Hou Tao wrote on Sat, Mar 02, 2019: > I can not find the patch in your tree [1], so will the patch be landed in v5.1 ? > Or are you expecting a v3 patch ? Hmm right, sorry - I just didn't have much time for this and hoped I could find time to sort the discussion we had on the second patch to take both at once, but this definitely won't happen for 5.1 so I should take the first patch for now. Will do that today after basic tests have finished. I should have taken it into -next earlier in principle but it can still make it in 5.1, thanks for the reminder. As for i_blocks, after some more thought and discussing off-list with Jeff (Layton), I'm more or less decided that the best would be to make this up straight from the size whenever size is updated. "blocks" on a network FS does not really make sense to me but I definitely want to avoid the case we have with i_blocks being 0 immediately after writing to a new file. After your first patch I think the simplest thing to do would be to just update i_blocks together with i_size in v9fs_i_size_write(). That's not a new bug however so I don't want to rush this in for 5.1 this late in the cycle so will take a more conservative approach on this one, feel free to send a patch otherwise I will do once the next cycle starts. -- Dominique
