linux-kernel - Re: [PATCH v2 0/2] Add QEMU virt-ctrl driver and update m68k virt

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <e7ddd671-6e25-4114-9d4c-21ecebe82ef5@vivier.eu>
Date: Wed, 4 Feb 2026 10:09:06 +0100
From: Laurent Vivier <laurent@...ier.eu>
To: Kuan-Wei Chiu <visitorckw@...il.com>
Cc: sre@...nel.org, jserv@...s.ncku.edu.tw, eleanor15x@...il.com,
 daniel@...f.com, linux-kernel@...r.kernel.org,
 linux-m68k@...ts.linux-m68k.org, linux-pm@...r.kernel.org,
 Geert Uytterhoeven <geert@...ux-m68k.org>
Subject: Re: [PATCH v2 0/2] Add QEMU virt-ctrl driver and update m68k virt

Le 04/02/2026 à 08:56, Geert Uytterhoeven a écrit :
> CC vivier

Thank you Geert

> On Tue, 3 Feb 2026 at 18:08, Kuan-Wei Chiu <visitorckw@...il.com> wrote:
>> Introduce a generic platform driver for the QEMU 'virt-ctrl' device [1]
>> and transitions the m68k 'virt' machine to use it, replacing
>> architecture-specific hooks.
>>
>> The new driver ('qemu-virt-ctrl') registers a restart handler and
>> populates the global 'pm_power_off' callback.
>>
>> On the m68k side, the platform initialization is updated to register
>> the 'qemu-virt-ctrl' platform device. Additionally, the 'mach_reset'
>> hook is bridged to 'do_kernel_restart()' to ensure the kernel's restart
>> handler chain is correctly invoked.
>>
>> Verified on QEMU m68k virt. Both system reset and power-off were
>> confirmed functional by invoking 'reboot(LINUX_REBOOT_CMD_RESTART)' and
>> 'reboot(LINUX_REBOOT_CMD_POWER_OFF)' from userspace.
>>
>> Link: https://gitlab.com/qemu-project/qemu/-/blob/v10.2.0/hw/misc/virt_ctrl.c [1]
>> ---
>> Changes in v2:
>> - Use devm_register_sys_off_handler() instead of register_restart_handler()
>>    and global pm_power_off.
>> - Switch Kconfig to tristate to support modular build.
>> - Add .id_table to platform_driver and use MODULE_DEVICE_TABLE() to correct
>>    module auto-loading.
>>
>> v1: https://lore.kernel.org/lkml/20260112182258.1851769-1-visitorckw@gmail.com/
>>
>> Kuan-Wei Chiu (2):
>>    power: reset: Add QEMU virt-ctrl driver
>>    m68k: virt: Switch to qemu-virt-ctrl driver
>>
>>   MAINTAINERS                          |  6 ++
>>   arch/m68k/virt/config.c              | 42 +------------
>>   arch/m68k/virt/platform.c            | 20 ++++++-
>>   drivers/power/reset/Kconfig          | 10 ++++
>>   drivers/power/reset/Makefile         |  1 +
>>   drivers/power/reset/qemu-virt-ctrl.c | 89 ++++++++++++++++++++++++++++
>>   6 files changed, 124 insertions(+), 44 deletions(-)
>>   create mode 100644 drivers/power/reset/qemu-virt-ctrl.c
>>
>> --
>> 2.53.0.rc1.225.gd81095ad13-goog

Some comments on the series:
- Please test LINUX_REBOOT_CMD_HALT as well. With mach_halt removed,
   HALT now falls back to machine_halt()’s infinite loop (no CMD_HALT),
   so behavior changes.
- Please add select POWER_RESET_QEMU_VIRT_CTRL under config VIRT
   in arch/m68k/Kconfig.machine to avoid a restart regression when the
   driver isn’t built-in.
- Endianness: QEMU’s virt-ctrl is DEVICE_NATIVE_ENDIAN, so BE on m68k but
   LE elsewhere. If the driver is meant to be generic, handle native endianness
   instead of hardcoding iowrite32be().

FWIW, I originally modeled virt-ctrl as a generic misc device, but since it only
exposes power/reset today, keeping the driver under drivers/power/reset is fine...

Please cc: me on the next version

Thanks,
Laurent