v4.7-rc2 kernel fails to boot when acpi=force

[G Gregory]

On 7 June 2016 at 12:05, Mark Rutland <mark.rutland at arm.com> wrote:
> On Tue, Jun 07, 2016 at 07:53:53PM +0900, Itaru Kitayama wrote:
>> Mark
>>
>> On 6/7/16 7:23 PM, Mark Rutland wrote:
>> >On Tue, Jun 07, 2016 at 05:06:39PM +0900, Itaru Kitayama wrote:
>> >>Upstream v4.7-rc2 kernel does not boot if parameter acpi is set to force
>> >>on Overdrive. (acpi=off gets to the prompt)
>> >
>> >Did acpi=force work on this board with an earlier kernel (e.g. v4.6)?
>>
>> v4.6 boots fine with acpi=foce.
>
> Ok.
>
>> >>When set to force it doesn't get to the Linux boot process and stalls.
>> >
>> >What output do you get before it stalls?
>> >
>> >I assume that you at least have output from the EFI stub?
>>
>> Yes.
>
> Do you get anything else? A full log woould be helpful.
>
>> >Which command line options are you passing?
>>
>> At the time of trying, only acpi=foce efi=debug
>
>> >If you get nothing after the stub, try also passing:
>> >
>> >earlycon=pl011,0xe1010000
>>

make defconfig
make Image.gz

Seems to boot fine here

cat /proc/cmdline
BOOT_IMAGE=(tftp)/arm64/Image console=ttyAMA0
earlycon=pl011,0xe1010000 acpi=force r8169.use_dac=1 ip=dhcp
root=/dev/nfs nfsroot=192.168.10.253:/srv/nfs/d02-debian rw

uname -r
4.7.0-rc2

Thanks

Graeme

>> Adding it in the /etc/default/grub file and updating grub via
>> mkgrub2-config seems to made the board always falls on the grub prompt.
>> I am able to switch to the EFI shell, but is there an easy way to
>> recover it? (I'm accessing the board remotely)
>
> If you can open GRUB, press 'e', and you can edit the command line.
>
> If GRUB doesn't work at all, you might be able to use the 'edit' command
> to modify the GRUB onfiguration file.
>
> Thanks,
> Mark.
>
>> >Which may give an indication of what's going on.
>> >
>> >>Here's the BIOS info:
>> >>
>> >>Version 2.17.1249. Copyright (C) 2015 American Megatrends, Inc.
>> >>BIOS Date: 09/03/2015 13:52:32 Ver: ROD0084E00
>> >
>> >I don't know if the FW from a year ago provided everything necessary. It
>> >may be something like the SPCR is simply missing, and there's no
>> >console= passed to the kernel.
>> >
>> >Thanks,
>> >Mark.
>> >
>>