Re: [9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)

Re: [9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)

[presotto]

[-- Attachment #1: Type: text/plain, Size: 95 bytes --]

Perhaps we're not getting memory to refresh itself when
we sleep and its contents is decaying?

[-- Attachment #2: Type: message/rfc822, Size: 4079 bytes --]

From: Axel Belinfante <Axel.Belinfante@cs.utwente.nl>
To: 9fans@cse.psu.edu
Subject: [9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)
Date: Tue, 22 Oct 2002 23:38:27 +0200
Message-ID: <200210222138.g9MLcRe12274@zamenhof.cs.utwente.nl>

For your information:
with my (admittedly hacked) kernel (based on the tar ball),
(and minor hacking of the dual sleeve code) I get:

without sleeve, light is slightly on before suspending;
suspending only once rio has started (booting completed):

 suspend followed by resume within ca. 2 seconds: OK
  (and I see the debugging code on the serial line console)

 suspend followed by resume later than ca. 2 seconds: not OK
  display comes back (most of the times) but light is not
  switched back on and screen/buttons don't respond, and
  debugging output on console is garbled (if it appears at all)


With the empty sleeve I get more or less the same effect
(i.e. quick suspend/resume within the ca 2 (3?) seconds works)

So, I have a short `window' in which I can resume; if I wait too long,
I can see the screen, but that seems to be all.

With the wavelan in the sleeve, suspend button seems to do almost
nothing -- except that ping no longer works, even though the
one led on the wavelan remains continously lit, and the other
still occasionally blinks. However, cat/net/ipselftab returns
nothing. Now running startip gives me:
'echo: write error: couldn't configure device'.

Axel.


> Ok. I'll build a fresh new kernel with the sources in the tar ball
> and let you know of any problem.
> In the mean time, you could try downloading the tar ball again, trying
> it, and letting me know the behaviour of suspend/resume on your bitsy.

Re: [9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)

[Fco.J.Ballesteros]

I'm having problems.

A fresh new kernel built here ends up causing
"conswrite: 0x200000000"
messages (apparently, devcons got a write with a bad qid.path).
I'm not sure, but I think that happen to me
when I first tried to boot the bitsy time ago.
I don't remember what I did to fix that. Ugh.

The funny thing is that it seems to properly suspend the
thing and resume it. (The messages seem to be caused from the
system calls made by boot in the bitsy).

I'll trace the problem when I get some time (don't know if I'll be able to
fix it this afternoon).

Re: [9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)

[Axel Belinfante]

> BTW, did you try the tar ball w/o your hacks?

I think so. But I could not run it ``out of the box'' because I had
to make minor changes to compile. My memory is fuzzy about suspend
resume working or not with it. I'm sorry. I can of course try it again.
If you have a 9bitsy that works for you, maybe I could try that?
(to rule out source/version differences).

Probably I should do one or more of the following:
 - try again the tar ball (with the minimal changes needed to compile)
 - try a compiled 9bitsy that works for you (instead of recompiling myself)
 - wait until you have had time to reintegrate your changes with the
   rest of the system

I won't have time for it before the evening anyway...
Axel.

Re: [9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)

[Fco.J.Ballesteros]

> So, I have a short `window' in which I can resume; if I wait too long,
> I can see the screen, but that seems to be all.

What you say is the behaviour I got before getting the suspend
code working for me. It had to do with some problems
in the power suspend procedure that could broke the ram refresh.

I'll try to get some time today to try a fresh kernel out of
the tar ball, to see if we can get somewhere.

BTW, did you try the tar ball w/o your hacks?

[9fans] bitsy suspend/resume (was: ipaq dual pcmcia sleeve)

[Axel Belinfante]

For your information:
with my (admittedly hacked) kernel (based on the tar ball),
(and minor hacking of the dual sleeve code) I get:

without sleeve, light is slightly on before suspending;
suspending only once rio has started (booting completed):

 suspend followed by resume within ca. 2 seconds: OK
  (and I see the debugging code on the serial line console)

 suspend followed by resume later than ca. 2 seconds: not OK
  display comes back (most of the times) but light is not
  switched back on and screen/buttons don't respond, and
  debugging output on console is garbled (if it appears at all)


With the empty sleeve I get more or less the same effect
(i.e. quick suspend/resume within the ca 2 (3?) seconds works)

So, I have a short `window' in which I can resume; if I wait too long,
I can see the screen, but that seems to be all.

With the wavelan in the sleeve, suspend button seems to do almost
nothing -- except that ping no longer works, even though the
one led on the wavelan remains continously lit, and the other
still occasionally blinks. However, cat/net/ipselftab returns
nothing. Now running startip gives me:
'echo: write error: couldn't configure device'.

Axel.


> Ok. I'll build a fresh new kernel with the sources in the tar ball
> and let you know of any problem.
> In the mean time, you could try downloading the tar ball again, trying
> it, and letting me know the behaviour of suspend/resume on your bitsy.