[Skiboot] [PATCH skiboot] hw/phb3/naples: Disable D-states

Stewart Smith stewart at linux.ibm.com
Thu Feb 21 16:31:12 AEDT 2019

David Gibson <david at gibson.dropbear.id.au> writes:
> On Mon, Feb 11, 2019 at 07:13:14PM +1100, Alexey Kardashevskiy wrote:
>> On 08/02/2019 16:33, David Gibson wrote:
>> > On Fri, Jan 11, 2019 at 03:06:52PM +1100, Alexey Kardashevskiy wrote:
>> >> Putting "Mellanox Technologies MT27700 Family [ConnectX-4] [15b3:1013]"
>> >> (more precisely, the second of 2 its PCI functions, no matter in what
>> >> order) into the D3 state causes EEH with the "PCT timeout" error.
>> >> This has been noticed on garrison machines only and firestones do not
>> >> seem to have this issue.
>> >>
>> >> This disables D-states changing for devices on root buses on Naples by
>> >> installing a config space access filter (copied from PHB4).
>> >>
>> >> Signed-off-by: Alexey Kardashevskiy <aik at ozlabs.ru>
>> > 
>> > While fixing this in firmware seems to make theoretical sense, from
>> > all I can tell it will take an awfully long time to get this merged
>> > into skiboot, built and rolled out to Garrison machines which need it.
>> > 
>> > I think we also need a kernel side workaround for this for the short
>> > to medium term.
>> Why not just have disable_idle_d3=1 in /etc/modprobe.d/xxx on garrisons?
> Uh.. I don't know any way to make the config conditional on the type
> of machine.

If there's a way to make things conditional, then it could be on the
compatible property in device tree (looking for ibm,garrison for
example), or on PVR... but those things seem more like in-kernel things
than anything else, which could be an okay-albeit-annoying hack to have
a fix in kernel "just work" along with a real firmware fix coming in $future.

Stewart Smith
OPAL Architect, IBM.

More information about the Skiboot mailing list