Re: 答复: [PATCH] staging: rtsx: Add support for RTS5260

[Greg KH]

On Mon, Oct 16, 2017 at 05:58:12AM +, 冯锐 wrote:
> > On Fri, Oct 13, 2017 at 10:54:22AM +0100, Lee Jones wrote:
> > > On Fri, 13 Oct 2017, Greg KH wrote:
> > >
> > > > On Fri, Oct 13, 2017 at 04:50:35PM +0800, rui_f...@realsil.com.cn wrote:
> > > > > From: rui_feng 
> > > > >
> > > > > Add support for new chip rts5260.
> > > > > In order to support rts5260，the definitions of some internal
> > > > > registers and workflow have to be modified and are different from
> > > > > its predecessors and OCP function is added for RTS5260.
> > > > > So we need this patch to ensure RTS5260 can work.
> > > > >
> > > > > Signed-off-by: Rui Feng 
> > > >
> > > > Your from and signed-off-by name does not match :(
> > > >
> > > > > ---
> > > > >  drivers/mfd/Makefile  |   4 +
> > > > >  drivers/mfd/rtsx_pcr.c| 127 ++-
> > > > >  drivers/mfd/rtsx_pcr.h|  12 +
> > > > >  drivers/staging/Kconfig   |   2 +
> > > > >  drivers/staging/rts5260/Kconfig   |   6 +
> > > > >  drivers/staging/rts5260/rts5260.c | 748
> > > > > ++
> > > > >  drivers/staging/rts5260/rts5260.h |  45 +++
> > > > >  include/linux/mfd/rtsx_pci.h  | 234 +++-
> > > >
> > > > I do not see a reason why this is a staging driver.  Where is the
> > > > TODO file listing what needs to be done to get it out of staging?
> > > > Why can it not just go into the "real" part of the kernel?
> > >
> > > It's not a staging driver.  Rui's focus appears to be to have this
> > > driver accepted into Mainline by hook or by crock.  He's tried MFD,
> > > Misc and now Staging!
> > 
> > Yeah, I've watched it too :)
> > 
> > > Background:
> > >
> > > There are a number of drivers in this family which currently reside in
> > > MFD.  These were accepted before my time.  After a recent review I've
> > > made the decision that these aren't MFD drivers at all.
> > >
> > > MFD drivers are ones which aid in registering and setting up shared
> > > resources for sub-devices which reside on the same piece of silicon.
> > > This driver does basically none of that.  Instead it *is* the (what we
> > > describe above as) sub-device.  It does everything.
> > 
> > I agree with your assessment.
> > 
> > > In the absence of a subsystem which covers this type of device, I
> > > suggested Misk as a good location to place these drivers.
> > 
> > What type of device is this thing?  I can't seem to figure that out.  If we 
> > can
> > determine that, then we can find the proper place for it in the kernel.
> > 
> > Rui, what does this hardware do?  What is the interface between the
> > hardware and userspace that this driver is creating?
> > 
> This driver is a pcie driver, it bridge mmc subsystem and memstick
> subsystem, and it does not deal with userspace.

Ok, sounds like drivers/misc/ is a good place for it.

thanks,

greg k-h

Re: 答复: [PATCH] staging: rtsx: Add support for RTS5260

[Greg KH]

On Mon, Oct 16, 2017 at 05:58:12AM +, 冯锐 wrote:
> > On Fri, Oct 13, 2017 at 10:54:22AM +0100, Lee Jones wrote:
> > > On Fri, 13 Oct 2017, Greg KH wrote:
> > >
> > > > On Fri, Oct 13, 2017 at 04:50:35PM +0800, rui_f...@realsil.com.cn wrote:
> > > > > From: rui_feng 
> > > > >
> > > > > Add support for new chip rts5260.
> > > > > In order to support rts5260，the definitions of some internal
> > > > > registers and workflow have to be modified and are different from
> > > > > its predecessors and OCP function is added for RTS5260.
> > > > > So we need this patch to ensure RTS5260 can work.
> > > > >
> > > > > Signed-off-by: Rui Feng 
> > > >
> > > > Your from and signed-off-by name does not match :(
> > > >
> > > > > ---
> > > > >  drivers/mfd/Makefile  |   4 +
> > > > >  drivers/mfd/rtsx_pcr.c| 127 ++-
> > > > >  drivers/mfd/rtsx_pcr.h|  12 +
> > > > >  drivers/staging/Kconfig   |   2 +
> > > > >  drivers/staging/rts5260/Kconfig   |   6 +
> > > > >  drivers/staging/rts5260/rts5260.c | 748
> > > > > ++
> > > > >  drivers/staging/rts5260/rts5260.h |  45 +++
> > > > >  include/linux/mfd/rtsx_pci.h  | 234 +++-
> > > >
> > > > I do not see a reason why this is a staging driver.  Where is the
> > > > TODO file listing what needs to be done to get it out of staging?
> > > > Why can it not just go into the "real" part of the kernel?
> > >
> > > It's not a staging driver.  Rui's focus appears to be to have this
> > > driver accepted into Mainline by hook or by crock.  He's tried MFD,
> > > Misc and now Staging!
> > 
> > Yeah, I've watched it too :)
> > 
> > > Background:
> > >
> > > There are a number of drivers in this family which currently reside in
> > > MFD.  These were accepted before my time.  After a recent review I've
> > > made the decision that these aren't MFD drivers at all.
> > >
> > > MFD drivers are ones which aid in registering and setting up shared
> > > resources for sub-devices which reside on the same piece of silicon.
> > > This driver does basically none of that.  Instead it *is* the (what we
> > > describe above as) sub-device.  It does everything.
> > 
> > I agree with your assessment.
> > 
> > > In the absence of a subsystem which covers this type of device, I
> > > suggested Misk as a good location to place these drivers.
> > 
> > What type of device is this thing?  I can't seem to figure that out.  If we 
> > can
> > determine that, then we can find the proper place for it in the kernel.
> > 
> > Rui, what does this hardware do?  What is the interface between the
> > hardware and userspace that this driver is creating?
> > 
> This driver is a pcie driver, it bridge mmc subsystem and memstick
> subsystem, and it does not deal with userspace.

