From: Stafford Horne via Libc-alpha <libc-alpha@sourceware.org>
To: Joseph Myers <joseph@codesourcery.com>
Cc: Openrisc <openrisc@lists.librecores.org>,
GLIBC patches <libc-alpha@sourceware.org>
Subject: Re: [PATCH 0/1] OpenRISC port
Date: Sat, 23 May 2020 06:01:54 +0900 [thread overview]
Message-ID: <20200522210154.GA75760@lianli.shorne-pla.net> (raw)
In-Reply-To: <alpine.DEB.2.21.2005221846590.10437@digraph.polyomino.org.uk>
On Fri, May 22, 2020 at 06:52:20PM +0000, Joseph Myers wrote:
> On Fri, 22 May 2020, Stafford Horne via Libc-alpha wrote:
>
> > Hi All,
> >
> > This is the patch for the OpenRISC port of glibc.
>
> Please see <https://sourceware.org/glibc/wiki/NewPorts> for information to
> include in a new port submission. That includes pointing to architecture
> and ABI manuals.
Thank you for the pointers, I will include this in v2. But the architecture
manual is here:
- Page: https://openrisc.io/architecture
- PDF: https://raw.githubusercontent.com/openrisc/doc/master/openrisc-arch-1.3-rev1.pdf
- Architecture Revision Required for Full Hard Float support:
- https://openrisc.io/proposals/p17-user-mode-fpcsr
> > The original patch was put together by Christian Scensson a few years ago. I
> > have taken the patch and finished all of the TODO items and run and fixed all of
> > the test failures.
>
> Do you mean Christian Svensson? (There's a copyright assignment from
> Christian Svensson for GCC GLIBC BINUTILS from 2014.)
Yes Christian Svensson, sorry for the typo. He is listed in the changelog on
the patch.
>
> > TEST RESULTS
> >
> > Pretty much all tests are passing however there are several math (fpu) tests
> > which are failing due to issues with setting underflow/overflow flags. These
>
> There should not be any such failures with a correct port and architecture
> implementation.
OK, I will look into it again. The last I remember there was somethign strange
happening with underflow exception not being set in the GCC FPU emulation code.
Note, the OpenRISC implementation I tested with has only 32-bit FPU hard
support. This meant that C API for log/cos/sin etc which internally are defined
to use 64-bit double's all end up using emulation.
> > I can provide more details.
>
> Full test results should be provided (see the above mentioned wiki page).
I will do with the next round of patches. This is all very helpful.
-Stafford
prev parent reply other threads:[~2020-05-22 21:02 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-22 11:36 [PATCH 0/1] OpenRISC port Stafford Horne via Libc-alpha
2020-05-22 11:36 ` [PATCH 1/1] Initial support for OpenRISC Stafford Horne via Libc-alpha
2020-05-22 12:56 ` Florian Weimer via Libc-alpha
2020-05-22 21:59 ` Stafford Horne via Libc-alpha
2020-05-22 22:02 ` Joseph Myers
2020-05-22 22:32 ` Stafford Horne via Libc-alpha
2020-06-10 20:17 ` Stafford Horne via Libc-alpha
2020-05-22 20:56 ` Joseph Myers
2020-05-22 21:31 ` Stafford Horne via Libc-alpha
2020-05-22 18:52 ` [PATCH 0/1] OpenRISC port Joseph Myers
2020-05-22 21:01 ` Stafford Horne via Libc-alpha [this message]
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
List information: https://www.gnu.org/software/libc/involved.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200522210154.GA75760@lianli.shorne-pla.net \
--to=libc-alpha@sourceware.org \
--cc=joseph@codesourcery.com \
--cc=openrisc@lists.librecores.org \
--cc=shorne@gmail.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.
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 read-only IMAP folder(s) and NNTP newsgroup(s).