[PATCH] sunrpc: on successful gss error pipe write, don't return error (try #3)

[PATCH] sunrpc: on successful gss error pipe write, don't return error (try #3)

[Jeff Layton]

When handling the gssd downcall, the kernel should distinguish between a
successful downcall that contains an error code and a failed downcall
(i.e. where the parsing failed or some other sort of problem occurred).

In the former case, gss_pipe_downcall should be returning the number of
bytes written to the pipe instead of an error. In the event of other
errors, we generally want the initiating task to retry the upcall so
we set msg.errno to -EAGAIN. An unexpected error code here is a bug
however, so BUG() in that case.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
---
 net/sunrpc/auth_gss/auth_gss.c |   17 ++++++++++++++++-
 1 files changed, 16 insertions(+), 1 deletions(-)

diff --git a/net/sunrpc/auth_gss/auth_gss.c b/net/sunrpc/auth_gss/auth_gss.c
index 3c3c50f..66cb89c 100644
--- a/net/sunrpc/auth_gss/auth_gss.c
+++ b/net/sunrpc/auth_gss/auth_gss.c
@@ -644,7 +644,22 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	p = gss_fill_context(p, end, ctx, gss_msg->auth->mech);
 	if (IS_ERR(p)) {
 		err = PTR_ERR(p);
-		gss_msg->msg.errno = (err == -EAGAIN) ? -EAGAIN : -EACCES;
+		switch (err) {
+		case -EACCES:
+			gss_msg->msg.errno = err;
+			err = mlen;
+			break;
+		case -EFAULT:
+		case -ENOMEM:
+		case -EINVAL:
+		case -ENOSYS:
+			gss_msg->msg.errno = -EAGAIN;
+			break;
+		default:
+			printk(KERN_CRIT "%s: bad return from "
+				"gss_fill_context: %ld\n", __func__, err);
+			BUG();
+		}
 		goto err_release_msg;
 	}
 	gss_msg->ctx = gss_get_ctx(ctx);
-- 
1.6.5.2

_______________________________________________
NFSv4 mailing list
NFSv4@linux-nfs.org
http://linux-nfs.org/cgi-bin/mailman/listinfo/nfsv4

Re: [PATCH] sunrpc: on successful gss error pipe write, don't return error (try #3)

[Trond Myklebust]

On Fri, 2009-12-18 at 16:23 -0500, Jeff Layton wrote: 
> When handling the gssd downcall, the kernel should distinguish between a
> successful downcall that contains an error code and a failed downcall
> (i.e. where the parsing failed or some other sort of problem occurred).
> 
> In the former case, gss_pipe_downcall should be returning the number of
> bytes written to the pipe instead of an error. In the event of other
> errors, we generally want the initiating task to retry the upcall so
> we set msg.errno to -EAGAIN. An unexpected error code here is a bug
> however, so BUG() in that case.

That looks good. I'll apply it after the two other fixes.

Trond