Skip to contentSkip to conversation

: Expand Xen blkfront for > 16 xvd

PATCH In review Chris Lalancette 2 participants 3 messages Aug 14, 2008 – Aug 19, 2008

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 : Expand Xen blkfront for > 16 xvd

Jeremy,
Until recently, the maximum number of xvd block devices you could attach to
a Xen domU was 16. This limitation turned out to be problematic for some users, so it was expanded to handle a much larger number of disks. However, this requires a couple of changes in the way that blkfront scans for disks. This functionality is already present in the Xen linux-2.6.18-xen.hg tree; the attached patch adds this functionality to the mainline xen-blkfront implementation. I successfully tested it on a 2.6.25 tree. I build tested it on 2.6.27-rc3, but couldn't get that tree to boot due to some other bug.

drivers/block/xen-blkfront.c +55 -21
index 3ca643c..bff602c 100644
@@ -105,15 +105,17 @@ static DEFINE_SPINLOCK(blkif_io_lock);
 #define GRANT_INVALID_REF	0
 
 #define PARTS_PER_DISK		16
+#define PARTS_PER_EXT_DISK      256
 
 #define BLKIF_MAJOR(dev) ((dev)>>8)
 #define BLKIF_MINOR(dev) ((dev) & 0xff)
 
-#define DEV_NAME	"xvd"	/* name in /dev */
+#define EXT_SHIFT 28
+#define EXTENDED (1<<EXT_SHIFT)
+#define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
+#define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
 
-/* Information about our VBDs. */
-#define MAX_VBDS 64
-static LIST_HEAD(vbds_list);
+#define DEV_NAME	"xvd"	/* name in /dev */
 
 static int get_id_from_freelist(struct blkfront_info *info)
 {
@@ -386,31 +388,60 @@ static int xlvbd_barrier(struct blkfront_info *info)
 }
 
 
-static int xlvbd_alloc_gendisk(int minor, blkif_sector_t capacity,
-			       int vdevice, u16 vdisk_info, u16 sector_size,
-			       struct blkfront_info *info)
+static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
+			       struct blkfront_info *info,
+			       u16 vdisk_info, u16 sector_size)
 {
 	struct gendisk *gd;
 	int nr_minors = 1;
 	int err = -ENODEV;
+	unsigned int offset;
+	int minor;
+	int nr_parts;
 
 	BUG_ON(info->gd != NULL);
 	BUG_ON(info->rq != NULL);
 
-	if ((minor % PARTS_PER_DISK) == 0)
-		nr_minors = PARTS_PER_DISK;
+	if ((info->vdevice>>EXT_SHIFT) > 1) {
+		/* this is above the extended range; something is wrong */
+		printk(KERN_WARNING "blkfront: vdevice 0x%x is above the extended range; ignoring\n", info->vdevice);
+		return -ENODEV;
+	}
+
+	if (!VDEV_IS_EXTENDED(info->vdevice)) {
+		minor = BLKIF_MINOR(info->vdevice);
+		nr_parts = PARTS_PER_DISK;
+	} else {
+		minor = BLKIF_MINOR_EXT(info->vdevice);
+		nr_parts = PARTS_PER_EXT_DISK;
+	}
+
+	if ((minor % nr_parts) == 0)
+		nr_minors = nr_parts;
 
 	gd = alloc_disk(nr_minors);
 	if (gd == NULL)
 		goto out;
 
