[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: 2.6.14 status summary, and upcoming 2.6.15 ...



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

On Mon, 28 Nov 2005 13:24:34 +0100
Frans Pop <aragorn@tiscali.nl> wrote:

> On Monday 28 November 2005 12:45, Jonas Smedegaard wrote:
> > The following problems are known for yaird currently in sid:
> 
> Add:
> * Does not work for drivers that don't have sysfs support, like
> BusLogic

Ahem - yes:

On Mon, 28 Nov 2005 12:45:42 +0100
Jonas Smedegaard <dr@jones.dk> wrote:

>  * Works only with drivers supporting sysfs.


> > We do not have _automated_ workarounds, which it seems you imply
> > and I clearly didn't above.
> 
> Manual workarounds are not always possible. In a lot of cases manual 
> workarounds are fine for upgrades, but they are impractical to say
> the least for new installations.
> Every case where a manual workaround is needed is one that makes
> yaird less attractive for use in Debian Installer.

I am aware of that:

It is not a bug in yaird to rely on the kernel providing sysfs info -
it's a design decision. And it's not a bug in the kernel to not provide
sysfs info for all drivers - it's a transition phase.

But the end result for Debian is that yaird may not be interesting to
use by default. We may want a ramdisk tool which keeps its own lists of
How the World Really Works, independent from the kernel.
Initramfs-tools stores some such info on its own and relies on udev for
other parts.


Correct me if I am wrong in above summary...


> > Only yaird problem I know of leading to problems booting (and
> > brick'ing if using bootloaders with no alternative) is with drivers
> > that does not support sysfs. We do not have an overview of the size
> > of that problem.
> 
> Correct. It would be very nice if that could be investigated.
> IMHO it should not be too hard to write a yaird module that checks if 
> modules that don't support sysfs are loaded on the running system
> and, if that is the case, adds them to the initrd.

I think it's more tricky than that: How to recognize if such module is
relevant for mounting the rootfs or not (without a static list)?

But I am not the expert - Erik is, so please file a bugreport and let
him and other clever people judge sanity of such improvement :-)


 - Jonas

- -- 
* Jonas Smedegaard - idealist og Internet-arkitekt
* Tlf.: +45 40843136  Website: http://dr.jones.dk/

 - Enden er nær: http://www.shibumi.org/eoti.htm
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (GNU/Linux)

iD8DBQFDiwBFn7DbMsAkQLgRAmt4AJ4xeaPlkgxHTXMkg3m8SsRThjxJUQCcCPrD
G5Mct5MPWRrJVzJnftgOCu4=
=AT/A
-----END PGP SIGNATURE-----



Reply to: