lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date:   Thu, 28 Jul 2022 17:20:13 +0300
From:   Laurent Pinchart <laurent.pinchart@...asonboard.com>
To:     paul.elder@...asonboard.com
Cc:     linux-media@...r.kernel.org, Dafna Hirschfeld <dafna@...tmail.com>,
        Mauro Carvalho Chehab <mchehab@...nel.org>,
        Heiko Stuebner <heiko@...ech.de>,
        linux-rockchip@...ts.infradead.org,
        linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] media: rkisp1: Add UYVY as an output format

Hi Paul,

On Thu, Jul 28, 2022 at 09:52:59PM +0900, paul.elder@...asonboard.com wrote:
> On Wed, Jul 20, 2022 at 01:29:57AM +0300, Laurent Pinchart wrote:
> > On Thu, Jul 14, 2022 at 08:26:03PM +0900, Paul Elder wrote:
> > > Add support for UYVY as an output format. The uv_swap bit in the
> > > MI_XTD_FORMAT_CTRL register that is used for the NV formats does not
> > > work for packed YUV formats. Thus, UYVY support is implemented via
> > > byte-swapping. This method clearly does not work for implementing
> > > support for YVYU and VYUY.
> > > 
> > > Signed-off-by: Paul Elder <paul.elder@...asonboard.com>
> > > 
> > > ---
> > > UYVY for the self path has not been tested because no device supports
> > > it. The rk3399 has a self path, but does not have the
> > > MI_OUTPUT_ALIGN_FORMAT register and thus does not support UYVY. The
> > > i.MX8MP does support UYVY, but does not have a self path.
> > 
> > I'm tempted to drop it then, as the code below isn't correct given that
> > you use the same register for both MP and SP. Let's address MP only for
> > now.
> 
> The same register is used for both MP and SP. They just have different
> bits. Which is why we'd need the read-write cycle, assuming that there
> exists a device that has both an SP and the MI_OUTPUT_ALIGN_FORMAT
> register.

Indeed, I had missed that. The documentation is confusing, the register
is described as "Output align format for main path", has both
mp_byte_swap and sp_byte_swap, but no sp equivalent to mp_lsb_alignment
(maybe the self path doesn't support raw outputs though ?).

I'm OK keeping support for both paths, but I think the
MI_OUTPUT_ALIGN_FORMAT register should then be initialized to a default
value somewhere.

