Re: [PATCH v1] Revert "ipmi:si: Gracefully handle if the BMC is non-functional"

[Guenter Roeck <linux@roeck-us.net>]

On 2/12/26 05:41, Rafael J. Wysocki wrote:
> From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> 
> Revert commit bc3a9d217755 ("ipmi:si: Gracefully handle if the BMC is
> non-functional") that attempted to improve the handling of the cases
> in which the BMC was not responsive, but did not succeed.
> 
> Instead, it introduced a regression causing AML in ACPI tables that use
> IMPI operation regions to block indefinitely on the tx_msg->tx_complete
> completion in acpi_ipmi_space_handler(), which may affect ACPI control
> methods on any system where IPMI is involved resulting in various types
> of breakage that is not straightforward to diagnose.
> 
> For example, on the system where the regression was first observed, it
> caused sysfs accesses to attributes exposed by the acpi_power_meter
> driver to block because they involved AML evaluation which is not
> super-easy to connect to IPMI.
> 
> This is a nasty and rather urgent problem with no viable fix in sight.
> 
> Note that AI was involved in diagnosing it, but didn't help much.
> 
> Fixes: bc3a9d217755 ("ipmi:si: Gracefully handle if the BMC is non-functional")
> Closes: https://lore.kernel.org/linux-acpi/CAK8fFZ6Vi4xayvdKh-_eLi-nDNMLuEoMsvwEnb33QqnwS7o4BA@mail.gmail.com/
> Reported-by: Jaroslav Pulchart <jaroslav.pulchart@gooddata.com>
> Tested-by: Jaroslav Pulchart <jaroslav.pulchart@gooddata.com>
> Cc: All applicable <stable@vger.kernel.org>
> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Acked-by: Guenter Roeck <linux@roeck-us.net>