Skip to contentSkip to conversation

nvme: zns: cap zone report nr_zones by DMA buffer size

PATCH Superseded Xixin Liu 3 participants 9 messages Jun 24 – Jul 6 NVMe

lore
Share & tools
lore

Xixin Liu

author
#0 lore raw

cover nvme: zns: cap zone report nr_zones by DMA buffer size

Hi,

The ZNS host driver issues Zone Management Receive with Partial Report (PR=1) fixed. The Number of Zones (NZ) field in the report header must reflect the zone descriptors actually transferred in the DMA buffer (ZNS Command Set Specification Rev 1.2, section 3.4.2).

nvme_ns_report_zones() caps the parse loop by the device-reported NZ and the caller's nr_zones limit, but not by the number of descriptors that fit in buflen. Cap nz with min3(device NZ, nr_zones - zone_idx, max_in_buf).

Thanks,
Xixin Liu

---

Xixin Liu (1):
nvme: zns: cap zone report nr_zones by DMA buffer size

 drivers/nvme/host/zns.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH 1/1 nvme: zns: cap zone report nr_zones by DMA buffer size

With Partial Report (PR=1), the Number of Zones (NZ) field in the report header must equal the number of zone descriptors fully transferred in the DMA buffer (ZNS Command Set Specification Rev 1.2, section 3.4.2).

nvme_ns_report_zones() does not cap the parse loop by max_in_buf derived from buflen. Cap nz with min3() over the device-reported count, nr_zones - zone_idx, and max_in_buf.

---
 drivers/nvme/host/zns.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
drivers/nvme/host/zns.c +8 -3
index 8ed1b6a33454..2a152e87bd76 100644
@@ -155,7 +155,8 @@ static int nvme_zone_parse_entry(struct nvme_ns *ns,
 	struct blk_zone zone = { };
 
 	if ((entry->zt & 0xf) != NVME_ZONE_TYPE_SEQWRITE_REQ) {
-		dev_err(ns->ctrl->device, "invalid zone type %#x\n", entry->zt);
+		dev_err(ns->ctrl->device, "invalid zone type %#x at zone %u\n",
+			entry->zt, idx);
 		return -EINVAL;
 	}
 
@@ -178,7 +179,7 @@ int nvme_ns_report_zones(struct nvme_ns *ns, sector_t sector,
 	struct nvme_zone_report *report;
 	struct nvme_command c = { };
 	int ret, zone_idx = 0;
-	unsigned int nz, i;
+	unsigned int max_in_buf, nz, i;
 	size_t buflen;
 
 	if (ns->head->ids.csi != NVME_CSI_ZNS)
@@ -188,6 +189,9 @@ int nvme_ns_report_zones(struct nvme_ns *ns, sector_t sector,
 	if (!report)
 		return -ENOMEM;
 
+	max_in_buf = (buflen - sizeof(struct nvme_zone_report)) /
+		sizeof(struct nvme_zone_descriptor);
+
 	c.zmr.opcode = nvme_cmd_zone_mgmt_recv;
 	c.zmr.nsid = cpu_to_le32(ns->head->ns_id);
 	c.zmr.numd = cpu_to_le32(nvme_bytes_to_numd(buflen));
@@ -207,7 +211,8 @@ int nvme_ns_report_zones(struct nvme_ns *ns, sector_t sector,
 			goto out_free;
 		}
 
-		nz = min((unsigned int)le64_to_cpu(report->nr_zones), nr_zones);
+		nz = min3((unsigned int)le64_to_cpu(report->nr_zones),
+			  nr_zones - zone_idx, max_in_buf);
 		if (!nz)
 			break;
 
Signature · 1 line
2.43.0
#2 lore raw
Quoted 18 lines · Xixin Liu

With Partial Report (PR=1), the Number of Zones (NZ) field in the report header must equal the number of zone descriptors fully transferred in the DMA buffer (ZNS Command Set Specification Rev 1.2, section 3.4.2).

nvme_ns_report_zones() does not cap the parse loop by max_in_buf derived from buflen. Cap nz with min3() over the device-reported count, nr_zones - zone_idx, and max_in_buf.

