linux-kernel - [PATCH v4] iommu/amd: Enhance "Completion-wait Time-out" error message

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>] [day] [month] [year] [list]

Message-ID: <20251105080342.820-1-dheerajkumar.srivastava@amd.com>
Date: Wed, 5 Nov 2025 13:33:42 +0530
From: Dheeraj Kumar Srivastava <dheerajkumar.srivastava@....com>
To: <joro@...tes.org>, <will@...nel.org>, <robin.murphy@....com>,
	<iommu@...ts.linux.dev>, <linux-kernel@...r.kernel.org>
CC: <suravee.suthikulpanit@....com>, <Vasant.Hegde@....com>,
	<Santosh.Shukla@....com>, Dheeraj Kumar Srivastava
	<dheerajkumar.srivastava@....com>
Subject: [PATCH v4] iommu/amd: Enhance "Completion-wait Time-out" error message

Current IOMMU driver prints "Completion-wait Time-out" error message with
insufficient information to further debug the issue.

Enhancing the error message as following:
1. Log IOMMU PCI device ID in the error message.
2. With "amd_iommu_dump=1" kernel command line option, dump entire
   command buffer entries including Head and Tail offset.

Dump the entire command buffer only on the first 'Completion-wait Time-out'
to avoid dmesg spam.

Signed-off-by: Dheeraj Kumar Srivastava <dheerajkumar.srivastava@....com>
---
Changes since v3:
-> Dump the entire command buffer only on the first 'Completion-wait Time-out'
   when amd_iommu_dump=1, instead of dumping it on every occurrence.

 drivers/iommu/amd/amd_iommu_types.h |  4 ++++
 drivers/iommu/amd/iommu.c           | 28 +++++++++++++++++++++++++++-
 2 files changed, 31 insertions(+), 1 deletion(-)

diff --git a/drivers/iommu/amd/amd_iommu_types.h b/drivers/iommu/amd/amd_iommu_types.h
index 95f63c5f6159..7576814f944d 100644
--- a/drivers/iommu/amd/amd_iommu_types.h
+++ b/drivers/iommu/amd/amd_iommu_types.h
@@ -247,6 +247,10 @@
 #define CMD_BUFFER_ENTRIES 512
 #define MMIO_CMD_SIZE_SHIFT 56
 #define MMIO_CMD_SIZE_512 (0x9ULL << MMIO_CMD_SIZE_SHIFT)
+#define MMIO_CMD_HEAD_MASK	GENMASK_ULL(18, 4)
+#define MMIO_CMD_BUFFER_HEAD(x) FIELD_GET(MMIO_CMD_HEAD_MASK, (x))
+#define MMIO_CMD_TAIL_MASK	GENMASK_ULL(18, 4)
+#define MMIO_CMD_BUFFER_TAIL(x) FIELD_GET(MMIO_CMD_TAIL_MASK, (x))
 
 /* constants for event buffer handling */
 #define EVT_BUFFER_SIZE		8192 /* 512 entries */
diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
index eb348c63a8d0..abce078d2323 100644
--- a/drivers/iommu/amd/iommu.c
+++ b/drivers/iommu/amd/iommu.c
@@ -1156,6 +1156,25 @@ irqreturn_t amd_iommu_int_handler(int irq, void *data)
  *
  ****************************************************************************/
 
+static void dump_command_buffer(struct amd_iommu *iommu)
+{
+	struct iommu_cmd *cmd;
+	int head, tail, i;
+
+	head = readl(iommu->mmio_base + MMIO_CMD_HEAD_OFFSET);
+	tail = readl(iommu->mmio_base + MMIO_CMD_TAIL_OFFSET);
+
+	pr_err("CMD Buffer head=%d tail=%d\n", (int)(MMIO_CMD_BUFFER_HEAD(head)),
+	       (int)(MMIO_CMD_BUFFER_TAIL(tail)));
+
+	for (i = 0; i < CMD_BUFFER_ENTRIES; i++) {
+		cmd = (struct iommu_cmd *)(iommu->cmd_buf + i * sizeof(*cmd));
+		pr_err("%3d: %08x %08x %08x %08x\n", i, cmd->data[0], cmd->data[1], cmd->data[2],
+		       cmd->data[3]);
+	}
+}
+
+
 static int wait_on_sem(struct amd_iommu *iommu, u64 data)
 {
 	int i = 0;
@@ -1166,7 +1185,14 @@ static int wait_on_sem(struct amd_iommu *iommu, u64 data)
 	}
 
 	if (i == LOOP_TIMEOUT) {
-		pr_alert("Completion-Wait loop timed out\n");
+
+		pr_alert("IOMMU %04x:%02x:%02x.%01x: Completion-Wait loop timed out\n",
+			 iommu->pci_seg->id, PCI_BUS_NUM(iommu->devid),
+			 PCI_SLOT(iommu->devid), PCI_FUNC(iommu->devid));
+
+		if (amd_iommu_dump)
+			DO_ONCE_LITE(dump_command_buffer, iommu);
+
 		return -EIO;
 	}
 
-- 
2.25.1