() is available since msvcr71.dll, so include
s
On Thu, 24 Oct 2024, Pali Rohár wrote:
While debugging this, I also noted that libc++ fails to build (when using
__USE_MINGW_ANSI_STDIO=1) after patch "crt: Add support for C95
(v)swprintf() functions when __USE_MINGW_ANSI_STDIO=0", because there's only
a 2 parameter version of swprintf, but it
在 2024-11-12 15:55, Martin Storsjö 写道:
The explanations around the volatile changes seem good to me now, and I trust you on the rest of the
x86 specific details here.
Thank you very much for the trust! Pushed this one now.
--
Best regards,
LIU Hao
OpenPGP_signature.asc
Description: OpenPGP
在 2024-11-12 15:53, Martin Storsjö 写道:
LGTM, these patches look reasonable to me.
Thanks. Pushed these two now.
--
Best regards,
LIU Hao
OpenPGP_signature.asc
Description: OpenPGP digital signature
___
Mingw-w64-public mailing list
Mingw-w64-pub
On Sat, 9 Nov 2024, LIU Hao wrote:
Add an explicit `volatile` qualifier for intrinsic functions that write to
TEB; do not rely on the implicit volatile semantics, just because they have
no output.
* If an asm statement reads memory that is not passed with "m", it shall
clobber "memory".
* I
On Tue, 12 Nov 2024, LIU Hao wrote:
--
LGTM, these patches look reasonable to me.
// Martin
___
Mingw-w64-public mailing list
Mingw-w64-public@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
25 matches
Mail list logo
