diff mbox series

[V3] nvme: enable ro namespace for ZNS without append

Message ID 20201110210708.5912-1-javier@samsung.com (mailing list archive)
State New, archived
Headers show
Series [V3] nvme: enable ro namespace for ZNS without append | expand

Commit Message

Javier González Nov. 10, 2020, 9:07 p.m. UTC 
From: Javier González <javier.gonz@samsung.com>

Allow ZNS NVMe SSDs to present a read-only namespace when append is not
supported, instead of rejecting the namespace directly.

This allows (i) the namespace to be used in read-only mode, which is not
a problem as the append command only affects the write path, and (ii) to
use standard management tools such as nvme-cli to choose a different
format or firmware slot that is compatible with the Linux zoned block
device.

This patch includes comments from Christoph, Niklas and Keith that
applied to a different approach setting capacity to 0
  https://www.spinics.net/lists/linux-block/msg60747.html

The reminder of the original patch will be submitted separately.

Changes since V2:
  - Fix small conflict with a queued patch from Sagi (from Christoph)
  - Fix indentation (from Niklas)
  - Refresh effects log page to account for FW changes (from Keith)

Changes since V1:
  - Change logic to use NVME_NS_ATTR_RO (from Christoph)
  - Set max_zone_append egen in RO. This allows the device to be
    properly revalidated and enables user-space tools such as blkzone to
    be used when interacting with this zoned device.

Signed-off-by: Javier González <javier.gonz@samsung.com>
---
 drivers/nvme/host/core.c |  5 ++---
 drivers/nvme/host/nvme.h |  2 ++
 drivers/nvme/host/zns.c  | 19 ++++++++++++++-----
 3 files changed, 18 insertions(+), 8 deletions(-)

Comments

Keith Busch Nov. 10, 2020, 10:09 p.m. UTC | #1 
On Tue, Nov 10, 2020 at 10:07:08PM +0100, javier@javigon.com wrote:
> -	if (id->nsattr & NVME_NS_ATTR_RO)
> +	if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags))
>  		set_disk_ro(disk, true);

If the FORCE_RO flag is set, the disk is set to read-only. If that flag
is later cleared, nothing clears the disk's read-only setting.

> +	/* Refresh effects log page to check for changes on append support */
> +	status = nvme_get_effects_log(ns->ctrl, ns->head->ids.csi, &ns->head->effects);

That function just returns the cached log; no refresh occurs there.
Sagi Grimberg Nov. 11, 2020, 1:41 a.m. UTC | #2 
>> -	if (id->nsattr & NVME_NS_ATTR_RO)
>> +	if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags))
>>   		set_disk_ro(disk, true);
> 
> If the FORCE_RO flag is set, the disk is set to read-only. If that flag
> is later cleared, nothing clears the disk's read-only setting.

Yea, that is true also for the non-force case, but before it broke
BLKROSET so I reverted that. We can use this FORCE_RO for BLKROSET as
well I think...
Christoph Hellwig Nov. 11, 2020, 8:15 a.m. UTC | #3 
On Tue, Nov 10, 2020 at 05:41:15PM -0800, Sagi Grimberg wrote:
>
>>> -	if (id->nsattr & NVME_NS_ATTR_RO)
>>> +	if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags))
>>>   		set_disk_ro(disk, true);
>>
>> If the FORCE_RO flag is set, the disk is set to read-only. If that flag
>> is later cleared, nothing clears the disk's read-only setting.
>
> Yea, that is true also for the non-force case, but before it broke
> BLKROSET so I reverted that. We can use this FORCE_RO for BLKROSET as
> well I think...

Let me prioritize the hard r/o setting.  mkp actually has an older patch
that did just that which I'll resurrect.
Sagi Grimberg Nov. 11, 2020, 8:36 a.m. UTC | #4 
>>>> -	if (id->nsattr & NVME_NS_ATTR_RO)
>>>> +	if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags))
>>>>    		set_disk_ro(disk, true);
>>>
>>> If the FORCE_RO flag is set, the disk is set to read-only. If that flag
>>> is later cleared, nothing clears the disk's read-only setting.
>>
>> Yea, that is true also for the non-force case, but before it broke
>> BLKROSET so I reverted that. We can use this FORCE_RO for BLKROSET as
>> well I think...
> 
> Let me prioritize the hard r/o setting.  mkp actually has an older patch
> that did just that which I'll resurrect.

