| 1 |
On 30 Jan 2003 01:37:46 -0500 |
| 2 |
Chad Huneycutt <chadh@g.o> wrote: |
| 3 |
|
| 4 |
<snip> |
| 5 |
|
| 6 |
> To address these problems, I have broken the pcmcia-cs package into |
| 7 |
> two separate pieces, pcmcia-cs-tools and pcmcia-cs-drivers. The idea |
| 8 |
> is that the split package will make setup a little more clear. It is |
| 9 |
> a little confusing to have your kernel configuration affect the |
| 10 |
> outcome of an emerge. |
| 11 |
|
| 12 |
Seems fine to me. I'm only using the kernel drivers, so I'll give this a |
| 13 |
test this evening (UK time). I don't want my laptop dying whilst I'm in |
| 14 |
the office. |
| 15 |
|
| 16 |
BTW I'm currently running pcmcia-cs-3.2.3-r1 (my entire notebooke is on |
| 17 |
~x86). I've set /etc/conf.d/pcmcia-cs to load yenta-socket |
| 18 |
(PCIC="yenta socket") and everything work here. My only comment is that |
| 19 |
the comments in this file need updating. |
| 20 |
|
| 21 |
> The pcmcia installation steps of a laptop now looks something like: |
| 22 |
> |
| 23 |
> 1. When configuring the kernel, if you need pcmcia support, make sure |
| 24 |
> to*enable* it (either as modules or built-in). If you need any |
| 25 |
> modules that *aren't provided by pcmcia-cs [probably need to provide a |
| 26 |
> pointer to pcmcia-cs.sf.net's list of supported cards], then enable |
| 27 |
> them(probably as modules, so they can be unloaded when you remove the |
| 28 |
> card). |
| 29 |
|
| 30 |
Reading the kernel configuration help, this looks like a problem for |
| 31 |
(very) old laptops, i.e. ones with the old 16 bit PCMCIA interface |
| 32 |
instead of the 32 bit CardBus interface. I don't have an old enough |
| 33 |
laptop to confirm whether the kernel drivers will work with old 16 bit |
| 34 |
PCMCIA interfaces. |
| 35 |
|
| 36 |
> 2. After installing your new kernel modules (make modules_install), if |
| 37 |
> you need any kernel modules provided by the pcmcia-cs project, then |
| 38 |
> emerge pcmcia-cs-drivers. NOTE: you will need to remerge this every |
| 39 |
> time you rebuild your kernel to rebuild the modules for your new |
| 40 |
> kernel and reinstall them. |
| 41 |
|
| 42 |
I won't use this unless I can't avoid it because I find it a real pain, |
| 43 |
especially when I have multiple kernels installed. |
| 44 |
|
| 45 |
> 3. If you want to use cardmgr to control starting and stopping of |
| 46 |
> pcmcia services, then emerge pcmcia-cs-tools && rc-update add pcmcia |
| 47 |
> default |
| 48 |
> |
| 49 |
> This scheme essentially means that we use yenta_socket and prefer |
| 50 |
> pcmcia-cs modules, although I think that if you configure a kernel |
| 51 |
> module, it will override the pcmcia-cs module. |
| 52 |
|
| 53 |
So you are saying if I already have the modules, it won't be overwritten |
| 54 |
by building the pcmcia-cs-kernel module? If this is the case, then I can |
| 55 |
test this for the PCMCIA modules I've built for my kernel. I'll report |
| 56 |
back on if it works later. |
| 57 |
|
| 58 |
> The only case that |
| 59 |
> isn't handled as I mentioned above is if you *need* the i82365 and ds |
| 60 |
> driver from pcmcia-cs. |
| 61 |
> |
| 62 |
> Thoughts, objections, discussion? |
| 63 |
|
| 64 |
As I say, the lack of support for 16 bit PCMCIA interfaces in the kernel |
| 65 |
may be a problem for people with (very) old notebooks, but I don't have |
| 66 |
anything old enough to test it on. In principal, what you've done will |
| 67 |
be fine for everyone else IMHO. |
| 68 |
-- |
| 69 |
Mark Gordon |
| 70 |
Paid to be a geek. |
| 71 |
Currently looking for a new job within commuting distance of Slough, |
| 72 |
Berks, U.K. |
| 73 |
|
| 74 |
-- |
| 75 |
gentoo-dev@g.o mailing list |
Archives