bus: Make BUS_TRANSLATE_RESOURCE behave more like other bus methods
ClosedPublic
Actions

Authored by markj on Nov 5 2021, 6:19 PM.

Details

Reviewers

Commits

Summary

Return an errno value upon failure, instead of 1.
Provide a bus_translate_resource() wrapper.
Implement the generic version, which traverses the hierarchy until a bus driver with a non-trivial implementation is found, in subr_bus.c like other similar default implementations.
Make ofw_pcib_translate_resource() return an error if a matching PCI address range is not found.
Make generic_pcie_translate_resource_common() return an int instead of a bool. Fix up callers.

No functional change intended.

Repository

rS FreeBSD src repository - subversion

Lint

Lint Passed

Unit

No Test Coverage

Build Status

Buildable 42680
Build 39568: arc lint + arc unit

markj requested review of this revision.Nov 5 2021, 6:19 PM

Thanks

This revision is now accepted and ready to land.Nov 9 2021, 5:27 PM

imp added inline comments.Nov 9 2021, 5:34 PM

sys/dev/ofw/ofw_pcib.c
489	I'd kill this.
514	and instead put *newstart = start; return (0); here. There's no locking issues to worry about for the single in single out pattern, nor are there likely to be.
521	and kill this else and just return ENOENT always.
sys/dev/pci/pci_host_generic.c
426	shouldn't this be != 0? If it was unsuccessful, we want to print the error message, no?
480	Do you need to undo the effects of rman_activate_resource() above here?

imp added inline comments.Nov 9 2021, 5:35 PM

sys/dev/pci/pci_host_generic.c
426	The original looks wrong as well if translate_resource returned true for an error, so either it's wrong or the description in the review is wrong.

markj marked 5 inline comments as done.Nov 9 2021, 5:58 PM

markj added inline comments.

sys/dev/pci/pci_host_generic.c
426	This should be != 0, thanks. The confusing bit is that generic_pcie_translate_resource_common() returned a bool, "found", and generic_pcie_translate_resource() inverted the result. The code was right before.
480	Hmm, probably yes. In practice I don't think this error can happen.

Handle Warner's notes:

Simplify ofw_pcib_translate_resource()
Fix an inverted test
Deactivate the resource in generic_pcie_activate_resource() if generic_pcie_translate_resource_common() fails

This revision now requires review to proceed.Nov 9 2021, 6:04 PM

markj marked an inline comment as done.Nov 9 2021, 6:05 PM

Thanks! Looks good to go now.

This revision is now accepted and ready to land.Nov 9 2021, 8:41 PM

This revision was automatically updated to reflect the committed changes.