convert nvme to atomic queue limits updates
Subsystems (5)
Christoph Hellwig
maintainerHi all,
this series converts nvme to the new atomic queue limit updates, and as part of that refactors a lot of the setup code. The first two patches are block layer patches already sent out as part of the md queue limits conversion and will hopefully get merged into Jens' tree soon.
Diffstat:
block/blk-settings.c | 43 ++++ drivers/nvme/host/apple.c | 18 +- drivers/nvme/host/core.c | 378 ++++++++++++++++++++---------------------- drivers/nvme/host/fc.c | 8 drivers/nvme/host/multipath.c | 13 - drivers/nvme/host/nvme.h | 12 - drivers/nvme/host/rdma.c | 13 - drivers/nvme/host/zns.c | 24 -- drivers/nvme/target/loop.c | 5 include/linux/blkdev.h | 3 10 files changed, 272 insertions(+), 245 deletions(-)
Christoph Hellwig
maintainerPATCH 1/21 block: add a queue_limits_set helper
Add a small wrapper around queue_limits_commit_update for stacking drivers that don't want to update existing limits, but set an entirely new set.
--- block/blk-settings.c | 18 ++++++++++++++++++ include/linux/blkdev.h | 1 + 2 files changed, 19 insertions(+)
block/blk-settings.c +18 -0
@@ -266,6 +266,24 @@ int queue_limits_commit_update(struct request_queue *q, }EXPORT_SYMBOL_GPL (queue_limits_commit_update); +/** +* queue_limits_commit_set - apply queue limits to queue +* @q: queue to update +* @lim: limits to apply +* +* Apply the limits in @lim that were freshly initialized to @q. +* To update existing limits use queue_limits_start_update() and +* queue_limits_commit_update() instead. +* +* Returns 0 if successful, else a negative error code. +*/ +int queue_limits_set (struct request_queue *q,struct queue_limits *lim) +{ +mutex_lock (&q->limits_lock); +return queue_limits_commit_update (q, lim); +} +EXPORT_SYMBOL_GPL (queue_limits_set); +/** * blk_queue_bounce_limit - set bounce buffer limit for queue * @q: the request queue for the device
include/linux/blkdev.h +1 -0
@@ -889,6 +889,7 @@ queue_limits_start_update(struct request_queue *q) }int queue_limits_commit_update (struct request_queue *q,struct queue_limits *lim); +int queue_limits_set (struct request_queue *q,struct queue_limits *lim);/* * Access functions for manipulating queue properties
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 2/21 block: add a queue_limits_stack_bdev helper
In reply to Christoph HellwigAdd a small wrapper around blk_stack_limits that allows passing a bdev for the bottom device and prints an error in case of misaligned device. The name fits into the new queue limits API and the intent is to eventually replace disk_stack_limits.
--- block/blk-settings.c | 25 +++++++++++++++++++++++++ include/linux/blkdev.h | 2 ++ 2 files changed, 27 insertions(+)
block/blk-settings.c +25 -0
@@ -891,6 +891,31 @@ int blk_stack_limits(struct queue_limits *t, struct queue_limits *b, }EXPORT_SYMBOL (blk_stack_limits); +/** +* queue_limits_stack_bdev - adjust queue_limits for stacked devices +* @t: the stacking driver limits (top device) +* @bdev: the underlying block device (bottom) +* @offset: offset to beginning of data within component device +* @pfx: prefix to use for warnings logged +* +* Description: +* This function is used by stacking drivers like MD and DM to ensure +* that all component devices have compatible block sizes and +* alignments. The stacking driver must provide a queue_limits +* struct (top) and then iteratively call the stacking function for +* all component (bottom) devices. The stacking function will +* attempt to combine the values and ensure proper alignment. +*/ +void queue_limits_stack_bdev (struct queue_limits *t,struct block_device *bdev, +sector_t offset,const char *pfx) +{ +if (blk_stack_limits (t, &bdev_get_queue (bdev)->limits, +get_start_sect (bdev) + offset)) +pr_notice ("%s: Warning: Device %pg is misaligned\n" , + pfx, bdev); +} +EXPORT_SYMBOL_GPL (queue_limits_stack_bdev); +/** * disk_stack_limits - adjust queue limits for stacked drivers * @disk: MD/DM gendisk (top)
include/linux/blkdev.h +2 -0
@@ -924,6 +924,8 @@ extern void blk_set_queue_depth(struct request_queue *q, unsigned int depth);extern void blk_set_stacking_limits (struct queue_limits *lim);extern int blk_stack_limits (struct queue_limits *t,struct queue_limits *b,sector_t offset); +void queue_limits_stack_bdev (struct queue_limits *t,struct block_device *bdev, +sector_t offset,const char *pfx);extern void disk_stack_limits (struct gendisk *disk,struct block_device *bdev,sector_t offset);extern void blk_queue_update_dma_pad (struct request_queue *,unsigned int );
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 3/21 nvme: set max_hw_sectors unconditionally
In reply to Christoph HellwigAll transports set a max_hw_sectors value in the nvme_ctrl, so make the code using it unconditional and clean it up using a little helper.
--- drivers/nvme/host/core.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/nvme/host/core.c +8 -8
@@ -1944,19 +1944,19 @@ static int nvme_configure_metadata(struct nvme_ctrl *ctrl,return 0 ; } +static u32 nvme_max_drv_segments (struct nvme_ctrl *ctrl) +{ +return ctrl->max_hw_sectors / (NVME_CTRL_PAGE_SIZE >>SECTOR_SHIFT ) +1 ; +} +static void nvme_set_queue_limits (struct nvme_ctrl *ctrl,struct request_queue *q) {bool vwc = ctrl->vwc &NVME_CTRL_VWC_PRESENT ; -if (ctrl->max_hw_sectors) { -u32 max_segments = - (ctrl->max_hw_sectors / (NVME_CTRL_PAGE_SIZE >>9 )) +1 ; - - max_segments =min_not_zero (max_segments, ctrl->max_segments); -blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors); -blk_queue_max_segments (q,min_t (u32, max_segments,USHRT_MAX )); - } +blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors); +blk_queue_max_segments (q,min_t (u32,USHRT_MAX , +min_not_zero (nvme_max_drv_segments (ctrl), ctrl->max_segments)));blk_queue_virt_boundary (q,NVME_CTRL_PAGE_SIZE -1 );blk_queue_dma_alignment (q,3 );blk_queue_write_cache (q, vwc, vwc);
Signature · 1 line
2.39.2
Re: [PATCH 03/21] nvme: set max_hw_sectors unconditionally
Quoted 30 lines · Christoph Hellwig
All transports set a
max_hw_sectorsvalue in thenvme_ctrl, so make the code using it unconditional and clean it up using a little helper.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)drivers/nvme/host/core.c +5
-4@@ -1944,19 +1944,19 @@ static int nvme_configure_metadata(struct nvme_ctrl *ctrl,return 0 ; } +static u32 nvme_max_drv_segments (struct nvme_ctrl *ctrl) +{ +return ctrl->max_hw_sectors / (NVME_CTRL_PAGE_SIZE >>SECTOR_SHIFT ) +1 ; +} +static void nvme_set_queue_limits (struct nvme_ctrl *ctrl,struct request_queue *q) {bool vwc = ctrl->vwc &NVME_CTRL_VWC_PRESENT ; -if (ctrl->max_hw_sectors) { -u32 max_segments = - (ctrl->max_hw_sectors / (NVME_CTRL_PAGE_SIZE >>9 )) +1 ; -
drivers/nvme/host/core.c static int nvme_configure_metadata(struct nvme_ctrl *ctrl, · line 1962
Quoted 10 lines
drivers/nvme/host/core.c +3
-4- max_segments =min_not_zero (max_segments, ctrl->max_segments); -blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors); -blk_queue_max_segments (q,min_t (u32, max_segments,USHRT_MAX )); - } +blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors); +blk_queue_max_segments (q,min_t (u32,USHRT_MAX , +min_not_zero (nvme_max_drv_segments (ctrl), ctrl->max_segments)));blk_queue_virt_boundary (q,NVME_CTRL_PAGE_SIZE -1 );blk_queue_dma_alignment (q,3 );blk_queue_write_cache (q, vwc, vwc);
Looks good,
Christoph Hellwig
maintainerPATCH 4/21 nvme: move NVME_QUIRK_DEALLOCATE_ZEROES out of nvme_config_discard
In reply to Christoph HellwigMove the handling of the NVME_QUIRK_DEALLOCATE_ZEROES quirk out of nvme_config_discard so that it is combined with the normal write_zeroes limit handling.
