[patch 11/49] relay: fix "full buffer with exactly full last subbuffer" accounting problem

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
From: Greg KH
Date: Monday, August 18, 2008 - 12:19 pm

2.6.25-stable review patch.  If anyone has any objections, please let us know.

------------------

From: Tom Zanussi <tzanussi@gmail.com>

commit 32194450330be327f3b25bf6b66298bd122599e9 upstream

In relay's current read implementation, if the buffer is completely full
but hasn't triggered the buffer-full condition (i.e. the last write
didn't cross the subbuffer boundary) and the last subbuffer is exactly
full, the subbuffer accounting code erroneously finds nothing available.
This patch fixes the problem.

Signed-off-by: Tom Zanussi <tzanussi@gmail.com>
Cc: Eduard - Gabriel Munteanu <eduard.munteanu@linux360.ro>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Mathieu Desnoyers <compudj@krystal.dyndns.org>
Cc: Andrea Righi <righi.andrea@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

---
 kernel/relay.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

--- a/kernel/relay.c
+++ b/kernel/relay.c
@@ -803,6 +803,10 @@ static void relay_file_read_consume(stru
 	size_t n_subbufs = buf->chan->n_subbufs;
 	size_t read_subbuf;
 
+	if (buf->subbufs_produced == buf->subbufs_consumed &&
+	    buf->offset == buf->bytes_consumed)
+		return;
+
 	if (buf->bytes_consumed + bytes_consumed > subbuf_size) {
 		relay_subbufs_consumed(buf->chan, buf->cpu, 1);
 		buf->bytes_consumed = 0;
@@ -834,6 +838,8 @@ static int relay_file_read_avail(struct 
 
 	relay_file_read_consume(buf, read_pos, 0);
 
+	consumed = buf->subbufs_consumed;
+
 	if (unlikely(buf->offset > subbuf_size)) {
 		if (produced == consumed)
 			return 0;
@@ -852,8 +858,12 @@ static int relay_file_read_avail(struct 
 	if (consumed > produced)
 		produced += n_subbufs * subbuf_size;
 
-	if (consumed == produced)
+	if (consumed == produced) {
+		if (buf->offset == subbuf_size &&
+		    buf->subbufs_produced > buf->subbufs_consumed)
+			return 1;
 		return 0;
+	}
 
 	return 1;
 }

-- 
--
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
[patch 00/49] 2.6.25-stable review, Greg KH, (Mon Aug 18, 12:18 pm)
[patch 07/49] SCSI: ses: fix VPD inquiry overrun, Greg KH, (Mon Aug 18, 12:19 pm)
[patch 11/49] relay: fix "full buffer with exactly full la ..., Greg KH, (Mon Aug 18, 12:19 pm)
[patch 12/49] radeonfb: fix accel engine hangs, Greg KH, (Mon Aug 18, 12:19 pm)
[patch 15/49] mlock() fix return values, Greg KH, (Mon Aug 18, 12:19 pm)
[patch 23/49] random32: seeding improvement, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 25/49] sparc64: FUTEX_OP_ANDN fix, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 27/49] uml: fix build when SLOB is enabled, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 30/49] uml: track and make up lost ticks, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 39/49] uml: Fix boot crash, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 40/49] uml: PATH_MAX needs limits.h, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 41/49] radeon: misc corrections, Greg KH, (Mon Aug 18, 12:20 pm)
[patch 46/49] x86: fix spin_is_contended(), Greg KH, (Mon Aug 18, 12:21 pm)