Trond Myklebust wrote:
> Just one comment: Isn't it easier to do this in generic_file_llseek()
> itself using inode->i_op->revalidate()? That would make it work for
> coda and smbfs too...
>
> Cheers,
> Trond
Yes, you are right. I've attached a patch which does the revalidate in
both default_llseek() and generic_file_llseek(). Also, it only happens
if i_size is going to be used. This makes NFS client, smbfs, opengfs, and coda
work right, among others. I copied do_revalidate() from fs/stat.c. It would be
nice if it was in a header file instead.
By the way, we are looking at the challenges of integrating a fully coherent
distributed/cluster filesystem into the Linux filesystem architecture.
--- linux-2.4.16.orig/fs/read_write.c Fri Dec 14 12:06:44 2001
+++ linux-2.4.16/fs/read_write.c Fri Dec 14 12:54:02 2001
@@ -20,6 +20,19 @@
mmap: generic_file_mmap,
};
+/*
+ * Revalidate the inode. This is required for proper NFS attribute caching.
+ * ARG! Copied from fs/stat.c (move to a header file)
+ */
+static __inline__ int
+do_revalidate(struct dentry *dentry)
+{
+ struct inode * inode = dentry->d_inode;
+ if (inode->i_op && inode->i_op->revalidate)
+ return inode->i_op->revalidate(dentry);
+ return 0;
+}
+
ssize_t generic_read_dir(struct file *filp, char *buf, size_t siz, loff_t *ppos)
{
return -EISDIR;
@@ -31,6 +44,8 @@
switch (origin) {
case 2:
+ if ((retval = do_revalidate(file->f_dentry)) < 0)
+ return retval;
offset += file->f_dentry->d_inode->i_size;
break;
case 1:
@@ -59,6 +74,8 @@
switch (origin) {
case 2:
+ if ((retval = do_revalidate(file->f_dentry)) < 0)
+ return retval;
offset += file->f_dentry->d_inode->i_size;
break;
case 1:
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Dec 15 2001 - 21:00:29 EST