--- drivers/nvme/host/core.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)
drivers/nvme/host/core.c +6 -5
@@ -1816,9 +1816,6 @@ static void nvme_config_discard(struct nvme_ctrl *ctrl, struct gendisk *disk,else blk_queue_max_discard_segments (queue,NVME_DSM_MAX_RANGES ); queue->limits.discard_granularity =queue_logical_block_size (queue); - -if (ctrl->quirks &NVME_QUIRK_DEALLOCATE_ZEROES ) -blk_queue_max_write_zeroes_sectors (queue,UINT_MAX ); }static bool nvme_ns_ids_equal (struct nvme_ns_ids *a,struct nvme_ns_ids *b)
@@ -2029,8 +2026,12 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,set_capacity_and_notify (disk, capacity);nvme_config_discard (ctrl, disk, head); -blk_queue_max_write_zeroes_sectors (disk->queue, - ctrl->max_zeroes_sectors); + +if (ctrl->quirks &NVME_QUIRK_DEALLOCATE_ZEROES ) +blk_queue_max_write_zeroes_sectors (disk->queue,UINT_MAX ); +else +blk_queue_max_write_zeroes_sectors (disk->queue, + ctrl->max_zeroes_sectors); }static bool nvme_ns_is_readonly (struct nvme_ns *ns,struct nvme_ns_info *info)
Signature · 1 line
2.39.2
Re: [PATCH 04/21] nvme: move NVME_QUIRK_DEALLOCATE_ZEROES out of nvme_config_discard
Quoted 28 lines · Christoph Hellwig
Move the handling of the
NVME_QUIRK_DEALLOCATE_ZEROESquirk out ofnvme_config_discardso that it is combined with the normalwrite_zeroeslimit handling.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-)drivers/nvme/host/core.c +0
-4@@ -1816,9 +1816,6 @@ static void nvme_config_discard(struct nvme_ctrl *ctrl, struct gendisk *disk,else blk_queue_max_discard_segments (queue,NVME_DSM_MAX_RANGES ); queue->limits.discard_granularity =queue_logical_block_size (queue); - -if (ctrl->quirks &NVME_QUIRK_DEALLOCATE_ZEROES ) -blk_queue_max_write_zeroes_sectors (queue,UINT_MAX ); }static bool nvme_ns_ids_equal (struct nvme_ns_ids *a,struct nvme_ns_ids *b)@@ -2029,8 +2026,12 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,set_capacity_and_notify (disk, capacity);nvme_config_discard (ctrl, disk, head); - blk_queue_max_write_zeroes_sectors(disk->queue,
drivers/nvme/host/core.c static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk, · line 2037
Quoted 10 lines
drivers/nvme/host/core.c +6
-1- ctrl->max_zeroes_sectors); + +if (ctrl->quirks &NVME_QUIRK_DEALLOCATE_ZEROES ) +blk_queue_max_write_zeroes_sectors (disk->queue,UINT_MAX ); +else +blk_queue_max_write_zeroes_sectors (disk->queue, + ctrl->max_zeroes_sectors); }static bool nvme_ns_is_readonly (struct nvme_ns *ns,struct nvme_ns_info *info)
Looks good,
Christoph Hellwig
maintainerPATCH 5/21 nvme: remove nvme_revalidate_zones
In reply to Christoph HellwigHandle setting the zone size / chunk_sectors and max_append_sectors limits together with the other ZNS limits, and just open code the call to blk_revalidate_zones in the current place.
--- drivers/nvme/host/core.c | 2 +- drivers/nvme/host/nvme.h | 1 - drivers/nvme/host/zns.c | 12 ++---------- 3 files changed, 3 insertions(+), 12 deletions(-)
drivers/nvme/host/core.c +1 -1
@@ -2154,7 +2154,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,blk_mq_unfreeze_queue (ns->disk->queue);if (blk_queue_is_zoned (ns->queue)) { - ret =nvme_revalidate_zones (ns); + ret =blk_revalidate_disk_zones (ns->disk,NULL );if (ret && !nvme_first_scan (ns->disk)) goto out; }
drivers/nvme/host/nvme.h +0 -1
@@ -1036,7 +1036,6 @@ static inline bool nvme_disk_is_ns_head(struct gendisk *disk) }#endif /* CONFIG_NVME_MULTIPATH */ -int nvme_revalidate_zones (struct nvme_ns *ns);int nvme_ns_report_zones (struct nvme_ns *ns,sector_t sector,unsigned int nr_zones,report_zones_cb cb,void *data);#ifdef CONFIG_BLK_DEV_ZONED
drivers/nvme/host/zns.c +2 -10
@@ -7,16 +7,6 @@#include <linux/vmalloc.h> #include "nvme.h" -int nvme_revalidate_zones (struct nvme_ns *ns) -{ -struct request_queue *q = ns->queue; - -blk_queue_chunk_sectors (q, ns->head->zsze); -blk_queue_max_zone_append_sectors (q, ns->ctrl->max_zone_append); - -return blk_revalidate_disk_zones (ns->disk,NULL ); -} -static int nvme_set_max_append (struct nvme_ctrl *ctrl) {struct nvme_command c = { };
@@ -113,6 +103,8 @@ int nvme_update_zone_info(struct nvme_ns *ns, unsigned lbaf)blk_queue_flag_set (QUEUE_FLAG_ZONE_RESETALL , q);disk_set_max_open_zones (ns->disk,le32_to_cpu (id->mor) +1 );disk_set_max_active_zones (ns->disk,le32_to_cpu (id->mar) +1 ); +blk_queue_chunk_sectors (ns->queue, ns->head->zsze); +blk_queue_max_zone_append_sectors (ns->queue, ns->ctrl->max_zone_append); free_data:kfree (id);return status;
Signature · 1 line
2.39.2
Re: [PATCH 05/21] nvme: remove nvme_revalidate_zones
Quoted 5 lines · Christoph Hellwig
Handle setting the zone size /
chunk_sectorsandmax_append_sectorslimits together with the other ZNS limits, and just open code the call toblk_revalidate_zonesin the current place.Signed-off-by Christoph Hellwig <[email protected]>
Looks good.
[...]
drivers/nvme/host/zns.c int nvme_update_zone_info(struct nvme_ns *ns, unsigned lbaf) · line 105
Quoted 7 lines
static int
nvme_set_max_append(struct nvme_ctrl *ctrl){drivers/nvme/host/zns.c +0
-0struct nvme_command c = { };@@ -113,6 +103,8 @@ int nvme_update_zone_info(struct nvme_ns *ns, unsigned lbaf)blk_queue_flag_set (QUEUE_FLAG_ZONE_RESETALL , q);disk_set_max_open_zones (ns->disk,le32_to_cpu (id->mor) +1 );disk_set_max_active_zones (ns->disk,le32_to_cpu (id->mar) +1 );
Not directly related to this patch, but I think that since you moved the max open & active limits to queue->limits, we really should rename these to blk_queue_max_open_zones() and blk_queue_max_active_zones() to be consistent with all the other limits. and may be same for disk_set_zoned() while we are at it.
