2.6.29-rc1 on zaurus (spitz)

2.6.29-rc1 on zaurus (spitz)

[Pavel Machek]

Hi!

2.6.29-rc1 was great step forward over previous kernels: it not only
displays something on screen but also mounts root
filesystem. Unfortunately it then panics with "attempted to kill
init".

I tried both with CONFIG_EABI=N and  CONFIG_EABI=Y, OABI_COMPAT=Y. Any
more tricks I should try?
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.29-rc1 on zaurus (spitz)

[Russell King]

On Wed, Jan 28, 2009 at 12:38:09PM +0100, Pavel Machek wrote:
> Hi!
> 
> 2.6.29-rc1 was great step forward over previous kernels: it not only
> displays something on screen but also mounts root
> filesystem. Unfortunately it then panics with "attempted to kill
> init".
> 
> I tried both with CONFIG_EABI=N and  CONFIG_EABI=Y, OABI_COMPAT=Y. Any
> more tricks I should try?

Don't enable EABI unless you have changed userspace to be EABI compliant.
Also check that you have NWFPE enabled if you're not using a soft-float
userspace.

You could also build the kernel with DEBUG_USER enabled and pass
user_debug=-1 on the kernel's command line, and see if you get any more
information.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Re: 2.6.29-rc1 on zaurus (spitz)

[Pavel Machek]

On Wed 2009-01-28 19:47:23, Russell King wrote:
> On Wed, Jan 28, 2009 at 12:38:09PM +0100, Pavel Machek wrote:
> > Hi!
> > 
> > 2.6.29-rc1 was great step forward over previous kernels: it not only
> > displays something on screen but also mounts root
> > filesystem. Unfortunately it then panics with "attempted to kill
> > init".
> 
> > I tried both with CONFIG_EABI=N and  CONFIG_EABI=Y, OABI_COMPAT=Y. Any
> > more tricks I should try?
> 
> Don't enable EABI unless you have changed userspace to be EABI compliant.
> Also check that you have NWFPE enabled if you're not using a soft-float
> userspace.

Probleem seems to be fixed in 2.6.29-rc3. 2.6.28 also works... (but
won't charge the battery :-().

> You could also build the kernel with DEBUG_USER enabled and pass
> user_debug=-1 on the kernel's command line, and see if you get any more
> information.

Thanks for support!

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.29-rc1 on zaurus (spitz)

[Cyril Hrubiš]

Hi,
2.6.29-rc4 boots on spitz and looks overall good but framebuffer
redraws some lines terribly slow. When I'm editing text in bash from
the middle in some cases I can see redrawing the text pixel by pixel.
And suspend to ram on pressing On/Off button does nothing now (some
missing config option?).
-- 
watashi |v|3+@n
Cyril Hrubis
metan@ucw.cz
http://atrey.karlin.mff.cuni.cz/~metan/

Re: 2.6.29-rc1 on zaurus (spitz)

[Cyril Hrubiš]

Hi,
The display problems seems to be cpufreq related. It changes type of
bad effects when varying cpu speed from slow and random line redraw to
flashing big white random patterns. And I've seen similar behaviour
with voipac pxa module where framebuffer with cpu set to 416Mhz moves
accidentally some pixels to the right under cpu load. Moreover several
speed changes leads to "hda: lost interrupt" on spitz.

-- 
watashi |v|3+@n
Cyril Hrubis
watashi.metan@gmail.com
http://atrey.karlin.mff.cuni.cz/~metan/

Re: 2.6.29-rc1 on zaurus (spitz)

[Pavel Machek]

Hi!

> And suspend to ram on pressing On/Off button does nothing now (some
> missing config option?).

It does not work for me either... but neither it does in 2.6.28...
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.29-rc1 on zaurus (spitz)

[Pavel Machek]

On Thu 2009-02-12 22:22:37, Cyril Hrubi? wrote:
> Hi,
> 2.6.29-rc4 boots on spitz and looks overall good but framebuffer
> redraws some lines terribly slow. When I'm editing text in bash from
> the middle in some cases I can see redrawing the text pixel by pixel.
> And suspend to ram on pressing On/Off button does nothing now (some
> missing config option?).

Hmm, it gets worse: echo mem > /sys/power/state sleeps 2.6.28, but
there's no way to resume it.

About hda timeouts... I remember seeing them under older kernels, too,
under loads such as kernel compile... plus keypresses tend to get lost
under high loads :-(.
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.29-rc1 on zaurus (spitz)

[Russell King]

On Sun, Feb 22, 2009 at 02:56:51PM +0100, Pavel Machek wrote:
> On Thu 2009-02-12 22:22:37, Cyril Hrubi? wrote:
> > Hi,
> > 2.6.29-rc4 boots on spitz and looks overall good but framebuffer
> > redraws some lines terribly slow. When I'm editing text in bash from
> > the middle in some cases I can see redrawing the text pixel by pixel.
> > And suspend to ram on pressing On/Off button does nothing now (some
> > missing config option?).
> 
> Hmm, it gets worse: echo mem > /sys/power/state sleeps 2.6.28, but
> there's no way to resume it.

Sounds like the on/off button isn't configured as a wakeup source.
Without knowing anything about the platform (iow, where the on/off
button is connected) it's hard to tell what could be the problem.

Has the on/off button ever worked?

> About hda timeouts... I remember seeing them under older kernels, too,
> under loads such as kernel compile... plus keypresses tend to get lost
> under high loads :-(.

Really, what we need is for someone to step up to take on sorting out
Zaurus.  Looking at this thread, it seems there's very few people here
willing.  Also, I think Richard Purdie has moved on from Zaurus stuff.

