On Mon, Feb 7, 2011 at 7:41 PM, Ted Ts'o <tytso@xxxxxxx> wrote: > On Sun, Feb 06, 2011 at 10:43:58AM +0200, Amir Goldstein wrote: >> When looking at alloc_sem, I realized that it is only needed to avoid >> race with adjacent group buddy initialization. > > Actually, alloc_sem is used to protect all of the block group specific > data structures; the buddy bitmap counters, adjusting the buddy bitmap > itself, the largest free order in a block group, etc. So even in the > case where block_size == page_size, alloc_sem is still needed! > This was my assumption by the name of the lock itself, but when I searched where down_write(alloc_sem) is called, I saw that it is called by 3 functions: mb_init_group - for lazy init of buddy cache, once per mount per blockgroup, before any alloc routines can access that blockgroup. add_groupblocks - for online resize, even before lazy init of buddy cache. init_inode_table - for lazy init of inode table, once per mkfs, per blockgroup. it seems to me like down_write(alloc_sem) may be taken in mb_init_group() *only* in case EXT4_MB_GRP_NEED_INIT(grp) and that all down_read(alloc_sem) calls in mballoc.c serve no purpose when block_size == page_size. I concluded that from the comment in mb_load_buddy(): /* Take the read lock on the group alloc * sem. This would make sure a parallel * ext4_mb_init_group happening on other * groups mapped by the page is blocked * till we are done with allocation */ it says alloc_sem protects against lazy init of adjacent groups and says nothing about protecting block group specific data structures... what am I missing??? what am I missing here??? -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
