From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Cc: Adam Pigg <adam@piggz.co.uk>,
linux-media@vger.kernel.org, yong.deng@magewell.com,
mchehab@kernel.org, linux-sunxi@lists.linux.dev
Subject: Re: [PATCH 1/3] media: sun6i-csi: merge sun6i_csi_formats and sun6i_csi_formats_match
Date: Sun, 26 Mar 2023 01:17:03 +0200 [thread overview]
Message-ID: <20230325231703.GI22214@pendragon.ideasonboard.com> (raw)
In-Reply-To: <Y/Xw7lsKOYFqaZEw@aptenodytes>
Hi Paul,
On Wed, Feb 22, 2023 at 11:39:42AM +0100, Paul Kocialkowski wrote:
> On Fri 17 Feb 23, 20:20, Laurent Pinchart wrote:
> > Could you share your opinion on the question below ?
>
> Yeah sorry for the long delay here. It's taken me a while to dive back into
> this topic.
>
> My first impression is that I would rather be in favor of keeping a dynamic
> approach like what's already in sun6i_csi_capture_link_validate and extract
> the mbus/pixfmt matching logic from there.
>
> I would be happy to craft a patch in that direction, but maybe you'd like
> to have a try at it (since it's your series). Just let me know, I think I can
> do it pretty quickly now that I have everything back in mind. I could also
> add some comment about the general hardware mechanism/limitations.
>
> The hardware is a bit odd in a few ways and I found that the explicit
> combinatory approach wasn't a very good fit (but obviously that's an open topic
> for discussions).
>
> The general idea is that the SUN6I_CSI_INPUT_FMT_YUV420/422 can only be used
> to produce outputs on 2 or 3 planes, but not packed YUV. There's also no
> explicit hardware reordering of the chroma components, so we need to lie about
> the input order (input_yuv_seq_invert) to achieve inverted chroma components
> on the output format.
>
> In order to produce packed data, we can only rely on SUN6I_CSI_INPUT_FMT_RAW
> which provides no reordering mechanism. This is why it made good sense to me
> to only have an explicit matching table for this case and rely on more general
> logic that reflects the hardware capabilities otherwise.
The explicit logic is more CPU-intensive so I have a small preference
for adding mbus_codes to the sun6i_csi_capture_format structure. It's
true that there will be some duplication of data, but I don't think it
will consume lots of memory. As mentioned in the review of the series
you've recently sent, I'm fine with both approaches, so you can decide.
> > On Sun, Jan 08, 2023 at 10:26:09PM +0200, Laurent Pinchart wrote:
> > > On Sun, Jan 08, 2023 at 06:23:56PM +0000, Adam Pigg wrote:
> > > > On Friday, 6 January 2023 23:31:48 GMT Laurent Pinchart wrote:
> > > > > On Fri, Jan 06, 2023 at 07:40:36PM +0000, adam@piggz.co.uk wrote:
>
> [...]
>
> > > > > > +#define SUN6I_BUS_FMTS(fmt...) (const u32[]) {fmt, 0}
>
> Cosmetic suggestion to stay consistent with the rest:
> #define SUN6I_CSI_CAPTURE_MBUS_CODES(mbus_codes...) \
> (const u32[]) { mbus_codes, 0 }
>
> Also it would look better in sun6i_csi_capture.h.
>
> But really I would be happier with a dynamic approach.
>
> [...]
