Skip to contentSkip to conversation

nvmet: Fix discover log page when offsets are used

PATCH Superseded Keith Busch 3 participants 4 messages Apr 8, 2019

Subsystems (2)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 nvmet: Fix discover log page when offsets are used

The nvme target hadn't been taking the Get Log Page offset parameter into consideration, and so has been returning corrupting logs pages when offsets are used. Since many tools, including nvme-cli, split the log request to 4k, we've been breaking discovery log responses when more than 3 subsystems exist.

Fix the log page by internally generating the entire discovery log page and copying only the requested bytes into the user buffer.

Cc Hannes Reinecke <hare at suse.de>
Signed-off-by Keith Busch <keith.busch at intel.com>
---
 drivers/nvme/target/admin-cmd.c | 10 +++++++
 drivers/nvme/target/discovery.c | 63 +++++++++++++++++++++++++----------------
 drivers/nvme/target/nvmet.h     |  1 +
 3 files changed, 50 insertions(+), 24 deletions(-)
drivers/nvme/target/admin-cmd.c +10 -0
index 76250181fee0..e7f3f6e71853 100644
@@ -24,6 +24,16 @@ u32 nvmet_get_log_page_len(struct nvme_command *cmd)
 	return len;
 }
 
+u64 nvmet_get_log_page_offset(struct nvme_command *cmd)
+{
+	u64 offset = le32_to_cpu(cmd->get_log_page.lpou);
+
+	offset <<= 32;
+	offset += le32_to_cpu(cmd->get_log_page.lpol);
+
+	return offset;
+}
+
 static void nvmet_execute_get_log_page_noop(struct nvmet_req *req)
 {
 	nvmet_req_complete(req, nvmet_zero_sgl(req, 0, req->data_len));
drivers/nvme/target/discovery.c +39 -24

100 lines not on this page · raw message

drivers/nvme/target/nvmet.h +1 -0
index 51e49efd7849..1653d19b187f 100644
@@ -428,6 +428,7 @@ u16 nvmet_copy_from_sgl(struct nvmet_req *req, off_t off, void *buf,
 u16 nvmet_zero_sgl(struct nvmet_req *req, off_t off, size_t len);
 
 u32 nvmet_get_log_page_len(struct nvme_command *cmd);
+u64 nvmet_get_log_page_offset(struct nvme_command *cmd);
 
 extern struct list_head *nvmet_ports;
 void nvmet_port_disc_changed(struct nvmet_port *port,
Signature · 1 line
2.14.4
maintainer
#1 lore raw

PATCH 1/1 nvmet: Fix discover log page when offsets are used

drivers/nvme/target/discovery.c static void nvmet_execute_get_disc_log_page(struct nvmet_req *req) · line 177

Quoted 13 lines · Keith Busch
drivers/nvme/target/discovery.c +4 -2
@@ -149,36 +168,33 @@ static void nvmet_execute_get_disc_log_page(struct nvmet_req *req)
 	 * If host provided data len is less than the header size, only the
 	 * number of bytes requested by host will be sent to host.
 	 */
-	hdr = kzalloc(alloc_len, GFP_KERNEL);
-	if (!hdr) {
+	down_read(&nvmet_config_sem);
+	alloc_len = sizeof(*hdr) + entry_size * discovery_log_entries(req);
+	buffer = kzalloc(alloc_len, GFP_KERNEL);
+	if (!buffer) {
 		status = NVME_SC_INTERNAL;
 		goto out;
 	}

There's an error here not unlocking the target control rw-semaphore on allocation failure. Going to send a v2 in 5 seconds, so please disregard this one.

#2 lore raw

PATCH 1/1 nvmet: Fix discover log page when offsets are used

In reply to Keith Busch
Quoted 9 lines · Keith Busch

The nvme target hadn't been taking the Get Log Page offset parameter into consideration, and so has been returning corrupting logs pages when offsets are used. Since many tools, including nvme-cli, split the log request to 4k, we've been breaking discovery log responses when more than 3 subsystems exist.

Fix the log page by internally generating the entire discovery log page and copying only the requested bytes into the user buffer.

Quoted 9 lines
Cc Hannes Reinecke <hare at suse.de>
Signed-off-by Keith Busch <keith.busch at intel.com>
---
  drivers/nvme/target/admin-cmd.c | 10 +++++++
  drivers/nvme/target/discovery.c | 63 +++++++++++++++++++++++++----------------
  drivers/nvme/target/nvmet.h     |  1 +
  3 files changed, 50 insertions(+), 24 deletions(-)
Reviewed-by ? James Smart? <james.smart at broadcom.com>

Looks good.

-- james

#3 lore raw

PATCH 1/1 nvmet: Fix discover log page when offsets are used

Quoted 23 lines · James Smart
Quoted 9 lines · Keith Busch

The nvme target hadn't been taking the Get Log Page offset parameter into consideration, and so has been returning corrupting logs pages when offsets are used. Since many tools, including nvme-cli, split the log request to 4k, we've been breaking discovery log responses when more than 3 subsystems exist.

Fix the log page by internally generating the entire discovery log page and copying only the requested bytes into the user buffer.

Quoted 10 lines
Cc Hannes Reinecke <hare at suse.de>
Signed-off-by Keith Busch <keith.busch at intel.com>

---
? drivers/nvme/target/admin-cmd.c | 10 +++++++ ? drivers/nvme/target/discovery.c | 63
+++++++++++++++++++++++++----------------
? drivers/nvme/target/nvmet.h???? |? 1 +
? 3 files changed, 50 insertions(+), 24 deletions(-)

Quoted 7 lines
Reviewed-by ? James Smart? <james.smart at broadcom.com>

Looks good.

-- james

?excepting the locking issue :)

-- james