Ok, sounds like drivers/misc/ is a good place for it.

thanks,

greg k-h

答复: [PATCH] staging: rtsx: Add support for RTS5260

[冯锐]

> On Fri, Oct 13, 2017 at 10:54:22AM +0100, Lee Jones wrote:
> > On Fri, 13 Oct 2017, Greg KH wrote:
> >
> > > On Fri, Oct 13, 2017 at 04:50:35PM +0800, rui_f...@realsil.com.cn wrote:
> > > > From: rui_feng 
> > > >
> > > > Add support for new chip rts5260.
> > > > In order to support rts5260，the definitions of some internal
> > > > registers and workflow have to be modified and are different from
> > > > its predecessors and OCP function is added for RTS5260.
> > > > So we need this patch to ensure RTS5260 can work.
> > > >
> > > > Signed-off-by: Rui Feng 
> > >
> > > Your from and signed-off-by name does not match :(
> > >
> > > > ---
> > > >  drivers/mfd/Makefile  |   4 +
> > > >  drivers/mfd/rtsx_pcr.c| 127 ++-
> > > >  drivers/mfd/rtsx_pcr.h|  12 +
> > > >  drivers/staging/Kconfig   |   2 +
> > > >  drivers/staging/rts5260/Kconfig   |   6 +
> > > >  drivers/staging/rts5260/rts5260.c | 748
> > > > ++
> > > >  drivers/staging/rts5260/rts5260.h |  45 +++
> > > >  include/linux/mfd/rtsx_pci.h  | 234 +++-
> > >
> > > I do not see a reason why this is a staging driver.  Where is the
> > > TODO file listing what needs to be done to get it out of staging?
> > > Why can it not just go into the "real" part of the kernel?
> >
> > It's not a staging driver.  Rui's focus appears to be to have this
> > driver accepted into Mainline by hook or by crock.  He's tried MFD,
> > Misc and now Staging!
> 
> Yeah, I've watched it too :)
> 
> > Background:
> >
> > There are a number of drivers in this family which currently reside in
> > MFD.  These were accepted before my time.  After a recent review I've
> > made the decision that these aren't MFD drivers at all.
> >
> > MFD drivers are ones which aid in registering and setting up shared
> > resources for sub-devices which reside on the same piece of silicon.
> > This driver does basically none of that.  Instead it *is* the (what we
> > describe above as) sub-device.  It does everything.
> 
> I agree with your assessment.
> 
> > In the absence of a subsystem which covers this type of device, I
> > suggested Misk as a good location to place these drivers.
> 
> What type of device is this thing?  I can't seem to figure that out.  If we 
> can
> determine that, then we can find the proper place for it in the kernel.
> 
> Rui, what does this hardware do?  What is the interface between the
> hardware and userspace that this driver is creating?
> 
This driver is a pcie driver, it bridge mmc subsystem and memstick subsystem, 
and it does not deal with userspace.

> thanks,
> 
> greg k-h
> 
> --Please consider the environment before printing this e-mail.

