2.4.17-pre[12]-ben0 trees hang on unsuspend

2.4.17-pre[12]-ben0 trees hang on unsuspend

[Derrik Pates]

I've been trying the latest rsync snaps from your kernel tree
(particularly to fix the time-warp bug that pops up in 2.4.16-ben0), but
I've discovered that these latest trees cause the iBook FireWire that I'm
using to hang up when I try to bring the unit out of suspend mode. The
"sleep" indicator stops pulsating, and the drive spins up - I can hear it
- but it never wakes up. The backlight never comes on, and typing at the
keyboard (into a root shell - I was at a console logged in as root when I
suspended) has no effect, apparently indicating it's really gone.
Rebooting into 2.4.16-ben0 and suspending still works as normal, so I know
it's not hardware.

One thing to note is that normally, when the system comes out of sleep,
the "sleep" indicator below the display flashes one last time before
coming up out of suspend - when using 2.4.17-pre[12]-ben0, I don't get
that "last flash" when I press a key to wake the system.

Derrik Pates      |   Sysadmin, Douglas School   |    #linuxOS on EFnet
dpates@dsdk12.net |     District (dsdk12.net)    |    #linuxOS on OPN


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Benjamin Herrenschmidt]

>I've been trying the latest rsync snaps from your kernel tree
>(particularly to fix the time-warp bug that pops up in 2.4.16-ben0), but
>I've discovered that these latest trees cause the iBook FireWire that I'm
>using to hang up when I try to bring the unit out of suspend mode. The
>"sleep" indicator stops pulsating, and the drive spins up - I can hear it
>- but it never wakes up. The backlight never comes on, and typing at the
>keyboard (into a root shell - I was at a console logged in as root when I
>suspended) has no effect, apparently indicating it's really gone.
>Rebooting into 2.4.16-ben0 and suspending still works as normal, so I know
>it's not hardware.
>
>One thing to note is that normally, when the system comes out of sleep,
>the "sleep" indicator below the display flashes one last time before
>coming up out of suspend - when using 2.4.17-pre[12]-ben0, I don't get
>that "last flash" when I press a key to wake the system.

Could be one of my sleep.S changes, let me check...

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Derrik Pates]

On Sun, 2 Dec 2001, Benjamin Herrenschmidt wrote:

> Could be one of my sleep.S changes, let me check...

I diff'd 2.4.16-ben0's sleep.S and 2.4.17-pre2-ben0's sleep.S, then
commented out the added stuff for register save/restore on 74x0 PPCs. That
seems to have corrected the problem - the system now wakes up correctly
from sleep, exactly as before. Any idea what's happening there? It looks
like that stuff shouldn't even be getting called for non-74x0 PPC
processors (/proc/cpuinfo IDs this system's CPU as a 750CX), so one of the
processor-type checks apparently isn't working quite right.

Derrik Pates      |   Sysadmin, Douglas School   |    #linuxOS on EFnet
dpates@dsdk12.net |     District (dsdk12.net)    |    #linuxOS on OPN


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Benjamin Herrenschmidt]

>On Sun, 2 Dec 2001, Benjamin Herrenschmidt wrote:
>
>> Could be one of my sleep.S changes, let me check...
>
>I diff'd 2.4.16-ben0's sleep.S and 2.4.17-pre2-ben0's sleep.S, then
>commented out the added stuff for register save/restore on 74x0 PPCs. That
>seems to have corrected the problem - the system now wakes up correctly
>from sleep, exactly as before. Any idea what's happening there? It looks
>like that stuff shouldn't even be getting called for non-74x0 PPC
>processors (/proc/cpuinfo IDs this system's CPU as a 750CX), so one of the
>processor-type checks apparently isn't working quite right.

Some bug in the assembly. I fixed it.

Ben.

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Derrik Pates]

On Thu, 3 Nov 2033, Benjamin Herrenschmidt wrote:

> Some bug in the assembly. I fixed it.

When will the fix show up in the rsync tree?

Derrik Pates      |   Sysadmin, Douglas School   |    #linuxOS on EFnet
dpates@dsdk12.net |     District (dsdk12.net)    |    #linuxOS on OPN


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Derrik Pates]

Never mind. I had to run rsync twice to get it to work. I think a
previous rsync may've been done under the influence of the 2.4.16-ben0
timewarp problem. The fix appears to be in there now.

