From: Sergei Khomenkov <shomenkow@gmail.com>
To: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Lucas Sinn <lucas.sinn@wolfvision.net>,
Mauro Carvalho Chehab <mchehab@kernel.org>,
Rob Herring <robh@kernel.org>,
Krzysztof Kozlowski <krzk+dt@kernel.org>,
Conor Dooley <conor+dt@kernel.org>,
Heiko Stuebner <heiko@sntech.de>,
Philipp Zabel <p.zabel@pengutronix.de>,
linux-media@vger.kernel.org, linux-rockchip@lists.infradead.org,
devicetree@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v7 2/4] media: rockchip: Add JPEG decoder driver
Date: Sat, 7 Nov 2026 15:38:56 +0200 [thread overview]
Message-ID: <596F6C9E-4A5D-4AE8-8D26-7A7247050D7D@gmail.com> (raw)
In-Reply-To: <20261022-rockchip-jpegdec-v7-2-3fd7dd55f408@pengutronix.de>
Hi Sascha,
> +/* Bytes per pixel used to size a coded buffer for a given resolution. */
> +#define RKJPEGD_CODED_MAX_DEPTH 2
A valid JPEG can be larger than 2 bytes per pixel. libjpeg-turbo sizes
its worst case in tj3JPEGBufSize() at 3 bytes per pixel for 4:2:0, 4
for 4:2:2 and 6 for 4:4:4, plus 2 KiB. Quality 100 noise comes close:
2.75 bytes per pixel at 4:2:0, both from libjpeg and from the Hantro H1
encoder on an RK3026. Such a frame does not fit the default buffer; by
your commit message a truncated one fails through BUF_EMPTY, so the
client gets an ERROR buffer for a valid picture. I can't run the VDPU720
to confirm.
The default only matters to clients that pass sizeimage 0. GStreamer
1.26 asks for max_width * max_height / 2 of the enumerated range; at
65528x65528 that wraps in its 32-bit calculate_max_sizeimage() to about
255 MiB, which the driver keeps, so 1.26 is not affected. Its main
branch (08ea5bc9fab9) asks for width * height bytes of the stream,
which the driver keeps too, so these frames would not fit. Both are
from reading the code; I did not run them on this driver.
Would 3 bytes per pixel be acceptable as the default, and as a floor
for smaller requests, like the 2 bytes per pixel floor in 6/7 of your
hantro-based v1? 4:4:4 at quality 100 would still not fit, but the
subsampling is not known before the header is parsed.
Thanks,
Sergei
next prev parent reply other threads:[~2026-11-07 13:39 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-22 11:06 [PATCH v7 0/4] " Sascha Hauer
2026-10-22 11:06 ` [PATCH v7 1/4] media: dt-bindings: Add Rockchip JPEG decoder Sascha Hauer
2026-10-22 11:06 ` [PATCH v7 2/4] media: rockchip: Add JPEG decoder driver Sascha Hauer
2026-10-22 17:00 ` Torsten Duwe
2026-10-23 6:59 ` Sascha Hauer
2026-10-28 19:53 ` Nicolas Dufresne
2026-11-07 13:38 ` Sergei Khomenkov [this message]
2026-10-22 11:06 ` [PATCH v7 3/4] arm64: dts: rockchip: rk3588: Add JPEG decoder node Sascha Hauer
2026-10-22 11:06 ` [PATCH v7 4/4] arm64: dts: rockchip: rk356x: " Sascha Hauer
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=596F6C9E-4A5D-4AE8-8D26-7A7247050D7D@gmail.com \
--to=shomenkow@gmail.com \
--cc=conor+dt@kernel.org \
--cc=devicetree@vger.kernel.org \
--cc=heiko@sntech.de \
--cc=krzk+dt@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=linux-rockchip@lists.infradead.org \
--cc=lucas.sinn@wolfvision.net \
--cc=mchehab@kernel.org \
--cc=p.zabel@pengutronix.de \
--cc=robh@kernel.org \
--cc=s.hauer@pengutronix.de \
/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
all inboxes | Powered by JetHome®