Re: [PATCH v2] 9p: fix i_size update race in getattr with writeback caching

[David Howells <dhowells@redhat.com>]

Pierre Barre <pierre@barre.sh> wrote:

> With writeback caching (cache=mmap), v9fs_stat2inode() and
> v9fs_stat2inode_dotl() unconditionally overwrite i_size from the server
> response, even when dirty pages may exist locally. This causes processes
> using lseek(SEEK_END) to see incorrect file sizes, leading to data
> corruption when extending files while concurrent stat() calls occur.
> 
> Fix by passing V9FS_STAT2INODE_KEEP_ISIZE when CACHE_WRITEBACK is
> enabled to preserve the client's authoritative i_size.

I think this might be the wrong approach.

A better way is probably, in v9fs_stat2inode() and v9fs_stat2inode_dotl(), if
the inode isn't new, compare the values for stat->mtime to inode->i_mtime and
stat->length to v9inode->netfs.remote_i_size and if they differ mark the inode
as being remotely modified, invalidate the pagecache and reset inode->i_size.

If stat->mtime == inode->i_mtime and stat->length ==
v9inode->netfs.remote_i_size, then don't alter inode->i_size.

David