I can send a patch if you want...
Quoted 5 lines
drivers/nvme/host/zns.c +2
-0+blk_queue_chunk_sectors (ns->queue, ns->head->zsze); +blk_queue_max_zone_append_sectors (ns->queue, ns->ctrl->max_zone_append); free_data:kfree (id);return status;
Signature · 2 lines
Damien Le Moal Western Digital Research
Christoph Hellwig
maintainerPATCH 6/21 nvme: move max_integrity_segments handling out of nvme_init_integrity
In reply to Christoph Hellwigmax_integrity_segments is just a hardware limit and doesn't need to be in nvme_init_integrity with the PI setup.
--- drivers/nvme/host/core.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-)
drivers/nvme/host/core.c +4 -7
@@ -1724,8 +1724,7 @@ int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo) }#ifdef CONFIG_BLK_DEV_INTEGRITY -static void nvme_init_integrity (struct gendisk *disk, -struct nvme_ns_head *head,u32 max_integrity_segments) +static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) {struct blk_integrity integrity = { };
@@ -1773,11 +1772,9 @@ static void nvme_init_integrity(struct gendisk *disk, integrity.tuple_size = head->ms; integrity.pi_offset = head->pi_offset; blk_integrity_register(disk, &integrity); - blk_queue_max_integrity_segments(disk->queue, max_integrity_segments); }#else -static void nvme_init_integrity (struct gendisk *disk, -struct nvme_ns_head *head,u32 max_integrity_segments) +static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) { }#endif /* CONFIG_BLK_DEV_INTEGRITY */
@@ -1954,6 +1951,7 @@ static void nvme_set_queue_limits(struct nvme_ctrl *ctrl,blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors);blk_queue_max_segments (q,min_t (u32,USHRT_MAX ,min_not_zero (nvme_max_drv_segments (ctrl), ctrl->max_segments))); +blk_queue_max_integrity_segments (q, ctrl->max_integrity_segments);blk_queue_virt_boundary (q,NVME_CTRL_PAGE_SIZE -1 );blk_queue_dma_alignment (q,3 );blk_queue_write_cache (q, vwc, vwc);
@@ -2017,8 +2015,7 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,if (head->ms) {if (IS_ENABLED (CONFIG_BLK_DEV_INTEGRITY ) && (head->features &NVME_NS_METADATA_SUPPORTED )) -nvme_init_integrity (disk, head, - ctrl->max_integrity_segments); +nvme_init_integrity (disk, head);else if (!nvme_ns_has_pi (head)) capacity =0 ; }
Signature · 1 line
2.39.2
Re: [PATCH 06/21] nvme: move max_integrity_segments handling out of nvme_init_integrity
Quoted 44 lines · Christoph Hellwig
max_integrity_segmentsis just a hardware limit and doesn't need to be innvme_init_integritywith the PI setup.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-)drivers/nvme/host/core.c +3
-5@@ -1724,8 +1724,7 @@ int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo) }#ifdef CONFIG_BLK_DEV_INTEGRITY -static void nvme_init_integrity (struct gendisk *disk, -struct nvme_ns_head *head,u32 max_integrity_segments) +static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) {struct blk_integrity integrity = { };@@ -1773,11 +1772,9 @@ static void nvme_init_integrity(struct gendisk *disk, integrity.tuple_size = head->ms; integrity.pi_offset = head->pi_offset; blk_integrity_register(disk, &integrity); - blk_queue_max_integrity_segments(disk->queue, max_integrity_segments); }#else -static void nvme_init_integrity (struct gendisk *disk, -struct nvme_ns_head *head,u32 max_integrity_segments) +static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) { }#endif /* CONFIG_BLK_DEV_INTEGRITY */ @@ -1954,6 +1951,7 @@ static void nvme_set_queue_limits(struct nvme_ctrl *ctrl,blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors);blk_queue_max_segments (q,min_t (u32,USHRT_MAX ,min_not_zero (nvme_max_drv_segments (ctrl), ctrl->max_segments))); +blk_queue_max_integrity_segments (q, ctrl->max_integrity_segments);blk_queue_virt_boundary (q,NVME_CTRL_PAGE_SIZE -1 );blk_queue_dma_alignment (q,3 );blk_queue_write_cache (q, vwc, vwc);
drivers/nvme/host/core.c static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk, · line 2021
Quoted 9 lines
drivers/nvme/host/core.c +1
-2@@ -2017,8 +2015,7 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,if (head->ms) {if (IS_ENABLED (CONFIG_BLK_DEV_INTEGRITY ) && (head->features &NVME_NS_METADATA_SUPPORTED )) -nvme_init_integrity (disk, head, - ctrl->max_integrity_segments); +nvme_init_integrity (disk, head);else if (!nvme_ns_has_pi (head)) capacity =0 ; }
Looks good,
Christoph Hellwig
maintainerPATCH 7/21 nvme: cleanup the nvme_init_integrity calling conventions
In reply to Christoph HellwigHandle the no metadata support case in nvme_init_integrity as well to simplify the calling convention and prepare for future changes in the area.
