[GIT,PULL,v2] FSI updates for 4.17

Message ID 5c7f9044fefed8bc0e4ca5688801ffa379f38f62.camel@kernel.crashing.org
State Not Applicable, archived
Headers show
Series
  • [GIT,PULL,v2] FSI updates for 4.17
Related show

Pull-request

https://git.kernel.org/pub/scm/linux/kernel/git/benh/linux-fsi.git tags/fsi-updates-2018-06-12

Message

Benjamin Herrenschmidt June 12, 2018, 10:54 a.m.
Hi Greg !

There are a first round of updates of the FSI stack, aiming at reducing/removing
he gap with the OpenBMC tree and a first step in getting dependent drivers
upstream.

These changes significantly improve the FSI bitbanging driver performance
and reliability, and add the new "sbefifo" driver for communicating with
the POWER9 Self Boot Engine (which will be needed for some upcoming
additional drivers).

These were posted to the list as patches a couple of weeks ago and were
prior to that tested in the OpenBMC tree and CI. Some subsequent driver
work depends on some of these, so merging early is probably the best.

The following changes since commit 8efcf34a263965e471e3999904f94d1f6799d42a:

  Merge tag 'armsoc-late' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc (2018-06-11 18:19:45 -0700)

are available in the Git repository at:

  https://git.kernel.org/pub/scm/linux/kernel/git/benh/linux-fsi.git tags/fsi-updates-2018-06-12

for you to fetch changes up to 9f4a8a2d7f9d71093f41c4bb0ef8707e8145bad3:

  fsi/sbefifo: Add driver for the SBE FIFO (2018-06-12 14:05:39 +1000)

----------------------------------------------------------------
FSI updates and sbefifo driver

----------------------------------------------------------------
Andrew Jeffery (2):
      fsi: gpio: Trace busy count
      fsi: gpio: Remove unused 'id' variable

Benjamin Herrenschmidt (8):
      fsi/fsi-master-gpio: Sample input data on different clock phase
      fsi/fsi-master-gpio: Add "no-gpio-delays" option
      fsi/fsi-master-gpio: Reduce turnaround clocks
      fsi/fsi-master-gpio: Reduce dpoll clocks
      fsi/fsi-master-gpio: Delay sampling of FSI data input
      fsi/fsi-master-gpio: Implement CRC error recovery
      fsi/fsi-master-gpio: More error handling cleanup
      fsi/sbefifo: Add driver for the SBE FIFO

Eddie James (1):
      fsi: scom: Remove PIB reset during probe

Jeremy Kerr (4):
      fsi: gpio: Use a mutex to protect transfers
      fsi/gpio: Include command build in locked section
      fsi/gpio: Use relative-addressing commands
      fsi/master-gpio: Replace bit_bit lock with IRQ disable/enable

 arch/arm/boot/dts/aspeed-bmc-opp-romulus.dts     |    1 +
 arch/arm/boot/dts/aspeed-bmc-opp-witherspoon.dts |    1 +
 arch/arm/boot/dts/aspeed-bmc-opp-zaius.dts       |    1 +
 drivers/fsi/Kconfig                              |    7 +
 drivers/fsi/Makefile                             |    1 +
 drivers/fsi/fsi-master-gpio.c                    |  349 ++++++++++++----
 drivers/fsi/fsi-sbefifo.c                        | 1005 +++++++++++++++++++++++++++++++++++++++++++++++
 drivers/fsi/fsi-scom.c                           |    8 -
 include/linux/fsi-sbefifo.h                      |   33 ++
 include/trace/events/fsi_master_gpio.h           |   43 ++
 10 files changed, 1362 insertions(+), 87 deletions(-)
 create mode 100644 drivers/fsi/fsi-sbefifo.c
 create mode 100644 include/linux/fsi-sbefifo.h

Comments

Greg Kroah-Hartman June 12, 2018, 11:18 a.m. | #1
On Tue, Jun 12, 2018 at 08:54:39PM +1000, Benjamin Herrenschmidt wrote:
> Hi Greg !
> 
> There are a first round of updates of the FSI stack, aiming at reducing/removing
> he gap with the OpenBMC tree and a first step in getting dependent drivers
> upstream.
> 
> These changes significantly improve the FSI bitbanging driver performance
> and reliability, and add the new "sbefifo" driver for communicating with
> the POWER9 Self Boot Engine (which will be needed for some upcoming
> additional drivers).
> 
> These were posted to the list as patches a couple of weeks ago and were
> prior to that tested in the OpenBMC tree and CI. Some subsequent driver
> work depends on some of these, so merging early is probably the best.

