On Mon, Dec 16, 2024 at 11:33:37AM +0530, Manivannan Sadhasivam wrote:
On Thu, Dec 12, 2024 at 10:25:53AM +0100, Niklas Cassel wrote:
If you need to respin this series, I strongly suggest that you send the Qcom fix separately. It is totally independent, and should be merged ASAP.
Even though it is an independent fix, it is needed to get Kselftests (also the legacy ones) passing without failures. That's why I kept it as patch 1. Otherwise, someone may test it and report failures.
If qcom needs this fix, then surely pcitest.sh is already failing for the BAR test for BAR 1 and BAR 3, for all qcom SoCs, and must have been doing so since the introduction of the introduction of the qcom-pcie-ep driver.
That should be fixed of course, but I do not see why converting the tests to kselftests is related in any way.
It seems cleaner if this series just converts the tests cases to kselftests, and nothing else. EPC drivers that was passing before should still pass after this conversion.
I understand that the qcom fix is important though, which is even bigger reason that it should be sent separately, so that it can go in ASAP, and not be blocked on this series landing. Because, AFAICT, you need that qcom fix to make the pcitest.sh test cases (even without converting to kselftests).
Kind regards, Niklas