On Wed, 14 Oct 2015 20:26:32 +0100
Ken Moffat <[email protected]> wrote:

> On Wed, Oct 14, 2015 at 08:01:49PM +0100, Hazel Russman wrote:
> > On Wed, 14 Oct 2015 19:33:49 +0100
> > Ken Moffat <[email protected]> wrote:
> > 
> > > On Wed, Oct 14, 2015 at 06:04:47PM +0100, Hazel Russman wrote:
> > > > > >
> > > > > > You can see that version 8.13 of coreutils does not have
> > > > > > this bug. 8.23 and 8.24 both have it, but only when running
> > > > > > in an xterm (and 8.23 did not have those tests so it
> > > > > > wouldn't have shown up). Did you perhaps do your runs in a
> > > > > > virtual console?
> > > > > 
> > > > > I run almost everything over ssh, but I went to my
> > > > > development system and ran it in both konsole and xterm on an
> > > > > xfce system and it worked properly for me.
> > > > > 
> > > > > Another wild guess -- a difference in kernel config?
> > > > > 
> > > > >    -- Bruce
> > > > 
> > > > Well, it must have something to do with devices. It might be
> > > > significant that Crux and LFS are the systems that have
> > > > home-made kernels; the Debian one is stock. What settings
> > > > should I be looking for? A whole kernel config file is a bit
> > > > big to post.
> > > > 
> > > > It would be interesting if a few other people could try this --
> > > > see if they agree with you or with me.
> > > > 
> > > > Hazel
> > > 
> > > I've built 7.8-rc (as it was) four times on real hardware, all
> > > x86_64.  In each case, my logs show:
> > > 
> > > PASS: tests/misc/seq-precision.sh
> > > 
> > > I did not keep a note of what sort of term I was using (didn't
> > > think I'd be asked :) but probably all were using rxvt-unicode -
> > > for the server, that was over ssh from a desktop box.
> > > 
> > > The desktops were variously: from 7.7, from 7.8 (a second system
> > > on my test box), from June svn.  The server build was from 7.6.
> > > 
> > > In the desktop boxes I put a new kernel on the host (similar to
> > > what I'm going to build for LFS - to try to find out any kernel
> > > problems before I build the new system) and use that.  For the
> > > server, I am using a 4.1.8 kernel and the old 7.6 system had
> > > already moved to 4.1.1.
> > > 
> > > At the moment, google is only finding this thread, and Frans de
> > > Boer's report from July -
> > > http://comments.gmane.org/gmane.linux.lfs.support/40574 and
> > > http://lists.opensuse.org/opensuse/2015-07/msg00380.html
> > > but of course it is possible that few people have run the tests.
> > > 
> > > Sorry I can't add anything useful.
> > > 
> > Those refs look spot on to me. I wonder why I couldn't find them!
> > Unfortunately they don't include any explanations.
> 
> I vaguely remembered the lfs post, but it did not show up for me in
> my first set of google results (or if it did, it was not near the
> top).
> 
> > 
> > Unlike you I build with the normal host kernel, which is 3.19 in
> > this case. I've diffed it with the corresponding kernel from Anduin
> > (not the one in Bruce's link, which is the later version) and I'm
> > attaching the result to this post. I hope your majordomo honours
> > attachments.
> > 
> > Hazel
> 
> One of the things I've always done is try newer kernels -
> fortunately, my altered .config for 4.2 now no longer forces
> everything onto the first CPU (that caught someone else this past
> week - I hit it weeks ago in an early -rc :)  Of course, the
> downside of testing development kernels is that I end up having to
> bisect - usually at a most inconvenient time.
> 
> If I read correctly, you have CONFIG_FHANDLE set - is this a systemd
> build ?  Nothing else jumps out, and I cannot see why that setting
> itself would be a problem.
> 
> ĸen

I tried it out on my laptop, which runs 32-bit systems: Mint Quiania
(coreutils-8.21) and LFS-7.7-systemd (coreutils-8.23). In both cases the
operation ended normally, no hanging about. It *has* to be something in
the kernel. Somehow that SIGPIPE is not getting sent. It doesn't
really matter a lot but I am going crazy trying to think of a logical
explanation.

I asked some of my friends at Linux Forums to try it. So far two have
(one with 8.24 on Arch), and both got a normal exit.

-- 

H Russman
-- 
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

Reply via email to