[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAHStOZ4DV8FbAgzMKhnbLOOPj7Gm8+5=KxMSGL6J1D00Ejk42g@mail.gmail.com>
Date: Thu, 2 Aug 2018 15:14:30 +0200
From: Maxime Jourdan <maxi.jourdan@...adoo.fr>
To: Neil Armstrong <narmstrong@...libre.com>
Cc: Maxime Jourdan <maxi.jourdan@...adoo.fr>,
Kevin Hilman <khilman@...libre.com>,
linux-amlogic <linux-amlogic@...ts.infradead.org>,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/4] soc: amlogic: add meson-canvas driver
Hi Neil,
2018-08-02 10:38 GMT+02:00 Neil Armstrong <narmstrong@...libre.com>:
> Please switch to the spdx header format here and in the .h.
> In the DRM driver these are updated in IRQ context, we should make sure we don't sleep
> in interrupt context if IRQ occurs when the VDEC updates it's canvases.
>
> Could you switch to spin_lock_irqsave() instead ?
Will do.
> Can you add the endian defines ?
>
> #define MESON_CANVAS_ENDIAN_SWAP16 0x1
> #define MESON_CANVAS_ENDIAN_SWAP32 0x3
> #define MESON_CANVAS_ENDIAN_SWAP64 0x7
> #define MESON_CANVAS_ENDIAN_SWAP128 0xf
>
> the SWAP64 is the one used in the VDEC and DRM Overlays.
Sure. Maybe I should push the flags (0x1, 0x2, 0x4, 0x8) rather than
the combinations?
Or both, since the combinations are more likely to be used in practice
more than the single flags.
> I rebased my DRM Overlay patch on top of this and :
> Tested-by: Neil Armstrong <narmstrong@...libre.com>
>
> You can keep this tag on future versions.
Thanks!
Powered by blists - more mailing lists