From: Rich Felker <dalias@libc.org>
To: musl@lists.openwall.com
Subject: Re: [PATCH] math: move i386 sqrtf to C
Date: Thu, 9 Jan 2020 21:07:47 -0500 [thread overview]
Message-ID: <20200110020747.GY30412@brightrain.aerifal.cx> (raw)
In-Reply-To: <20200109231858.GP23985@port70.net>
On Fri, Jan 10, 2020 at 12:18:58AM +0100, Szabolcs Nagy wrote:
> * Rich Felker <dalias@libc.org> [2020-01-09 17:00:14 -0500]:
> > On Thu, Jan 09, 2020 at 10:00:06PM +0100, Szabolcs Nagy wrote:
> > > > Note that eval_as_float only helps if -ffloat-store is used, which is
> > > > a nasty hack and also nonconforming, arguably worse than the behavior
> > > > without it, so we should probably drop use of that as a fallback, and
> > > > use fp_barrier[f] instead if needed.
> > >
> > > i think -ffloat-store almost always drops excess precision
> > > including returns and assignments, so with that no
> > > annotation is needed. but yes the way the annotation is
> > > defined now is not useful against broken compilers or
> > > non-standard excess precision setting, in glibc the
> > > annotation is defined differently (with inline asm).
> >
> > I was thinking in the context of wanting to remove from configure the:
> >
> > || { test "$ARCH" = i386 && tryflag CFLAGS_C99FSE -ffloat-store ; }
> >
> > which is probably doing more harm than good. Do you know if there are
> > things that'd break if we did that? I think eval_as_float should
> > probably be defined as fp_barrierf to make it safe in your code,
> > conditional on FLT_EVAL_METHOD>0 (and likewise >1 for eval_as_double).
>
> i think -fexcess-precision=standard was introduced in
> gcc 4.5 and to get reliable behaviour before that we
> needed -ffloat-store.
I don't think the behavior was "reliable" with -ffloat-store; it's
wrong with respect to the defined meaning of FLT_EVAL_METHOD.
> since we had -ffloat-store i turned off the volatile
> hacks in commit 6d3f1a39c14b12026df84f386875b094e3652990
> and later completely removed the annotations in commit
> 9b0fcb441a44456c7b071c7cdaf90403f81ec05a
Thanks for these references.
> on new compilers -fexcess-precision=standard is used,
> but that turned out to do too many stores on the fdlibm
> code (which is why glibc kept using =fast), so in commit
> e216951f509b71da193da2fc63e25b998740d58b i started using
> float_t and double_t to get fast code in standard mode.
> (of course this made things worse for -ffloat-store).
This was the right thing to do, and I think it largely but not
entirely eliminates the need for caring about how the compiler handles
this, except in a few cases. It could probably be eliminated in more.
For example the argument reduction code cited above could use the
right constants for double_t rather than double to avoid the need to
store/load to drop excess precision.
> i think we would need to add back the old annotations
> to make old compilers safe without -ffloat-store.
> (fdlibm often raises fenv exceptions via a final rounding
> before return, those could be often handled more cleanly
> by __math_oflow etc helpers, but since it was not designed
> for inline errno handling some normal return paths can
> raise fp exceptions too and thus need eval_as_* annotation).
I think I asked you before, but from a standpoint of fenv stuff, I'm
confused why the eval_as_* things are useful at all; it looks like you
would need fp_barrier* to ensure they're actually evaluated (e.g. in
the presence of LTO with a compiler that doesn't honor fenv right).
But I think it's also useful to distinguish between possibility of
wrong exceptions being raised, which is a rather minor issue since
some widely-used compilers don't even support fenv reasonably at at
all, and the possibility of wrong values being returned for functions
where the result is required to be correctly rounded. I would deem it
a serious problem for sqrt[f] or fma[f] to return the wrong value when
compiled with gcc3 or pcc. I don't think I would particularly care if
exceptions failed to be raised properly when compiled with gcc3 or
pcc, though. So I probably would like to ensure that, whatever code we
end up with in i386 sqrt[f].c, it it ends up working even if the
compiler does not handle excess precision correctly.