> > > ---
> > >  .../platform/rockchip/rkisp1/rkisp1-capture.c | 40 +++++++++++++++++++
> > >  1 file changed, 40 insertions(+)
> > > 
> > > diff --git a/drivers/media/platform/rockchip/rkisp1/rkisp1-capture.c b/drivers/media/platform/rockchip/rkisp1/rkisp1-capture.c
> > > index 85fd85fe208c..77496ccef7ec 100644
> > > --- a/drivers/media/platform/rockchip/rkisp1/rkisp1-capture.c
> > > +++ b/drivers/media/platform/rockchip/rkisp1/rkisp1-capture.c
> > > @@ -97,6 +97,12 @@ static const struct rkisp1_capture_fmt_cfg rkisp1_mp_fmts[] = {
> > >  		.uv_swap = 0,
> > >  		.write_format = RKISP1_MI_CTRL_MP_WRITE_YUVINT,
> > >  		.mbus = MEDIA_BUS_FMT_YUYV8_2X8,
> > > +	}, {
> > > +		.fourcc = V4L2_PIX_FMT_UYVY,
> > > +		.uv_swap = 0,
> > > +		.yc_swap = 1,
> > > +		.write_format = RKISP1_MI_CTRL_MP_WRITE_YUVINT,
> > > +		.mbus = MEDIA_BUS_FMT_YUYV8_2X8,
> > >  	}, {
> > >  		.fourcc = V4L2_PIX_FMT_YUV422P,
> > >  		.uv_swap = 0,
> > > @@ -231,6 +237,13 @@ static const struct rkisp1_capture_fmt_cfg rkisp1_sp_fmts[] = {
> > >  		.write_format = RKISP1_MI_CTRL_SP_WRITE_INT,
> > >  		.output_format = RKISP1_MI_CTRL_SP_OUTPUT_YUV422,
> > >  		.mbus = MEDIA_BUS_FMT_YUYV8_2X8,
> > > +	}, {
> > > +		.fourcc = V4L2_PIX_FMT_UYVY,
> > > +		.uv_swap = 0,
> > > +		.yc_swap = 1,
> > > +		.write_format = RKISP1_MI_CTRL_SP_WRITE_INT,
> > > +		.output_format = RKISP1_MI_CTRL_SP_OUTPUT_YUV422,
> > > +		.mbus = MEDIA_BUS_FMT_YUYV8_2X8,
> > >  	}, {
> > >  		.fourcc = V4L2_PIX_FMT_YUV422P,
> > >  		.uv_swap = 0,
> > > @@ -464,6 +477,20 @@ static void rkisp1_mp_config(struct rkisp1_capture *cap)
> > >  		rkisp1_write(rkisp1, RKISP1_CIF_MI_XTD_FORMAT_CTRL, reg);
> > >  	}
> > >  
> > > +	/*
> > > +	 * uv swapping with the MI_XTD_FORMAT_CTRL register only works for
> > 
> > s@uv@U/V@
> > 
> > > +	 * NV12/NV21 and NV16/NV61, so instead use byte swap to support UYVY.
> > > +	 * YVYU and VYUY cannot be supported with this method.
> > > +	 */
> > > +	if (rkisp1->info->features & RKISP1_FEATURE_MI_OUTPUT_ALIGN) {
> > > +		reg = rkisp1_read(rkisp1, RKISP1_CIF_MI_OUTPUT_ALIGN_FORMAT);
> > > +		if (cap->pix.cfg->yc_swap)
> > > +			reg |= RKISP1_CIF_OUTPUT_ALIGN_FORMAT_MP_BYTE_SWAP_BYTES;
> > > +		else
> > > +			reg &= ~RKISP1_CIF_OUTPUT_ALIGN_FORMAT_MP_BYTE_SWAP_BYTES;
> > > +		rkisp1_write(rkisp1, RKISP1_CIF_MI_OUTPUT_ALIGN_FORMAT, reg);
> > 
> > As the register is not initialized anywhere, it would be better to write
> > it fully here instead of a read-modify-write cycle. Same comments below.
> > 
> > > +	}
> > > +
> > >  	rkisp1_mi_config_ctrl(cap);
> > >  
> > >  	reg = rkisp1_read(rkisp1, RKISP1_CIF_MI_CTRL);
> > > @@ -505,6 +532,19 @@ static void rkisp1_sp_config(struct rkisp1_capture *cap)
> > >  		rkisp1_write(rkisp1, RKISP1_CIF_MI_XTD_FORMAT_CTRL, reg);
> > >  	}
> > >  
> > > +	/*
> > > +	 * uv swapping with the MI_XTD_FORMAT_CTRL register only works for
> > > +	 * NV12/NV21 and NV16/NV61, so instead use byte swap to support UYVY.
> > > +	 * YVYU and VYUY cannot be supported with this method.
> > > +	 */
> > > +	if (rkisp1->info->features & RKISP1_FEATURE_MI_OUTPUT_ALIGN) {
> > > +		reg = rkisp1_read(rkisp1, RKISP1_CIF_MI_OUTPUT_ALIGN_FORMAT);
> > > +		if (cap->pix.cfg->yc_swap)
> > > +			reg |= RKISP1_CIF_OUTPUT_ALIGN_FORMAT_SP_BYTE_SWAP_BYTES;
> > > +		else
> > > +			reg &= ~RKISP1_CIF_OUTPUT_ALIGN_FORMAT_SP_BYTE_SWAP_BYTES;
> > > +		rkisp1_write(rkisp1, RKISP1_CIF_MI_OUTPUT_ALIGN_FORMAT, reg);
> > > +	}
> > 
> > Missing blank line.
> > 
> > Reviewed-by: Laurent Pinchart <laurent.pinchart@...asonboard.com>
> > 
> > >  	rkisp1_mi_config_ctrl(cap);
> > >  
> > >  	mi_ctrl = rkisp1_read(rkisp1, RKISP1_CIF_MI_CTRL);

-- 
Regards,

Laurent Pinchart

Powered by blists - more mailing lists