This one time, at band camp, Gabor HALASZ said:
> Stephen Gran wrote:
> >
> >Did you see the first paragraph of "Some problems with udev and module
> >loading", where it talks about setting ROOTFS and adding a second
> >symlink?  That is the part that appears to apply to you.  There is also
> >a discussion of why this has to be the default (reported file system
> >corruption).
> >
> 
> I think, the data corruption caused by hw or kernel bug.

You're welcome to think it.  I have decided that it is worse to have data
or filesystem corruption than to have a slow drive until you do a little
extra work (in this case, setting up a symlink and editing a file that has
to be edited anyway).  Unless you're prepared to assure me that this bug
in some other thing will never again be triggered by someone innocently
running an init script at boot, I will continue to think the way I do.

> Your sollution is only a workaround for a nasty thing, and violates the 
> standard boot methods which works well for a long-long time. This is a 
> wrong way and if somebody have a problem with dma or other hw failures, 
> must use an alternative configurations.

No, we already have one script mount root, and another mount all the other
file systems.  My approach mirrors that - run the hdparm script once for
the root file system, wait for modules to be loaded, and run it again.
This is not some strange new thing here.

If you have a solution that:
a) Doesn't cause data corruption by running on a drive that is mounted
b) Works with udev/hotplug setups, but also without them
c) also only runs after modules are loaded (but doesn't run a second
   time on the root drive, since it has been set up already)

Point me to it.  I'm happy to consider patches or new ideas.  I've about
bent this script as far as it will bend, and I'd be happy to go for
something simpler, but I can't seem to think of a new way to come at
the problem that solves the 3 problems above in a more elegant way.

By the way, please use whatever equivalent of reply-to-all your mailer
uses to keep the bug report in the email trail.

Thanks,
-- 
 -----------------------------------------------------------------
|   ,''`.                                            Stephen Gran |
|  : :' :                                        [EMAIL PROTECTED] |
|  `. `'                        Debian user, admin, and developer |
|    `-                                     http://www.debian.org |
 -----------------------------------------------------------------

Attachment: signature.asc
Description: Digital signature

Reply via email to