On Tue, Nov 29, 2011 at 04:09:56PM -0500, Phillip Susi wrote: > On 11/29/2011 11:02 AM, Petr Uzel wrote: > >I was also thinking about some new BLKPG-like ioctl()s. Could you > >please elaborate more on your new ioctl() proposal? Perhaps we could > >combine the ideas... > > I just added one to update the length of the partition. Attaching the patch > so far ( untested ). > > >Features of the new ioctl()/s I was thinking about are: > >- get largest number of partition known to the kernel > >- atomically add/remove more than one partition > > What benefit would those provide?
1/ largest partition: to stop messing with range, ext_range, ped_disk_get_max_supported_partition_count() etc. if the only thing we need to do is to BLKPG_DEL_PARTITION all partitions known to the kernel 2/ add/remove multiple partitions: I didn't think about this too much yet, but I think that with clever design, this could help with timing issues we saw in disk_sync_part_table.
