> From: Felipe Balbi [mailto:balbi@xxxxxx] > Sent: Wednesday, February 20, 2013 12:29 AM > On Tue, Feb 19, 2013 at 06:50:03PM -0800, Paul Zimmerman wrote: > > Here is v4 of the DWC2 patch set. I made most of the changes you asked > > for, except for the following: > > - I did not convert to a threaded IRQ handler. I would like to postpone > > that for now. > > without any reasoning ? The HSOTG core enables SOF interrupts in some cases, so it has to respond in less than 125us to interrupts. I'm afraid with a threaded IRQ handler the latency for handling SOFs could be too large. I'm not saying I won't do it eventually, I'd just rather not open that possible can of worms right now. > > - I did not convert to devm_request_and_ioremap() in the probe function, > > because that requires a struct resource, which you don't have in a PCI > > driver. Perhaps I'm missing something. > > look at line 287 on include/linux/pci.h Ah, that's what I was missing. Yes, that makes the probe() code much nicer, thanks. > > - I removed a few unneeded module parameters, but left the rest for now. > > I discovered that some of the parameters need to be set to non- > > hardware-default values, or else the driver doesn't work correctly. > > isn't that something which is wrong with your FPGA model ? IIRC you can > change register default values via coreConsultant. Building and validating a new FPGA image is a lengthy task. If I can make things work by tweaking some module parameters, so much the better. Plus, as part of our validation testing we try with different values for things like FIFO sizes and DMA modes. Building a new FPGA image to tweak stuff like that is just not practical. -- Paul -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html