Derrik Pates      |   Sysadmin, Douglas School   |    #linuxOS on EFnet
dpates@dsdk12.net |     District (dsdk12.net)    |    #linuxOS on OPN


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Jörn Nettingsmeier]

Benjamin Herrenschmidt wrote:
>
> >On Sun, 2 Dec 2001, Benjamin Herrenschmidt wrote:
> >
> >> Could be one of my sleep.S changes, let me check...
> >
> >I diff'd 2.4.16-ben0's sleep.S and 2.4.17-pre2-ben0's sleep.S, then
> >commented out the added stuff for register save/restore on 74x0 PPCs. That
> >seems to have corrected the problem - the system now wakes up correctly
> >from sleep, exactly as before. Any idea what's happening there? It looks
> >like that stuff shouldn't even be getting called for non-74x0 PPC
> >processors (/proc/cpuinfo IDs this system's CPU as a 750CX), so one of the
> >processor-type checks apparently isn't working quite right.
>
> Some bug in the assembly. I fixed it.

any guesses why i'm seeing the bug in 2.4.16-ben0 ?


--
Jörn Nettingsmeier
home://Kurfürstenstr.49.45138.Essen.Germany
phone://+49.201.491621
http://spunk.dnsalias.org
http://www.linuxdj.com/audio/lad/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Jörn Nettingsmeier]

Derrik Pates wrote:
>
> I've been trying the latest rsync snaps from your kernel tree
> (particularly to fix the time-warp bug that pops up in 2.4.16-ben0), but
> I've discovered that these latest trees cause the iBook FireWire that I'm
> using to hang up when I try to bring the unit out of suspend mode. The
> "sleep" indicator stops pulsating, and the drive spins up - I can hear it
> - but it never wakes up. The backlight never comes on, and typing at the
> keyboard (into a root shell - I was at a console logged in as root when I
> suspended) has no effect, apparently indicating it's really gone.
> Rebooting into 2.4.16-ben0 and suspending still works as normal, so I know
> it's not hardware.

i'm seeing the same, but on 2.4.16-ben0 on a pismo powerbook.
it does not happen every time, but often enough to be almost
unusable.

i *think* it gets triggered when i hit a button immediately after
the screen has blanked. when i wait a couple of seconds until the
suspend led has started blinking, i get the machine back most of the
time.

when it dies, the hd comes up, but the screen remains black. i tried
blind-typing "reboot" in a root shell, but nothing happened. i could
not ssh into the box from outside. there are no ping echos.



--
Jörn Nettingsmeier
home://Kurfürstenstr.49.45138.Essen.Germany
phone://+49.201.491621
http://spunk.dnsalias.org
http://www.linuxdj.com/audio/lad/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Benjamin Herrenschmidt]

>
>i'm seeing the same, but on 2.4.16-ben0 on a pismo powerbook.
>it does not happen every time, but often enough to be almost
>unusable.
>
>i *think* it gets triggered when i hit a button immediately after
>the screen has blanked. when i wait a couple of seconds until the
>suspend led has started blinking, i get the machine back most of the
>time.
>
>when it dies, the hd comes up, but the screen remains black. i tried
>blind-typing "reboot" in a root shell, but nothing happened. i could
>not ssh into the box from outside. there are no ping echos.
>

Try the latest rsync and tell me if that still happens.

ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Jörn Nettingsmeier]

Benjamin Herrenschmidt wrote:
>
> >
> >i'm seeing the same, but on 2.4.16-ben0 on a pismo powerbook.
> >it does not happen every time, but often enough to be almost
> >unusable.
> >
> >i *think* it gets triggered when i hit a button immediately after
> >the screen has blanked. when i wait a couple of seconds until the
> >suspend led has started blinking, i get the machine back most of the
> >time.
> >
> >when it dies, the hd comes up, but the screen remains black. i tried
> >blind-typing "reboot" in a root shell, but nothing happened. i could
> >not ssh into the box from outside. there are no ping echos.
> >
>
> Try the latest rsync and tell me if that still happens.

so far, it has not died after resume. but it only went through 3
sleep cycles so far...
btw, alsa compiles now, the feature.h header is there.

thanks,


jörn


--
Jörn Nettingsmeier
home://Kurfürstenstr.49.45138.Essen.Germany
phone://+49.201.491621
http://spunk.dnsalias.org
http://www.linuxdj.com/audio/lad/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Benjamin Herrenschmidt]

