Re: `halt` doesn’t work on Lenovo T60 with coreboot and GRUB payload

[Andrei Borzenkov <arvidjaar@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 2352 bytes --]

15.04.2017 19:32, Paul Menzel пишет:
> Dear GRUB folks,
> 
> 
> On a Lenovo T60 with coreboot and the GRUB payload, version 2.02-rc1,
> entering `halt` in the GRUB command line, nothing happens. The cursor
> goes one line below, and everything stays that way. After that, the
> system can only be powered off by pressing the power button for ten
> seconds.
> 
> Setting `debug=all` before that – `debug=halt` didn’t work – the last
> lines are below.
> 
> ```
> […]
> commands/acpihalt.c:107: data type = 0x12
> commands/acpihalt.c:241: Opcode 0x8
> commands/acpihalt.c:242: Tell 2dbd
> commands/acpihalt.c:107: data type = 0x12
> commands/acpihalt.c:241: Opcode 0x8
> commands/acpihalt.c:242: Tell 2dcd
> commands/acpihalt.c:269: S5 found
> commands/acpihalt.c:444: SLP_TYP = 7, port = 0x504
> ```
> 

So grub found how to power off system from ACPI table. Next it attempts
to do it. If it fails, you should see at least error message "ACPI
shutdown failed". If you do not see it, it looks like it is stuck
somewhere in firmware. It is quite possible that firmware expects us to
do something else before actually entering S5 state. In particular,
there are several ACPI methods that are expected be executed by OS
before actually performing state transition.

Could you make available ACPI tables (/sys/firmware/acpi/tables/* on
modern Linux kernel)?

> I would have expected at least the monitor to go dark, and maybe also
> the system to power off as there is no specific command `poweroff`
> [1]. 
> 
>> The command halts the computer. If the --no-apm option is specified,
>> no APM BIOS call is performed. Otherwise, the computer is shut down
>> using APM.
> 
> I heard, that SysV implemented `halt` “incorrectly”, so that it also

It is irrelevant here. GRUB halt attempts to call ACPI to initiate S5
(power off) transition.

> powered off the system. Only `halt -p` or `poweroff` was supposed to
> that. so I don’t know, how GRUB’s implementation of `halt` is supposed
> to work.
> 
> 
> Kind regards,
> 
> Paul
> 
> 
> [1] https://www.gnu.org/software/grub/manual/html_node/halt.html#halt
> 
> 
> 
> _______________________________________________
> Grub-devel mailing list
> Grub-devel@gnu.org
> https://lists.gnu.org/mailman/listinfo/grub-devel
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]