答复: [PATCH] staging: rtsx: Add support for RTS5260

[冯锐]

> On Fri, Oct 13, 2017 at 10:54:22AM +0100, Lee Jones wrote:
> > On Fri, 13 Oct 2017, Greg KH wrote:
> >
> > > On Fri, Oct 13, 2017 at 04:50:35PM +0800, rui_f...@realsil.com.cn wrote:
> > > > From: rui_feng 
> > > >
> > > > Add support for new chip rts5260.
> > > > In order to support rts5260，the definitions of some internal
> > > > registers and workflow have to be modified and are different from
> > > > its predecessors and OCP function is added for RTS5260.
> > > > So we need this patch to ensure RTS5260 can work.
> > > >
> > > > Signed-off-by: Rui Feng 
> > >
> > > Your from and signed-off-by name does not match :(
> > >
> > > > ---
> > > >  drivers/mfd/Makefile  |   4 +
> > > >  drivers/mfd/rtsx_pcr.c| 127 ++-
> > > >  drivers/mfd/rtsx_pcr.h|  12 +
> > > >  drivers/staging/Kconfig   |   2 +
> > > >  drivers/staging/rts5260/Kconfig   |   6 +
> > > >  drivers/staging/rts5260/rts5260.c | 748
> > > > ++
> > > >  drivers/staging/rts5260/rts5260.h |  45 +++
> > > >  include/linux/mfd/rtsx_pci.h  | 234 +++-
> > >
> > > I do not see a reason why this is a staging driver.  Where is the
> > > TODO file listing what needs to be done to get it out of staging?
> > > Why can it not just go into the "real" part of the kernel?
> >
> > It's not a staging driver.  Rui's focus appears to be to have this
> > driver accepted into Mainline by hook or by crock.  He's tried MFD,
> > Misc and now Staging!
> 
> Yeah, I've watched it too :)
> 
> > Background:
> >
> > There are a number of drivers in this family which currently reside in
> > MFD.  These were accepted before my time.  After a recent review I've
> > made the decision that these aren't MFD drivers at all.
> >
> > MFD drivers are ones which aid in registering and setting up shared
> > resources for sub-devices which reside on the same piece of silicon.
> > This driver does basically none of that.  Instead it *is* the (what we
> > describe above as) sub-device.  It does everything.
> 
> I agree with your assessment.
> 
> > In the absence of a subsystem which covers this type of device, I
> > suggested Misk as a good location to place these drivers.
> 
> What type of device is this thing?  I can't seem to figure that out.  If we 
> can
> determine that, then we can find the proper place for it in the kernel.
> 
> Rui, what does this hardware do?  What is the interface between the
> hardware and userspace that this driver is creating?
> 
This driver is a pcie driver, it bridge mmc subsystem and memstick subsystem, 
and it does not deal with userspace.

> thanks,
> 
> greg k-h
> 
> --Please consider the environment before printing this e-mail.

答复: [PATCH] staging: rtsx: Add support for RTS5260

[冯锐]

Loop Wei Wang

> On Fri, Oct 13, 2017 at 11:11:15AM +0100, Lee Jones wrote:
> > On Fri, 13 Oct 2017, Greg KH wrote:
> >
> > > On Fri, Oct 13, 2017 at 10:54:22AM +0100, Lee Jones wrote:
> > > > On Fri, 13 Oct 2017, Greg KH wrote:
> > > >
> > > > > On Fri, Oct 13, 2017 at 04:50:35PM +0800, rui_f...@realsil.com.cn
> wrote:
> > > > > > From: rui_feng 
> > > > > >
> > > > > > Add support for new chip rts5260.
> > > > > > In order to support rts5260，the definitions of some internal
> > > > > > registers and workflow have to be modified and are different
> > > > > > from its predecessors and OCP function is added for RTS5260.
> > > > > > So we need this patch to ensure RTS5260 can work.
> > > > > >
> > > > > > Signed-off-by: Rui Feng 
> > > > >
> > > > > Your from and signed-off-by name does not match :(
> > > > >
> > > > > > ---
> > > > > >  drivers/mfd/Makefile  |   4 +
> > > > > >  drivers/mfd/rtsx_pcr.c| 127 ++-
> > > > > >  drivers/mfd/rtsx_pcr.h|  12 +
> > > > > >  drivers/staging/Kconfig   |   2 +
> > > > > >  drivers/staging/rts5260/Kconfig   |   6 +
> > > > > >  drivers/staging/rts5260/rts5260.c | 748
> > > > > > ++
> > > > > >  drivers/staging/rts5260/rts5260.h |  45 +++
> > > > > >  include/linux/mfd/rtsx_pci.h  | 234 +++-
> > > > >
> > > > > I do not see a reason why this is a staging driver.  Where is
> > > > > the TODO file listing what needs to be done to get it out of
> > > > > staging?  Why can it not just go into the "real" part of the kernel?
> > > >
> > > > It's not a staging driver.  Rui's focus appears to be to have this
> > > > driver accepted into Mainline by hook or by crock.  He's tried
> > > > MFD, Misc and now Staging!
> > >
> > > Yeah, I've watched it too :)
> > >
> > > > Background:
> > > >
> > > > There are a number of drivers in this family which currently
> > > > reside in MFD.  These were accepted before my time.  After a
> > > > recent review I've made the decision that these aren't MFD drivers at 
> > > > all.
> > > >
> > > > MFD drivers are ones which aid in registering and setting up
> > > > shared resources for sub-devices which reside on the same piece of
> silicon.
> > > > This driver does basically none of that.  Instead it *is* the
> > > > (what we describe above as) sub-device.  It does everything.
> > >
> > > I agree with your assessment.
> > >
> > > > In the absence of a subsystem which covers this type of device, I
> > > > suggested Misk as a good location to place these drivers.
> >
> > Sorry, that should have been "Misc"
> >
> > > What type of device is this thing?  I can't seem to figure that out.
> > > If we can determine that, then we can find the proper place for it
> > > in the kernel.
> >
> > It's a card (MMC, (u)SD, etc) reader.
> 
> Ok, how does it hook up to the hardware to talk to the reader?
> 
> And we have other card readers in drivers/misc/ so maybe that is the right
> place for it, depending on the api to the hardware.
> 
> thanks,
> 
> greg k-h
> 
> --Please consider the environment before printing this e-mail.

