Greetings, and questions about work flow
laforge at openmoko.org
Tue May 15 13:40:29 CEST 2007
On Tue, May 15, 2007 at 12:32:59PM +0200, Pierre Ossman wrote:
> Hi everyone,
> For those of you not recognizing my name, I'm the maintainer of the MMC layer in
> the kernel (which makes it possible to read the included SD card).
yes, thanks for the introduction ;)
> I thought I'd get up to speed on how you guys work. "Normal" kernel works
> usually involves a git repo, but I couldn't find one on the wiki. Do you just
> keep replacing the patches in the svn repo when you modify stuff?
As Ben Dooks also has pointed out, git trees don't really work if you
have something like 20 to 40 disjunct features that at some point you
want to submit as per-feature patches to mainline. Having that many local
branches and syncing/merging them all the tiem is a major PITA. I've
tried this for netfilter/iptables before and was not very happy with it.
> Secondly, I'm curious as to what kind of contact you have with
> Samsung. The wiki states you got the SD driver from TomTom, not from
> Samsung, so I would assume not so good.
The SD driver is from Maintech, whom I introduced to TomTom some time
ago to write Samsung S3c2410 related code. The author of s3cmci, Thoms
Kleffel, is currently in the final schedule of working on his thesis,
while at the same time co-running a embedded company, so he doesn't
really have much time on merging it mainline now [after the first
attempts failed some time ago].
> That driver is not a beauty btw,
Well, I actually think it is quite fine, but then I'm not the MMC/SD
maintianer ;). Please let me know what kind of concerns you have.
> so my first task will probably be to sanitize it. Hopefully we can
> also get it in mainline.
> That requires a maintainer though, and I do not have time to take on
> that role.
I'm happy to take that. Thomas Kleffel already indicated that he would
be happy with me working on this. I understand the s3cmci driver very
well, and I've written u-boot s3c2410 sd card code, so I'm faimilar with
the controller hardware, too.
> Samsung would be preferable, but if they're not available,
> is there anyone else that might be interested?
Samsung is not really actively supporting Linux mainline and/or tracking
We have made contact and had a meeting where we discussed this issue and
explained the Linux kernel development model to them, and outlined how
much it is in their own interest to make sure the best possible support
for their SoC's is in mainline.
I got the feeling that they seem to agree, but that Linux currently is
not percieved as an important market for them - which is probably true,
the S3c24xx are mostly used by windows mobile pda/smart phones today...
In any case, we're trying to get this message to the respective senior
people inside Samsung LSI department, and are hoping to improve it.
However, I doubt this would change anything with regard to their old
chips (s3c2410/s3c2440 are quite old). I know that for their new chips
they're working on 2.6.20 based drivers.
OpenMoko will do a lot of 2410/2440/2443 and later maybe s3c6400 related
development, so I'm happy to maintain the respective drivers/code if we
cannot get more cooperation from Samsung.
- Harald Welte <laforge at openmoko.org> http://openmoko.org/
Software for the world's first truly open Free Software mobile phone
More information about the openmoko-kernel