public inbox for gentoo-dev@lists.gentoo.org
 help / color / mirror / Atom feed
* [gentoo-dev] which contribution prefered - to gentoo or to project itself?
@ 2003-05-11 18:36 Eric Noack
  2003-05-11 19:03 ` Peter Johanson
  0 siblings, 1 reply; 5+ messages in thread
From: Eric Noack @ 2003-05-11 18:36 UTC (permalink / raw
  To: gentoo-dev


Hi there. i have a question about a to-be patch to pcmcia-cs.

while providing almost any pcmcia driver as well as 
patched orinoco drivers, for example, i think most gentoo people choose
pcmcia-cs drivers over kernel pcmcia support.

however unfortunately i needed support for an isdn pcmcia card, and
pcmcia-cs has no isdn support or drivers at all.

on the other side, adding the support to pcmcia_cs is as easy as copying a single driver file
(something_cs.c) from linux/drivers/isdn/hisax/ into the pcmcia-cs source tree
and add it to the make file, if only the kernel hisax driver has been compiled with support
for the pcmcia chip

(which is possible without enabling kernel pcmcia, only the carddriver itself
cant be compiled without)

so it would be easy to write a patch to pcmcia_cs that would check whether the kernel has
isdn and hisax support, and which pcmcia-chip-protocols have been compiled into the hisax driver,
and copy & compile the needed *_cs.c from the hisax dir

however this "patch" could be completely provided by a modified ebuild, maybe with support for an
"isdn" USEflag or such, too, eventually more easily than modifying the pcmcia_cs configure script

so before writing this modification as either a patch to pcmcia_cs or the ebuild or both (ebuild that adds the patch)

what do you think would be prefered?

Eric

--
gentoo-dev@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [gentoo-dev] which contribution prefered - to gentoo or to project itself?
  2003-05-11 18:36 [gentoo-dev] which contribution prefered - to gentoo or to project itself? Eric Noack
@ 2003-05-11 19:03 ` Peter Johanson
  2003-05-11 19:45   ` Eric Noack
  0 siblings, 1 reply; 5+ messages in thread
From: Peter Johanson @ 2003-05-11 19:03 UTC (permalink / raw
  To: Eric Noack; +Cc: gentoo-dev, chadh

On Sun, May 11, 2003 at 08:36:23PM +0200, Eric Noack wrote:
> however unfortunately i needed support for an isdn pcmcia card, and
> pcmcia-cs has no isdn support or drivers at all.

):

> 
> on the other side, adding the support to pcmcia_cs is as easy as copying a single driver file
> (something_cs.c) from linux/drivers/isdn/hisax/ into the pcmcia-cs source tree
> and add it to the make file, if only the kernel hisax driver has been compiled with support
> for the pcmcia chip
> 
> (which is possible without enabling kernel pcmcia, only the carddriver itself
> cant be compiled without)
> 
> so it would be easy to write a patch to pcmcia_cs that would check whether the kernel has
> isdn and hisax support, and which pcmcia-chip-protocols have been compiled into the hisax driver,
> and copy & compile the needed *_cs.c from the hisax dir

The check would definitely be good, but i'm hesitant to blindly copy
files over from /usr/src/linux (which may contain any bastardization of
the linux kernel with unknown patches, etc) and attempting to compile
it. i think it'd be better to push a packaged tarball to gentoo mirrors
that hold the necessary driver pieces. That way, we can test that things
actually work. If we copy over from /usr/src/linux, we have no way of
verifying things will work. 
> 
> however this "patch" could be completely provided by a modified ebuild, maybe with support for an
> "isdn" USEflag or such, too, eventually more easily than modifying the pcmcia_cs configure script

i definitely don't think this warrants a new USE flag, unless there's
quite a few more ebuilds that would benefit from it. As long as we can
verify that no compilation errors will occur, i see no problems with
doing this by default (given the above tests pass of course).
> 
> so before writing this modification as either a patch to pcmcia_cs or the ebuild or both (ebuild that adds the patch)
If we have a seperate tarball, We can do the testing in the ebuild and
apply a patch to the pcmcia directory including the configure edit and
the new *_cs.c files.

also, i haven't scoured the pcmcia-cs m/l for this particular topic, but
does david hinds intend to add this support eventually?

chad, what do you think about this approach? any other ideas or
implementations you see as being superior?

-pete

-- 
Peter Johanson
<latexer@gentoo.org>

--
gentoo-dev@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [gentoo-dev] which contribution prefered - to gentoo or to project itself?
  2003-05-11 19:03 ` Peter Johanson
@ 2003-05-11 19:45   ` Eric Noack
  2003-05-11 19:53     ` Peter Johanson
  0 siblings, 1 reply; 5+ messages in thread
From: Eric Noack @ 2003-05-11 19:45 UTC (permalink / raw
  To: gentoo-dev



Hmm

I agree that extra useflags arent such a good idea ;)

Since the "Configure" of pcmcia-cs is not autogenerated but handmade, and
the kernel configuration is allready read there, the hisax support of the interesting cards should
be easy to check, so the ebuild will only do an extra patch.
So my original question has been answered by the code already :)


However its the question if it would be better to provide a patch "with" the sources of
these *_cs.c pcmcia client drivers, that could break with the hisax part of the driver,
or copy the kernel _cs.c files of this clients, which could possibly break with the pcmcia part.

