(no top-posting, please. It's the 2nd time I'm asking) Hi, Jerry Huang <jerry.huang@xxxxxxx> writes: > Hi, Balbi, > Sorry for later responding. > I tested Linux 4.8, 4.7, 4.6 and 4.5, has the same issue: > [ 0.756166] xhci-hcd: probe of xhci-hcd.0.auto failed with error -5 > > Then I found one patch (dwc3 initiated xhci probe problem in arm64 4.4 kernel due to DMA setup) as below link, and I applied this patch to Linux-4.5, the USB3.0 worked well! Therefore I applied this patch to Linux-4.9-rc7, and the USB3.0 works well, too! Which used of_dma_configure(&xhci->dev, dwc->dev->of_node) to set dma. > https://patches.linaro.org/patch/65892/ > > However, in Linux-4.9-rc8, I found these 6 patches applied: > https://patchwork.kernel.org/patch/9434163/ > https://patchwork.kernel.org/patch/9434155/ > https://patchwork.kernel.org/patch/9434131/ > https://patchwork.kernel.org/patch/9434165/ > https://patchwork.kernel.org/patch/9434167/ > https://patchwork.kernel.org/patch/9434133/ > and the commit ID are: > 9863f279c1d48e366a179e76683247f239b30975 > 0abcfafea32452f1d07901320efa8c17470dbeed > ec3a36d441f2adcb8a0d3f6534ac488f2202c87a > e9d10c536b26732d38d59fa3639b85bcd819837e > 60478aead82d761c71a2a402b3fd2817af9e57ac > 21f57eb38129042d2910d1e9b5fd36f3eea93637 > > And I got the below error when I attached one USB3.0 disk (Just USB2.0 disk can work). From above 6 patches, it seems remove dma_set_coherent_mask from file drivers/usb/dwc3/host.c. > > usb 4-1: new SuperSpeed USB device number 2 using xhci-hcd > [ 16.224264] usb 4-1: device descriptor read/8, error -61 > [ 16.336179] usb 4-1: new SuperSpeed USB device number 2 using xhci-hcd > [ 16.360257] usb 4-1: device descriptor read/8, error -61 > [ 16.652196] usb 4-1: new SuperSpeed USB device number 3 using xhci-hcd > [ 16.676256] usb 4-1: device descriptor read/8, error -61 > [ 16.788179] usb 4-1: new SuperSpeed USB device number 3 using xhci-hcd > [ 16.812255] usb 4-1: device descriptor read/8, error -61 > > So, how to handle this issue? oh, it's the DMA thing. We have patches going to next for this. Arnd, do you know if next has all missing pieces? There was something pending on XHCI IIRC, right? (keeping context below) > -----Original Message----- > From: Felipe Balbi [mailto:felipe.balbi@xxxxxxxxxxxxxxx] > Sent: Monday, December 05, 2016 7:00 PM > To: Jerry Huang <jerry.huang@xxxxxxx>; stern@xxxxxxxxxxxxxxxxxxx; linux-usb@xxxxxxxxxxxxxxx; Mathias Nyman <mathias.nyman@xxxxxxxxxxxxxxx> > Cc: Sriram Dash <sriram.dash@xxxxxxx> > Subject: RE: USB3.0 DWC3 can't work with latest Linux-4.9.rc7 > > > (no top-posting, please) > > Hi, > > it helps if you include maintainers for correct drivers. scripts/get_maintainer.pl helps a lot. > > Jerry Huang <jerry.huang@xxxxxxx> writes: >> And I tested USB2.0 disk, with Sriram's 6 patches, it works well. So Just USB3.0 can't work. >> >> BTW, >> The board with USB3.0 DWC3 controller and the USB3.0 disk work well with Linux-4.1.8. > > alright, in that case you can bisect. Please do so and find the first commit when this started happening. > >> -----Original Message----- >> From: Jerry Huang >> Sent: Monday, December 05, 2016 6:11 PM >> To: stern@xxxxxxxxxxxxxxxxxxx; linux-usb@xxxxxxxxxxxxxxx >> Cc: Sriram Dash <sriram.dash@xxxxxxx> >> Subject: USB3.0 DWC3 can't work with latest Linux-4.9.rc7 >> >> Hi, Guys, >> 1. I tested the USB3.0 DWC3 with latest Linux-4.9-rc7 on LS1043ardb board. >> But I got the below errors, this error is returned from dma_supported(dev, mask) while running " dma_set_mask_and_coherent" in function " xhci_plat_probe". >> ... >> [ 0.756166] xhci-hcd: probe of xhci-hcd.0.auto failed with error -5 >> [ 0.762438] xhci-hcd: probe of xhci-hcd.1.auto failed with error -5 >> [ 0.768712] xhci-hcd: probe of xhci-hcd.2.auto failed with error -5 >> [ 0.775049] usbcore: registered new interface driver usb-storage >> ... >> >> 2. Then, I applied 6 patches from Sriram: >> https://patchwork.kernel.org/patch/9434163/ >> https://patchwork.kernel.org/patch/9434155/ >> https://patchwork.kernel.org/patch/9434131/ >> https://patchwork.kernel.org/patch/9434165/ >> https://patchwork.kernel.org/patch/9434167/ >> https://patchwork.kernel.org/patch/9434133/ >> >> however, I got another error as below shows when I attach one USB disk to this board. >> It seems USB device return the 8 bytes descriptor, but all 8 byte data driver read are zero from the buffer received. >> Has any comment for this issue? >> >> usb 2-1: new SuperSpeed USB device number 2 using xhci-hcd [ >> 164.323396] usb 2-1: device descriptor read/8, error -61 > > hmm, connection refused. Mathias, any idea? > > BTW, are you sure this board worked with *upstream* (as in, Linus' tree) > v4.1 or were you using some vendor kernel which is based off of v4.1? > > Care to check v4.8, v4.7 and v4.6? -- balbi
Attachment:
signature.asc
Description: PGP signature
