drivers/i3c/master/svc-i3c-master.c | 27 +++++++++++---------------- 1 file changed, 11 insertions(+), 16 deletions(-)
This patchset reduces the IBI transaction time by the following improvements. 1. Receive the request in interrupt context. 2. Emit the STOP as soon as possible. Stanley Chu (2): i3c: master: svc: Receive IBI requests in interrupt context i3c: master: svc: Emit STOP asap in the IBI transaction drivers/i3c/master/svc-i3c-master.c | 27 +++++++++++---------------- 1 file changed, 11 insertions(+), 16 deletions(-) -- 2.34.1
On Tue, 15 Apr 2025 13:18:06 +0800, Stanley Chu wrote:
> This patchset reduces the IBI transaction time by the following
> improvements.
> 1. Receive the request in interrupt context.
> 2. Emit the STOP as soon as possible.
>
> Stanley Chu (2):
> i3c: master: svc: Receive IBI requests in interrupt context
> i3c: master: svc: Emit STOP asap in the IBI transaction
>
> [...]
Applied, thanks!
[1/2] i3c: master: svc: Receive IBI requests in interrupt context
https://git.kernel.org/abelloni/c/8d29fa6d921c
[2/2] i3c: master: svc: Emit STOP asap in the IBI transaction
https://git.kernel.org/abelloni/c/81f2a9af9821
Best regards,
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
On 15/04/2025 at 13:18:06 +08, Stanley Chu <stanley.chuys@gmail.com> wrote: > This patchset reduces the IBI transaction time by the following > improvements. > 1. Receive the request in interrupt context. I initially had a few concerns about that, especially since the wait periods were bounded to 1s, but actually we are already in the irqsave situation when running this code, so your series might not have such a huge system-wide performance impact in the end. > 2. Emit the STOP as soon as possible. Acked-by: Miquel Raynal <miquel.raynal@bootlin.com> Thanks, Miquèl
On Tue, Apr 15, 2025 at 4:31 PM Miquel Raynal <miquel.raynal@bootlin.com> wrote: > > On 15/04/2025 at 13:18:06 +08, Stanley Chu <stanley.chuys@gmail.com> wrote: > > > This patchset reduces the IBI transaction time by the following > > improvements. > > 1. Receive the request in interrupt context. > > I initially had a few concerns about that, especially since the wait > periods were bounded to 1s, but actually we are already in the irqsave Hi Miquel, The poll timeout is 1 ms. Normally, the IBI can be completed in a few us. Thanks, Stanley > situation when running this code, so your series might not have > such a huge system-wide performance impact in the end. > > > 2. Emit the STOP as soon as possible. > > Acked-by: Miquel Raynal <miquel.raynal@bootlin.com> > > Thanks, > Miquèl
© 2016 - 2025 Red Hat, Inc.