Philip Webb writes:

> There is a problem somewhere & it mb a small fix for my machine
> & it wb nice to track it down if others are not experiencing it.

Try 'strace xterm' in a terminal with long scrollback history, press the
enter key a couple of times when it hangs, and analyze the output at these
positions later. Or send these parts to the list, maybe someone here will
spot what's going on.

Wonko

On Wed, 2009-11-11 at 16:41 -0500, Philip Webb wrote:
> 091111 Mick wrote:
> > On Wednesday 11 November 2009 13:34:49 Philip Webb wrote:
> >> 091111 Mick wrote:
> >>> xpdf takes less that 1 second to pop up when started from a terminal
> >>> in Fluxbox, but I am not running xorg-x11.
> >> What are you running to support graphics for Fluxbox ??
> > Is this what you mean, or are you talking about the drivers that I use?
> > ==========================================
> > [i] x11-base/xorg-server

...
> So you are running Xorg-x11 ! -- sorry for the quibble.

For the record xorg-x11 is the meta-package that pulls in lots of stuff,
xorg-server is the minimal ebuild that xorg-x11 would pull in.

Phillip, what version did xorg-server go from/to when you upgraded?

the testing nvidia-drivers look a bit suspicious, is it worth seeing
what the problem was with the stable version?

Why did you pick 185.18.36-r1? Is there a requirement for the 18x
series? The latest "unstable" (but not masked) is 190.42-r2.

--
Iain Buchanan <iaindb at netspace dot net dot au>

"God is a comedian playing to an audience too afraid to laugh."
- Voltaire

On Wed, 2009-11-11 at 12:12 +0000, Mick wrote:

> The localhost issue in /etc/hosts only affected KDE as far as I
> recall.

Unfortunately not. I've had it a few times with gnome, but not since
the early days... I think it was an X issue, not related to the DE,
afair

--
Iain Buchanan <iaindb at netspace dot net dot au>

This door is baroquen, please wiggle Handel.
(If I wiggle Handel, will it wiggle Bach?)
-- Found on a door in the MSU music building

On Thursday 12 November 2009 00:40:16 Alex Schuster wrote:
> Philip Webb writes:
> > There is a problem somewhere & it mb a small fix for my machine
> > & it wb nice to track it down if others are not experiencing it.
>
> Try 'strace xterm' in a terminal with long scrollback history, press the
> enter key a couple of times when it hangs, and analyze the output at these
> positions later. Or send these parts to the list, maybe someone here will
> spot what's going on.
>
> Wonko
>


I did that earlier just for fun. The output is huge :-)

I see here that while xterm is starting, the output stutters quite a lot and
each time there is a "resource unavailable" message just before. It's trying
to read fd3 which turns out to be the Unix socket in /tmp

If anyone feels like trawling through strace output, I can post what happens
here. xterm is now taking 4 seconds to start whereas before it was almost
instant


--
alan dot mckinnon at gmail dot com

091112 Alan McKinnon wrote:
> On Thursday 12 November 2009 00:40:16 Alex Schuster wrote:
>> Philip Webb writes:
>>> There is a problem somewhere & it mb a small fix for my machine
>>> & it wb nice to track it down if others are not experiencing it.
>> Try 'strace xterm' in a terminal with long scrollback history, ...

Or better, use the '-o outputfile' option.

>> ... press Enter a couple of times when it hangs
>> and analyze the output at these positions later.
> I did that earlier just for fun. The output is huge :-)

> 2400 lines when I did it with the same result as yours.

> I see here that while xterm is starting, the output stutters quite a lot
> and each time there is a "resource unavailable" message just before.
> It's trying to read fd3 which turns out to be the Unix socket in /tmp .

Yes, there are 3 lock file/dirs there :

root:527 tmp> ls -la
total 5
drwxrwxrwt 7 root root 200 2009-11-17 16:41 .
drwxr-xr-x 23 root root 560 2009-07-04 05:53 ..
drwxrwxrwt 2 root root 88 2009-11-17 12:50 .ICE-unix
drwx------ 3 purslow purslow 192 2009-11-17 16:30 kde-purslow
drwx------ 2 purslow purslow 240 2009-11-17 16:32 ksocket-purslow
-r--r--r-- 1 root purslow 11 2009-11-17 12:47 .X0-lock
drwxrwxrwt 2 root root 72 2009-11-17 12:47 .X11-unix

Are you using Nvidia ? -- someone else mentioned that as a possible cause.

Otherwise, does anyone have further suggestions ?