---
 drivers/nvme/host/zns.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
drivers/nvme/host/zns.c +0 -0
index 8ed1b6a33454..2a152e87bd76 100644

drivers/nvme/host/zns.c static int nvme_zone_parse_entry(struct nvme_ns *ns, · line 161

Quoted 8 lines
drivers/nvme/host/zns.c +2 -1
@@ -155,7 +155,8 @@ static int nvme_zone_parse_entry(struct nvme_ns *ns,
 	struct blk_zone zone = { };
 
 	if ((entry->zt & 0xf) != NVME_ZONE_TYPE_SEQWRITE_REQ) {
-		dev_err(ns->ctrl->device, "invalid zone type %#x\n", entry->zt);
+		dev_err(ns->ctrl->device, "invalid zone type %#x at zone %u\n",
+			entry->zt, idx);
 		return -EINVAL;
 	}

This looks fine, but unrelated. Please split it into a separate patch.

The rest looks good as well.

author
#3 lore raw

v2 cover nvme: zns: cap zone report nr_zones by DMA buffer size

In reply to Xixin Liu

Hi Christoph,

Thanks for the review. Split the zone index logging change into its own patch (2/2). The nr_zones / buflen fix is unchanged in 1/2.

Thanks,
Xixin Liu

---

Xixin Liu (2):
nvme: zns: cap zone report nr_zones by DMA buffer size
nvme: zns: include zone index in invalid zone type error

 drivers/nvme/host/zns.c | 12 ++++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
Signature · 1 line
2.43.0
author
#4 lore raw

PATCH v2 1/2 nvme: zns: cap zone report nr_zones by DMA buffer size

With Partial Report (PR=1), the Number of Zones (NZ) field in the report header must equal the number of zone descriptors fully transferred in the DMA buffer (ZNS Command Set Specification Rev 1.2, section 3.4.2).

nvme_ns_report_zones() does not cap the parse loop by max_in_buf derived from buflen. Cap nz with min3() over the device-reported count, nr_zones - zone_idx, and max_in_buf.

---
 drivers/nvme/host/zns.c | 9 ++++++++---
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/nvme/host/zns.c +6 -2
index 8ed1b6a33454..a1b2c3d4e5f6 100644
@@ -178,7 +179,7 @@ int nvme_ns_report_zones(struct nvme_ns *ns, sector_t sector,
 	struct nvme_zone_report *report;
 	struct nvme_command c = { };
 	int ret, zone_idx = 0;
-	unsigned int nz, i;
+	unsigned int max_in_buf, nz, i;
 	size_t buflen;
 
 	if (ns->head->ids.csi != NVME_CSI_ZNS)
@@ -188,6 +189,9 @@ int nvme_ns_report_zones(struct nvme_ns *ns, sector_t sector,
 	if (!report)
 		return -ENOMEM;
 
+	max_in_buf = (buflen - sizeof(struct nvme_zone_report)) /
+		sizeof(struct nvme_zone_descriptor);
+
 	c.zmr.opcode = nvme_cmd_zone_mgmt_recv;
 	c.zmr.nsid = cpu_to_le32(ns->head->ns_id);
 	c.zmr.numd = cpu_to_le32(nvme_bytes_to_numd(buflen));
@@ -207,7 +211,8 @@ int nvme_ns_report_zones(struct nvme_ns *ns, sector_t sector,
 			goto out_free;
 		}
 
-		nz = min((unsigned int)le64_to_cpu(report->nr_zones), nr_zones);
+		nz = min3((unsigned int)le64_to_cpu(report->nr_zones),
+			  nr_zones - zone_idx, max_in_buf);
 		if (!nz)
 			break;
 
Signature · 1 line
2.43.0
author
#6 lore raw

PATCH v2 2/2 nvme: zns: include zone index in invalid zone type error

In reply to Xixin Liu

Include the zone index when reporting an invalid zone type during zone descriptor parsing.

---
 drivers/nvme/host/zns.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/nvme/host/zns.c +2 -1
index 8ed1b6a33454..b2c3d4e5f6a7 100644
@@ -155,7 +155,8 @@ static int nvme_zone_parse_entry(struct nvme_ns *ns,
 	struct blk_zone zone = { };
 
 	if ((entry->zt & 0xf) != NVME_ZONE_TYPE_SEQWRITE_REQ) {
-		dev_err(ns->ctrl->device, "invalid zone type %#x\n", entry->zt);
+		dev_err(ns->ctrl->device, "invalid zone type %#x at zone %u\n",
+			entry->zt, idx);
 		return -EINVAL;
 	}
 
Signature · 1 line
2.43.0
maintainer Applied
#8 lore raw
In reply to Xixin Liu
Quoted 4 lines · Xixin Liu

Hi Christoph,

Thanks for the review. Split the zone index logging change into its own patch (2/2). The nr_zones / buflen fix is unchanged in 1/2.

Thanks, applied to nvme-7.3.