ata: core: fix irq description on AHCI single irq systems

On my machine with single irq AHCI just the PCI id is printed as
description in /proc/interrupts.
I found a related discussion from beginning of this year:
http://www.gossamer-threads.com/lists/linux/kernel/2117335

Seems like 4f37b504 ("libata: Use dev_name() for request_irq() to
distinguish devices") tried to fix displaying a proper interrupt
description for one scenario but broke it for another one.

The mentioned discussion ended in the current situation being
considered as broken but w/o a patch to fix it.

The following patch is based on a proposal in this mail thread.
Now the interrupt is properly described as:
PCI-MSI 512000-edge      ahci[0000:00:1f.2]

By combining both values also the scenario that commit 4f37b504
("libata: Use dev_name() for request_irq() to distinguish devices")
refers to should still be fine. There it should look like this now:
ahci[20100000.ide]

Using managed memory allocation ensures that the irq description
lives at least as long as the interrupt.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>

drivers/ata/libata-core.c

@@ -6223,6 +6223,7 @@ int ata_host_activate(struct ata_host *host, int irq,
 		      struct scsi_host_template *sht)
 {
 	int i, rc;
+	char *irq_desc;
 
 	rc = ata_host_start(host);
 	if (rc)
@@ -6234,8 +6235,14 @@ int ata_host_activate(struct ata_host *host, int irq,
 		return ata_host_register(host, sht);
 	}
 
+	irq_desc = devm_kasprintf(host->dev, GFP_KERNEL, "%s[%s]",
+				  dev_driver_string(host->dev),
+				  dev_name(host->dev));
+	if (!irq_desc)
+		return -ENOMEM;
+
 	rc = devm_request_irq(host->dev, irq, irq_handler, irq_flags,
-			      dev_name(host->dev), host);
+			      irq_desc, host);
 	if (rc)
 		return rc;