On Sun, 2007-09-23 at 13:15 +0800, Yan Zheng wrote: > 2007/9/18, Chris Mason <[EMAIL PROTECTED]>: > > On Tue, 18 Sep 2007 15:18:24 +0800 > > "Yan Zheng" <[EMAIL PROTECTED]> wrote: > > > > > Hello > > > > > > When file size is 1, both the 'start' and 'end' of a extent_state are > > > set to 0 by set_extent_uptodate. > > > > I saw this one too last night, but ended up fixing it a little > > differently. I changed btrfs_get_extent to insert things into the > > extent maps at a block granularity. This is more in line with how the > > rest of the code works. I'm running fsx-linux on it now (an important > > test for these kinds of changes). > >
[ BTW, I pushed out my fix for this over the weekend ] > Hi > > Inline data are not truncated in btrfs_truncate_in_trans, maybe we > need check inode->i_size here to determine the start of data area that > need fill zero. Another option is truncate inline data in > btrfs_truncate_in_trans, I can work on this if you agree. truncating the inline data would be great. -chris _______________________________________________ Btrfs-devel mailing list [email protected] http://oss.oracle.com/mailman/listinfo/btrfs-devel