Rich
next prev parent reply other threads:[~2020-01-10 2:07 UTC|newest]
Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-05 16:35 math patches for moving bare asm to C inline asm Alexander Monakov
2020-01-05 16:36 ` [PATCH] math: move x86_64 fabs, fabsf to C with " Alexander Monakov
2020-01-05 20:05 ` Rich Felker
2020-01-05 21:32 ` Alexander Monakov
2020-01-05 22:43 ` Rich Felker
2020-01-06 8:17 ` Alexander Monakov
2020-01-06 8:40 ` [PATCH] math: move more x86-family fabs functions to C Alexander Monakov
2020-03-21 17:06 ` [musl] " Rich Felker
2020-01-06 16:50 ` [PATCH] math: move trivial x86-family sqrt " Alexander Monakov
2020-01-06 17:43 ` [PATCH] math: move i386 sqrtf " Alexander Monakov
2020-01-06 18:32 ` Pascal Cuoq
2020-01-09 15:55 ` Alexander Monakov
2020-01-09 17:00 ` Rich Felker
2020-01-09 21:00 ` Szabolcs Nagy
2020-01-09 22:00 ` Rich Felker
2020-01-09 23:18 ` Szabolcs Nagy
2020-01-10 2:07 ` Rich Felker [this message]
2020-01-10 9:17 ` Szabolcs Nagy
2020-01-14 17:59 ` [musl] " Alexander Monakov
2020-01-14 18:47 ` Szabolcs Nagy
2020-01-07 13:06 ` [PATCH] math: move i386 sqrt " Alexander Monakov
2020-01-08 7:26 ` Rich Felker
2020-03-21 17:53 ` [musl] " Rich Felker
2020-03-21 17:57 ` Rich Felker
2020-03-21 20:30 ` Alexander Monakov
2020-01-11 15:06 ` [PATCH] math: move x86_64 (l)lrint(f) functions " Alexander Monakov
2020-01-11 15:23 ` [PATCH] math: move more x86-family lrint " Alexander Monakov
2020-01-11 16:07 ` Rich Felker
2020-01-11 16:22 ` Rich Felker
2020-01-14 11:54 ` [musl] [PATCH] math: move x86-family rint " Alexander Monakov
2020-01-14 18:17 ` [musl] Q: dealing with missing removal of excess precision Alexander Monakov
2020-01-14 18:50 ` Szabolcs Nagy
2020-01-14 18:58 ` Rich Felker
2020-01-14 19:53 ` Alexander Monakov
2020-02-06 14:51 ` Rich Felker
2020-02-06 17:15 ` Alexander Monakov
2020-02-06 17:46 ` Rich Felker
2020-02-06 19:03 ` Rich Felker
2020-02-06 20:02 ` Rich Felker
2020-02-06 22:08 ` Szabolcs Nagy
2020-02-22 19:59 ` Rich Felker
2020-02-22 20:21 ` Alexander Monakov
2020-02-23 0:19 ` Rich Felker
2020-02-23 16:14 ` Alexander Monakov
2020-03-20 18:12 ` Rich Felker
2020-03-22 1:19 ` Rich Felker
2020-03-22 17:40 ` Alexander Monakov
2020-03-22 17:53 ` Rich Felker
2020-03-22 18:51 ` Alexander Monakov
2020-03-22 19:10 ` Rich Felker
2020-03-22 19:46 ` Alexander Monakov
2020-01-14 20:41 ` [musl] [PATCH] math: move x86-family remainder functions to C Alexander Monakov
2020-01-15 6:54 ` Szabolcs Nagy
2020-01-15 15:44 ` [musl] [PATCH] math: move x86-family fmod " Alexander Monakov
2020-01-16 21:00 ` [musl] [PATCH] math: add x86_64 remquol Alexander Monakov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200110020747.GY30412@brightrain.aerifal.cx \
--to=dalias@libc.org \
--cc=musl@lists.openwall.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://git.vuxu.org/mirror/musl/
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).