[SLOF] [PATCH slof] pci: Align PCI node names with QEMU
Thomas Huth
thuth at redhat.com
Tue Oct 22 17:47:03 AEDT 2019
On 22/10/2019 01.28, Alexey Kardashevskiy wrote:
>
>
> On 21/10/2019 18:41, Thomas Huth wrote:
>> On 21/10/2019 09.02, Alexey Kardashevskiy wrote:
>>> During the ibm,client-arhitecture-support client call, we rely on QEMU
>>> providing a full device tree which SLOF then merged into its internal
>>> tree so we rely on both SLOF and QEMU using the same node names for
>>> devices of the same type.
>>>
>>> This changes device tree node names to what QEMU uses.
>>>
>>> The change was triggered by "(unknown-)legacy-device" which is used by
>>> virtio-balloon; other changes either fix typos or remove devices which
>>> are very unlikely to be implemented by QEMU for pseries, or even if they
>>> are, we are ok with using generic class names anyway.
>>>
>>> Signed-off-by: Alexey Kardashevskiy <aik at ozlabs.ru>
>>> ---
>>>
>>> This is to fix https://bugs.launchpad.net/qemu/+bug/1847440?comments=all
>>>
>>>
>>> ---
>>> slof/fs/pci-class-code-names.fs | 41 +++++----------------------------
>>> 1 file changed, 6 insertions(+), 35 deletions(-)
>>>
>>> diff --git a/slof/fs/pci-class-code-names.fs b/slof/fs/pci-class-code-names.fs
>>> index 6cf3b8fd083c..cd9395fb47ba 100644
>>> --- a/slof/fs/pci-class-code-names.fs
>>> +++ b/slof/fs/pci-class-code-names.fs
>>> @@ -13,7 +13,7 @@
>>> : pci-class-name-00 ( addr -- str len )
>>> pci-class@ 8 rshift FF and CASE
>>> 01 OF s" display" ENDOF
>>> - dup OF s" unknown-legacy-device" ENDOF
>>> + dup OF s" legacy-device" ENDOF
>>> ENDCASE
>>> ;
>>>
>>> @@ -47,7 +47,6 @@
>>> : pci-class-name-03 ( addr -- str len )
>>> pci-class@ FFFF and CASE
>>> 0000 OF s" vga" ENDOF
>>> - 0001 OF s" 8514-compatible" ENDOF
>>> 0100 OF s" xga" ENDOF
>>> 0200 OF s" 3d-controller" ENDOF
>>> dup OF s" display" ENDOF
>>> @@ -91,23 +90,9 @@
>>> : pci-class-name-07 ( addr -- str len )
>>> pci-class@ FFFF and CASE
>>> 0000 OF s" serial" ENDOF
>>> - 0001 OF s" 16450-serial" ENDOF
>>> - 0002 OF s" 16550-serial" ENDOF
>>> - 0003 OF s" 16650-serial" ENDOF
>>> - 0004 OF s" 16750-serial" ENDOF
>>> - 0005 OF s" 16850-serial" ENDOF
>>> - 0006 OF s" 16950-serial" ENDOF
>>> 0100 OF s" parallel" ENDOF
>>> - 0101 OF s" bi-directional-parallel" ENDOF
>>> - 0102 OF s" ecp-1.x-parallel" ENDOF
>>> - 0103 OF s" ieee1284-controller" ENDOF
>>> - 01FE OF s" ieee1284-device" ENDOF
>>> 0200 OF s" multiport-serial" ENDOF
>>> 0300 OF s" modem" ENDOF
>>> - 0301 OF s" 16450-modem" ENDOF
>>> - 0302 OF s" 16550-modem" ENDOF
>>> - 0303 OF s" 16650-modem" ENDOF
>>> - 0304 OF s" 16750-modem" ENDOF
>>> 0400 OF s" gpib" ENDOF
>>> 0500 OF s" smart-card" ENDOF
>>> dup OF s" communication-controller" ENDOF
>>> @@ -118,18 +103,9 @@
>>> : pci-class-name-08 ( addr -- str len )
>>> pci-class@ FFFF and CASE
>>> 0000 OF s" interrupt-controller" ENDOF
>>> - 0001 OF s" isa-pic" ENDOF
>>> - 0002 OF s" eisa-pic" ENDOF
>>> - 0010 OF s" io-apic" ENDOF
>>> - 0020 OF s" iox-apic" ENDOF
>>> 0100 OF s" dma-controller" ENDOF
>>> - 0101 OF s" isa-dma" ENDOF
>>> - 0102 OF s" eisa-dma" ENDOF
>>> 0200 OF s" timer" ENDOF
>>> - 0201 OF s" isa-system-timer" ENDOF
>>> - 0202 OF s" eisa-system-timer" ENDOF
>>> 0300 OF s" rtc" ENDOF
>>> - 0301 OF s" isa-rtc" ENDOF
>>> 0400 OF s" hot-plug-controller" ENDOF
>>> 0500 OF s" sd-host-conrtoller" ENDOF
>>> dup OF s" system-periphal" ENDOF
>>> @@ -156,10 +132,7 @@
>>
>> I think I'd rather not remove the entries from class 7 and 8 ... for
>> example, board-js2x uses a 16550 UART IIRC ... yeah, js2x is broken
>> anyway, but in case somebody ever restores this board, we might need
>> this again.
>
> Why might we _need_ this? Does anything rely on the exact node name for these devices? I am not removing them, I am just
> changing names, "compatible" should still be used to match the driver. What do I miss?
Theoretically, people could use the device tree paths (including these
names) in their NVRAM variables. Whether the NVRAM batteries of the YDL
Powerstations or JS20/JS21 boards are still working is another question,
of course ;-)
Well, I saw that you already committed the patch, so never mind. I guess
we can deal with that later in case we ever re-activate board-js2x
(which is very unlikely anyway).
Thomas
More information about the SLOF
mailing list