On Thu, May 28, 2020 at 03:28:17AM +0100, Ken Moffat wrote:
> On Thu, May 28, 2020 at 02:08:08AM +0100, Ken Moffat wrote:
> > On Wed, May 27, 2020 at 10:13:12PM +0100, Ken Moffat wrote:
> > > On Wed, May 27, 2020 at 12:43:26PM +0800, Xi Ruoyao wrote:
> > > >
> > >
> > > When the -O0 override in glibc errored, I changed that to -O1 (still
> > > with -march=native). Tests on bash still showed the SIGFPE falure,
> > > but I suspect I need to boot the system to properly test it.
> > > Unfortunately it locked up overnight (fairly common on that box) so
> > > I have not yet got to that stage.
> > >
> > > Just to be clear, at the moment I'm only trying to detune glibc.
> > >
> >
> > Result! In chroot bash still had the SIGFPE failures, and check had
> > the two check_check and check_check_exports failures which prompted
> > all of this.
> >
>
> Too much editing and revision before sending, failed to read it -
>
> > Booted. System is trunk, built from an April build. FLAGS of -O3
> > and -march=native, with hardening where appropriate (not glibc!),
> > but for glibc I reduced the FLAGS to -O1 -march=native.
> >
>
> *Host* system was from an April build, the new system is trunk.
>
> > Built check and ran the tests, all passed.
> >
> > This is on my ryzen 1300X and it's clear that having a "weird or
> > defective" glibc underneath chroot makes running these tests there
> > not useful
> >
> > I'm not at all convinced that I want to continue with -O1 in glibc.
> > I'll throw this one away and start another build using glibc's
> > default flags (-O2 -g without -march=native).
> >
> > To be continued, I hope ;-) Meanwhile, thanks to everyone for the
> > help and suggestions.
> >
Building glibc with '-O2 -g' the check tests succeeded after I
booted. Will keep this system for a few days (I need a 4 core
gcc-10 machine for firefox-68.9.0), then I'll see if adding
-march=native affects things : hopefully, I'll use this system as
the host, so that I can run the tests in chroot.
But still no closer to working out why this was breaking in 2018
when we first added 'check'. I had an old comment that I forced -O3
in glibc on x86_64 (I think I based it on an old suggestion from
Greg Schafer, but google can't find anything relevant), but the added
flags had dropped back to only -pipe several years ago. And then at
times -march=native appeared in my glibc build.
And I suppose I must remember to update my last year's notes on
'tuning' when I understand exactly what caused my raising of sigfpe
to break.
ĸen
--
Do you not know that, what you belittle by the name tree is but the
mere four-dimensional analogue of a whole multidimensional universe
which - no, I can see you do not. -- Druellae (a Dryad)
--
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
Do not top post on this list.
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
http://en.wikipedia.org/wiki/Posting_style
