All of lore.kernel.org
 help / color / mirror / Atom feed
* Couple more fixes for Linux raid metadata 1.x support
@ 2010-07-24 10:22 Doug Nazar
  2010-07-26 17:20 ` Lennart Sorensen
  0 siblings, 1 reply; 3+ messages in thread
From: Doug Nazar @ 2010-07-24 10:22 UTC (permalink / raw)
  To: grub-devel, Colin Watson, Felix Zielcke, Peter Henn

[-- Attachment #1: Type: text/plain, Size: 442 bytes --]

  While doing some in-depth testing came across a few more issues. The 
second one really threw me for a loop. The LVM wouldn't come up because 
it couldn't find some of the raid arrays but when I hexdumped the raid 
arrays it showed the correct uuid. Applies on top of previous patch.

- Ignore spare and faulty drives
- Set a dynamic array->number. It's used for the disk->id and is important
   for the cache subsystem that it be unique.


[-- Attachment #2: grub-update-raid-metadata-1x-support-2.diff --]
[-- Type: text/plain, Size: 2931 bytes --]

=== modified file 'disk/mdraid_linux.c'
--- disk/mdraid_linux.c	2010-07-23 05:25:00 +0000
+++ disk/mdraid_linux.c	2010-07-24 05:53:53 +0000
@@ -363,6 +363,19 @@
     return grub_error (GRUB_ERR_OUT_OF_RANGE, "csum invalid");
   }
 
+  if (grub_le_to_cpu32 (real_sb->dev_number) <
+      grub_le_to_cpu32 (real_sb->max_dev))
+    array->index = grub_le_to_cpu16
+      (real_sb->dev_roles[grub_le_to_cpu32 (real_sb->dev_number)]);
+  else
+    array->index = 0xffff;  /* disk will be later not used! */
+    
+  if (array->index == 0xffff || array->index == 0xfffe)
+  {
+    grub_free (real_sb);
+    return grub_error (GRUB_ERR_OUT_OF_RANGE, "spare or faulty drive");
+  }
+    
   array->name = grub_strdup (real_sb->set_name);
   if (! array->name)
     {
@@ -375,14 +388,11 @@
   array->layout = grub_le_to_cpu32 (real_sb->layout);
   array->total_devs = grub_le_to_cpu32 (real_sb->raid_disks);
   array->disk_size = grub_le_to_cpu64 (real_sb->size);
+  if (!array->disk_size)
+    /* Level 0 doesn't seem to set sb->size */
+    array->disk_size = grub_le_to_cpu64 (real_sb->data_size);
   array->chunk_size = grub_le_to_cpu32 (real_sb->chunksize);
   array->freshness = grub_le_to_cpu32(real_sb->events);
-  if (grub_le_to_cpu32 (real_sb->dev_number) <
-      grub_le_to_cpu32 (real_sb->max_dev))
-    array->index = grub_le_to_cpu16
-      (real_sb->dev_roles[grub_le_to_cpu32 (real_sb->dev_number)]);
-  else
-    array->index = 0xffff;  /* disk will be later not used! */
   array->uuid_len = 16;
   array->uuid = grub_malloc (16);
   if (!array->uuid)

=== modified file 'disk/raid.c'
--- disk/raid.c	2010-07-23 05:25:00 +0000
+++ disk/raid.c	2010-07-24 09:57:05 +0000
@@ -555,25 +555,28 @@
       grub_memset (&array->device, 0, sizeof (array->device));
       grub_memset (&array->start_sector, 0, sizeof (array->start_sector));
 
+      /* mdraid 1.x superblocks don't have a number but we need a unique id
+         for the cache system. Start numbering down from the top. */
       if (array->name)
-	goto skip_duplicate_check;
+	array->number = -1;
+	
       /* Check whether we don't have multiple arrays with the same number.  */
       for (p = array_list; p != NULL; p = p->next)
         {
-          if (! p->name && p->number == array->number) 
+          if (p->number == array->number) 
 	    break;
         }
 
       if (p)
         {
           /* The number is already in use, so we need to find a new one.  */
-          int i = 0;
+	  int i = array->name ? -2 : 0;
 
 	  while (1)
 	    {
 	      for (p = array_list; p != NULL; p = p->next)
 		{
-		  if (! p->name && p->number == i)
+		  if (p->number == i)
 		    break;
 		}
 
@@ -584,10 +587,10 @@
 		  break;
 		}
 
-	      i++;
+	      i += array->name ? -1 : 1;
 	    }
 	}
-    skip_duplicate_check:
+	
       /* mdraid 1.x superblocks have only a name stored not a number.
 	 Use it directly as GRUB device.  */
       if (! array->name)


^ permalink raw reply	[flat|nested] 3+ messages in thread
end of thread, other threads:[~2010-07-26 21:07 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-07-24 10:22 Couple more fixes for Linux raid metadata 1.x support Doug Nazar
2010-07-26 17:20 ` Lennart Sorensen
2010-07-26 21:07   ` Big Endian fix patch (was: Re: Couple more fixes for Linux raid metadata 1.x support) Lennart Sorensen

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.