Re: [PATCH 00/15] fsi: Overall improvements and new SBE fifo driver
On Sun, Jun 10, 2018 at 04:46:53PM +1000, Benjamin Herrenschmidt wrote: > (Greg, see below) > > On Tue, 2018-05-29 at 22:00 +0930, Joel Stanley wrote: > > On 29 May 2018 at 11:00, Benjamin Herrenschmidt > > wrote: > > > This series brings in a number of improvements to our FSI stack > > > (one of the service interfaces for communicating between a BMC chip and > > > our POWER processors). > > > > > > The GPIO based "Soft FSI" performance is significantly improved, and > > > it's reliability as well. > > > > > > The SBE fifo driver provides the interface to the processor "Self Boot > > > Engine" > > > > > > Some of these patches have been simmering for a while in the OpenBMC > > > tree. > > > > I run this series atop of 4.17-rc7 today, and they look solid. > > > > Tested-by: Joel Stanley > > So how do we proceed ? I have more coming on top of this, but we should > start with getting this series merged. > > Greg, should we create a FSI git repo with a 3-member maintainer team > (Jeremy, Chris and myself) and send you pull requests ? Or are you > happy to continue picking up patches ? Which ever is best/easiest for you is fine with me, I can work either way just fine. > We have more stuff to come in there, including support for a HW master > contoller in a future BMC chip, some more slave drivers, etc.. > > It also looks like the s390 guys might start using some of that as well > (their chips use FSI as well, so far they used our old service > processor which uses an ancient non-uptream set of drivers & stack, but > that's likely to change). Then maybe a git tree is easiest for all of you to work against? thanks, greg k-h
Re: [PATCH 00/15] fsi: Overall improvements and new SBE fifo driver
On Sun, Jun 10, 2018 at 04:46:53PM +1000, Benjamin Herrenschmidt wrote: > (Greg, see below) > > On Tue, 2018-05-29 at 22:00 +0930, Joel Stanley wrote: > > On 29 May 2018 at 11:00, Benjamin Herrenschmidt > > wrote: > > > This series brings in a number of improvements to our FSI stack > > > (one of the service interfaces for communicating between a BMC chip and > > > our POWER processors). > > > > > > The GPIO based "Soft FSI" performance is significantly improved, and > > > it's reliability as well. > > > > > > The SBE fifo driver provides the interface to the processor "Self Boot > > > Engine" > > > > > > Some of these patches have been simmering for a while in the OpenBMC > > > tree. > > > > I run this series atop of 4.17-rc7 today, and they look solid. > > > > Tested-by: Joel Stanley > > So how do we proceed ? I have more coming on top of this, but we should > start with getting this series merged. > > Greg, should we create a FSI git repo with a 3-member maintainer team > (Jeremy, Chris and myself) and send you pull requests ? Or are you > happy to continue picking up patches ? Which ever is best/easiest for you is fine with me, I can work either way just fine. > We have more stuff to come in there, including support for a HW master > contoller in a future BMC chip, some more slave drivers, etc.. > > It also looks like the s390 guys might start using some of that as well > (their chips use FSI as well, so far they used our old service > processor which uses an ancient non-uptream set of drivers & stack, but > that's likely to change). Then maybe a git tree is easiest for all of you to work against? thanks, greg k-h
Re: [PATCH 00/15] fsi: Overall improvements and new SBE fifo driver
(Greg, see below) On Tue, 2018-05-29 at 22:00 +0930, Joel Stanley wrote: > On 29 May 2018 at 11:00, Benjamin Herrenschmidt > wrote: > > This series brings in a number of improvements to our FSI stack > > (one of the service interfaces for communicating between a BMC chip and > > our POWER processors). > > > > The GPIO based "Soft FSI" performance is significantly improved, and > > it's reliability as well. > > > > The SBE fifo driver provides the interface to the processor "Self Boot > > Engine" > > > > Some of these patches have been simmering for a while in the OpenBMC > > tree. > > I run this series atop of 4.17-rc7 today, and they look solid. > > Tested-by: Joel Stanley So how do we proceed ? I have more coming on top of this, but we should start with getting this series merged. Greg, should we create a FSI git repo with a 3-member maintainer team (Jeremy, Chris and myself) and send you pull requests ? Or are you happy to continue picking up patches ? We have more stuff to come in there, including support for a HW master contoller in a future BMC chip, some more slave drivers, etc.. It also looks like the s390 guys might start using some of that as well (their chips use FSI as well, so far they used our old service processor which uses an ancient non-uptream set of drivers & stack, but that's likely to change). Cheers, Ben. > Cheers, > > Joel
Re: [PATCH 00/15] fsi: Overall improvements and new SBE fifo driver
(Greg, see below) On Tue, 2018-05-29 at 22:00 +0930, Joel Stanley wrote: > On 29 May 2018 at 11:00, Benjamin Herrenschmidt > wrote: > > This series brings in a number of improvements to our FSI stack > > (one of the service interfaces for communicating between a BMC chip and > > our POWER processors). > > > > The GPIO based "Soft FSI" performance is significantly improved, and > > it's reliability as well. > > > > The SBE fifo driver provides the interface to the processor "Self Boot > > Engine" > > > > Some of these patches have been simmering for a while in the OpenBMC > > tree. > > I run this series atop of 4.17-rc7 today, and they look solid. > > Tested-by: Joel Stanley So how do we proceed ? I have more coming on top of this, but we should start with getting this series merged. Greg, should we create a FSI git repo with a 3-member maintainer team (Jeremy, Chris and myself) and send you pull requests ? Or are you happy to continue picking up patches ? We have more stuff to come in there, including support for a HW master contoller in a future BMC chip, some more slave drivers, etc.. It also looks like the s390 guys might start using some of that as well (their chips use FSI as well, so far they used our old service processor which uses an ancient non-uptream set of drivers & stack, but that's likely to change). Cheers, Ben. > Cheers, > > Joel
Re: [PATCH 00/15] fsi: Overall improvements and new SBE fifo driver
On 29 May 2018 at 11:00, Benjamin Herrenschmidt wrote: > This series brings in a number of improvements to our FSI stack > (one of the service interfaces for communicating between a BMC chip and > our POWER processors). > > The GPIO based "Soft FSI" performance is significantly improved, and > it's reliability as well. > > The SBE fifo driver provides the interface to the processor "Self Boot > Engine" > > Some of these patches have been simmering for a while in the OpenBMC > tree. I run this series atop of 4.17-rc7 today, and they look solid. Tested-by: Joel Stanley Cheers, Joel
Re: [PATCH 00/15] fsi: Overall improvements and new SBE fifo driver
On 29 May 2018 at 11:00, Benjamin Herrenschmidt wrote: > This series brings in a number of improvements to our FSI stack > (one of the service interfaces for communicating between a BMC chip and > our POWER processors). > > The GPIO based "Soft FSI" performance is significantly improved, and > it's reliability as well. > > The SBE fifo driver provides the interface to the processor "Self Boot > Engine" > > Some of these patches have been simmering for a while in the OpenBMC > tree. I run this series atop of 4.17-rc7 today, and they look solid. Tested-by: Joel Stanley Cheers, Joel
