[PATCH 3/3] ACPI: ARM64: support for ACPI_TABLE_UPGRADE

Jon Masters jcm at redhat.com
Mon May 23 11:29:06 PDT 2016


On 05/23/2016 01:56 PM, Lorenzo Pieralisi wrote:
> On Tue, May 17, 2016 at 12:48:53PM -0400, Jon Masters wrote:
>> On 05/17/2016 12:44 PM, Jon Masters wrote:
>>
>>> 1). During development of a platform, it is much easier to debug
>>> problems with tables if you can test replacement ones without having to
>>> respin the firmware. In the server world, you usually don't have the
>>> firmware source code, so to get it respun could be days-weeks even if
>>> you are working with the authors closely. We have practically used this
>>> feature on a number of platforms already and it will continue.
>>
>> For example, on one platform we were unable to fully boot RHEL(SA) due
>> to a bug in one of the ACPI tables. But I was able to boot the system to
>> a ramdisk containing a uuencode library and then write out the content
>> of the tables over the serial port, then decompile/patch/recompile, and
>> override replacement tables on the system. Then we beat the vendor up
>> with the fixes and the official firmware was corrected.
> 
> Can you explain to me please why you can't do it with GRUB ?

It's doable with GRUB (if you rebuild GRUB modules to include it)

> I am using mainline GRUB and its acpi command all the time to update
> static ACPI tables for testing new features (ie IORT) and it works
> just fine for me (and you can still override the DSDT, which is
> likely to be the main source of bugs, through the CONFIG_ACPI_CUSTOM_DSDT
> config option, that works on ARM in mainline with no changes required).
> 
> I just want to understand if there is really a compelling reason
> for adding this stuff when we can easily implement its features
> through something that is usable today without any kernel changes.

We're looking for x86 feature parity in the base platform. Every time
there's a gratuitous differentiation it's a waste of time for folks
trying to figure out what is different on an ARM system. So if we
completely remove the ACPI override feature from the kernel and make
everyone use GRUB instead, then fine, but ARM shouldn't be special.

Jon.

-- 
Computer Architect | Sent from my Fedora powered laptop




More information about the linux-arm-kernel mailing list