On 10/28/20 9:18 AM, harshad shirwadkar wrote:
Actually the simpler fix for this in case of fast commits is to check if the inode is on the fast commit list or not. Since we clear the fast commit list after every fast and / or full commit, it's always true that if the inode is not on the list, that means it isn't dirty. This will simplify the logic here and then we can probably get rid of i_fc_committed_subtid field altogether. I'll test this and send out a patch.
Yes, sounds like a better solution. Thanks! -ritesh