--- drivers/nvme/host/core.c | 29 +++++++++++++++-------------- 1 file changed, 15 insertions(+), 14 deletions(-)
drivers/nvme/host/core.c +15 -14
@@ -1723,11 +1723,21 @@ int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo)return 0 ; } -#ifdef CONFIG_BLK_DEV_INTEGRITY -static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) +static bool nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) {struct blk_integrity integrity = { }; +if (!head->ms) +return true; + +/* +* PI can always be supported as we can ask the controller to simply +* insert/strip it, which is not possible for other kinds of metadata. +*/ +if (!IS_ENABLED (CONFIG_BLK_DEV_INTEGRITY ) || + !(head->features &NVME_NS_METADATA_SUPPORTED )) +return nvme_ns_has_pi (head); +switch (head->pi_type) {case NVME_NS_DPS_PI_TYPE3 :switch (head->guard_type) {
@@ -1772,12 +1782,8 @@ static void nvme_init_integrity(struct gendisk *disk, struct nvme_ns_head *head) integrity.tuple_size = head->ms; integrity.pi_offset = head->pi_offset;blk_integrity_register (disk, &integrity); +return true; } -#else -static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) -{ -} -#endif /* CONFIG_BLK_DEV_INTEGRITY */ static void nvme_config_discard (struct nvme_ctrl *ctrl,struct gendisk *disk,struct nvme_ns_head *head)
@@ -2012,13 +2018,8 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,* I/O to namespaces with metadata except when the namespace supports * PI, as it can strip/insert in that case. */ -if (head->ms) { -if (IS_ENABLED (CONFIG_BLK_DEV_INTEGRITY ) && - (head->features &NVME_NS_METADATA_SUPPORTED )) -nvme_init_integrity (disk, head); -else if (!nvme_ns_has_pi (head)) - capacity =0 ; - } +if (!nvme_init_integrity (disk, head)) + capacity =0 ;set_capacity_and_notify (disk, capacity);
Signature · 1 line
2.39.2
Re: [PATCH 07/21] nvme: cleanup the nvme_init_integrity calling conventions
Quoted 57 lines · Christoph Hellwig
Handle the no metadata support case in
nvme_init_integrityas well to simplify the calling convention and prepare for future changes in the area.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 29 +++++++++++++++-------------- 1 file changed, 15 insertions(+), 14 deletions(-)drivers/nvme/host/core.c +13
-9@@ -1723,11 +1723,21 @@ int nvme_getgeo(struct block_device *bdev, struct hd_geometry *geo)return 0 ; } -#ifdef CONFIG_BLK_DEV_INTEGRITY -static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) +static bool nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) {struct blk_integrity integrity = { }; +if (!head->ms) +return true; + +/* +* PI can always be supported as we can ask the controller to simply +* insert/strip it, which is not possible for other kinds of metadata. +*/ +if (!IS_ENABLED (CONFIG_BLK_DEV_INTEGRITY ) || + !(head->features &NVME_NS_METADATA_SUPPORTED )) +return nvme_ns_has_pi (head); +switch (head->pi_type) {case NVME_NS_DPS_PI_TYPE3 :switch (head->guard_type) {@@ -1772,12 +1782,8 @@ static void nvme_init_integrity(struct gendisk *disk, struct nvme_ns_head *head) integrity.tuple_size = head->ms; integrity.pi_offset = head->pi_offset;blk_integrity_register (disk, &integrity); +return true; } -#else -static void nvme_init_integrity (struct gendisk *disk,struct nvme_ns_head *head) -{ -} -#endif /* CONFIG_BLK_DEV_INTEGRITY */ static void nvme_config_discard (struct nvme_ctrl *ctrl,struct gendisk *disk,struct nvme_ns_head *head)@@ -2012,13 +2018,8 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,* I/O to namespaces with metadata except when the namespace supports * PI, as it can strip/insert in that case. */ -if (head->ms) { - if (IS_ENABLED(CONFIG_BLK_DEV_INTEGRITY) &&
drivers/nvme/host/core.c static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk, · line 2025
Quoted 10 lines
drivers/nvme/host/core.c +2
-5- (head->features & NVME_NS_METADATA_SUPPORTED)) -nvme_init_integrity (disk, head); - else if (!nvme_ns_has_pi(head)) - capacity =0 ; - } +if (!nvme_init_integrity (disk, head)) + capacity =0 ;set_capacity_and_notify (disk, capacity);
Looks good,
Christoph Hellwig
maintainerPATCH 8/21 nvme: move blk_integrity_unregister into nvme_init_integrity
In reply to Christoph HellwigMove uneregistering the existing integrity profile into the helper dealing with all the other integrity / metadata setup.
--- drivers/nvme/host/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/nvme/host/core.c +2 -2
@@ -1727,6 +1727,8 @@ static bool nvme_init_integrity(struct gendisk *disk, struct nvme_ns_head *head) {struct blk_integrity integrity = { }; +blk_integrity_unregister (disk); +if (!head->ms)return true;
@@ -1980,8 +1982,6 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk, bs = (1 <<9 ); } -blk_integrity_unregister (disk); - atomic_bs = phys_bs = bs;if (id->nabo ==0 ) {/*
Signature · 1 line
2.39.2
Re: [PATCH 08/21] nvme: move blk_integrity_unregister into nvme_init_integrity
Quoted 22 lines · Christoph Hellwig
Move uneregistering the existing integrity profile into the helper dealing with all the other integrity / metadata setup.
Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/nvme/host/core.c +2
-0@@ -1727,6 +1727,8 @@ static bool nvme_init_integrity(struct gendisk *disk, struct nvme_ns_head *head) {struct blk_integrity integrity = { }; +blk_integrity_unregister (disk); +if (!head->ms)return true;
drivers/nvme/host/core.c static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk, · line 1987
Quoted 8 lines
drivers/nvme/host/core.c +0
-2@@ -1980,8 +1982,6 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk, bs = (1 <<9 ); } -blk_integrity_unregister (disk); - atomic_bs = phys_bs = bs;if (id->nabo ==0 ) {/*
Looks good,
Christoph Hellwig
maintainerPATCH 9/21 nvme: don't use nvme_update_disk_info for the multipath disk
In reply to Christoph HellwigCurrently nvme_update_ns_info_block calls nvme_update_disk_info both for the namespace attached disk, and the multipath one (if it exists). This is very different from how other stacking drivers work, and leads to a lot of complexity.
Switch to setting the disk capacity and initializing the integrity profile, and let blk_stack_limits which already is called just below deal with updating the other limits.
--- drivers/nvme/host/core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/nvme/host/core.c +2 -1
@@ -2159,7 +2159,8 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,if (nvme_ns_head_multipath (ns->head)) {blk_mq_freeze_queue (ns->head->disk->queue); -nvme_update_disk_info (ns->ctrl, ns->head->disk, ns->head, id); +nvme_init_integrity (ns->head->disk, ns->head); +set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk));set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info));nvme_mpath_revalidate_paths (ns); blk_stack_limits(&ns->head->disk->queue->limits,
Signature · 1 line
2.39.2
Re: [PATCH 09/21] nvme: don't use nvme_update_disk_info for the multipath disk
Quoted 8 lines · Christoph Hellwig
Currently
nvme_update_ns_info_blockcallsnvme_update_disk_infoboth for the namespace attached disk, and the multipath one (if it exists). This is very different from how other stacking drivers work, and leads to a lot of complexity.Switch to setting the disk capacity and initializing the integrity profile, and let
blk_stack_limitswhich already is called just below deal with updating the other limits.
where does the blk_stack_limits called in the nvme/core ? Is it part of some preparation series ?
Quoted 20 lines
Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)drivers/nvme/host/core.c +2
-1@@ -2159,7 +2159,8 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,if (nvme_ns_head_multipath (ns->head)) {blk_mq_freeze_queue (ns->head->disk->queue); -nvme_update_disk_info (ns->ctrl, ns->head->disk, ns->head, id); +nvme_init_integrity (ns->head->disk, ns->head); +set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk));set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info));nvme_mpath_revalidate_paths (ns); blk_stack_limits(&ns->head->disk->queue->limits,
Re: [PATCH 09/21] nvme: don't use nvme_update_disk_info for the multipath disk
Quoted 8 lines · Max Gurtovoy
Quoted 5 lines · Christoph Hellwig
Currently
nvme_update_ns_info_blockcallsnvme_update_disk_infoboth for the namespace attached disk, and the multipath one (if it exists). This is very different from how other stacking drivers work, and leads to a lot of complexity.
Quoted 6 lines
Quoted 3 lines
Switch to setting the disk capacity and initializing the integrity profile, and let
blk_stack_limitswhich already is called just below deal with updating the other limits.where does the
blk_stack_limitscalled in the nvme/core ? Is it part of some preparation series ?
please ignore, I missed the last line in the chunk..
Quoted 22 lines
Quoted 21 lines
Signed-off-by Christoph Hellwig <[email protected]>---
drivers/nvme/host/core.c| 3 ++-1 file changed, 2 insertions(+), 1 deletion(-)drivers/nvme/host/core.c +2
-1@@ -2159,7 +2159,8 @@ static int nvme_update_ns_info_block(structnvme_ns *ns, if (nvme_ns_head_multipath(ns->head)) { blk_mq_freeze_queue(ns->head->disk->queue); - nvme_update_disk_info(ns->ctrl, ns->head->disk, ns->head, id); + nvme_init_integrity(ns->head->disk, ns->head); + set_capacity_and_notify(ns->head->disk, get_capacity(ns->disk)); set_disk_ro(ns->head->disk, nvme_ns_is_readonly(ns, info)); nvme_mpath_revalidate_paths(ns); blk_stack_limits(&ns->head->disk->queue->limits,
Christoph Hellwig
maintainerPATCH 10/21 nvme: move a few things out of nvme_update_disk_info
In reply to Christoph HellwigMove setting up the integrity profile and setting the disk capacity out of nvme_update_disk_info to get nvme_update_disk_info into a shape where it just sets queue_limits eventually.
