On Tue, Feb 14, 2012 at 06:55:22PM +0900, Kyungmin Park wrote: > On 2/14/12, Felipe Balbi <balbi@xxxxxx> wrote: > > Hi, > > > > On Tue, Feb 14, 2012 at 06:36:59PM +0900, Kyungmin Park wrote: > >> On 2/14/12, Felipe Balbi <balbi@xxxxxx> wrote: > >> > Hi, > >> > > >> > On Fri, Feb 10, 2012 at 10:12:19AM +0100, Lukasz Majewski wrote: > >> >> From: Joonyoung Shim <jy0922.shim@xxxxxxxxxxx> > >> >> > >> >> The s3c-hsotg driver controls S3C64XX specific registers directly but > >> >> this driver can be to EXYNOS also. This removes arch specific parts > >> >> from > >> > > >> > what is this EXYNOS ? Last time I saw a patch for this controller it was > >> > basically Synopsys DWC USB3 controller for which we already have a > >> > driver under drivers/usb/dwc3/ > >> > >> Make it clear to know. > >> S3C series, S5P series, Exynos4 series use dwc2. > >> Exynos5 series use dwc3. > >> > >> So it's different USB controller. > >> > >> Exynos is brand name of samsung SoC. > > > > thanks for the clarification. But hey, you guys are using a usb2 core > > which was sourced from another company and there has been other people > > using the same core and trying to get a driver merged. > > > > It's quite alarming that noone pointed that company to this driver since > > it's the same core they were using. > > > > We need to kill code duplication guys, so I suggest re-factoring > > samsung-specific details out of this driver and renaming it to dwc2.c or > > something similar. > > > > When writing the dwc3 driver we could have chosen to write a omap5 usb3 > > controller driver, but instead we opted to make the driver as re-usable > > as possible and separated all OMAP5-specific details out of the core DWC > > USB3 driver. I hope you guys have it in you to do the same for this > > driver at some point. > > > > That's the whole idea of having an open source kernel to start with > > (well, one of them). We want to come up with re-usable solutions in > > order to decrease the maintainability overhead, in order to socialize > > bug-fixes, in order to have (in the long run) optimized solutions which > > work fine on several use-cases, but for that we need the help of driver > > writters. We need driver writters to start thinking about other users. > > > > If you source an IP from someone else, try to write the driver so that > > there's a core driver for the IP and a parent device/driver for the > > SoC-specific details. Keep in mind that if someone starts using your > > driver, you will get bugfixes for free. > > Exactly. that's I want. No more SoC vendor specific drivers codes. > Okay it seems you have plan or no objection for merging dwc2 codes if > it's acceptable for mainline. sure, if it's clean, reusable and compiles on all arches I have no objection. Specially if it also helps removing duplicated code from the tree :-) > Before that it needs to discuss the original author of dwc2 patches > how to improve or/and co-work for mainlining. that would be great. Thanks for taking that task so quickly :-) -- balbi
Attachment:
signature.asc
Description: Digital signature
