RFC: GSM flowcontrol handling, GTA01 and GTA02

Andy Green andy at openmoko.com
Mon May 26 20:21:34 CEST 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

| Patch tarball: http://moko.mwester.net/mw09_kernel_patches.tgz

OK I stuck these into andy branch in git -- compiles OK but I didn't
test gsm performance with this lot in.

Thanks for the work.

Some meta-comments about the patchset rather than the patches

~ - Rather than a script, use a series file to list the order the patches
should apply in.  If you use stgit, this kind of stuff "just happens".

~ - If you add

From: Me <me at here.com>
my-patch-name.patch
Signed-off-by: Me <me at here.com>

at the top of each patch, it saves me doing it... and it is much better
if the originator does the Signed-off-by

~ - The first thing that gets done with incoming patches now is to run
them through ./scripts/checkpatch.pl provided with mainline kernel.  It
found a lot of things to moan about.  Some of them aren't really your
fault as you are just working with what was already the "style in the
file", eg, it complains about using extern in a .c file.  I fixed the
rest of them in a patch on top of your patchset I will post to the
kernel list in a moment.  But you can get a free "proofread" of the
patches before sending by using ./scripts/checkpatch.pl yourself, we
have to be increasingly hardcore about it if we are not to go upstream
with our pants down.

On the patches themselves, nothing about "neo" or "gsm" or "gta" should
appear in ./drivers/serial/s3c2410.c.  For example, in that file the
patches seem to hardcode different behaviour for port #0 because it "is
the GSM port".  This will definitely get its ass kicked if we tried to
push it upstream.  Somehow the "special handling" needs to be done by an
attribute or function pointer through the platform stuff, so
mach-gta01.c gets to implement any knowledge about port usage on our
stuff, either by function handler called at the right time or just by
setting a platform attribute on the port or some such.

Same goes for magic statics in ./drivers/serial/s3c2410.c that hardcode
knowledge about gta01-specific devices.

If you fancy to improve the "upstream compatability" and reissue, please
also look at the style fixes patch or better run ./scripts/checkpatch.pl
on them yourself too.

Despite these comments I really appreciate the work you have put in and
stuck this first version of the patches on the "andy" branch straight away.

- -Andy
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkg6/64ACgkQOjLpvpq7dMqIKwCfURiRWgro910OoNlWZhsnjF7S
T44An2fWEiVfkE3fkVoeoWEQXx0AiJ1U
=ys9m
-----END PGP SIGNATURE-----



More information about the openmoko-devel mailing list