Re: [PATCH v5 04/12] i3c: master: Add support for devices using SETAASA

[Alexandre Belloni <alexandre.belloni@bootlin.com>]

On 03/07/2026 09:08:55+0000, Akhil R wrote:
> On Thu, 25 Jun 2026 07:42:03 -0500 Frank Li wrote:
> > On Thu, Jun 25, 2026 at 09:38:15AM +0000, Akhil R wrote:
> >> On Wed, 24 Jun 2026 13:57:46 -0400, Frank Li wrote:
> >> ...
> >> ...
> >> >> [Severity: High]
> >> >> Is it possible that sending the SETAASA broadcast before direct SETDASA
> >> >> assignments breaks initialization for devices that natively support SETAASA
> >> >> but are configured for SETDASA?
> >> >>
> >> >> According to the I3C specification, any device on the bus natively supporting
> >> >> SETAASA will respond to this broadcast by adopting its static address as its
> >> >> dynamic address.
> >> >>
> >> >> After this broadcast, the driver iterates through devices and attempts to
> >> >> assign custom dynamic addresses via direct SETDASA commands:
> >> >>
> >> >> drivers/i3c/master.c:i3c_master_early_i3c_dev_add() {
> >> >>     ...
> >> >> 	ret = i3c_master_setdasa_locked(master, i3cdev->info.static_addr,
> >> >> 					i3cdev->boardinfo->init_dyn_addr);
> >> >>     ...
> >> >> }
> >> >>
> >> >> Since the target device already adopted its dynamic address during the
> >> >> SETAASA broadcast, it is no longer in the unassigned state and will NACK
> >> >> the subsequent SETDASA command.
> >> >
> >> > Look like correct, but I am not sure if target will NACK SETDASA. Or should
> >> > use SETNEWDA for SETAASA method.
> >>
> >> Yes, this looks valid for mixed device buses. I can move
> >> i3c_master_setaasa_locked() after the SETDASA handling and before
> >> i3c_master_do_daa() in the same function, so SETDASA-assigned devices will
> >> ignore the later SETAASA broadcast. Does that sound good to you?
> > 
> > yes, try it to follow spec.
> 
> I just noticed that the specification says: if both bits 0 and 1 are set,
> meaning both SETDASA and SETAASA are supported, the I3C Bus Controller should
> use SETDASA first.
> 
> So moving i3c_master_setaasa_locked() after SETDASA handling follows the spec.
> 
> I will wait a few more days before sending v6, to see if there are any other
> concerns.

No other concerns on my side.

> 
> Best Regards,
> Akhil
> 
> -- 
> linux-i3c mailing list
> linux-i3c@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-i3c

-- 
linux-i3c mailing list
linux-i3c@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-i3c