>
> > > > > > /* YUV422 */
> > > > > > {
> > > > > >
> > > > > > @@ -123,6 +139,8 @@ static const struct sun6i_csi_capture_format
> > > > > > sun6i_csi_capture_formats[] = {>
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_RAW_8,
> > > > > > .input_format_raw = true,
> > > > > > .hsize_len_factor = 2,
> > > > > >
> > > > > > + .mbus_codes = SUN6I_BUS_FMTS(MEDIA_BUS_FMT_YUYV8_2X8,
> > > > > > + MEDIA_BUS_FMT_YUYV8_1X16),
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_YVYU,
> > > > > >
> > > > > > @@ -130,6 +148,8 @@ static const struct sun6i_csi_capture_format
> > > > > > sun6i_csi_capture_formats[] = {>
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_RAW_8,
> > > > > > .input_format_raw = true,
> > > > > > .hsize_len_factor = 2,
> > > > > >
> > > > > > + .mbus_codes = SUN6I_BUS_FMTS(MEDIA_BUS_FMT_YVYU8_2X8,
> > > > > > + MEDIA_BUS_FMT_YVYU8_1X16),
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_UYVY,
> > > > > >
> > > > > > @@ -137,6 +157,8 @@ static const struct sun6i_csi_capture_format
> > > > > > sun6i_csi_capture_formats[] = {
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_RAW_8,
> > > > > > .input_format_raw = true,
> > > > > > .hsize_len_factor = 2,
> > > > > >
> > > > > > + .mbus_codes = SUN6I_BUS_FMTS(MEDIA_BUS_FMT_UYVY8_2X8,
> > > > > > + MEDIA_BUS_FMT_UYVY8_1X16),
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_VYUY,
> > > > > >
> > > > > > @@ -144,57 +166,68 @@ static const struct sun6i_csi_capture_format
> > > > > > sun6i_csi_capture_formats[] = {>
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_RAW_8,
> > > > > > .input_format_raw = true,
> > > > > > .hsize_len_factor = 2,
> > > > > >
> > > > > > + .mbus_codes = SUN6I_BUS_FMTS(MEDIA_BUS_FMT_VYUY8_2X8,
> > > > > > + MEDIA_BUS_FMT_VYUY8_1X16),
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_NV16,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV422SP,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV422SP,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > >
> > > > > I don't think this is correct. To produce semi-planar or multi-planar
> > > > > YUV formats, I believe the CSI needs YUV input. This should thus be
> > > > > (unless I'm mistaken)
>
> You are correct.
>
> > > > >
> > > > > .mbus_codes = SUN6I_BUS_FMTS(MEDIA_BUS_FMT_UYVY8_2X8,
> > > > > MEDIA_BUS_FMT_UYVY8_1X16,
> > > > > MEDIA_BUS_FMT_VYUY8_2X8,
> > > > > MEDIA_BUS_FMT_VYUY8_1X16,
> > > > > MEDIA_BUS_FMT_YUYV8_2X8,
> > > > > MEDIA_BUS_FMT_YUYV8_1X16,
> > > > > MEDIA_BUS_FMT_YVYU8_2X8,
> > > > > MEDIA_BUS_FMT_YVYU8_1X16),
> > > > >
> > > > > and same below.
>
> All of the YUV420/422 pixel formats on 2 or 3 planes can take all of the
> supports packed 16-bit YUV bus formats, which is why it doesn't seem very
> graceful to have an explicit list.
>
> > > > >
> > > > Hi Laurent
> > > >
> > > > Thanks for the help and tips. Ive made all the other changes, which can be
> > > > viewed here until i resubmit them https://github.com/sailfish-on-dontbeevil/
> > > > kernel-megi/commits/pinephone-libcamera
> > > >
> > > > Im just not quite sure on this one. I think my implementation of merging the
> > > > arrays keeps the previous mapping right? In sun6i_csi_capture_format_matches
> > > > there is no mapping for the *NV formats, and the remaining ones ive set to 0?
> > >
> > > The current implementation allows writing multi-planar formats (e.g.
> > > NV12) to memory when the input of the CSI is a YUV media bus format
> > > (e.g. YUYV8_1X16). This patch doesn't change that, but it will prevent
> > > NV12 from being enumerated when using media bus-based enumeration of
> > > pixel formats, so userspace won't see NV12 as being available.
> > >
> > > It would be fine fixing that issue in a separate patch on top of this
> > > one, but I though you could as well do both in one go.
> >
> > Adam, you mentioned that NV12 and NV16 "don't work". Could you elaborate
> > and explain what you've tried exactly ?
>
> We definitely need to keep the ability to produce NV12, NV16 and friends from
> YUV bus formats.
>
> > > > > Paul, could you confirm this ?
> > > > >
> > > > > I'm a bit surprised that the CSI can't shuffle the YUV components for
> > > > > packed YUYV formats, but so be it if that's a hardware limitation.
>
> Yep that is correct, it's a hardware limitation.
