Subject: Re: raidframe: parity dirty with read-only file systems?
To: None <firstname.lastname@example.org>
From: Greg Oster <email@example.com>
Date: 07/13/2006 12:58:32
"Martin J. Laubach" writes:
> | that might live on top of a RAID set, RAIDframe is stuck in thinking
> | that whoever opened a partition might want to write to it... and
> | thus it gets marked "dirty" as soon as it's opened, and doesn't get
> | marked "clean" until the last open partition is closed.
> How about marking it dirty only as soon as the first write occurs?
Something like that might work... it'd delay the first write a bit,
but probably would be implemented with just a few lines of code
in raidstrategy()... (hm... actually, I see now that
rf_markalldirty() is called in a few other places too, not just on
the first open... I need to think about why that might be, and
whether or not those other calls are "extra" and can be removed... )