[PATCH] Atomic lock

HIBINO Kei (ex8k-hbn@asahi-net.or.jp)
Mon, 08 Jan 2001 02:46:51 +0900


I think atomic lock is needed for SMP code.

diff -ur linux-2.2.19pre6/drivers/block/ll_rw_blk.c linux-2.2.19pre6+lockfix/drivers/block/ll_rw_blk.c
--- linux-2.2.19pre6/drivers/block/ll_rw_blk.c Mon Dec 11 09:49:41 2000
+++ linux-2.2.19pre6+lockfix/drivers/block/ll_rw_blk.c Mon Jan 8 02:40:37 2001
@@ -578,11 +578,9 @@
count = bh->b_size >> 9;
sector = bh->b_rsector;

- /* Uhhuh.. Nasty dead-lock possible here.. */
- if (buffer_locked(bh))
+ if(test_and_set_bit(BH_Lock, &bh->b_state)) { /* Does atomic lock resolv problems? */
return;
- /* Maybe the above fixes it, and maybe it doesn't boot. Life is interesting */
- lock_buffer(bh);
+ }

if (blk_size[major]) {
unsigned long maxsector = (blk_size[major][MINOR(bh->b_rdev)] << 1) + 1;
diff -ur linux-2.2.19pre6/drivers/block/md.c linux-2.2.19pre6+lockfix/drivers/block/md.c
--- linux-2.2.19pre6/drivers/block/md.c Tue Sep 5 02:39:16 2000
+++ linux-2.2.19pre6+lockfix/drivers/block/md.c Mon Jan 8 02:41:19 2001
@@ -778,9 +778,9 @@
int md_make_request (int minor, int rw, struct buffer_head * bh)
{
if (md_dev [minor].pers->make_request) {
- if (buffer_locked(bh))
+ if(test_and_set_bit(BH_Lock, &bh->b_state)) { /* atomic lock */
return 0;
- set_bit(BH_Lock, &bh->b_state);
+ }
if (rw == WRITE || rw == WRITEA) {
if (!buffer_dirty(bh)) {
bh->b_end_io(bh, test_bit(BH_Uptodate, &bh->b_state));
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/