I know there are a small number of people on linux-arm-kernel who have
access to some Zaurus platforms, but I don't know what they are.  Really,
you should be posting to that list if you seriously want help, because
I doubt they're subscribed to LKML.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Re: 2.6.29-rc1 on zaurus (spitz)

[Richard Purdie]

On Sun, 2009-02-22 at 14:23 +0000, Russell King wrote:
> On Sun, Feb 22, 2009 at 02:56:51PM +0100, Pavel Machek wrote:
> > On Thu 2009-02-12 22:22:37, Cyril Hrubi? wrote:
> > > Hi,
> > > 2.6.29-rc4 boots on spitz and looks overall good but framebuffer
> > > redraws some lines terribly slow. When I'm editing text in bash from
> > > the middle in some cases I can see redrawing the text pixel by pixel.
> > > And suspend to ram on pressing On/Off button does nothing now (some
> > > missing config option?).
> > 
> > Hmm, it gets worse: echo mem > /sys/power/state sleeps 2.6.28, but
> > there's no way to resume it.
> 
> Sounds like the on/off button isn't configured as a wakeup source.
> Without knowing anything about the platform (iow, where the on/off
> button is connected) it's hard to tell what could be the problem.

Its connected in through GPIO 0.

> Has the on/off button ever worked?

Yes, this always used to work fine for both suspending and resuming.

> > About hda timeouts... I remember seeing them under older kernels, too,
> > under loads such as kernel compile... plus keypresses tend to get lost
> > under high loads :-(.
> 
> Really, what we need is for someone to step up to take on sorting out
> Zaurus.  Looking at this thread, it seems there's very few people here
> willing.  Also, I think Richard Purdie has moved on from Zaurus stuff.

Its really a time problem. I couldn't keep up with the PXA architecture
changes and give enough time to review and test them all effectively and
things have slipped from there. I've tried to provide input where I can
but would certainly welcome any help. My new employer means I'll be
doing less ARM work in general too for some reason...

Cheers,

Richard

-- 
Richard Purdie
Intel Open Source Technology Centre

Re: 2.6.29-rc1 on zaurus (spitz)

[Pavel Machek]

> > About hda timeouts... I remember seeing them under older kernels, too,
> > under loads such as kernel compile... plus keypresses tend to get lost
> > under high loads :-(.
> 
> Really, what we need is for someone to step up to take on sorting out
> Zaurus.  Looking at this thread, it seems there's very few people here
> willing.  Also, I think Richard Purdie has moved on from Zaurus stuff.
> 
> I know there are a small number of people on linux-arm-kernel who have
> access to some Zaurus platforms, but I don't know what they are.  Really,
> you should be posting to that list if you seriously want help, because
> I doubt they're subscribed to LKML.

linux-arm-kernel seems subscriber-only :-(.

Anyway, new problem blocking testing is that somewhere between
2.6.29-rc3 and -rc7, pcmcia support stopped working => hda is no
longer detected.
								Pavel  

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 2.6.29-rc1 on zaurus (spitz)

[Russell King]

On Sat, Mar 14, 2009 at 05:24:39PM +0100, Pavel Machek wrote:
> > > About hda timeouts... I remember seeing them under older kernels, too,
> > > under loads such as kernel compile... plus keypresses tend to get lost
> > > under high loads :-(.
> > 
> > Really, what we need is for someone to step up to take on sorting out
> > Zaurus.  Looking at this thread, it seems there's very few people here
> > willing.  Also, I think Richard Purdie has moved on from Zaurus stuff.
> > 
> > I know there are a small number of people on linux-arm-kernel who have
> > access to some Zaurus platforms, but I don't know what they are.  Really,
> > you should be posting to that list if you seriously want help, because
> > I doubt they're subscribed to LKML.
> 
> linux-arm-kernel seems subscriber-only :-(.

Yes.

> Anyway, new problem blocking testing is that somewhere between
> 2.6.29-rc3 and -rc7, pcmcia support stopped working => hda is no
> longer detected.

lalala, I'm not listening.

Really, if you want help, I've pointed you at the place where you can
get that.  I'm not able to give you any other help with your Zaurus
problems.

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Re: 2.6.29-rc1 on zaurus (spitz)

[Russell King]

And, btw, please drop Pavel Urban's email address from future replies,
it seems his address is not valid.

  pavel.urban@ct.cz
    SMTP error from remote mail server after RCPT TO:<pavel.urban@ct.cz>:
    host m-in.ct.cz [194.228.96.22]: 553 5.3.0 <pavel.urban@ct.cz>... Not in DB

-- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:

Re: 2.6.29-rc1 on zaurus (spitz)

[Pavel Machek]

Hi!

> On Sat, Mar 14, 2009 at 05:24:39PM +0100, Pavel Machek wrote:
> > > > About hda timeouts... I remember seeing them under older kernels, too,
> > > > under loads such as kernel compile... plus keypresses tend to get lost
> > > > under high loads :-(.
> > > 
> > > Really, what we need is for someone to step up to take on sorting out
> > > Zaurus.  Looking at this thread, it seems there's very few people here
> > > willing.  Also, I think Richard Purdie has moved on from Zaurus stuff.
> > > 
> > > I know there are a small number of people on linux-arm-kernel who have
> > > access to some Zaurus platforms, but I don't know what they are.  Really,
> > > you should be posting to that list if you seriously want help, because
> > > I doubt they're subscribed to LKML.
> > 
> > linux-arm-kernel seems subscriber-only :-(.
> 
> Yes.
> 
> > Anyway, new problem blocking testing is that somewhere between
> > 2.6.29-rc3 and -rc7, pcmcia support stopped working => hda is no
> > longer detected.
> 
> lalala, I'm not listening.

lalala... I think I tried to subscribe to linux-arm-kernel, but all I
got is automated reply and no flood of messages. Is subscription
moderated, too?


								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html