From: Arun Kumar K <arunkk.samsung@gmail.com>
To: Sylwester Nawrocki <sylvester.nawrocki@gmail.com>
Cc: Arun Kumar K <arun.kk@samsung.com>,
LMML <linux-media@vger.kernel.org>,
Sylwester Nawrocki <s.nawrocki@samsung.com>,
kilyeon.im@samsung.com, shaik.ameer@samsung.com,
linux-samsung-soc <linux-samsung-soc@vger.kernel.org>
Subject: Re: [RFC v2 03/10] exynos5-fimc-is: Adds common driver header files
Date: Fri, 21 Jun 2013 12:44:14 +0530 [thread overview]
Message-ID: <CALt3h7_r3QLQs1urkhQO+1fCv1J+RaWSQ511RJELxE91cgEUdA@mail.gmail.com> (raw)
In-Reply-To: <51C3865A.4050701@gmail.com>
Hi Sylwester,
On Fri, Jun 21, 2013 at 4:16 AM, Sylwester Nawrocki
<sylvester.nawrocki@gmail.com> wrote:
> Guys, I was wondering how difficult would be to make a common driver
> for the Exynos4 and Exynos5 FIMC-IS ? My feeling is that it would allow
> to save significant amount of code, since the hardware has many
> similarities. I imagine it would be a lot of work, and testing would have
> been a bit difficult. But would it really to troublesome to make a common
> driver ? Could you list some arguments against it ? For the MFC we have
> same driver, handling different firmware versions. Similarly for the other
> media IPs. Only the FIMC-IS subsystems would have separate drivers.
> My intentions is really only to reduce the amount of code we would have
> to merge with this new driver, nothing else. But I'm not going to push
> for the common driver if this is too much trouble.
We have thought about it while starting the development and major
arguments against common driver are :
- FIMC-IS IP has significantly changed from Exynos4.
In Exynos4, it has sub-components ISP, DRC and FD where as in exynos5,
it has ISP, DRC, SCC, ODC, DIS. 3DNR, SCP and FD.
- The FW design has changed considerably to make use of camera2 api
interface. Most of the code in the new driver is for this FW interface
which are done in fimc_is_pipeline.* and fimc_is_interface.*. This is the
major reason against a common driver as the new FW expects each
input frame to be passed along with the controls in a SHOT command.
It is a request-response mode handled per-frame by the FW which is
a major design philosophy change from exynos4.
- Two scalers introduced in the pipeline capable of DMA out which
again changes the pipeline design considerably compared to exynos4.
- The only common part of code between exynos 4 and 5 now is in
the fimc-isp.c and fimc-is-sensor.c and some control structures
in header files. If re-used, only some user controls part can be
re-used and most of the code will still be different.
>From the exynos5 driver, still the fimc-is-scaler.*, fimc-is-pipeline.*,
fimc-is-interface.* has to be retained which constitutes majority of the LOC.
Regards
Arun
next prev parent reply other threads:[~2013-06-21 7:14 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-31 13:03 [RFC v2 00/10] Exynos5 FIMC-IS driver Arun Kumar K
2013-05-31 13:03 ` [RFC v2 01/10] exynos5-fimc-is: Add Exynos5 FIMC-IS device tree bindings documentation Arun Kumar K
2013-06-20 22:45 ` Sylwester Nawrocki
2013-07-09 11:08 ` Arun Kumar K
2013-07-16 21:23 ` Sylwester Nawrocki
2013-07-17 4:42 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 02/10] exynos5-fimc-is: Adds fimc-is driver core files Arun Kumar K
2013-06-06 5:20 ` Sachin Kamat
2013-06-07 10:26 ` Arun Kumar K
2013-06-20 22:46 ` Sylwester Nawrocki
2013-07-09 11:10 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 03/10] exynos5-fimc-is: Adds common driver header files Arun Kumar K
2013-06-20 22:46 ` Sylwester Nawrocki
2013-06-21 7:14 ` Arun Kumar K [this message]
2013-07-09 11:20 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 04/10] exynos5-fimc-is: Adds the register definition and context header Arun Kumar K
2013-06-06 6:24 ` Sachin Kamat
2013-06-07 10:27 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 05/10] exynos5-fimc-is: Adds the sensor subdev Arun Kumar K
2013-06-06 6:39 ` Sachin Kamat
2013-06-07 10:30 ` Arun Kumar K
2013-06-20 23:04 ` Sylwester Nawrocki
2013-07-09 12:01 ` Arun Kumar K
2013-06-26 7:27 ` Hans Verkuil
2013-07-09 12:04 ` Arun Kumar K
2013-07-16 22:03 ` Sylwester Nawrocki
2013-07-17 4:55 ` Arun Kumar K
2013-07-17 14:14 ` Sylwester Nawrocki
2013-07-18 4:35 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 06/10] exynos5-fimc-is: Adds isp subdev Arun Kumar K
2013-06-06 6:18 ` Sachin Kamat
2013-06-07 10:28 ` Arun Kumar K
2013-06-20 23:25 ` Sylwester Nawrocki
2013-07-09 11:42 ` Arun Kumar K
2013-07-16 22:11 ` Sylwester Nawrocki
2013-07-17 4:56 ` Arun Kumar K
2013-08-02 4:31 ` Arun Kumar K
2013-08-03 21:38 ` Sylwester Nawrocki
2013-06-26 7:15 ` Hans Verkuil
2013-07-09 11:42 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 07/10] exynos5-fimc-is: Adds scaler subdev Arun Kumar K
2013-06-06 6:45 ` Sachin Kamat
2013-06-26 7:13 ` Hans Verkuil
2013-07-09 11:30 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 08/10] exynos5-fimc-is: Adds the hardware pipeline control Arun Kumar K
2013-05-31 13:03 ` [RFC v2 09/10] exynos5-fimc-is: Adds the hardware interface module Arun Kumar K
2013-06-21 11:23 ` Andrzej Hajda
2013-07-09 11:26 ` Arun Kumar K
2013-05-31 13:03 ` [RFC v2 10/10] exynos5-fimc-is: Adds the Kconfig and Makefile Arun Kumar K
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=CALt3h7_r3QLQs1urkhQO+1fCv1J+RaWSQ511RJELxE91cgEUdA@mail.gmail.com \
--to=arunkk.samsung@gmail.com \
--cc=arun.kk@samsung.com \
--cc=kilyeon.im@samsung.com \
--cc=linux-media@vger.kernel.org \
--cc=linux-samsung-soc@vger.kernel.org \
--cc=s.nawrocki@samsung.com \
--cc=shaik.ameer@samsung.com \
--cc=sylvester.nawrocki@gmail.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;
as well as URLs for NNTP newsgroup(s).