Re: [PATCH v2] media: intel/ipu6: remove buttress ish structure

[Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>]

On Fri, Nov 01, 2024 at 05:07:25PM +0800, Bingbu Cao wrote:
> 
> On 11/1/24 4:19 PM, Sakari Ailus wrote:
> > Hi Bingbu,
> > 
> > On Fri, Nov 01, 2024 at 03:47:54PM +0800, Bingbu Cao wrote:
> >> Sakari and Stanislaw,
> >>
> >> On 11/1/24 3:46 PM, Sakari Ailus wrote:
> >>>>>> @@ -386,10 +382,8 @@ irqreturn_t ipu6_buttress_isr(int irq, void *isp_ptr)
> >>>>>>  		}
> >>>>>>  
> >>>>>>  		if (irq_status & BUTTRESS_ISR_IPC_FROM_ISH_IS_WAITING) {
> >>>>>> -			dev_dbg(&isp->pdev->dev,
> >>>>>> -				"BUTTRESS_ISR_IPC_FROM_ISH_IS_WAITING\n");
> >>>>>> -			ipu6_buttress_ipc_recv(isp, &b->ish, &b->ish.recv_data);
> >>>>>> -			complete(&b->ish.recv_complete);
> >>>>>> +			dev_warn(&isp->pdev->dev,
> >>>>>> +				 "BUTTRESS_ISR_IPC_FROM_ISH_IS_WAITING\n");
> >>>>>
> >>>>> I think this is a unrelated change, right?
> >>>>
> >>>> I mean the change from dev_dbg() to dev_warn().
> >>>
> >>> We're not handling these interrupts anymore in any way.
> >>>
> >>> I wonder if the ipu6_buttress_ipc_recv() call should still remain in place,
> >>> even if we really do nothing with these. It looks like some kind of an
> >>> acknowledgement mechanism.
> >>
> >> I just confirm that IPC_FROM_ISH_IS_WAITING and IPC_EXEC_DONE_BY_ISH are
> >> not valid anymore from IPU6, I think the handling here and below could be
> >> removed.
> > 
> > Do you know which IPU version still needed it?
> > 
> > There are folks who'd like to add IPU4 support to the driver but they can
> > add it back if it's needed.
> >
> 
> I know that ISH IPC was added from IPU4, but I am not sure IPU4 really
> need that now.

Ok, I think on v3, I'll remove handling of BUTTRESS_ISR_IPC_FROM_ISH_IS_WAITING
and BUTTRESS_ISR_IPC_EXEC_DONE_BY_ISH from isr, but will keep the BIT's 
definitions just in case.

Regards
Stanislaw