On Thu, 2018-02-01 at 20:28 +0100, Nikola Ciprich wrote: > 7d8905d064058f4b65057e0101588f362f288bc0 is the first bad commit > commit 7d8905d064058f4b65057e0101588f362f288bc0 > Author: Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> > Date: Mon Jul 24 20:28:32 2017 +0300 > > serial: 8250_pci: Enable device after we check black list > > Andy, this commit causes BrainBoxes 4100 multiport serial card to get > incorrect IRQ assigned since 4.14 He-he. It looks like one more broken PCI card found. Have you chance to read this discussion [1] and this solution [2]? Yes, it pretty much not the same case, though it might give a clue how to fix it. > here's card lspci entry: > > 02:00.0 0702: 135a:4100 (rev 02) > Subsystem: 135a:0443 > Flags: fast devsel, IRQ 18 > Memory at f7d02000 (32-bit, non-prefetchable) [size=128] > I/O ports at e000 [size=128] > I/O ports at e080 [size=64] > I/O ports at e0c0 [size=16] > Memory at f7d01000 (32-bit, non-prefetchable) [size=128] > Memory at f7d00000 (32-bit, non-prefetchable) [size=128] > Capabilities: [50] MSI: Enable- Count=1/4 Maskable- 64bit+ > Capabilities: [78] Power Management version 3 > Capabilities: [80] Express Legacy Endpoint, MSI 01 > Capabilities: [100] Virtual Channel > Capabilities: [800] Advanced Error Reporting > > after the commit, it gets IRQ 11 and doesn't work anymore.. > any idea why could that be? I'm not sure I'm reading blacklist > definition correctly, but it seems to me it's not listed there.. > > would you mind having a look at it? > > If I could be of any assistance, please let me know.. [1]: https://www.spinics.net/lists/linux-parisc/msg08179.html [2]: https://www.spinics.net/lists/linux-parisc/msg08187.html > > > By no means I suggest to use polling and be done with it. > > > So the patch I sent fixes the list_add issue with auto-probing. > > > That is > > > good. However it seems autoprobing did not find the right IRQ for > > > the > > > device. This seems to be the only problem because it works (or > > > pretends > > > to work) with polling. > > > I have no idea how to fix autoprobing. If you dig the old irq > > > numbers > > > then they should work with 4.14. The other way might be to figure > > > out > > > why the driver does not set proper irq numbers. > > > > OK, I suppose git bisect will be my friend here again :) Because something in PCI core remaps IRQ for this device based on whatever clues it has got from around. -- Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> Intel Finland Oy -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
