Re: procfs vs. kernel threads (was: Re: test1-ac10 and bttv)

From: Alexander Viro (viro@math.psu.edu)
Date: Wed Jun 07 2000 - 02:55:15 EST


On Wed, 7 Jun 2000 almesber@lrc.di.epfl.ch wrote:

> Here's the promised patch. fs->root and fs->rootmnt are NULL when a
> kernel thread is launched before current->fs is properly initialized.
> (E.g. msp3400 when not using modules.) After applying John's patch to
> videodev.c, my patch to i2c-old.c, and this one, I can watch TV
> again ;-)

Wrong fix. Right one is to have a mini-fs (quite possibly only with the
root directory) and set it from the very beginning. Then all this
special-case crap will go away. I'll do that - BTW, it will simplify the
mount_root() big way. Come to think of that, we might put a file there and
do union-mount of real root over the original one. File being something
like /shutdown that would close all descriptors, umount / (really umount,
none of this remount-ro stuff) and call reboot(9). That's a trivial static
binary and we could let init(8) just exec in the end of everything.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Wed Jun 07 2000 - 21:00:27 EST