Re: implementation of shrink_dcache_sb

Bill Hawes (whawes@star.net)
Wed, 17 Sep 1997 08:18:16 -0400


Keith Owens wrote:

> Does this mean that all inodes and the super block for a device are now
> purged when a device is unmounted? Previously the inodes and sb would
> be kept and only purged when the device was remounted *and* a media
> change was flagged.

Hi Keith,
Currently at unmount time shrink_dcache() is called, and the inodes for
the device are invalidated. My patch doesn't change any of this policy;
it just allows shrinking only the dcache for the particular device.

I'm not sure what is done at remount time, so I'll have to look into
that. But I agree with you that it would be safer to invalidate all
inodes if the media might have changed.

Regards,
Bill