Re: Re: fr1-2.16 patch for 2.6.10 kernel

"Also sprach denis bonnenfant:"
> >     if (page) memset(page, 0, PAGE_SIZE);
> > 
> > The alloc is not guaranteed.

I haven't chased that down, but you might want to turn the bitmap off
after resync by adding the following to close_sync() in raid1.c, which
seems t be run at end of resync.

Untested (I'm not sure if it is right - what if the sync is aborted?
Will this fn still be run?).


 static void close_sync(conf_t *conf)
 {
         spin_lock_irq(&conf->resync_lock);
         wait_event_lock_irq(conf->wait_resume, !conf->barrier, conf->resync_lock);
         spin_unlock_irq(&conf->resync_lock);
 
         if (conf->barrier) BUG();
         if (waitqueue_active(&conf->wait_idle)) BUG();
 
         mempool_destroy(conf->r1buf_pool);
         conf->r1buf_pool = NULL;
 
+#ifdef CONFIG_MD_FR1
+        if (conf->bitmap) {
+                struct bitmap *bitmap = conf->bitmap;
+                bitmap->stop(bitmap); // flag bitmap inactive
+        }
+#endif /* CONFIG_MD_FR1 */
 }

I don't know if that's dangerous.  It doesn't look so. It clears a flag
in the bitmap and that's all. I can see why I didn't bother.

To be braver, one could flush the bitmap pages by running

                 remove_bitmap(conf);
                 create_bitmap(conf);

which readies it again. That will change conf->bitmap, so nothing had
better be holding that value.

After create, a start call is still required in order to turn it on.
That's issued by the fault code. The start call sets the event count.

Peter