On Tue, Dec 31, 2024 at 08:33:57PM +0100, Niklas Cassel wrote:
On 31 December 2024 20:18:12 CET, Manivannan Sadhasivam manivannan.sadhasivam@linaro.org wrote:
On Tue, Dec 31, 2024 at 06:42:42PM +0100, Niklas Cassel wrote:
On Tue, Dec 31, 2024 at 06:43:41PM +0530, Manivannan Sadhasivam wrote:
(...)
- # RUN pci_ep_data_transfer.dma.COPY_TEST ...
- # OK pci_ep_data_transfer.dma.COPY_TEST
- ok 11 pci_ep_data_transfer.dma.COPY_TEST
- # PASSED: 11 / 11 tests passed.
- # Totals: pass:11 fail:0 xfail:0 xpass:0 skip:0 error:0
+Testcase 11 (pci_ep_data_transfer.dma.COPY_TEST) will fail for most of the DMA +capable endpoint controllers due to the absence of the MEMCPY over DMA. For such +controllers, it is advisable to skip the forementioned testcase using below +command::
Hm.. this is strictly not correct. If will currently fail because pci-epf-test.c does: if ((reg->flags & FLAG_USE_DMA) && epf_test->dma_private) return -EINVAL;
So even if a DMA driver has support for the DMA_MEMCPY cap, if the DMA driver also has the DMA_PRIVATE cap, this test will fail because of the code in pci-epf-test.c.
Right. But I think the condition should be changed to test for the MEMCPY capability instead. Like,
diff --git a/drivers/pci/endpoint/functions/pci-epf-test.c b/drivers/pci/endpoint/functions/pci-epf-test.c index ef6677f34116..0b211d60a85b 100644 --- a/drivers/pci/endpoint/functions/pci-epf-test.c +++ b/drivers/pci/endpoint/functions/pci-epf-test.c @@ -328,7 +328,7 @@ static void pci_epf_test_copy(struct pci_epf_test *epf_test, void *copy_buf = NULL, *buf;
if (reg->flags & FLAG_USE_DMA) {
if (epf_test->dma_private) {
if (!dma_has_cap(DMA_MEMCPY, epf_test->dma_chan_tx->device->cap_mask)) { dev_err(dev, "Cannot transfer data using DMA\n"); ret = -EINVAL; goto set_status;That check does seem to make more sense than the code that is currently there. (Perhaps send this as a proper patch?)
Will do.
Note that I'm not an expert at dmaengine.
I have some patches that adds DMA_MEMCPY to dw-edma, but I'm not sure if the DWC eDMA hardware supports having both src and dst as PCI addresses, or if only one of them can be a PCI address (with the other one being a local address).
If only one of them can be a PCI address, then I'm not sure if your suggested patch is correct.
I don't see why that would be an issue. DMA_MEMCPY is independent of PCI/local addresses. If a dmaengine driver support doing MEMCPY, then the dma cap should be sufficient. As you said, if a controller supports both SLAVE and MEMCPY, the test currently errors out, which is wrong.
- Mani