Messages in this thread | | | From | Flash Liu (劉炳傳) <> | Subject | Re: [PATCH v2] firmware: arm_scmi: Avoid to call mbox_client_txdone on txdone_irq mode | Date | Fri, 2 Feb 2024 12:36:49 +0000 |
| |
Hi Cristian,
Thanks for kindly review and explains. Yeah, I have ever tried another way to skip the call (i.e. let mark_txdone be null). However, it looks not easy and also to backport.
Awaiting your test results and further suggestions, thanks.
Regards, Pin-Chuan
On Fri, 2024-02-02 at 10:56 +0000, Cristian Marussi wrote: > > External email : Please do not click links or open attachments until > you have verified the sender or the content. > On Thu, Feb 01, 2024 at 05:57:52PM +0800, Pin-Chuan Liu wrote: > > On txdone_irq mode, tx_tick is done from mbox_chan_txdone. > > Calling to mbox_client_txdone could get error message > > and return directly, add a check to avoid this. > > > > Signed-off-by: Pin-Chuan Liu <flash.liu@mediatek.com> > > Hi Pin-Chuan, > > thanks for this, it was indeed sort of on my todo-list too, to allow > MHUs > equipped with Tx-Ack IRQ to work with SCMI. > > Having said that, this looks good to me in general, my only pain > points > are: the fact that we have to peek into the controller structures to > know > how it is configured, BUT I wouldn't know how to do it in any other > way in fact as pof now...; and the fact that there is a constant > runtime > conditional check for each message sent even though the tx_ack irq > presence > can be detected once for all at setup time, BUT even this is not > easily > solvable as of now in the SCMI stack. > > So, after all of this babbling of mine, I would say that your patch > is fine > as it is, also because it is easy to backport; next week when I am > back I'll > give it a go on a couple of platforms and get back to you with a > proper > review/test. > > Thanks again ! > Cristian
| |