>
> Cheers,
>
> Paul
>
> > > > > I'm also thinking that a subsequent patch could drop the raw check from
> > > > > sun6i_csi_capture_link_validate():
> > > > >
> > > > > - /* With raw input mode, we need a 1:1 match between input and output. */
> > > > > - if (bridge_format->input_format == SUN6I_CSI_INPUT_FMT_RAW ||
> > > > > - capture_format->input_format_raw) {
> > > > > - match = sun6i_csi_capture_format_match(pixelformat,
> > > > > - fmt.format.code);
> > > > > - if (!match)
> > > > > - goto invalid;
> > > > > - }
> > > > > + /* Make sure the media bus code and pixel format are compatible. */
> > > > > + match = sun6i_csi_capture_format_match(pixelformat, fmt.format.code);
> > > > > + if (!match)
> > > > > + goto invalid;
> > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_NV61,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV422SP,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV422SP,
> > > > > > .input_yuv_seq_invert = true,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_YUV422P,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV422P,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV422P,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > > /* YUV420 */
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_NV12_16L16,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV420MB,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV420MB,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_NV12,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV420SP,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV420SP,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_NV21,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV420SP,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV420SP,
> > > > > > .input_yuv_seq_invert = true,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > >
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_YUV420,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV420P,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV420P,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_YVU420,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_YUV420P,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_YUV420P,
> > > > > > .input_yuv_seq_invert = true,
> > > > > >
> > > > > > + .mbus_codes = 0,
> > > > > >
> > > > > > },
> > > > > > /* Compressed */
> > > > > > {
> > > > > >
> > > > > > .pixelformat = V4L2_PIX_FMT_JPEG,
> > > > > > .output_format_frame = SUN6I_CSI_OUTPUT_FMT_FRAME_RAW_8,
> > > > > > .output_format_field = SUN6I_CSI_OUTPUT_FMT_FIELD_RAW_8,
> > > > > >
> > > > > > + .mbus_codes = SUN6I_BUS_FMTS(MEDIA_BUS_FMT_JPEG_1X8),
> > > > > >
> > > > > > },
> > > > > >
> > > > > > };
> > > > > >
> > > > > > @@ -210,118 +243,20 @@ struct sun6i_csi_capture_format
> > > > > > *sun6i_csi_capture_format_find(u32 pixelformat)>
> > > > > > return NULL;
> > > > > >
> > > > > > }
> > > > > >
> > > > > > -/* RAW formats need an exact match between pixel and mbus formats. */
> > > > > > -static const
> > > > > > -struct sun6i_csi_capture_format_match sun6i_csi_capture_format_matches[]
> > > > > > = { - /* YUV420 */
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_YUYV,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_YUYV8_2X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_YUYV,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_YUYV8_1X16,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_YVYU,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_YVYU8_2X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_YVYU,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_YVYU8_1X16,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_UYVY,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_UYVY8_2X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_UYVY,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_UYVY8_1X16,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_VYUY,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_VYUY8_2X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_VYUY,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_VYUY8_1X16,
> > > > > > - },
> > > > > > - /* RGB */
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_RGB565,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_RGB565_2X8_LE,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_RGB565X,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_RGB565_2X8_BE,
> > > > > > - },
> > > > > > - /* Bayer */