--- drivers/nvme/host/core.c | 46 +++++++++++++++++++++++----------------- 1 file changed, 26 insertions(+), 20 deletions(-)
drivers/nvme/host/core.c +26 -20
@@ -1965,12 +1965,13 @@ static void nvme_set_queue_limits(struct nvme_ctrl *ctrl,blk_queue_write_cache (q, vwc, vwc); } -static void nvme_update_disk_info (struct nvme_ctrl *ctrl,struct gendisk *disk, -struct nvme_ns_head *head,struct nvme_id_ns *id) +static bool nvme_update_disk_info (struct nvme_ns *ns,struct nvme_id_ns *id) { -sector_t capacity =nvme_lba_to_sect (head,le64_to_cpu (id->nsze)); +struct gendisk *disk = ns->disk; +struct nvme_ns_head *head = ns->head;u32 bs =1U << head->lba_shift;u32 atomic_bs, phys_bs, io_opt =0 ; +bool valid = true;/* * The block layer can't support LBA sizes larger than the page size
@@ -1978,8 +1979,8 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,* allow block I/O. */ if (head->lba_shift >PAGE_SHIFT || head->lba_shift <SECTOR_SHIFT ) { - capacity =0 ; bs = (1 <<9 ); + valid = false; } atomic_bs = phys_bs = bs;
@@ -1992,7 +1993,7 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,if (id->nsfeat &NVME_NS_FEAT_ATOMICS && id->nawupf) atomic_bs = (1 +le16_to_cpu (id->nawupf)) * bs;else - atomic_bs = (1 + ctrl->subsys->awupf) * bs; + atomic_bs = (1 + ns->ctrl->subsys->awupf) * bs; }if (id->nsfeat &NVME_NS_FEAT_IO_OPT ) {
@@ -2012,24 +2013,14 @@ static void nvme_update_disk_info(struct nvme_ctrl *ctrl, struct gendisk *disk,blk_queue_io_min (disk->queue, phys_bs);blk_queue_io_opt (disk->queue, io_opt); -/* -* Register a metadata profile for PI, or the plain non-integrity NVMe -* metadata masquerading as Type 0 if supported, otherwise reject block -* I/O to namespaces with metadata except when the namespace supports -* PI, as it can strip/insert in that case. -*/ -if (!nvme_init_integrity (disk, head)) - capacity =0 ; - -set_capacity_and_notify (disk, capacity); - -nvme_config_discard (ctrl, disk, head); +nvme_config_discard (ns->ctrl, disk, head); -if (ctrl->quirks &NVME_QUIRK_DEALLOCATE_ZEROES ) +if (ns->ctrl->quirks &NVME_QUIRK_DEALLOCATE_ZEROES )blk_queue_max_write_zeroes_sectors (disk->queue,UINT_MAX );else blk_queue_max_write_zeroes_sectors (disk->queue, - ctrl->max_zeroes_sectors); + ns->ctrl->max_zeroes_sectors); +return valid; }static bool nvme_ns_is_readonly (struct nvme_ns *ns,struct nvme_ns_info *info)
@@ -2103,6 +2094,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, struct nvme_ns_info *info) {struct nvme_id_ns *id; +sector_t capacity;unsigned lbaf;int ret;
@@ -2121,6 +2113,8 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, lbaf =nvme_lbaf_index (id->flbas); ns->head->lba_shift = id->lbaf[lbaf].ds; ns->head->nuse =le64_to_cpu (id->nuse); + capacity =nvme_lba_to_sect (ns->head,le64_to_cpu (id->nsze)); +nvme_set_queue_limits (ns->ctrl, ns->queue); ret =nvme_configure_metadata (ns->ctrl, ns->head, id);
@@ -2129,7 +2123,19 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, goto out; }nvme_set_chunk_sectors (ns, id); -nvme_update_disk_info (ns->ctrl, ns->disk, ns->head, id); +if (!nvme_update_disk_info (ns, id)) + capacity =0 ; + +/* +* Register a metadata profile for PI, or the plain non-integrity NVMe +* metadata masquerading as Type 0 if supported, otherwise reject block +* I/O to namespaces with metadata except when the namespace supports +* PI, as it can strip/insert in that case. +*/ +if (!nvme_init_integrity (ns->disk, ns->head)) + capacity =0 ; + +set_capacity_and_notify (ns->disk, capacity);if (ns->head->ids.csi ==NVME_CSI_ZNS ) { ret =nvme_update_zone_info (ns, lbaf);
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 11/21 nvme: move setting the write cache flags out of nvme_set_queue_limits
In reply to Christoph Hellwignvme_set_queue_limits is used on the admin queue and all gendisks including hidden ones that don't support block I/O. The write cache setting on the other hand only makes sense for block I/O. Move the blk_queue_write_cache call to nvme_update_ns_info_block instead.
--- drivers/nvme/host/core.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/nvme/host/core.c +2 -3
@@ -1954,15 +1954,12 @@ static u32 nvme_max_drv_segments(struct nvme_ctrl *ctrl)static void nvme_set_queue_limits (struct nvme_ctrl *ctrl,struct request_queue *q) { -bool vwc = ctrl->vwc &NVME_CTRL_VWC_PRESENT ; -blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors);blk_queue_max_segments (q,min_t (u32,USHRT_MAX ,min_not_zero (nvme_max_drv_segments (ctrl), ctrl->max_segments)));blk_queue_max_integrity_segments (q, ctrl->max_integrity_segments);blk_queue_virt_boundary (q,NVME_CTRL_PAGE_SIZE -1 );blk_queue_dma_alignment (q,3 ); -blk_queue_write_cache (q, vwc, vwc); }static bool nvme_update_disk_info (struct nvme_ns *ns,struct nvme_id_ns *id)
@@ -2093,6 +2090,7 @@ static int nvme_update_ns_info_generic(struct nvme_ns *ns,static int nvme_update_ns_info_block (struct nvme_ns *ns,struct nvme_ns_info *info) { +bool vwc = ns->ctrl->vwc &NVME_CTRL_VWC_PRESENT ;struct nvme_id_ns *id;sector_t capacity;unsigned lbaf;
@@ -2154,6 +2152,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,if ((id->dlfeat &0x7 ) ==0x1 && (id->dlfeat & (1 <<3 ))) ns->head->features |=NVME_NS_DEAC ;set_disk_ro (ns->disk,nvme_ns_is_readonly (ns, info)); +blk_queue_write_cache (ns->disk->queue, vwc, vwc);set_bit (NVME_NS_READY , &ns->flags);blk_mq_unfreeze_queue (ns->disk->queue);
Signature · 1 line
2.39.2
Re: [PATCH 11/21] nvme: move setting the write cache flags out of nvme_set_queue_limits
Quoted 39 lines · Christoph Hellwig
nvme_set_queue_limitsis used on the admin queue and all gendisks including hidden ones that don't support block I/O. The write cache setting on the other hand only makes sense for block I/O. Move theblk_queue_write_cachecall tonvme_update_ns_info_blockinstead.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)drivers/nvme/host/core.c +1
-3@@ -1954,15 +1954,12 @@ static u32 nvme_max_drv_segments(struct nvme_ctrl *ctrl)static void nvme_set_queue_limits (struct nvme_ctrl *ctrl,struct request_queue *q) { -bool vwc = ctrl->vwc &NVME_CTRL_VWC_PRESENT ; -blk_queue_max_hw_sectors (q, ctrl->max_hw_sectors);blk_queue_max_segments (q,min_t (u32,USHRT_MAX ,min_not_zero (nvme_max_drv_segments (ctrl), ctrl->max_segments)));blk_queue_max_integrity_segments (q, ctrl->max_integrity_segments);blk_queue_virt_boundary (q,NVME_CTRL_PAGE_SIZE -1 );blk_queue_dma_alignment (q,3 ); -blk_queue_write_cache (q, vwc, vwc); }static bool nvme_update_disk_info (struct nvme_ns *ns,struct nvme_id_ns *id)@@ -2093,6 +2090,7 @@ static int nvme_update_ns_info_generic(struct nvme_ns *ns,static int nvme_update_ns_info_block (struct nvme_ns *ns,struct nvme_ns_info *info) { +bool vwc = ns->ctrl->vwc &NVME_CTRL_VWC_PRESENT ;struct nvme_id_ns *id;sector_t capacity;unsigned lbaf;
drivers/nvme/host/core.c static int nvme_update_ns_info_block(struct nvme_ns *ns, · line 2157
Quoted 6 lines
drivers/nvme/host/core.c +1
-0@@ -2154,6 +2152,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,if ((id->dlfeat &0x7 ) ==0x1 && (id->dlfeat & (1 <<3 ))) ns->head->features |=NVME_NS_DEAC ;set_disk_ro (ns->disk,nvme_ns_is_readonly (ns, info)); +blk_queue_write_cache (ns->disk->queue, vwc, vwc);set_bit (NVME_NS_READY , &ns->flags);blk_mq_unfreeze_queue (ns->disk->queue);
Looks good,
Quoted 1 line
Christoph Hellwig
maintainerPATCH 12/21 nvme: move common logic into nvme_update_ns_info
In reply to Christoph Hellwignvme_update_ns_info_generic and nvme_update_ns_info_block share a fair amount of logic related to not fully supported namespace formats and updating the multipath information. Move this logic into the common caller.