Sounds good.
Javier González Nov. 11, 2020, 9:21 a.m. UTC | #5 
On 11.11.2020 00:36, Sagi Grimberg wrote:
>
>>>>>-	if (id->nsattr & NVME_NS_ATTR_RO)
>>>>>+	if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags))
>>>>>   		set_disk_ro(disk, true);
>>>>
>>>>If the FORCE_RO flag is set, the disk is set to read-only. If that flag
>>>>is later cleared, nothing clears the disk's read-only setting.
>>>
>>>Yea, that is true also for the non-force case, but before it broke
>>>BLKROSET so I reverted that. We can use this FORCE_RO for BLKROSET as
>>>well I think...
>>
>>Let me prioritize the hard r/o setting.  mkp actually has an older patch
>>that did just that which I'll resurrect.
>
>Sounds good.

Cool. I'll repost fixing the log page update. I can then rebase on the
patches you send for this - or you can put them on top if it is easier.

Thanks!
diff mbox series

Patch

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index fff90200497c..8a224a6f2473 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -2083,7 +2083,7 @@  static void nvme_update_disk_info(struct gendisk *disk,
 	nvme_config_discard(disk, ns);
 	nvme_config_write_zeroes(disk, ns);
 
-	if (id->nsattr & NVME_NS_ATTR_RO)
+	if (id->nsattr & NVME_NS_ATTR_RO || test_bit(NVME_NS_FORCE_RO, &ns->flags))
 		set_disk_ro(disk, true);
 }
 
@@ -2951,8 +2951,7 @@  int nvme_get_log(struct nvme_ctrl *ctrl, u32 nsid, u8 log_page, u8 lsp, u8 csi,
 	return nvme_submit_sync_cmd(ctrl->admin_q, &c, log, size);
 }
 
-static int nvme_get_effects_log(struct nvme_ctrl *ctrl, u8 csi,
-				struct nvme_effects_log **log)
+int nvme_get_effects_log(struct nvme_ctrl *ctrl, u8 csi, struct nvme_effects_log **log)
 {
 	struct nvme_cel *cel = xa_load(&ctrl->cels, csi);
 	int ret;
diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index 83fb30e317e0..857fca95f016 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -449,6 +449,7 @@  struct nvme_ns {
 #define NVME_NS_REMOVING	0
 #define NVME_NS_DEAD     	1
 #define NVME_NS_ANA_PENDING	2
+#define NVME_NS_FORCE_RO	3
 
 	struct nvme_fault_inject fault_inject;
 
@@ -638,6 +639,7 @@  int nvme_delete_ctrl(struct nvme_ctrl *ctrl);
 
 int nvme_get_log(struct nvme_ctrl *ctrl, u32 nsid, u8 log_page, u8 lsp, u8 csi,
 		void *log, size_t size, u64 offset);
+int nvme_get_effects_log(struct nvme_ctrl *ctrl, u8 csi, struct nvme_effects_log **log);
 struct nvme_ns *nvme_get_ns_from_disk(struct gendisk *disk,
 		struct nvme_ns_head **head, int *srcu_idx);
 void nvme_put_ns_from_disk(struct nvme_ns_head *head, int idx);
diff --git a/drivers/nvme/host/zns.c b/drivers/nvme/host/zns.c
index 67e87e9f306f..47679a90795c 100644
--- a/drivers/nvme/host/zns.c
+++ b/drivers/nvme/host/zns.c
@@ -54,13 +54,22 @@  int nvme_update_zone_info(struct nvme_ns *ns, unsigned lbaf)
 	struct nvme_id_ns_zns *id;
 	int status;
 
+	/* Refresh effects log page to check for changes on append support */
+	status = nvme_get_effects_log(ns->ctrl, ns->head->ids.csi, &ns->head->effects);
+	if (status)
+		return status;
+
 	/* Driver requires zone append support */
-	if (!(le32_to_cpu(log->iocs[nvme_cmd_zone_append]) &
-			NVME_CMD_EFFECTS_CSUPP)) {
+	if ((le32_to_cpu(log->iocs[nvme_cmd_zone_append]) & NVME_CMD_EFFECTS_CSUPP)) {
+		if (test_and_clear_bit(NVME_NS_FORCE_RO, &ns->flags))
+			dev_warn(ns->ctrl->device,
+				 "append supported for zoned namespace:%d. Remove read-only mode\n",
+				 ns->head->ns_id);
+	} else {
+		set_bit(NVME_NS_FORCE_RO, &ns->flags);
 		dev_warn(ns->ctrl->device,
-			"append not supported for zoned namespace:%d\n",
-			ns->head->ns_id);
-		return -EINVAL;
+			 "append not supported for zoned namespace:%d. Forcing to read-only mode\n",
+			 ns->head->ns_id);
 	}
 
 	/* Lazily query controller append limit for the first zoned namespace */