On Mon, 16 Feb 2009, Dmitry Nezhevenko wrote: > On Mon, Feb 16, 2009 at 12:56:53PM +0100, maximilian attems wrote: > > an experienced user using rootfstype may aswell know about rootdelay. > > if not it is easy to look up in man initramfs-tools. > > > > Thats looks wrong. rootdelay and rootfstype are different options.
sure they are in principle, but it is wise to check if you see an fs on the corresponding device before trying to mount it as devices may appear *before* they can be really accessed. > This bug report was opened because rootfstype is non-working for > filesystems that are not supported yet by udev. > > Yes. This may be workarounded with rootdelay=15 or something like this. > But this is workaround of existing bug. > > If rootfstype was not designed to specify custom filesystem type, why it > exists at all? Any use case example? Why it even tries to check/detect > filesystem type if correct one was provided by experienced user? And any > reason why initramfs should fail and wait for "rootdelay" timeout if root > device exists and filesystem type was specified at boot time? > > For example, rootdelay may be set to some big value to avoid booting > troubles for slow usb devices that needs some time to appear. But why I > should wait 5-10 seconds on machine where USB works fast enough and > initramfs already polls for root device every 0.1 second? Wasn't such poll > introduced to reduce boot time, avoid waiting and boot ASAP? guy come on. you are using an bleeding edge fs, it isn't that hard to pass 2 interconnected boot commands at once. if you insist i may add a notice to man initramfs-tools, but that is the most that i fell that is appropriate. kind regards -- maks -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org