On Wed, Sep 19, 2012 at 02:41:14PM -0400, Theodore Ts'o wrote: > On Wed, Aug 22, 2012 at 02:05:39PM +0800, Zheng Liu wrote: > > + * Extents status encompass delayed extents and extent locks > > I've looked over this patch set, and as near as I can tell, we aren't > (yet) using the extents status structure to do extent-level locking. > I could imagine using that in the future so we aren't using i_data_sem > to lock out the entire tree, so block allocations could happen in > parallel, but that's not here yet. > > Is there something I'm missing, or was something else meant by "extent > locks" here? Hi Ted, In my plan, the first step for extent status tree is only to record delay extent in memory. It can simplify the implementation of fiemap and bigalloc, and introduce lseek SEEK_DATA/SEEK_HOLE support. That is all I want to finish in first step. I am very glad to see that these patches has been applied into dev branch even though it still has some problems. I will look at these problem ASAP. Indeed, in this patch set, extent-level locking is still not used because that is my next step in my plan [1]. I will try to use it after these patches are fully tested and applied. Thanks. 1. http://www.spinics.net/lists/linux-fsdevel/msg56291.html Regards, Zheng -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
