NFSv4: Fix handling of non-atomic change attrbute updates

Browse source

If the change attribute update is declared to be non-atomic by the
server, or our cached value does not match the server's value before the
operation was performed, then we should declare the inode cache invalid.

On the other hand, if the change to the directory raced with a lookup or
getattr which already updated the change attribute, then optimise away
the revalidation.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

...

This commit is contained in:

Trond Myklebust 2021-06-25 15:49:31 -04:00

parent 213bb58475

commit 20cf7d4ea4

1 changed files with 15 additions and 18 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

37

fs/nfs/nfs4proc.c

View file