linux-kernel - [RFC patch] return early if all dmar HW unit ignored

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 31 Jul 2009 16:29:01 +0800
From:	Luming Yu <luming.yu@...il.com>
To:	LKML <linux-kernel@...r.kernel.org>
Subject: [RFC patch] return early if all dmar HW unit ignored

Hello list,

When debugging an IOMMU problem, I noticed it should be much more safe
to return early than late if all remapping HW unit are ignored.
To figure out what device causes the iommu problem on my linux box:
hp-compaq dc7800, I tried to mark all dmar HW
unit ignored,but still got a lot of unxepcted dmar_fault. So I think
the proposed patch make sense. The next step is to
add a boot option to make dmar HW units selectable enable/disable individually.

Please review.

**The patch is enclosed in text attachment*
**Using web client to send the patch* *
**below is for review, please apply attached  patch*/

Thanks,
Luming


Signed-off-by: Yu Luming <luming.yu@...el.com>

 intel-iommu.c |   26 +++++++++++++++++---------
 1 file changed, 17 insertions(+), 9 deletions(-)

diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index ebc9b8d..713e206 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -2994,9 +2994,10 @@ static void __init iommu_exit_mempool(void)

 }

-static void __init init_no_remapping_devices(void)
+static int __init init_no_remapping_devices(void)
 {
 	struct dmar_drhd_unit *drhd;
+	int drhd_total = 0,drhd_ignored = 0;

 	for_each_drhd_unit(drhd) {
 		if (!drhd->include_all) {
@@ -3008,32 +3009,34 @@ static void __init init_no_remapping_devices(void)
 			if (i == drhd->devices_cnt)
 				drhd->ignored = 1;
 		}
+		drhd_total++;
 	}
-
-	if (dmar_map_gfx)
-		return;
-
 	for_each_drhd_unit(drhd) {
 		int i;
-		if (drhd->ignored || drhd->include_all)
+		if (drhd->ignored)
+			drhd_ignored++;
+		if (drhd->ignored || drhd->include_all) {
+			continue;
+		}
+		if (dmar_map_gfx)
 			continue;
-
 		for (i = 0; i < drhd->devices_cnt; i++)
 			if (drhd->devices[i] &&
 				!IS_GFX_DEVICE(drhd->devices[i]))
 				break;
-
 		if (i < drhd->devices_cnt)
 			continue;

 		/* bypass IOMMU if it is just for gfx devices */
 		drhd->ignored = 1;
+		drhd_ignored++;
 		for (i = 0; i < drhd->devices_cnt; i++) {
 			if (!drhd->devices[i])
 				continue;
 			drhd->devices[i]->dev.archdata.iommu = DUMMY_DEVICE_DOMAIN_INFO;
 		}
 	}
+	return (drhd_total == drhd_ignored);
 }

 #ifdef CONFIG_SUSPEND
@@ -3200,7 +3203,12 @@ int __init intel_iommu_init(void)
 	iommu_init_mempool();
 	dmar_init_reserved_ranges();

-	init_no_remapping_devices();
+	if(init_no_remapping_devices()) {
+		printk(KERN_ERR "IOMMU: all dmar HW unit ignored\n");
+		put_iova_domain(&reserved_iova_list);
+		iommu_exit_mempool();
+		return ret;
+	}

 	ret = init_dmars();
 	if (ret) {

Download attachment "0.patch" of type "application/octet-stream" (1743 bytes)