Re: [PATCH v11 00/11] Rockchip ISP Driver

From: Ezequiel Garcia
Date: Mon Nov 18 2019 - 13:52:26 EST


Hi Hans,

Thanks for taking care of this.

On Thu, 2019-11-14 at 09:42 +0100, Hans Verkuil wrote:
> On 11/14/19 6:12 AM, Helen Koike wrote:
> > Hello,
> >
> > This series adds the Rockchip Image Signal Processing Unit v1 driver to
> > staging.
> >
> > The main reason to be in staging is that people are already using it from the
> > mailing list (including libcamera), and having it in mainline makes the workflow
> > easier. Also, it is easier for other people to contribute back (with code
> > or testing the driver).
> >
> > We plan to actively work on this driver to get it our of staging.
> >
> > This patchset is also available at:
> > https://gitlab.collabora.com/koike/linux/tree/rockchip/isp/v11
> >
> > Libcamera patched to work with this version:
> > https://gitlab.collabora.com/koike/libcamera
> > (also sent to the mailing list)
> >
> > The major difference in v11 are:
> > - Fixed compiling warnings found with W=1
> > - Fixed checkpatch errors
> > - Add clock-names values in dt-bindings
>
> Looking at checkpatch I see a few remaining issues that I believe should be
> fixed before merging this:
>
> CHECK: spinlock_t definition without comment
> #575: FILE: drivers/staging/media/rkisp1/isp_stats.h:43:
> + spinlock_t irq_lock;
>
> CHECK: struct mutex definition without comment
> #581: FILE: drivers/staging/media/rkisp1/isp_stats.h:49:
> + struct mutex wq_lock;
>
> CHECK: spinlock_t definition without comment
> #1648: FILE: drivers/staging/media/rkisp1/isp_params.h:25:
> + spinlock_t config_lock;
>
> CHECK: spinlock_t definition without comment
> #2058: FILE: drivers/staging/media/rkisp1/capture.h:145:
> + spinlock_t vbq_lock;
>

I'd rather merge this as-is, adding a TODO entry stating
we need to revisit locking specifically, because I'd like
to take a close look at these spinlocks/mutex,
instead of just addding comments for then.

> Once this is done together with the Jacob Chen email clarification
> it is ready to be merged for v5.6.
>

I'll find out more about this.

> It passes all the sparse/smatch tests, so that's very good.
>

Great!

Thanks,
Ezequiel