On 2011年07月31日 15:08, Joel Becker Wrote: > On Sat, Jul 30, 2011 at 03:25:32PM +0800, Coly Li wrote: >> On 2011年07月29日 21:19, Joel Becker Wrote: >>> On Fri, Jul 29, 2011 at 03:48:45AM -0600, Andreas Dilger wrote: >>>> On 2011-07-28, at 4:07 PM, Joel Becker wrote: >>>>> We use ethernet crc32 in ocfs2. btrfs uses crc32c. Frankly, I >>>>> could have used crc32c if I'd really thought about the hardware >>>>> acceleration benefits. I think it's a good idea for ext4. >>>> >>>> The problem with crc32[c] is that if you don't have hardware acceleration >>>> it is terribly slow. >>> >>> We find ethernet crc32 just fine in ocfs2. I use the kernel's >>> implementation, which survives everyone's network traffic, and of course >>> we added the triggers to jbd2 so we only have to do the calculations on >>> read and write. >>> >> >> Ext4 supports non-journal mode, and there are a few users (Google, Taobao, etc.). >> A trigger of jbd2 may not work well for non-journal Ext4 ... >> >> And in non-journal mode, there is not copy of any meta data block in jbd2, we need to be >> more careful in check summing, e.g. inode/block bitmap blocks... > > Sure, but you could use a trigger in journaled mode and then do > the checksums directly in the __ext4_handle_journal_dirty_*() functions > in non-journaled mode. Sure, it would be a little more CPU time, but > the user picked "checksums + no journal" at mkfs time. > Yes, my idea was similar to you. One thing not clear to me is, in non-journal mode, how to make the page of bitmap block being stable. Because bits setting in Ext4 bitmap is non-locking, it might be possible that new bit setting after check sum is calculated. Coly -- 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