--
========================,,======================== ====================
SUPPORT ___________//___, Philip Webb
ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto
TRANSIT `-O----------O---' purslowatchassdotutorontodotca

On Tuesday 17 November 2009 23:56:00 Philip Webb wrote:
> 091112 Alan McKinnon wrote:
> > On Thursday 12 November 2009 00:40:16 Alex Schuster wrote:
> >> Philip Webb writes:
> >>> There is a problem somewhere & it mb a small fix for my machine
> >>> & it wb nice to track it down if others are not experiencing it.
> >>
> >> Try 'strace xterm' in a terminal with long scrollback history, ...
>
> Or better, use the '-o outputfile' option.
>
> >> ... press Enter a couple of times when it hangs
> >> and analyze the output at these positions later.
> >
> > I did that earlier just for fun. The output is huge :-)
> >
> > 2400 lines when I did it with the same result as yours.
> >
> > I see here that while xterm is starting, the output stutters quite a lot
> > and each time there is a "resource unavailable" message just before.
> > It's trying to read fd3 which turns out to be the Unix socket in /tmp .
>
> Yes, there are 3 lock file/dirs there :
>
> root:527 tmp> ls -la
> total 5
> drwxrwxrwt 7 root root 200 2009-11-17 16:41 .
> drwxr-xr-x 23 root root 560 2009-07-04 05:53 ..
> drwxrwxrwt 2 root root 88 2009-11-17 12:50 .ICE-unix
> drwx------ 3 purslow purslow 192 2009-11-17 16:30 kde-purslow
> drwx------ 2 purslow purslow 240 2009-11-17 16:32 ksocket-purslow
> -r--r--r-- 1 root purslow 11 2009-11-17 12:47 .X0-lock
> drwxrwxrwt 2 root root 72 2009-11-17 12:47 .X11-unix
>
> Are you using Nvidia ? -- someone else mentioned that as a possible cause.

Yes, nvidia-190.42-r3

I've noticed recently that kdm takes looooooong to start - more than 30
seconds, and xterm is still taking over 3 seconds to load.

I'll try downgrade nvidia tomorrow when I have a chance and see what happens

>
> Otherwise, does anyone have further suggestions ?
>

--
alan dot mckinnon at gmail dot com

091118 Alan McKinnon wrote:
> On Tuesday 17 November 2009 23:56:00 Philip Webb wrote:
>> Are you using Nvidia ? -- someone else mentioned that as a possible cause.
> Yes, nvidia-190.42-r3
> I'll try downgrade nvidia tomorrow and see what happens

I'm using 185 & couldn't get 180 to compile with Kernel 2.6.31 ,
so it looks as if Nvidia isn't the problem.

> I've noticed recently that kdm takes looooooong to start -
> more than 30 seconds, and xterm is still taking over 3 seconds to load.

I've tried starting it from another Xterm & as root & there's no difference.
I don't use Kdm, but 'startx' from a raw terminal into Fluxbox.

It looks like a bug for the developers of X .

--
========================,,======================== ====================
SUPPORT ___________//___, Philip Webb
ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto
TRANSIT `-O----------O---' purslowatchassdotutorontodotca

On Tuesday 17 November 2009 23:00:23 Alan McKinnon wrote:
> On Tuesday 17 November 2009 23:56:00 Philip Webb wrote:
> > Are you using Nvidia ? -- someone else mentioned that as a possible
> > cause.
>
> Yes, nvidia-190.42-r3
>
> I've noticed recently that kdm takes looooooong to start - more than 30
> seconds, and xterm is still taking over 3 seconds to load.
>
> I'll try downgrade nvidia tomorrow when I have a chance and see what
> happens

I've tried switching to the nv kernel driver and it made no difference here.
I'm pretty sure it's an X problem. I'd better try reverting X to an older
version.

--
Rgds
Peter

091118 Peter Humphrey wrote:
> I've tried switching to the nv kernel driver and it made no difference.
> I'm pretty sure it's an X problem.

Extensive Googling found 1 useful hint of where the problem might lie:

http://stackoverflow.com/questions/1087580/resource-temporarily-unavailable-in-boost-asio

"'Resource temporarily unavailable' is normally the text description
for EAGAIN, indicating that the operation should be retried ...
It's generally worth looking at the man page
for the underlying libc function; which is recvfrom in this case"

Yes, 'man recvfrom' has rather opaque info which mb relevant ... (sigh)

--
========================,,======================== ====================
SUPPORT ___________//___, Philip Webb
ELECTRIC /] [] [] [] [] []| Cities Centre, University of Toronto
TRANSIT `-O----------O---' purslowatchassdotutorontodotca

On Thursday 19 November 2009 04:33:14 Philip Webb wrote:

> Extensive Googling found 1 useful hint of where the problem might lie:
>
>
> http://stackoverflow.com/questions/1087580/resource-temporarily-unavaila
> ble-in-boost-asio
>
> "'Resource temporarily unavailable' is normally the text description
> for EAGAIN, indicating that the operation should be retried ...
> It's generally worth looking at the man page
> for the underlying libc function; which is recvfrom in this case"
>
> Yes, 'man recvfrom' has rather opaque info which mb relevant ... (sigh)

Relevant, perhaps, but not useful - to me, at least. I'll just have to wait
for the next release of X. Thanks anyway.

--
Rgds
Peter