>
>so far, it has not died after resume. but it only went through 3
>sleep cycles so far...
>btw, alsa compiles now, the feature.h header is there.
>
>thanks,

feature.h ? That one is gone and have been replaced by pmac_feature.h
for some time now. Maybe Alsa was fixed to handle the new mecanism.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Jörn Nettingsmeier]

Benjamin Herrenschmidt wrote:
>
> >Benjamin Herrenschmidt wrote:
> >>
> >> >
> >> >so far, it has not died after resume. but it only went through 3
> >> >sleep cycles so far...
> >
> >after a few days of frequent use, i can say that sleep/wakeup
> >bahaviour has improved vastly.
> >however there was again one hard lockup yesterday.
> >i was on the road when it hit, so i haven't been able to ping or ssh
> >into the machine.
> >i think it only happens when you strike a key immediately after
> >sleep, before the standby led starts glowing.
>
> Ok, I'll do some experiments and see if I can find something.

let me know if can test something for you.
is there any bulletproof method to capture kernel messages on a
pismo when it goes bang ?
has somebody used the modem port as a serial console ? i have a
spare modem lying around, perhaps it might be worth trying...

> >> >btw, alsa compiles now, the feature.h header is there.
> >> >
> >> >thanks,
> >>
> >> feature.h ? That one is gone and have been replaced by pmac_feature.h
> >> for some time now. Maybe Alsa was fixed to handle the new mecanism.
> >
> >oops, my fault. i had manually copied the feature.h from the
> >"official tree" into yours some days ago, and of course it remained
> >there when i rsynced again....
> >yes, alsa compiles even without that old header. looks like it has
> >been fixed.
> >
> >anyway, it seems that the kernel.org tree still has feature.h, and
> >not pmac-feature.h. if the new mechanism is consensus, it should be
> >propagated to the official tree asap, so that port maintainers
> >become aware of it. my reports about the compile problems to
> >alsa-dev were ignored, because people could not make sense of me
> >whining about a missing header that was obviously there.
>
> Patches are in marcelo's hands, they should be in 2.4.17 before
> it goes final.

good news !


thanks for your quick responses and your help.

best,

jörn

--
Jörn Nettingsmeier
home://Kurfürstenstr.49.45138.Essen.Germany
phone://+49.201.491621
http://spunk.dnsalias.org
http://www.linuxdj.com/audio/lad/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Benjamin Herrenschmidt]

>> >after a few days of frequent use, i can say that sleep/wakeup
>> >bahaviour has improved vastly.
>> >however there was again one hard lockup yesterday.
>> >i was on the road when it hit, so i haven't been able to ping or ssh
>> >into the machine.
>> >i think it only happens when you strike a key immediately after
>> >sleep, before the standby led starts glowing.
>>
>> Ok, I'll do some experiments and see if I can find something.
>
>let me know if can test something for you.
>is there any bulletproof method to capture kernel messages on a
>pismo when it goes bang ?
>has somebody used the modem port as a serial console ? i have a
>spare modem lying around, perhaps it might be worth trying...

xmon_printf (on screen) is usually a working properly if you add
your own messages, but if you are debugging sleep, you may have to
disable display sleep to see the messages.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Michael Schmitz]

> is there any bulletproof method to capture kernel messages on a
> pismo when it goes bang ?
> has somebody used the modem port as a serial console ? i have a
> spare modem lying around, perhaps it might be worth trying...

You should be able to use the modem port as serial console - just use some
terminal emulator or dialer to initiate the connection after the machine
has come up (something as basic as cu will do, as long as you can quit it
without the modem hanging up). Two modems connected with a piece of wire,
one initiating the connection the other answering, is enough.

	Michael


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.17-pre[12]-ben0 trees hang on unsuspend

[Jörn Nettingsmeier]

Michael Schmitz wrote:
>
> > is there any bulletproof method to capture kernel messages on a
> > pismo when it goes bang ?
> > has somebody used the modem port as a serial console ? i have a
> > spare modem lying around, perhaps it might be worth trying...
>
> You should be able to use the modem port as serial console - just use some
> terminal emulator or dialer to initiate the connection after the machine
> has come up (something as basic as cu will do, as long as you can quit it
> without the modem hanging up). Two modems connected with a piece of wire,
> one initiating the connection the other answering, is enough.

thanks. i will try that.


--
Jörn Nettingsmeier
home://Kurfürstenstr.49.45138.Essen.Germany
phone://+49.201.491621
http://spunk.dnsalias.org
http://www.linuxdj.com/audio/lad/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/