On Tue, Feb 12, 2013 at 12:01:01AM +0000, Paul Zimmerman wrote: > > From: Felipe Balbi [mailto:balbi@xxxxxx] > > Sent: Monday, February 11, 2013 3:05 AM > > > > On Sat, Feb 09, 2013 at 07:37:47PM -0800, Paul Zimmerman wrote: > > > - The PCI bus driver has a couple dozen module parameters to control > > > various optional features of the controller. I realize module parameters > > > > optional features ? Are those something you can detect by reading some > > configuration registers just like we do on dwc3 ? > > The PCI bus driver is intended for our FPGA platform, where developers > may want to enable/disable things at boot time for testing purposes. For > a production platform, these should be set via DT or something, and not > be end user configurable. > > And yes, a lot of these things (like the type of Phy used) are discoverable > from the GHWCFG registers. So if you give -1 for the parameter, the driver > will use the GHWCFG values for the default. Hmm, but it looks like that is > broken for some of the parameters, I will fix that in the next version. would be nicer to just do that by default and remove the parameters, don't you think ? I mean, you can always change the bitfile on the FPGA and have the driver working differently. > But, just because the core supports a particular feature, doesn't mean > that feature is wanted on every platform. So I think some of these should > remain selectable regardless. Just not end user selectable on a production > platform right, that's ok > > > - There is quite a bit of debug code included. We would like to keep that > > > until the integration with s3c-hsotg is complete, then most of it can be > > > stripped out. > > > > Just make sure people don't depend on it. Maybe using dev_vdbg() is one > > way to actually keep it in the driver even after full cleanup is done. > > See how I implemented dev_vdbg() support for dwc3. > > For this version, I am using dev_vdbg for a lot of the messages. I am > keeping dev_dbg for some things that you may want to see even if you > are not in verbose mode, like core configuration values and unexpected > runtime conditions. I got rid of almost all of the dev_info messages, > although now I see there are a few more that should probably be > dev_dbg. that's alright then ;-) -- balbi
Attachment:
signature.asc
Description: Digital signature
