On Sat, May 31, 2014 at 12:05:15AM +0000, Paul Zimmerman wrote: > > From: Felipe Balbi [mailto:balbi@xxxxxx] > > Sent: Friday, May 30, 2014 4:42 PM > > > > On Fri, May 23, 2014 at 11:39:24AM -0700, Paul Zimmerman wrote: > > > Newer DWC3 controllers can be built for USB 2.0-only mode, where > > > most of the USB 3.0 circuitry is left out. To support this mode, > > > the driver must limit the speed programmed into the DCFG register > > > to Hi-Speed or lower. > > > > > > Reads and writes to the PIPECTL register are left as-is, since > > > they should be no-ops in USB 2.0-only mode. Calls to phy_init() > > > etc. for the USB3 phy are also left as-is, since the no-op USB3 > > > phy should be used for USB 2.0-only mode controllers. > > > > > > Signed-off-by: Paul Zimmerman <paulz@xxxxxxxxxxxx> > > > --- > > > Hi Felipe, > > > > > > Does this look OK to you? I think it is fine to leave the PIPECTL > > > accesses and the phy_init() calls as-is, but if you would prefer > > > that I also conditionalize those I can do that. We have at least > > > one customer who will need this feature fairly soon, so we would > > > like to get this in without too much delay, although I guess we > > > missed the 3.16 merge window. > > > > I like this a lot :-) Very nice of Synopsys to support this > > configuration. Could you just let me know which versions of the core > > support this configuration ? We have AM437x which has this sort of > > "quirk" although, I think it's done using a TI-specific "modification", > > perhaps ? > > It has been officially supported since 2.60a. But it's possible that > customers have hacked up something like this on their own with previous > versions, so the exact version number might not mean much. The patch > should work for all versions of the core, because the > DWC_USB3_SSPHY_INTERFACE bits have always been there, going back to > preproduction versions of the core. > > It has been pointed out to me that DT can already be used to limit the > max speed, using the 'maximum-speed' property (duh). But I think we > still want the patch for non-DT platforms like dwc3-pci. certainly, this patch is very much apreciated. I'll get back to merging as soon as I catch up with all other email noise I have waiting for me. cheers -- balbi
Attachment:
signature.asc
Description: Digital signature
