On Tue, 2015-12-01 at 23:49 +0100, Santiago Garcia Mantinan wrote:
> Hi!
>
> A friend gave me this tablet which has a broken glass + digitizer so I
> thought I could maybe make some use of it under Debian.
>
> The device is a 32 bits Atom based uefi machine, so I got first jessie's
> installer and then a weekly stretch build but on both of them I could get to
> the first menu but then keyboard stopped working (keyboard and boot pen
> drive are plugged to a usb HUB).
>
> To try to get more info on the USB problem I thought I could copy an
> installed uefi debian to the usb pen drive that I was using, so I installed
> it on a uefi virtualbox and then copied the raw disk to the pen drive.
>
> The pen boots but after it loads kernel, initrd it fails to locate the root
> image (it is on USB and we have no USB, so this was expected) but the
> machine also has an sd card reader which seemed detected, so I copied the
> same virtualbox image to the SD and booted of the USB (doesn't seem to allow
> to boot from the SD) then the kernel and initrd are loaded and root is
> found and debian booted.
>
> The image and kernel of Debian I'm running are Jessie's and I had an
> rc.local which I was intending to gather some info, but unfortunately the
> machine seems to hang without even finishing booting.
>
> I'll try to paste here what I've got till now, what seemed weird to me is
> that I didn't see any trace of the USB or the WiFi devices.
The firmware appears to have set up the PCI root and ACPI tables
wrongly:
[...]
> Dec 2 09:29:28 debian kernel: [ 0.115143] PCI: MMCONFIG for domain 0000 [bus 00-ff] at [mem 0xe0000000-0xefffffff] (base 0xe0000000)
> Dec 2 09:29:28 debian kernel: [ 0.120549] [Firmware Info]: PCI: MMCONFIG at [mem 0xe0000000-0xefffffff] not reserved in ACPI motherboard resources
> Dec 2 09:29:28 debian kernel: [ 0.120555] PCI: not using MMCONFIG
[...]
> Dec 2 09:29:28 debian kernel: [ 0.139017] pci 0000:00:02.0: can't claim BAR 2 [mem 0xe0000000-0xefffffff]: no compatible bridge window
[...]
> I have some ideas of how to continue, installing a 4.3 kernel, maybe the 686
> branch, not the 586 one,
You should definitely use 686-pae on this processor, not 586. But
that's not likely to change the behaviour.
It is possible that more recent kernel versions have a workaround but I
don't see one.
> but I thought commenting here first as I was
> expecting things to be on the pci bus but doesn't seem to be the case.
>
> Any ideas of how to continue?
>
> Any other info I can try to gather and provide?
Some kernel parameters to try:
pci=assign-busses
pci=realloc
or together:
pci=assign-busses,realloc
Ben.
--
Ben Hutchings
Theory and practice are closer in theory than in practice.
- John Levine, moderator of comp.compilersAttachment:
signature.asc
Description: This is a digitally signed message part