The last version had a problem with xfstests generic/130, generic/169, generic/246, generic/309, and generic/393. Other code, namely orangefs_setattr_size (truncate), looks at i_size without knowledge that it was never set. This doesn't make decisions whether to actually do the truncate based on the possibly not set (or stale!) i_size. I'm still not entirely happy with it though. I feel like we may be missing other code which looks at i_size. Martin Brandenburg (2): orangefs: implement statx orangefs: do not check possibly stale size on truncate fs/orangefs/file.c | 6 +++-- fs/orangefs/inode.c | 18 ++++++++----- fs/orangefs/namei.c | 3 +++ fs/orangefs/orangefs-kernel.h | 4 ++- fs/orangefs/orangefs-utils.c | 63 +++++++++++++++++++++++++++++-------------- 5 files changed, 64 insertions(+), 30 deletions(-) -- 2.1.4