--- drivers/nvme/host/core.c | 79 +++++++++++++++++++--------------------- 1 file changed, 37 insertions(+), 42 deletions(-)
drivers/nvme/host/core.c +37 -42
@@ -2070,21 +2070,8 @@ static int nvme_update_ns_info_generic(struct nvme_ns *ns,set_disk_ro (ns->disk,nvme_ns_is_readonly (ns, info));blk_mq_unfreeze_queue (ns->disk->queue); -if (nvme_ns_head_multipath (ns->head)) { -blk_mq_freeze_queue (ns->head->disk->queue); -set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); -nvme_mpath_revalidate_paths (ns); -blk_stack_limits (&ns->head->disk->queue->limits, - &ns->queue->limits,0 ); - ns->head->disk->flags |=GENHD_FL_HIDDEN ; -blk_mq_unfreeze_queue (ns->head->disk->queue); - } -/* Hide the block-interface for these devices */ - ns->disk->flags |=GENHD_FL_HIDDEN ; -set_bit (NVME_NS_READY , &ns->flags); - -return 0 ; +return -ENODEV ; } static int nvme_update_ns_info_block(struct nvme_ns *ns,
@@ -2104,7 +2091,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,/* namespace not allocated or attached */ info->is_removed = true; ret = -ENODEV ; - goto error; + goto out; }blk_mq_freeze_queue (ns->disk->queue);
@@ -2162,54 +2149,62 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, goto out; } -if (nvme_ns_head_multipath (ns->head)) { -blk_mq_freeze_queue (ns->head->disk->queue); -nvme_init_integrity (ns->head->disk, ns->head); -set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk)); -set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); -nvme_mpath_revalidate_paths (ns); -blk_stack_limits (&ns->head->disk->queue->limits, - &ns->queue->limits,0 ); -disk_update_readahead (ns->head->disk); -blk_mq_unfreeze_queue (ns->head->disk->queue); - } - ret =0 ; out: -/* -* If probing fails due an unsupported feature, hide the block device, -* but still allow other access. -*/ -if (ret == -ENODEV ) { - ns->disk->flags |=GENHD_FL_HIDDEN ; -set_bit (NVME_NS_READY , &ns->flags); - ret =0 ; - } - -error:kfree (id);return ret; }static int nvme_update_ns_info (struct nvme_ns *ns,struct nvme_ns_info *info) { +int ret; +switch (info->ids.csi) {case NVME_CSI_ZNS :if (!IS_ENABLED (CONFIG_BLK_DEV_ZONED )) {dev_info (ns->ctrl->device,"block device for nsid %u not supported without CONFIG_BLK_DEV_ZONED\n" , info->nsid); -return nvme_update_ns_info_generic (ns, info); + ret =nvme_update_ns_info_generic (ns, info); +break ; } -return nvme_update_ns_info_block (ns, info); + ret =nvme_update_ns_info_block (ns, info); +break ;case NVME_CSI_NVM : -return nvme_update_ns_info_block (ns, info); + ret =nvme_update_ns_info_block (ns, info); +break ;default :dev_info (ns->ctrl->device,"block device for nsid %u not supported (csi %u)\n" , info->nsid, info->ids.csi); -return nvme_update_ns_info_generic (ns, info); + ret =nvme_update_ns_info_generic (ns, info); +break ; + } + +/* +* If probing fails due an unsupported feature, hide the block device, +* but still allow other access. +*/ +if (ret == -ENODEV ) { + ns->disk->flags |=GENHD_FL_HIDDEN ; +set_bit (NVME_NS_READY , &ns->flags); + ret =0 ; + } + +if (!ret &&nvme_ns_head_multipath (ns->head)) { +blk_mq_freeze_queue (ns->head->disk->queue); +if (!(ns->disk->flags &GENHD_FL_HIDDEN )) +nvme_init_integrity (ns->head->disk, ns->head); +set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk)); +set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); +nvme_mpath_revalidate_paths (ns); +blk_stack_limits (&ns->head->disk->queue->limits, + &ns->queue->limits,0 ); +disk_update_readahead (ns->head->disk); +blk_mq_unfreeze_queue (ns->head->disk->queue); } + +return ret; }#ifdef CONFIG_BLK_SED_OPAL
Signature · 1 line
2.39.2
Re: [PATCH 12/21] nvme: move common logic into nvme_update_ns_info
Quoted 113 lines · Christoph Hellwig
nvme_update_ns_info_genericandnvme_update_ns_info_blockshare a fair amount of logic related to not fully supported namespace formats and updating the multipath information. Move this logic into the common caller.Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 79 +++++++++++++++++++--------------------- 1 file changed, 37 insertions(+), 42 deletions(-)drivers/nvme/host/core.c +18
-42@@ -2070,21 +2070,8 @@ static int nvme_update_ns_info_generic(struct nvme_ns *ns,set_disk_ro (ns->disk,nvme_ns_is_readonly (ns, info));blk_mq_unfreeze_queue (ns->disk->queue); -if (nvme_ns_head_multipath (ns->head)) { -blk_mq_freeze_queue (ns->head->disk->queue); -set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); -nvme_mpath_revalidate_paths (ns); -blk_stack_limits (&ns->head->disk->queue->limits, - &ns->queue->limits,0 ); - ns->head->disk->flags |=GENHD_FL_HIDDEN ; -blk_mq_unfreeze_queue (ns->head->disk->queue); - } -/* Hide the block-interface for these devices */ - ns->disk->flags |=GENHD_FL_HIDDEN ; -set_bit (NVME_NS_READY , &ns->flags); - -return 0 ; +return -ENODEV ; } static int nvme_update_ns_info_block(struct nvme_ns *ns,@@ -2104,7 +2091,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns,/* namespace not allocated or attached */ info->is_removed = true; ret = -ENODEV ; - goto error; + goto out; }blk_mq_freeze_queue (ns->disk->queue);@@ -2162,54 +2149,62 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, goto out; } -if (nvme_ns_head_multipath (ns->head)) { -blk_mq_freeze_queue (ns->head->disk->queue); -nvme_init_integrity (ns->head->disk, ns->head); -set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk)); -set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); -nvme_mpath_revalidate_paths (ns); -blk_stack_limits (&ns->head->disk->queue->limits, - &ns->queue->limits,0 ); -disk_update_readahead (ns->head->disk); -blk_mq_unfreeze_queue (ns->head->disk->queue); - } - ret =0 ; out: -/* -* If probing fails due an unsupported feature, hide the block device, -* but still allow other access. -*/ -if (ret == -ENODEV ) { - ns->disk->flags |=GENHD_FL_HIDDEN ; -set_bit (NVME_NS_READY , &ns->flags); - ret =0 ; - } - -error:kfree (id);return ret; }static int nvme_update_ns_info (struct nvme_ns *ns,struct nvme_ns_info *info) { +int ret; +switch (info->ids.csi) {case NVME_CSI_ZNS :if (!IS_ENABLED (CONFIG_BLK_DEV_ZONED )) {dev_info (ns->ctrl->device,"block device for nsid %u not supported without CONFIG_BLK_DEV_ZONED\n" , info->nsid); -return nvme_update_ns_info_generic (ns, info); + ret =nvme_update_ns_info_generic (ns, info); +break ; } -return nvme_update_ns_info_block (ns, info); + ret =nvme_update_ns_info_block (ns, info); +break ;case NVME_CSI_NVM : -return nvme_update_ns_info_block (ns, info); + ret =nvme_update_ns_info_block (ns, info); +break ;default :dev_info (ns->ctrl->device,"block device for nsid %u not supported (csi %u)\n" , info->nsid, info->ids.csi); -return nvme_update_ns_info_generic (ns, info); + ret =nvme_update_ns_info_generic (ns, info); +break ; + } + +/* +* If probing fails due an unsupported feature, hide the block device, +* but still allow other access. +*/
drivers/nvme/host/core.c static int nvme_update_ns_info_block(struct nvme_ns *ns, · line 2197
Quoted 10 lines
drivers/nvme/host/core.c +10
-0+if (ret == -ENODEV ) { + ns->disk->flags |=GENHD_FL_HIDDEN ; +set_bit (NVME_NS_READY , &ns->flags); + ret =0 ; + } + +if (!ret &&nvme_ns_head_multipath (ns->head)) { +blk_mq_freeze_queue (ns->head->disk->queue); +if (!(ns->disk->flags &GENHD_FL_HIDDEN )) +nvme_init_integrity (ns->head->disk, ns->head);
the logic in "nvme_update_ns_info_generic()" today is to hide also ns->head->disk..
drivers/nvme/host/core.c static int nvme_update_ns_info_block(struct nvme_ns *ns, · line 2203
Quoted 6 lines
drivers/nvme/host/core.c +6
-0+set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk)); +set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); +nvme_mpath_revalidate_paths (ns); +blk_stack_limits (&ns->head->disk->queue->limits, + &ns->queue->limits,0 ); +disk_update_readahead (ns->head->disk);
we don't call disk_update_readahead() in "nvme_update_ns_info_generic()" today..
Quoted 7 lines
drivers/nvme/host/core.c +3
-0+blk_mq_unfreeze_queue (ns->head->disk->queue); } + +return ret; }#ifdef CONFIG_BLK_SED_OPAL
Christoph Hellwig
maintainerRe: [PATCH 12/21] nvme: move common logic into nvme_update_ns_info
Quoted 9 lines · Max Gurtovoy
Quoted 9 lines
drivers/nvme/host/core.c +9
-0+/* +* If probing fails due an unsupported feature, hide the block device, +* but still allow other access. +*/ +if (ret == -ENODEV ) { + ns->disk->flags |=GENHD_FL_HIDDEN ; +set_bit (NVME_NS_READY , &ns->flags); + ret =0 ; + }
Quoted 8 lines
drivers/nvme/host/core.c static int nvme_update_ns_info_block(struct nvme_ns *ns, · line 2197
Quoted 5 lines
drivers/nvme/host/core.c +5
-0+ +if (!ret &&nvme_ns_head_multipath (ns->head)) { +blk_mq_freeze_queue (ns->head->disk->queue); +if (!(ns->disk->flags &GENHD_FL_HIDDEN )) +nvme_init_integrity (ns->head->disk, ns->head);the logic in "
nvme_update_ns_info_generic()" today is to hide alsons->head->disk..
Indeed, that got lost.
Quoted 11 lines
drivers/nvme/host/core.c static int nvme_update_ns_info_block(struct nvme_ns *ns, · line 2203
Quoted 8 lines
drivers/nvme/host/core.c +6
-0+set_capacity_and_notify (ns->head->disk,get_capacity (ns->disk)); +set_disk_ro (ns->head->disk,nvme_ns_is_readonly (ns, info)); +nvme_mpath_revalidate_paths (ns); +blk_stack_limits (&ns->head->disk->queue->limits, + &ns->queue->limits,0 ); +disk_update_readahead (ns->head->disk);we don't call
disk_update_readahead()in "nvme_update_ns_info_generic()" today..
Yes, but it's harmless as it just propagates the limits from the queue_limits to the bdi. That beeing said I've failed to send out a last patch that actually removes it by switching to the queue_limits_set API for the multipath node as well.
Christoph Hellwig
maintainerPATCH 13/21 nvme: split out a nvme_identify_ns_nvm helper
In reply to Christoph HellwigSplit the logic to query the Identify Namespace Data Structure, NVM Command Set into a separate helper.
--- drivers/nvme/host/core.c | 38 ++++++++++++++++++++++++++------------ 1 file changed, 26 insertions(+), 12 deletions(-)
drivers/nvme/host/core.c +26 -12
@@ -1831,12 +1831,35 @@ static bool nvme_ns_ids_equal(struct nvme_ns_ids *a, struct nvme_ns_ids *b) a->csi == b->csi; } +static int nvme_identify_ns_nvm (struct nvme_ctrl *ctrl,unsigned int nsid, +struct nvme_id_ns_nvm **nvmp) +{ +struct nvme_command c = { + .identify.opcode = nvme_admin_identify, + .identify.nsid =cpu_to_le32 (nsid), + .identify.cns =NVME_ID_CNS_CS_NS , + .identify.csi =NVME_CSI_NVM , + }; +struct nvme_id_ns_nvm *nvm; +int ret; + + nvm =kzalloc (sizeof (*nvm),GFP_KERNEL ); +if (!nvm) +return -ENOMEM ; + + ret =nvme_submit_sync_cmd (ctrl->admin_q, &c, nvm,sizeof (*nvm)); +if (ret) +kfree (nvm); +else + *nvmp = nvm; +return ret; +} +static int nvme_init_ms (struct nvme_ctrl *ctrl,struct nvme_ns_head *head,struct nvme_id_ns *id) {bool first = id->dps &NVME_NS_DPS_PI_FIRST ;unsigned lbaf =nvme_lbaf_index (id->flbas); -struct nvme_command c = { };struct nvme_id_ns_nvm *nvm;int ret =0 ;u32 elbaf;
@@ -1849,18 +1872,9 @@ static int nvme_init_ms(struct nvme_ctrl *ctrl, struct nvme_ns_head *head, goto set_pi; } - nvm =kzalloc (sizeof (*nvm),GFP_KERNEL ); -if (!nvm) -return -ENOMEM ; - - c.identify.opcode = nvme_admin_identify; - c.identify.nsid =cpu_to_le32 (head->ns_id); - c.identify.cns =NVME_ID_CNS_CS_NS ; - c.identify.csi =NVME_CSI_NVM ; - - ret =nvme_submit_sync_cmd (ctrl->admin_q, &c, nvm,sizeof (*nvm)); + ret =nvme_identify_ns_nvm (ctrl, head->ns_id, &nvm);if (ret) - goto free_data; + goto set_pi; elbaf =le32_to_cpu (nvm->elbaf[lbaf]);
Signature · 1 line
2.39.2
Re: [PATCH 13/21] nvme: split out a nvme_identify_ns_nvm helper
Quoted 59 lines · Christoph Hellwig
Split the logic to query the Identify Namespace Data Structure, NVM Command Set into a separate helper.
