[PATCH v6 00/24] PCI: Convert all dynamic sysfs attributes to static

Krzysztof Wilczyński kwilczynski at kernel.org
Fri Apr 24 05:00:44 AEST 2026


Hello,

> >   Sashiko's feedback:
> >     - https://sashiko.dev/#/patchset/20260422161407.118748-1-kwilczynski%40kernel.org
> >     - https://sashiko.dev/#/patchset/20260416180107.777065-1-kwilczynski%40kernel.org
> >     - https://sashiko.dev/#/patchset/20260411080148.471335-1-kwilczynski%40kernel.org
> >     - https://sashiko.dev/#/patchset/20260410055040.39233-1-kwilczynski%40kernel.org
> 
> Thanks for all this work!  I think this will be a great step forward.
> I assume you've probably looked at the sashiko feedback and concluded
> that no changes are needed, e.g., the feedback is out of scope for
> this series?  Your thoughts there would help me out.

The only actionable feedback would be to update commit message of patch 18,
so that it mentions pci_stop_dev() rather than pci_stop_bus_device(),
something we can do later.

One could also drop #if guard from the struct pci_dev as added in the patch 9,
which is later removed completely anyway.  There is no ODR violation or ABI
break there.

We won't be fixing any more kernel lock-down issues at this time, and the
value of legacy I/O space size (the 0xffff) is something that was there
from 2005, so not something this series introduced or changed.

Using HAVE_PCI_MMAP and/or ARCH_GENERIC_PCI_MMAP_RESOURCE to stop the
resizable BAR resources from being included on platforms that do not
provide either, like Alpha, should not be a problem.

So, no.  No serious issues there.  But let me know if you want to change
something, so I can send next version if needed.

Thank you!

	Krzysztof


More information about the Linuxppc-dev mailing list