Re: [uml-devel] Partitions with 2.5.38-1

[Mike A. <an...@us...>]

James or Jeff,
	Is there a work around or patch for this yet. I assume I am
	hitting the same problem. I am seeing that uml / ubd does not
	do a set_capacity before add_disk which I believe was indicated
	in a previous mail. I tried some quick hacks, but it did not
	improve my situation much.

-andmike

James W McMechan [mcm...@ju...] wrote:
> I have been looking at partitions in 2.5.38-1 :)
> The ubd_add function no longer calls ubd_revalidate
> The common layers don't seem to ever call ubd_open and never try to read
> the
> partition table by opening the device it will not try to open it even
> when
> the 0x6201 root device is directly specified.
> 
> I don't know what Al Viro said is the correct method but it is not
> working
> yet :(
> 
> This may be a problem specific to ubd as the ubd driver really does want
> the
> device opened before any attempts to use the block queue for read or
> write
> but other kinds of drivers will allow for unopened access to the device.
> 
> That is why the old ubd_revalidate had that somewhat silly looking
> open/close pair. so that the block queue would not fail due to a unopened
> device. As I recall open/close mostly applied to mount/unmount and the
> partition read happens before that.-andmike
--
Michael Anderson
an...@us...