From: Prarit Bhargava <prarit@redhat.com>
Date: Wed, 20 Jan 2010 15:31:36 -0500
Subject: [acpi] fix NULL pointer panic in acpi_run_os
Message-id: <20100120152331.8192.35093.sendpatchset@prarit.bos.redhat.com>
Patchwork-id: 22678
O-Subject: [RHEL5]: ACPI: Fix NULL pointer panic in acpi_run_osc()
Bugzilla: 547733
RH-Acked-by: Don Dutile <ddutile@redhat.com>
RH-Acked-by: Andrew Jones <drjones@redhat.com>
RH-Acked-by: Dean Nelson <dnelson@redhat.com>

Backport of f8123381bae8d581d81f24b55719db3d9a6a262c

Test for a NULL length _OSC object.

On the 206m series IBM machines, 5 _OSC objects are declared but there are
only 4.  This leads to a panic when looking at the last object.

We do not have the IBM HW to reproduce this issue, so this was only compile
tested.  In the BZ, IBM says that this patch does resolve the NULL pointer
panic.

Successfully compiled by me.

Resolves BZ 547733.

Signed-off-by: Jarod Wilson <jarod@redhat.com>

diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
index 0186070..334e7c8 100644
--- a/drivers/pci/pci-acpi.c
+++ b/drivers/pci/pci-acpi.c
@@ -83,6 +83,9 @@ static acpi_status acpi_run_osc(acpi_handle handle,
 	if (ACPI_FAILURE(status))
 		return status;
 
+	if (!output.length)
+		return AE_NULL_OBJECT;
+
 	out_obj = output.pointer;
 	if (out_obj->type != ACPI_TYPE_BUFFER) {
 		printk(KERN_DEBUG "Evaluate _OSC returns wrong type\n");