Re: [PATCH v2 2/4] media: qcom: camss: Add support for MSM8939

[Vincent Knecht <vincent.knecht@mailoo.org>]

Le lundi 26 mai 2025 à 09:20 +0100, Bryan O'Donoghue a écrit :
> On 25/05/2025 20:25, Vincent Knecht via B4 Relay wrote:
> > From: Vincent Knecht <vincent.knecht@mailoo.org>
> > 
> > The camera subsystem for the MSM8939 is the same as MSM8916 except with
> > 3 CSID instead of 2, and some higher clock rates.
> > 
> > As a quirk, this SoC needs writing values to 2 VFE VBIF registers
> > (see downstream msm8939-camera.dtsi vbif-{regs,settings} properties).
> > This fixes black stripes across sensor and garbage in CSID TPG outputs.
> > 
> > Add support for the MSM8939 camera subsystem.
> > 
> > Signed-off-by: Vincent Knecht <vincent.knecht@mailoo.org>
> > ---
> >   drivers/media/platform/qcom/camss/camss-csiphy.c   |   1 +
> >   drivers/media/platform/qcom/camss/camss-ispif.c    |   8 +-
> >   drivers/media/platform/qcom/camss/camss-vfe-vbif.c |   7 +
> >   drivers/media/platform/qcom/camss/camss-vfe.c      |   1 +
> >   drivers/media/platform/qcom/camss/camss.c          | 157 +++++++++++++++++++++
> >   drivers/media/platform/qcom/camss/camss.h          |   1 +
> >   6 files changed, 173 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/media/platform/qcom/camss/camss-csiphy.c b/drivers/media/platform/qcom/camss/camss-csiphy.c
> > index c622efcc92ff3781d7fc3ace0253c2d64c91e847..6311fc2975aa1345e430a477c8a6476f1d7e5663 100644
> > --- a/drivers/media/platform/qcom/camss/camss-csiphy.c
> > +++ b/drivers/media/platform/qcom/camss/camss-csiphy.c
> > @@ -605,6 +605,7 @@ int msm_csiphy_subdev_init(struct camss *camss,
> >   		return PTR_ERR(csiphy->base);
> >   
> >   	if (camss->res->version == CAMSS_8x16 ||
> > +	    camss->res->version == CAMSS_8x39 ||
> >   	    camss->res->version == CAMSS_8x53 ||
> >   	    camss->res->version == CAMSS_8x96) {
> >   		csiphy->base_clk_mux =
> > diff --git a/drivers/media/platform/qcom/camss/camss-ispif.c b/drivers/media/platform/qcom/camss/camss-ispif.c
> > index 2dc585c6123dd248a5bacd9c7a88cb5375644311..aaf3caa42d33dcb641651e7f5bc0c2a564d85bfa 100644
> > --- a/drivers/media/platform/qcom/camss/camss-ispif.c
> > +++ b/drivers/media/platform/qcom/camss/camss-ispif.c
> > @@ -1112,6 +1112,8 @@ int msm_ispif_subdev_init(struct camss *camss,
> >   	/* Number of ISPIF lines - same as number of CSID hardware modules */
> >   	if (camss->res->version == CAMSS_8x16)
> >   		ispif->line_num = 2;
> > +	else if (camss->res->version == CAMSS_8x39)
> > +		ispif->line_num = 3;
> 
> > +		.interrupt = { "vfe0" },
> > +		.vfe = {
> > +			.line_num = 3,
> 
> Hmm should we really be setting line_num inline in the code ?
> 
> I don't believe we should.
> 
> These parameters should be passed from the resources structures.
> 
> ---
> bod

I've just followed suit, no strong opinion about it.
Can we agree this could be changed in another series ?