Re: [9fans] pm in the bitsy

Re: [9fans] pm in the bitsy

[presotto]

Any wakeup should appear as a reset to the chip causing it to jump
to the flash bootloader.  The reason the code is turned off is that
we haven't had the guts to change the boot loader yet - one mistake and
it's get a new bitsy.

The linux source has some pointers and all the hardware manuals are on
line pointed to by handhelds.org.

Re: [9fans] pm in the bitsy

[forsyth]

>>You'll also have to learn Ken's loader which may be even harder.  It
>>still confuses me occasionally.

you'll need to dig in sooner rather than later:
the latest bootstrap on the ipaq i've got doesn't
understand the -H5 header 5l generates
because they've changed it again.
i'm trying to work out exactly what the new one
needs.  my last hurried attempts a few days ago ought to
have worked, but didn't.  i'll go through it again more slowly!

at vita we've also got some small but important bug fixes to 5l that i'll
retrofit.

Re: [9fans] pm in the bitsy

[Sape Mullender]

Nemo,

> I only took a quick look, so this could be wrong; but I think that
> rs232, lcd, audio, and of course sa1100 are doing a power off and have
> the code to get back on. Perhaps the pcmcia/wavelan will not do so (I
> didn't check that yet).

Peter Bosch — who joined our group a few months ago — is working on
code for all of our architectures to support dynamic pcmcia and cardbus card
insertion and removal.  Power management will fall out of that.  If you want,
you can do a quick hack for the Wavelan, but there will be better code at
some point.  The power code in the other drivers does exist but hasn't been
tested yet.  I do turn power off on close in the audio driver, so powering
in/off is not a problem, but I haven't tested how open-file state survives
power-off followed by power-on.

> Regarding the power-on entry point, my boot loader is already jumping
> (well, not sure yet, but looks like), to the address kept in
> powerregs->pspr.  The linux code was kind of a tutorial here.

> So, if I'm not so mistaken, I think I'd only need to save all the
> processor state and write an assembly routine to reload it (placing its
> address into powerregs->pspr in sa1100_power_off).

That's the impression I got.  A good idea might be to make the first
instruction of pwer-restore be to reload powerregs->pspr with the
cold-boot address.  If things go wrong in the power-up code, at least
you'll come back to life by recycling power.

> My only problem is that I'm learning arm assembler while trying to get
> this working, so I'm not too optimistic regarding the final result.

You'll also have to learn Ken's loader which may be even harder.  It
still confuses me occasionally.

> In any case, if you still think I'm missing something (well, probably
> many things) and you have a minute to let me know, that would be great.

We'd love it if you could get started on power management for the BItsy and
we'll definitely help you where we can.

By the way, during the conversion to 9P2000, I added the power and
configure entry points to the devtab for all of our systems — they only
existed for the bitsy.  Peter Bosch will have a lot to say about how we use
the configure entry point; we can dictate the use of the power entry point.

	Sape

Re: [9fans] pm in the bitsy

[F.J.Ballesteros]

[-- Attachment #1: Type: text/plain, Size: 943 bytes --]

I only took a quick look, so this could be wrong; but I think that
rs232, lcd, audio, and of course sa1100 are doing a power off and have
the code to get back on. Perhaps the pcmcia/wavelan will not do so (I
didn't check that yet).

Regarding the power-on entry point, my boot loader is already jumping
(well, not sure yet, but looks like), to the address kept in
powerregs->pspr.  The linux code was kind of a tutorial here.

So, if I'm not so mistaken, I think I'd only need to save all the
processor state and write an assembly routine to reload it (placing its
address into powerregs->pspr in sa1100_power_off).

My only problem is that I'm learning arm assembler while trying to get
this working, so I'm not too optimistic regarding the final result.

In any case, if you still think I'm missing something (well, probably
many things) and you have a minute to let me know, that would be great.

thanks for your mail


[-- Attachment #2: Type: message/rfc822, Size: 3734 bytes --]

[-- Attachment #2.1.1: Type: text/plain, Size: 475 bytes --]

We haven't finished writing the code to suspend/resume the bitsy yet.
We still have to add the code to all the drivers that will make them power off
the piece of hardware they control and power that back on after resume.
Then we need to write the code that turns off memory and turns it back on
after resume and, finally, we need to write the code that deals with the
power-on entry point.  We have done none of this so I'm not surprised you
have problems.

	Sape


[-- Attachment #2.1.2: Type: message/rfc822, Size: 1605 bytes --]

From: F.J.Ballesteros <nemo@gsyc.escet.urjc.es>
To: 9fans@cse.psu.edu
Subject: [9fans] pm in the bitsy
Date: Fri, 1 Jun 2001 16:53:15 +0200
Message-ID: <20010601144450.A35CC19A27@mail.cse.psu.edu>

Hi,

	I think I managed to get it suspended using the
code #undefined in power.c. 
	My only problem is that my ignorance regarding the bitsy
doesn't let me know how to make it resume. It seems that the processor
is no longer waking up (I must power cycle it); that's despite
one line that seems to prepare the on/off button to interrupt the
processor and wake it up again. Any hint, pointer,
or direction? 

thanks

Re: [9fans] pm in the bitsy

[Sape Mullender]

[-- Attachment #1: Type: text/plain, Size: 475 bytes --]

We haven't finished writing the code to suspend/resume the bitsy yet.
We still have to add the code to all the drivers that will make them power off
the piece of hardware they control and power that back on after resume.
Then we need to write the code that turns off memory and turns it back on
after resume and, finally, we need to write the code that deals with the
power-on entry point.  We have done none of this so I'm not surprised you
have problems.

	Sape


[-- Attachment #2: Type: message/rfc822, Size: 1605 bytes --]

From: F.J.Ballesteros <nemo@gsyc.escet.urjc.es>
To: 9fans@cse.psu.edu
Subject: [9fans] pm in the bitsy
Date: Fri, 1 Jun 2001 16:53:15 +0200
Message-ID: <20010601144450.A35CC19A27@mail.cse.psu.edu>

Hi,

	I think I managed to get it suspended using the
code #undefined in power.c. 
	My only problem is that my ignorance regarding the bitsy
doesn't let me know how to make it resume. It seems that the processor
is no longer waking up (I must power cycle it); that's despite
one line that seems to prepare the on/off button to interrupt the
processor and wake it up again. Any hint, pointer,
or direction? 

thanks

[9fans] pm in the bitsy

[F.J.Ballesteros]

Hi,

	I think I managed to get it suspended using the
code #undefined in power.c. 
	My only problem is that my ignorance regarding the bitsy
doesn't let me know how to make it resume. It seems that the processor
is no longer waking up (I must power cycle it); that's despite
one line that seems to prepare the on/off button to interrupt the
processor and wake it up again. Any hint, pointer,
or direction? 

thanks