Yeah, but as they are not "regression fixes", I'll put them into my
-next branch for merging with 4.19-rc1.  I'll do that once 4.18-rc1 is
released, can't touch my trees until that happens.

thanks,

greg k-h
Benjamin Herrenschmidt June 12, 2018, 11:25 a.m. | #2
On Tue, 2018-06-12 at 13:18 +0200, Greg Kroah-Hartman wrote:
> On Tue, Jun 12, 2018 at 08:54:39PM +1000, Benjamin Herrenschmidt wrote:
> > Hi Greg !
> > 
> > There are a first round of updates of the FSI stack, aiming at reducing/removing
> > he gap with the OpenBMC tree and a first step in getting dependent drivers
> > upstream.
> > 
> > These changes significantly improve the FSI bitbanging driver performance
> > and reliability, and add the new "sbefifo" driver for communicating with
> > the POWER9 Self Boot Engine (which will be needed for some upcoming
> > additional drivers).
> > 
> > These were posted to the list as patches a couple of weeks ago and were
> > prior to that tested in the OpenBMC tree and CI. Some subsequent driver
> > work depends on some of these, so merging early is probably the best.
> 
> Yeah, but as they are not "regression fixes", I'll put them into my
> -next branch for merging with 4.19-rc1.  I'll do that once 4.18-rc1 is
> released, can't touch my trees until that happens.

We are in regression only mode even before -rc1 these days ? including
for code that is basically completely unused upstream (which is what
this is trying to fix) ? :-) That's rather full-on ...

Anyway no biggie, it will just slow down some getting some of the other
drivers I suppose, though we can always explicitly list the dependency
on this tag to work around it.

Cheers,
Ben.
Greg Kroah-Hartman June 12, 2018, 11:44 a.m. | #3
On Tue, Jun 12, 2018 at 09:25:11PM +1000, Benjamin Herrenschmidt wrote:
> On Tue, 2018-06-12 at 13:18 +0200, Greg Kroah-Hartman wrote:
> > On Tue, Jun 12, 2018 at 08:54:39PM +1000, Benjamin Herrenschmidt wrote:
> > > Hi Greg !
> > > 
> > > There are a first round of updates of the FSI stack, aiming at reducing/removing
> > > he gap with the OpenBMC tree and a first step in getting dependent drivers
> > > upstream.
> > > 
> > > These changes significantly improve the FSI bitbanging driver performance
> > > and reliability, and add the new "sbefifo" driver for communicating with
> > > the POWER9 Self Boot Engine (which will be needed for some upcoming
> > > additional drivers).
> > > 
> > > These were posted to the list as patches a couple of weeks ago and were
> > > prior to that tested in the OpenBMC tree and CI. Some subsequent driver
> > > work depends on some of these, so merging early is probably the best.
> > 
> > Yeah, but as they are not "regression fixes", I'll put them into my
> > -next branch for merging with 4.19-rc1.  I'll do that once 4.18-rc1 is
> > released, can't touch my trees until that happens.
> 
> We are in regression only mode even before -rc1 these days ? including
> for code that is basically completely unused upstream (which is what
> this is trying to fix) ? :-) That's rather full-on ...

For code that goes into -rc1, it has to be in linux-next _BEFORE_ -final
comes out.  That's been the rule for years, you should know that...

> Anyway no biggie, it will just slow down some getting some of the other
> drivers I suppose, though we can always explicitly list the dependency
> on this tag to work around it.

There should not be a rush here, remember, kernel development does not
have deadlines :)

greg k-h
Benjamin Herrenschmidt June 12, 2018, 12:40 p.m. | #4
On Tue, 2018-06-12 at 13:44 +0200, Greg Kroah-Hartman wrote:
> 
> > We are in regression only mode even before -rc1 these days ? including
> > for code that is basically completely unused upstream (which is what
> > this is trying to fix) ? :-) That's rather full-on ...
> 
> For code that goes into -rc1, it has to be in linux-next _BEFORE_ -final
> comes out.  That's been the rule for years, you should know that...

Oh I know and I also know that the rule was always somewhat flexible
for "new" drivers ;-) But heh, it's not biggie.

> > Anyway no biggie, it will just slow down some getting some of the other
> > drivers I suppose, though we can always explicitly list the dependency
> > on this tag to work around it.
> 
> There should not be a rush here, remember, kernel development does not
> have deadlines :)

Sort-of, it does make things difficult to have so much crap piling up
out of tree or in vendor trees as you well know, so the faster that
pile is sorted, the better everybody is. But I can deal with the
dependencies for 4.19 (assuming we get the 2 or 3 other drivers I want
to see upstream lined up as well in the next few weeks).

Cheers,
Ben.
Greg Kroah-Hartman July 7, 2018, 3:03 p.m. | #5
On Tue, Jun 12, 2018 at 08:54:39PM +1000, Benjamin Herrenschmidt wrote:
> Hi Greg !
> 
> There are a first round of updates of the FSI stack, aiming at reducing/removing
> he gap with the OpenBMC tree and a first step in getting dependent drivers
> upstream.
> 
> These changes significantly improve the FSI bitbanging driver performance
> and reliability, and add the new "sbefifo" driver for communicating with
> the POWER9 Self Boot Engine (which will be needed for some upcoming
> additional drivers).
> 
> These were posted to the list as patches a couple of weeks ago and were
> prior to that tested in the OpenBMC tree and CI. Some subsequent driver
> work depends on some of these, so merging early is probably the best.
> 
> The following changes since commit 8efcf34a263965e471e3999904f94d1f6799d42a:
> 
>   Merge tag 'armsoc-late' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc (2018-06-11 18:19:45 -0700)
> 
> are available in the Git repository at:
> 
>   https://git.kernel.org/pub/scm/linux/kernel/git/benh/linux-fsi.git tags/fsi-updates-2018-06-12

Sorry for the delay, now merged.

greg k-h