lkml.org 
[lkml]   [2003]   [Mar]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
Subject[Patch 2/8] 2.4: Fix for enormous numbers of buffers on BUF_LOCKED
Buffer_head syncing places BUF_DIRTY buffers on the BUF_LOCKED, but
buffers can stay there indefinitely, upsetting buffer accounting. The
symptoms are benign, mostly just nonsensical numbers in alt-sysrq-m
output, but it's pretty cheap to clean these up in kupdated
nonetheless, just by refiling any unlocked buffers which happen to be
at the head of the BUF_LOCKED list. (We stop at the first locked
buffer rather than walking the whole list, avoiding walking any
buffers more than once.)

--- linux-2.4-ext3push/fs/buffer.c.=K0001=.orig 2003-03-25 10:59:15.000000000 +0000
+++ linux-2.4-ext3push/fs/buffer.c 2003-03-25 10:59:15.000000000 +0000
@@ -2958,6 +2958,30 @@ int bdflush(void *startup)
}
}

+
+/*
+ * Do some IO post-processing here!!!
+ */
+void do_io_postprocessing(void)
+{
+ int i;
+ struct buffer_head *bh, *next;
+
+ spin_lock(&lru_list_lock);
+ bh = lru_list[BUF_LOCKED];
+ if (bh) {
+ for (i = nr_buffers_type[BUF_LOCKED]; i-- > 0; bh = next) {
+ next = bh->b_next_free;
+
+ if (!buffer_locked(bh))
+ __refile_buffer(bh);
+ else
+ break;
+ }
+ }
+ spin_unlock(&lru_list_lock);
+}
+
/*
* This is the kernel update daemon. It was used to live in userspace
* but since it's need to run safely we want it unkillable by mistake.
@@ -3009,6 +3033,7 @@ int kupdate(void *startup)
#ifdef DEBUG
printk(KERN_DEBUG "kupdate() activated...\n");
#endif
+ do_io_postprocessing();
sync_old_buffers();
run_task_queue(&tq_disk);
}
-
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/
\
 
 \ /
  Last update: 2005-03-22 13:34    [W:0.094 / U:1.124 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site