[PATCH 4/5] intel-iommu: Reinstate RMRRs if a device is removed frompassthrough domain

From: Alex Williamson
Date: Mon Oct 26 2009 - 19:29:54 EST


When a device is setup for passthrough it has full access to memory
so processing the RMRRs is unnecessary. However, if we remove the device
from the si_domain, we need to reinstate the associated RMRRs.

Signed-off-by: Alex Williamson <alex.williamson@xxxxxx>
---

drivers/pci/intel-iommu.c | 33 +++++++++++++++++++++++++++++++++
1 files changed, 33 insertions(+), 0 deletions(-)

diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index 19f10ae..a7f4476 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -2538,6 +2538,10 @@ static int iommu_no_mapping(struct device *dev)
if (iommu_should_identity_map(pdev, 0))
return 1;
else {
+ struct dmar_rmrr_unit *rmrr;
+ struct dmar_domain *domain;
+ int i, ret;
+
/*
* Devices that cannot support identity mapping
* are removed from si_domain and fall back to
@@ -2546,6 +2550,35 @@ static int iommu_no_mapping(struct device *dev)
domain_remove_one_dev_info(si_domain, pdev);
printk(KERN_INFO "%s uses non-identity mapping\n",
pci_name(pdev));
+
+ domain = get_valid_domain_for_dev(pdev);
+ if (!domain) {
+ printk(KERN_ERR
+ "Allocating domain for %s failed",
+ pci_name(pdev));
+ return 0;
+ }
+
+ ret = domain_add_dev_info(domain, pdev,
+ CONTEXT_TT_MULTI_LEVEL);
+ if (ret) {
+ printk(KERN_ERR
+ "Attaching %s to domain failed",
+ pci_name(pdev));
+ return 0;
+ }
+
+ for_each_rmrr_units(rmrr) {
+ for (i = 0; i < rmrr->devices_cnt; i++) {
+ if (pdev != rmrr->devices[i])
+ continue;
+ ret = iommu_prepare_rmrr_dev(rmrr,
+ pdev);
+ if (ret)
+ printk(KERN_ERR
+ "IOMMU: mapping reserved region failed\n");
+ }
+ }
return 0;
}
} else {

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/