On Wed, 2012-11-21 at 07:20 -0300, Ezequiel Garcia wrote: > Yes, I thought of this. > > For now, I decided to keep the implementation as simple as possible.
WRT atomic LEB caches and barriers - I guess if there are users who would prefer higher I/O speed to the power-cut tolerance, you could: 1. ignore I/O barriers 2. use plain leb_unmap/leb_write instead of atomic leb change, which is faster. There could be a switch, probably an ioctl? But this is not very important now, I am just throwing an idea. -- Best Regards, Artem Bityutskiy
signature.asc
Description: This is a digitally signed message part