答复: [PATCH] staging: rtsx: Add support for RTS5260

[冯锐]

Loop Wei Wang

> On Fri, Oct 13, 2017 at 11:11:15AM +0100, Lee Jones wrote:
> > On Fri, 13 Oct 2017, Greg KH wrote:
> >
> > > On Fri, Oct 13, 2017 at 10:54:22AM +0100, Lee Jones wrote:
> > > > On Fri, 13 Oct 2017, Greg KH wrote:
> > > >
> > > > > On Fri, Oct 13, 2017 at 04:50:35PM +0800, rui_f...@realsil.com.cn
> wrote:
> > > > > > From: rui_feng 
> > > > > >
> > > > > > Add support for new chip rts5260.
> > > > > > In order to support rts5260，the definitions of some internal
> > > > > > registers and workflow have to be modified and are different
> > > > > > from its predecessors and OCP function is added for RTS5260.
> > > > > > So we need this patch to ensure RTS5260 can work.
> > > > > >
> > > > > > Signed-off-by: Rui Feng 
> > > > >
> > > > > Your from and signed-off-by name does not match :(
> > > > >
> > > > > > ---
> > > > > >  drivers/mfd/Makefile  |   4 +
> > > > > >  drivers/mfd/rtsx_pcr.c| 127 ++-
> > > > > >  drivers/mfd/rtsx_pcr.h|  12 +
> > > > > >  drivers/staging/Kconfig   |   2 +
> > > > > >  drivers/staging/rts5260/Kconfig   |   6 +
> > > > > >  drivers/staging/rts5260/rts5260.c | 748
> > > > > > ++
> > > > > >  drivers/staging/rts5260/rts5260.h |  45 +++
> > > > > >  include/linux/mfd/rtsx_pci.h  | 234 +++-
> > > > >
> > > > > I do not see a reason why this is a staging driver.  Where is
> > > > > the TODO file listing what needs to be done to get it out of
> > > > > staging?  Why can it not just go into the "real" part of the kernel?
> > > >
> > > > It's not a staging driver.  Rui's focus appears to be to have this
> > > > driver accepted into Mainline by hook or by crock.  He's tried
> > > > MFD, Misc and now Staging!
> > >
> > > Yeah, I've watched it too :)
> > >
> > > > Background:
> > > >
> > > > There are a number of drivers in this family which currently
> > > > reside in MFD.  These were accepted before my time.  After a
> > > > recent review I've made the decision that these aren't MFD drivers at 
> > > > all.
> > > >
> > > > MFD drivers are ones which aid in registering and setting up
> > > > shared resources for sub-devices which reside on the same piece of
> silicon.
> > > > This driver does basically none of that.  Instead it *is* the
> > > > (what we describe above as) sub-device.  It does everything.
> > >
> > > I agree with your assessment.
> > >
> > > > In the absence of a subsystem which covers this type of device, I
> > > > suggested Misk as a good location to place these drivers.
> >
> > Sorry, that should have been "Misc"
> >
> > > What type of device is this thing?  I can't seem to figure that out.
> > > If we can determine that, then we can find the proper place for it
> > > in the kernel.
> >
> > It's a card (MMC, (u)SD, etc) reader.
> 
> Ok, how does it hook up to the hardware to talk to the reader?
> 
> And we have other card readers in drivers/misc/ so maybe that is the right
> place for it, depending on the api to the hardware.
> 
> thanks,
> 
> greg k-h
> 
> --Please consider the environment before printing this e-mail.