linux-kernel - [PATCH] iommu/amd: Do not flush IRTE when only updating isRun and destination fields

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:   Tue, 17 Oct 2023 09:42:36 -0500
From:   Suravee Suthikulpanit <suravee.suthikulpanit@....com>
To:     <linux-kernel@...r.kernel.org>, <iommu@...ts.linux.dev>
CC:     <joro@...tes.org>, <mlevitsk@...hat.com>, <seanjc@...gle.com>,
        <vasant.hegde@....com>, <jon.grimm@....com>,
        <santosh.shukla@....com>, <joao.m.martins@...cle.com>,
        <alejandro.j.jimenez@...cle.com>, <boris.ostrovsky@...cle.com>,
        Suravee Suthikulpanit <suravee.suthikulpanit@....com>
Subject: [PATCH] iommu/amd: Do not flush IRTE when only updating isRun and destination fields

According to the recent update in the AMD IOMMU spec [1], the IsRun and
Destination fields of the Interrupt Remapping Table Entry (IRTE) are not
cached by the IOMMU hardware.

Therefore, do not issue the INVALIDATE_INTERRUPT_TABLE command when
updating IRTE[IsRun] and IRTE[Destination] when IRTE[GuestMode]=1, which
should help improve IOMMU AVIC/x2AVIC performance.

References:
[1] AMD IOMMU Spec Revision (Rev 3.08-PUB)
(Link: https://www.amd.com/content/dam/amd/en/documents/processor-tech-docs/specifications/48882_IOMMU.pdf)

Cc: Joao Martins <joao.m.martins@...cle.com>
Cc: Alejandro Jimenez <alejandro.j.jimenez@...cle.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@....com>
---
 drivers/iommu/amd/iommu.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
index 089886485895..d63590563d3e 100644
--- a/drivers/iommu/amd/iommu.c
+++ b/drivers/iommu/amd/iommu.c
@@ -2970,8 +2970,8 @@ static int alloc_irq_index(struct amd_iommu *iommu, u16 devid, int count,
 	return index;
 }
 
-static int modify_irte_ga(struct amd_iommu *iommu, u16 devid, int index,
-			  struct irte_ga *irte)
+static int __modify_irte_ga(struct amd_iommu *iommu, u16 devid, int index,
+			    struct irte_ga *irte)
 {
 	struct irq_remap_table *table;
 	struct irte_ga *entry;
@@ -2998,6 +2998,18 @@ static int modify_irte_ga(struct amd_iommu *iommu, u16 devid, int index,
 
 	raw_spin_unlock_irqrestore(&table->lock, flags);
 
+	return 0;
+}
+
+static int modify_irte_ga(struct amd_iommu *iommu, u16 devid, int index,
+			  struct irte_ga *irte)
+{
+	bool ret;
+
+	ret = __modify_irte_ga(iommu, devid, index, irte);
+	if (ret)
+		return ret;
+
 	iommu_flush_irt_and_complete(iommu, devid);
 
 	return 0;
@@ -3681,8 +3693,8 @@ int amd_iommu_update_ga(int cpu, bool is_run, void *data)
 	}
 	entry->lo.fields_vapic.is_run = is_run;
 
-	return modify_irte_ga(ir_data->iommu, ir_data->irq_2_irte.devid,
-			      ir_data->irq_2_irte.index, entry);
+	return __modify_irte_ga(ir_data->iommu, ir_data->irq_2_irte.devid,
+				ir_data->irq_2_irte.index, entry);
 }
 EXPORT_SYMBOL(amd_iommu_update_ga);
 #endif
-- 
2.34.1