Re: [PATCH] AFS: Implement shared-writable mmap [try #2]

[Nick Piggin <nickpiggin@yahoo.com.au>]

David Howells wrote:
> Implement shared-writable mmap for AFS.
> 
> The key with which to access the file is obtained from the VMA at the point
> where the PTE is made writable by the page_mkwrite() VMA op and cached in the
> affected page.
> 
> If there's an outstanding write on the page made with a different key, then
> page_mkwrite() will flush it before attaching a record of the new key.
> 
> [try #2] Only flush the page if the page is still part of the mapping (truncate
> may have discarded it).

Couple more issues...

> Signed-off-by: David Howells <dhowells@redhat.com>

> +/*
> + * notification that a previously read-only page is about to become writable
> + * - if it returns an error, the caller will deliver a bus error signal
> + *
> + * we use this to make a record of the key with which the writeback should be
> + * performed and to flush any outstanding writes made with a different key
> + *
> + * the key to be used is attached to the file pinned by the VMA
> + */
> +int afs_page_mkwrite(struct vm_area_struct *vma, struct page *page)
> +{
> +	struct afs_vnode *vnode = AFS_FS_I(vma->vm_file->f_mapping->host);
> +	struct key *key = vma->vm_file->private_data;
> +	int ret;
> +
> +	_enter("{{%x:%u},%x},{%lx}",
> +	       vnode->fid.vid, vnode->fid.vnode, key_serial(key), page->index);
> +
> +	lock_page(page);
> +	if (page->mapping == vma->vm_file->f_mapping)
> +		ret = afs_prepare_write(vma->vm_file, page, 0, 0);

I would strongly suggest you used (0, PAGE_CACHE_SIZE) for the range, and
have your nopage function DTRT.

Minor issue: you can just check for `if (!page->mapping)` for truncation,
which is the usual signal to tell the reader you're checking for truncate.
Then you can remove the comment...


> +	else
> +		ret = 0; /* seems truncate interfered - let the caller deal
> +			  * with it (presumably the PTE changed too) */

Rather than add this (not always correct) comment about the VM workings, I'd
just add a directive in the page_mkwrite API documentation that the filesystem
is to return 0 if the page has been truncated.


> +	unlock_page(page);
> +
> +	_leave(" = %d", ret);
> +	return ret;
> +}

-- 
SUSE Labs, Novell Inc.