Re: [jornada] Re: Kernel source? (Power managememt for HP Jornada 720)

The power management support required a few changes in the bootldr
to get back from sleep. (The main problem was that the code
supporting resume was all conditionalized under H3600).
If you put in a new bootldr, it should come back ok. (I will verify this
early next week and put a bootldr binary file on the handhelds
site).

I will try the "hda:lost interrupt". I am not sure what is causing
this.

The "eth0: interrupt(s) dropped" message seems like
it might (!) be a bug in the driver. When I try it, the frequency
that it appears (when I am using a socketcom ethernet card)
is about the same as the frequency of the message on the iPaq.
This message is printed when the driver is polled using card
services and finds an unserviced interrupt (I think). It is somewhat
interesting that this clock interrupt based polling allows the
driver to keep on working, even if the CF slot interrupt is
completely dead!

More later,
jca

Greg Berenfield wrote:

> Just adding my experiences with 2.4.18-rmk3 with John's pm routines
> (checked out today, 7-10-2002).
>
> I too see the "hda: lost interrupt" messages and can not get my
> microdrive to work due to these.
>
> I also get "eth0: interrupt(s) dropped!" messages but the nic in the
> pcmcia slot still functions ok.
>
> As for power management, I can seemingly suspend my jornada.
> Here's what the serial console shows when pressing the power button:
> -------------------------------------
> power [7f]
> power [ff]
> run_sbin_pm_helper:71 pm_helper_path=/sbin/pm_helper
> jornada720_pcmcia_configure_socket(): config socket 0 vcc 50 vpp 0
> jornada720_pcmcia_configure_socket(): config socket 0 vcc 0 vpp 0
> [sa1100_jornada720.c:216] jornada720_pcmcia_socket_suspend() stub
> jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 0
> jornada720_pcmcia_configure_socket(): config socket 1 vcc 0 vpp 0
> [sa1100_jornada720.c:216] jornada720_pcmcia_socket_suspend() stub
> -------------------------------------
>
> But resume doesn't seem to work :(
>
> Nothing happens when pressing power button again (tried pressing and
> holding down for up to 10s) and the recessed reset button has no effect
> either. The only way to 'reset' is by removing the battery for a cold
> boot.
>
> FYI, I am using a jornada with linux on the flashrom.
>
> Also, to get it to compile, I used the def-configs/jornada720 but had to
> remove the #ifdef stuff in arch/arm/mach-sa1100/pm.c around the
> #include <asm/arch/h3600_asic.h> line to get definitions for the H3800
> compilation complaints.
>
> Tried removing microdrive and ethernet nic, rebooted and had same
> results for trying to resume.
>
> So.. I'll head back to 2.4.17-rmk5 (the hda problems are critical for
> me).
>
> Also I wonder if tweaking to the bootloader may be necessary. I'm happy
> to flash it if an updated version is available.
>
> Greg
>
> On Mon, Jun 24, 2002 at 07:53:39PM +0200, Peter Gerwinski wrote:
> > Christopher Hoover wrote:
> > > Please check the archives. Hpcboot will *NOT* directly boot a linux elf
> > > kernel image w/o modification to either hpcboot or the kernel startup
> > > code. Details in the archive.
> >
> > I did the necessary modifications to the kernel. So far my roadmap
> > is as follows:
> >
> > - Get the current kernel from handhelds.org.
> > - Comment out 3 lines concerning "invalid architecture?"
> > in arch/arm/kernel/head-armv.S
> > (see http://acid.ch.pw.edu.pl/~sq5bpf/mylinux/linux-jornada720/).
> > - Comment out the test "if (mount_initrd && ROOT_DEV != real_root_dev
> > && MAJOR(ROOT_DEV) == RAMDISK_MAJOR && MINOR(ROOT_DEV == 0)"
> > in init/main.c; make the following action take place unconditionally
> > (see BootLinuxWithoutFlashing.html written for the IPAQ,
> > somewhere under handhelds.org).
> > - Apply the CIBD patch.
> > - Correct the obvious error in drivers/block/makeh, or else use
> > some other mechanism to put a 2 MB RamDisk into the CIBD. (More
> > about this in a separate email.)
> > - The 2 MB RamDisk must be set up to serve as an initrd and contain
> > a linuxrc that starts PCMCIA services, accesses a CF IDE card and
> > mounts one of its partitions as the root file system.
> > (I am uploading my current - gzipped - disk image to:
> > http://www.peter.gerwinski.de/linux-2048.initrd-20020624.gz)
> >
> > (Hey - this can be a starting point for a HOWTO!:-)
> >
> > I got this working for linux-2.4.18-rmk2-hh2 (without power
> > management) - see my previous email. However for the current kernel
> > (with power management) I get the following:
> >
> > [...]
> > Loading modules ...
> > Using /lib/modules/2.4.18-rmk3/kernel/drivers/ide/ide-disk.o
> > Using /lib/modules/2.4.18-rmk3/kernel/drivers/ide/ide-cs.o
> > Running cardmgr ...
> > cardmgr[13]: starting, version is 3.1.20
> > cardmgr[13]: watching 2 sockets
> > cardmgr[13]: Card Services release does not match
> > cardmgr[13]: initializing socket 1
> > cardmgr[13]: socket 1: ATA/IDE Fixed Disk
> > cardmgr[13]: product info: "CF Card", "CFA 256MB CHH", "", jornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 0
> > ""
> > cardjornada720_pcmcia_configure_socket(): config socket 1 vcc 33 vpp 0
> > mgr[13]: manfid: 0x0007, 0x0000 function: 4 (fixed disk)
> > Trying to free nonexistent resource <c2820000-c282000f>
> > hda: 256MB CHH, ATA DISK drive
> > ide0 at 0xc2820000-0xc2820007,0xc282000e on irq 115
> > hda: 500640 sectors (256 MB) w/16KiB Cache, CHS=1043/15/32
> > Partition check:
> > /dev/ide/host0/bus0/target0/lun0:hda: lost interrupt
> > hda: lost interrupt
> > hda: lost interrupt
> > [...]
> >
> > This was no problem with linux-2.4.18-rmk2-hh2. Any idea what is
> > going wrong and what to do against it?
> >
> > Peter
> > --
Received on Wed Jul 10 2002 - 16:04:20 EDT