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SBGGR8,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SBGGR8_1X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SGBRG8,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SGBRG8_1X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SGRBG8,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SGRBG8_1X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SRGGB8,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SRGGB8_1X8,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SBGGR10,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SBGGR10_1X10,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SGBRG10,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SGBRG10_1X10,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SGRBG10,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SGRBG10_1X10,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SRGGB10,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SRGGB10_1X10,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SBGGR12,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SBGGR12_1X12,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SGBRG12,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SGBRG12_1X12,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SGRBG12,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SGRBG12_1X12,
> > > > > > - },
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_SRGGB12,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_SRGGB12_1X12,
> > > > > > - },
> > > > > > - /* Compressed */
> > > > > > - {
> > > > > > - .pixelformat = V4L2_PIX_FMT_JPEG,
> > > > > > - .mbus_code = MEDIA_BUS_FMT_JPEG_1X8,
> > > > > > - },
> > > > > > -};
> > > > > > -
> > > > > >
> > > > > > static bool sun6i_csi_capture_format_match(u32 pixelformat, u32
> > > > > > mbus_code)
> > > > > > {
> > > > > >
> > > > > > - unsigned int i;
> > > > > > -
> > > > > > - for (i = 0; i < ARRAY_SIZE(sun6i_csi_capture_format_matches); i++) {
> > > > > > - const struct sun6i_csi_capture_format_match *match =
> > > > > > - &sun6i_csi_capture_format_matches[i];
> > > > > > -
> > > > > > - if (match->pixelformat == pixelformat &&
> > > > > > - match->mbus_code == mbus_code)
> > > > > > - return true;
> > > > > > + unsigned int i, j;
> > > > > > +
> > > > > > + for (i = 0; i < ARRAY_SIZE(sun6i_csi_capture_formats); i++) {
> > > > > > + const struct sun6i_csi_capture_format *format =
> > > > > > + &sun6i_csi_capture_formats[i];
> > > > > > +
> > > > > > + if (format->pixelformat == pixelformat) {
> > > > > > + for (j = 0; format->mbus_codes[j]; j++) {
> > > > > > + if (mbus_code == format->mbus_codes[j])
> > > > > > + return true;
> > > > > > + }
> > > > > > + }
> > > > > >
> > > > > > }
> > > > > >
> > > > > > return false;
> > > > > >
> > > > > > diff --git a/drivers/media/platform/sunxi/sun6i-csi/sun6i_csi_capture.h
> > > > > > b/drivers/media/platform/sunxi/sun6i-csi/sun6i_csi_capture.h index
> > > > > > 3ee5ccefbd10..0484942834e3 100644
> > > > > > --- a/drivers/media/platform/sunxi/sun6i-csi/sun6i_csi_capture.h
> > > > > > +++ b/drivers/media/platform/sunxi/sun6i-csi/sun6i_csi_capture.h
> > > > > > @@ -27,11 +27,7 @@ struct sun6i_csi_capture_format {
> > > > > >
> > > > > > bool input_yuv_seq_invert;
> > > > > > bool input_format_raw;
> > > > > > u32 hsize_len_factor;
> > > > > >
> > > > > > -};
> > > > > > -
> > > > > > -struct sun6i_csi_capture_format_match {
> > > > > > - u32 pixelformat;
> > > > > > - u32 mbus_code;
> > > > > > + const u32 *mbus_codes;
> > > > > >
> > > > > > };
> > > > > >
> > > > > > #undef current
--
Regards,
Laurent Pinchart
next prev parent reply other threads:[~2023-03-25 23:16 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-06 19:40 [PATCH 0/3] suns6-csi changes to support libcamera adam
2023-01-06 19:40 ` [PATCH 1/3] media: sun6i-csi: merge sun6i_csi_formats and sun6i_csi_formats_match adam
2023-01-06 23:31 ` Laurent Pinchart
2023-01-08 18:23 ` Adam Pigg
2023-01-08 20:26 ` Laurent Pinchart
2023-02-17 18:20 ` Laurent Pinchart
2023-02-22 10:39 ` Paul Kocialkowski
2023-03-21 17:50 ` Adam Pigg
2023-03-23 12:55 ` Paul Kocialkowski
2023-03-25 23:17 ` Laurent Pinchart [this message]
2023-01-06 19:40 ` [PATCH 2/3] media: sun6i-csi: implement V4L2_CAP_IO_MC adam
2023-01-06 23:34 ` Laurent Pinchart
2023-01-06 19:40 ` [PATCH 3/3] media: sun6i-csi: implement vidioc_enum_framesizes adam
2023-01-06 23:35 ` Laurent Pinchart
2023-02-22 10:43 ` Paul Kocialkowski
2023-03-25 23:12 ` Laurent Pinchart
2023-01-06 23:10 ` [PATCH 0/3] suns6-csi changes to support libcamera piggz1
2023-01-07 8:43 ` Adam Pigg
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230325231703.GI22214@pendragon.ideasonboard.com \
--to=laurent.pinchart@ideasonboard.com \
--cc=adam@piggz.co.uk \
--cc=linux-media@vger.kernel.org \
--cc=linux-sunxi@lists.linux.dev \
--cc=mchehab@kernel.org \
--cc=paul.kocialkowski@bootlin.com \
--cc=yong.deng@magewell.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox