On Thu, May 28, 2020 at 6:47 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
>
> case S_IFREG:
> inode->i_op = &dlmfs_file_inode_operations;
> inode->i_fop = &dlmfs_file_operations;
>
> i_size_write(inode, DLM_LVB_LEN);
> is the only thing that does anything to size of that sucker. IOW, that
> i_size_read() might as well had been an explicit 64.
Heh. Indeed. I did actually grep for i_size_write() use in ocfs2 and
saw several. But I didn't realize to limit it to just the dlmfs part.
So it does that crazy sequence number lock dance on 32-bit just to
read a constant value.
Oh well.
It would be nice to get those follow-up cleanups eventually, but I
guess the general user access cleanups are more important than this
very odd special case silliness.
Linus