I personally would prefer trying to copy & compile the sources from the kernel dir, to fit against
latest patches to the isdn and HiSax subsystem.

I dont think the pcmcia system in the 2.4 kernel would be changed to a level where
a _cs.c file from the kernel wouldnt compile with the pcmcia_cs pcmcia sub-system anymore,
since that would affect the cardmgr and such, which comes always with pcmcia-cs
and would have to comply with kernel-pcmcia in that case, too.
It is theoretically possible, but a change to the isdn system or comunication between the
hisax part and the _cs part of the same card driver is more likely, and would break _cs parts
supplied by a patch to pcmcia_cs directly.

The alternative would be to provide almost the whole isdn or at least hisax support with
pcmcia-cs like its done with WLAN card support.
but since only 3 of the about 2 dozen ISDN cards supported by HiSax are PCMCIAs,
this wouldnt be such a good idea i think.

Eric

--
gentoo-dev@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [gentoo-dev] which contribution prefered - to gentoo or to project itself?
  2003-05-11 19:45   ` Eric Noack
@ 2003-05-11 19:53     ` Peter Johanson
  2003-05-11 22:21       ` Eric Noack
  0 siblings, 1 reply; 5+ messages in thread
From: Peter Johanson @ 2003-05-11 19:53 UTC (permalink / raw
  To: Eric Noack; +Cc: gentoo-dev

On Sun, May 11, 2003 at 09:45:21PM +0200, Eric Noack wrote:
> Hmm
> 
> I agree that extra useflags arent such a good idea ;)
> 
> Since the "Configure" of pcmcia-cs is not autogenerated but handmade, and
> the kernel configuration is allready read there, the hisax support of the interesting cards should
> be easy to check, so the ebuild will only do an extra patch.
> So my original question has been answered by the code already :)

good point. We can just add another fun sed line based on the
success/failure of the test on the kernel config.

> 
> 
> However its the question if it would be better to provide a patch "with" the sources of
> these *_cs.c pcmcia client drivers, that could break with the hisax part of the driver,
> or copy the kernel _cs.c files of this clients, which could possibly break with the pcmcia part.
> 
> I personally would prefer trying to copy & compile the sources from the kernel dir, to fit against
> latest patches to the isdn and HiSax subsystem.

yeah. i agree now that i think more on it.

> I dont think the pcmcia system in the 2.4 kernel would be changed to a level where
> a _cs.c file from the kernel wouldnt compile with the pcmcia_cs pcmcia sub-system anymore,
> since that would affect the cardmgr and such, which comes always with pcmcia-cs
> and would have to comply with kernel-pcmcia in that case, too.
> It is theoretically possible, but a change to the isdn system or comunication between the
> hisax part and the _cs part of the same card driver is more likely, and would break _cs parts
> supplied by a patch to pcmcia_cs directly.

yeah. i thought of that right after i sent the email. Make up a version
of the ebuild copying over from /usr/src/linux, and lets see how well
that works out.

> The alternative would be to provide almost the whole isdn or at least hisax support with
> pcmcia-cs like its done with WLAN card support.
> but since only 3 of the about 2 dozen ISDN cards supported by HiSax are PCMCIAs,
> this wouldnt be such a good idea i think.

i agree; i think this would would be extra work, and a waste of time for people
when it's a small enough thing just to include in the pcmcia-cs ebuild.

-- 
Peter Johanson
<latexer@gentoo.org>

--
gentoo-dev@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [gentoo-dev] which contribution prefered - to gentoo or to project itself?
  2003-05-11 19:53     ` Peter Johanson
@ 2003-05-11 22:21       ` Eric Noack
  0 siblings, 0 replies; 5+ messages in thread
From: Eric Noack @ 2003-05-11 22:21 UTC (permalink / raw
  To: gentoo-dev


ok so much for that,
we have a patch to compile that module and detect the card on the pcmcia bus
and an ebuild to include it now.

if interested, look at

http://gemia.de/pcmcia-cs-hisax/pcmcia-cs-3.2.4-r1.ebuild
http://gemia.de/pcmcia-cs-hisax/pcmcia-cs-3.2.4-hisax-patch.diff

however if someone is there to test that, i have really no idea how to
initialize that card to get some interface or such, so
/etc/pcmcia/isdn, isdn.opts and what else configuration would be needed,
is a book with 7 seals 4 me until I rtfm on isdn4linux...

if anyone could test and-or gimme hints that would be cool.

have fun.


Eric

--
gentoo-dev@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2003-05-11 22:21 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-05-11 18:36 [gentoo-dev] which contribution prefered - to gentoo or to project itself? Eric Noack
2003-05-11 19:03 ` Peter Johanson
2003-05-11 19:45   ` Eric Noack
2003-05-11 19:53     ` Peter Johanson
2003-05-11 22:21       ` Eric Noack

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox