Re: [PATCH v2 3/4] soc: qcom: llcc-qcom: Add support for SM8750

[Melody Olvera]

On 3/11/2025 3:17 AM, Konrad Dybcio wrote:
> On 3/4/25 11:23 PM, Melody Olvera wrote:
> > Add system cache table and configs for SM8750 SoCs.
> >
> > Signed-off-by: Melody Olvera <quic_molvera@xxxxxxxxxxx>
> > ---
> [...]
>
> > +		.usecase_id = LLCC_MODPE,
> > +		.slice_id = 29,
> > +		.max_cap = 256,
> > +		.priority = 1,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xf0000000,
> > +		.ovcap_prio = true,
> ovcap_prio = false, alloc_oneway_en = true
>
> > +	}, {
> > +		.usecase_id = LLCC_WRCACHE,
> > +		.slice_id = 31,
> > +		.max_cap = 512,
> > +		.priority = 1,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> .activate_on_init = true,
>
> [...]
>
> > +		.usecase_id = LLCC_LCPDARE,
> > +		.slice_id = 30,
> > +		.max_cap = 128,
> > +		.priority = 5,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.activate_on_init = true,
> > +		.ovcap_prio = true,
> ovcap_prio = false, alloc_oneway_en = true
>
> [...]
>
> > +		.usecase_id = LLCC_VIDVSP,
> > +		.slice_id = 4,
> > +		.max_cap = 256,
> > +		.priority = 4,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +	}, {
> > +		.usecase_id = LLCC_VIDDEC,
> > +		.slice_id = 5,
> > +		.max_cap = 6144,
> > +		.priority = 4,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.cache_mode = 2,
> > +		.vict_prio = true,
> .vict_prio = false, .overcap_prio = true
>
> > +	}, {
> > +		.usecase_id = LLCC_CAMOFE,
> > +		.slice_id = 33,
> > +		.max_cap = 6144,
> > +		.priority = 4,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.mru_uncap_en = true,
> > +		.vict_prio = true,
> .mru_uncap_en = false, stale_en = true
> .vict_prio = false, .overcap_prio = true
>
> > +	}, {
> > +		.usecase_id = LLCC_CAMRTIP,
> > +		.slice_id = 13,
> > +		.max_cap = 1024,
> > +		.priority = 4,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.mru_uncap_en = true,
> > +		.vict_prio = true,
> same
>
> > +	}, {
> > +		.usecase_id = LLCC_CAMSRTIP,
> > +		.slice_id = 14,
> > +		.max_cap = 6144,
> > +		.priority = 4,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.mru_uncap_en = true,
> > +		.vict_prio = true,
> same
>
> > +	}, {
> > +		.usecase_id = LLCC_CAMRTRF,
> > +		.slice_id = 7,
> > +		.max_cap = 3584,
> > +		.priority = 1,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.mru_uncap_en = true,
> > +		.vict_prio = true,
> same
>
> > +	}, {
> > +		.usecase_id = LLCC_CAMSRTRF,
> > +		.slice_id = 21,
> > +		.max_cap = 6144,
> > +		.priority = 1,
> > +		.fixed_size = true,
> > +		.bonus_ways = 0xffffffff,
> > +		.mru_uncap_en = true,
> > +		.vict_prio = true,
> same

Ack for all the above. Looks like I was looking at an outdated data 
table unfortunately.

> Apart from that, it looks like there's some sort of grouping / parent-child
> relationships involved in this thing - do we need more sw changes for that?

Yes; I originally thought that wasn't relevant for this version but upon 
further review it seems
it is. I believe the functionality changes belong in the previous patch 
so I'll add them there
and add the relevant data to this patch.

Thanks,
Melody