-	if (nr_minors > 1)
-		sprintf(gd->disk_name, "%s%c", DEV_NAME,
-			'a' + minor / PARTS_PER_DISK);
-	else
-		sprintf(gd->disk_name, "%s%c%d", DEV_NAME,
-			'a' + minor / PARTS_PER_DISK,
-			minor % PARTS_PER_DISK);
+	offset = minor / nr_parts;
+
+	if (nr_minors > 1) {
+		if (offset < 26)
+			sprintf(gd->disk_name, "%s%c", DEV_NAME, 'a' + offset);
+		else
+			sprintf(gd->disk_name, "%s%c%c", DEV_NAME,
+				'a' + ((offset / 26)-1), 'a' + (offset % 26));
+	} else {
+		if (offset < 26)
+			sprintf(gd->disk_name, "%s%c%d", DEV_NAME,
+				'a' + offset,
+				minor & (nr_parts - 1));
+		else
+			sprintf(gd->disk_name, "%s%c%c%d", DEV_NAME,
+				'a' + ((offset / 26) - 1),
+				'a' + (offset % 26),
+				minor & (nr_parts - 1));
+	}
 
 	gd->major = XENVBD_MAJOR;
 	gd->first_minor = minor;
@@ -699,8 +730,13 @@ static int blkfront_probe(struct xenbus_device *dev,
 	err = xenbus_scanf(XBT_NIL, dev->nodename,
 			   "virtual-device", "%i", &vdevice);
 	if (err != 1) {
-		xenbus_dev_fatal(dev, err, "reading virtual-device");
-		return err;
+		/* go looking in the extended area instead */
+		err = xenbus_scanf(XBT_NIL, dev->nodename, "virtual-device-ext",
+				   "%i", &vdevice);
+		if (err != 1) {
+			xenbus_dev_fatal(dev, err, "reading virtual-device");
+			return err;
+		}
 	}
 
 	info = kzalloc(sizeof(*info), GFP_KERNEL);
@@ -861,9 +897,7 @@ static void blkfront_connect(struct blkfront_info *info)
 	if (err)
 		info->feature_barrier = 0;
 
-	err = xlvbd_alloc_gendisk(BLKIF_MINOR(info->vdevice),
-				  sectors, info->vdevice,
-				  binfo, sector_size, info);
+	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size);
 	if (err) {
 		xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
 				 info->xbdev->otherend);

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#1 lore raw
Quoted 8 lines · Chris Lalancette

Jeremy,
Until recently, the maximum number of xvd block devices you could attach to
a Xen domU was 16. This limitation turned out to be problematic for some users, so it was expanded to handle a much larger number of disks. However, this requires a couple of changes in the way that blkfront scans for disks. This functionality is already present in the Xen linux-2.6.18-xen.hg tree; the attached patch adds this functionality to the mainline xen-blkfront implementation.

I haven't tested this yet. You have tested it OK with some pvops kernel? If so, send it to Jens Axboe with my ack.

Quoted 2 lines

I successfully tested it on a 2.6.25 tree. I build tested it
on 2.6.27-rc3, but couldn't get that tree to boot due to some other bug.

What other bug?

J

#2 lore raw
Quoted 10 lines · Jeremy Fitzhardinge
Quoted 8 lines · Chris Lalancette

Jeremy,
Until recently, the maximum number of xvd block devices you could attach to
a Xen domU was 16. This limitation turned out to be problematic for some users, so it was expanded to handle a much larger number of disks. However, this requires a couple of changes in the way that blkfront scans for disks. This functionality is already present in the Xen linux-2.6.18-xen.hg tree; the attached patch adds this functionality to the mainline xen-blkfront implementation.

Quoted 2 lines

I haven't tested this yet. You have tested it OK with some pvops kernel? If so, send it to Jens Axboe with my ack.

Yes, I tested it with a 2.6.25-something Fedora 9 pv-ops kernel (i386). I'll send it along to Jens.

Quoted 5 lines
Quoted 3 lines

I successfully tested it on a 2.6.25 tree. I build tested it
on 2.6.27-rc3, but couldn't get that tree to boot due to some other bug.

What other bug?

This is http://bugzilla.redhat.com/show_bug.cgi?id=459067, that I think you've already looked at. Basically any F-10 or upstream git kernel is crashing on an i386 RHEL-5 HV. We were a little confused by your comment in that bug, however; we were under the impression that the fix you mentioned was specifically a 32-on-64 fix, not for 32-on-32. If we were wrong, please point it out.

Thanks,
Chris Lalancette