Signed-off-by Christoph Hellwig <[email protected]>--- drivers/nvme/host/core.c | 38 ++++++++++++++++++++++++++------------ 1 file changed, 26 insertions(+), 12 deletions(-)drivers/nvme/host/core.c +24
-7@@ -1831,12 +1831,35 @@ static bool nvme_ns_ids_equal(struct nvme_ns_ids *a, struct nvme_ns_ids *b) a->csi == b->csi; } +static int nvme_identify_ns_nvm (struct nvme_ctrl *ctrl,unsigned int nsid, +struct nvme_id_ns_nvm **nvmp) +{ +struct nvme_command c = { + .identify.opcode = nvme_admin_identify, + .identify.nsid =cpu_to_le32 (nsid), + .identify.cns =NVME_ID_CNS_CS_NS , + .identify.csi =NVME_CSI_NVM , + }; +struct nvme_id_ns_nvm *nvm; +int ret; + + nvm =kzalloc (sizeof (*nvm),GFP_KERNEL ); +if (!nvm) +return -ENOMEM ; + + ret =nvme_submit_sync_cmd (ctrl->admin_q, &c, nvm,sizeof (*nvm)); +if (ret) +kfree (nvm); +else + *nvmp = nvm; +return ret; +} +static int nvme_init_ms (struct nvme_ctrl *ctrl,struct nvme_ns_head *head,struct nvme_id_ns *id) {bool first = id->dps &NVME_NS_DPS_PI_FIRST ;unsigned lbaf =nvme_lbaf_index (id->flbas); -struct nvme_command c = { };struct nvme_id_ns_nvm *nvm;int ret =0 ;u32 elbaf;@@ -1849,18 +1872,9 @@ static int nvme_init_ms(struct nvme_ctrl *ctrl, struct nvme_ns_head *head, goto set_pi; } - nvm =kzalloc (sizeof (*nvm),GFP_KERNEL ); -if (!nvm) -return -ENOMEM ; - - c.identify.opcode = nvme_admin_identify; - c.identify.nsid =cpu_to_le32 (head->ns_id);
drivers/nvme/host/core.c static int nvme_init_ms(struct nvme_ctrl *ctrl, struct nvme_ns_head *head, · line 1879
Quoted 10 lines
drivers/nvme/host/core.c +2
-5- c.identify.cns =NVME_ID_CNS_CS_NS ; - c.identify.csi =NVME_CSI_NVM ; - - ret =nvme_submit_sync_cmd (ctrl->admin_q, &c, nvm,sizeof (*nvm)); + ret =nvme_identify_ns_nvm (ctrl, head->ns_id, &nvm);if (ret) - goto free_data; + goto set_pi; elbaf =le32_to_cpu (nvm->elbaf[lbaf]);
I actually had a similar logic in one of the patches of the PI series I've sent :)
BTW, we need only the elbaf field from the nvm response structure so we can ease the code a bit and free the nvm inside the nvme_identify_ns_nvm() instead of in the caller...
Otherwise, looks good
Quoted 1 line
Christoph Hellwig
maintainerRe: [PATCH 13/21] nvme: split out a nvme_identify_ns_nvm helper
Quoted 5 lines · Max Gurtovoy
I actually had a similar logic in one of the patches of the PI series I've sent :)
BTW, we need only the elbaf field from the nvm response structure so we can ease the code a bit and free the nvm inside thenvme_identify_ns_nvm()instead of in the caller...
Ithought about that, and while it might be a tad cleaner currently, as soon as the identify_ns_nvm structure grows more fields that we'll use that'll get uglier.
Christoph Hellwig
maintainerPATCH 14/21 nvme: don't query identify data in configure_metadata
In reply to Christoph HellwigMove reading the Identify Namespace Data Structure, NVM Command Set out of configure_metadata into the caller. This allows doing the identify call outside the frozen I/O queues, and prepares for using data from the Identify data structure for other purposes.
--- drivers/nvme/host/core.c | 49 ++++++++++++++++------------------------ 1 file changed, 19 insertions(+), 30 deletions(-)
drivers/nvme/host/core.c +19 -30
@@ -1855,32 +1855,26 @@ static int nvme_identify_ns_nvm(struct nvme_ctrl *ctrl, unsigned int nsid,return ret; } -static int nvme_init_ms (struct nvme_ctrl *ctrl,struct nvme_ns_head *head, -struct nvme_id_ns *id) +static void nvme_init_ms (struct nvme_ctrl *ctrl,struct nvme_ns_head *head, +struct nvme_id_ns *id,struct nvme_id_ns_nvm *nvm) {bool first = id->dps &NVME_NS_DPS_PI_FIRST ;unsigned lbaf =nvme_lbaf_index (id->flbas); -struct nvme_id_ns_nvm *nvm; -int ret =0 ;u32 elbaf; head->pi_size =0 ; head->ms =le16_to_cpu (id->lbaf[lbaf].ms); -if (!(ctrl->ctratt &NVME_CTRL_ATTR_ELBAS )) { +if (!nvm || !(ctrl->ctratt &NVME_CTRL_ATTR_ELBAS )) { head->pi_size =sizeof (struct t10_pi_tuple ); head->guard_type =NVME_NVM_NS_16B_GUARD ; goto set_pi; } - ret =nvme_identify_ns_nvm (ctrl, head->ns_id, &nvm); -if (ret) - goto set_pi; - elbaf =le32_to_cpu (nvm->elbaf[lbaf]);/* no support for storage tag formats right now */ if (nvme_elbaf_sts (elbaf)) - goto free_data; + goto set_pi; head->guard_type =nvme_elbaf_guard_type (elbaf);switch (head->guard_type) {
@@ -1894,8 +1888,6 @@ static int nvme_init_ms(struct nvme_ctrl *ctrl, struct nvme_ns_head *head,break ; } -free_data: -kfree (nvm); set_pi:if (head->pi_size && head->ms >= head->pi_size) head->pi_type = id->dps &NVME_NS_DPS_PI_MASK ;
@@ -1906,22 +1898,17 @@ static int nvme_init_ms(struct nvme_ctrl *ctrl, struct nvme_ns_head *head, head->pi_offset =0 ; else head->pi_offset = head->ms - head->pi_size; - -return ret; } -static int nvme_configure_metadata (struct nvme_ctrl *ctrl, -struct nvme_ns_head *head,struct nvme_id_ns *id) +static void nvme_configure_metadata (struct nvme_ctrl *ctrl, +struct nvme_ns_head *head,struct nvme_id_ns *id, +struct nvme_id_ns_nvm *nvm) { -int ret; - - ret =nvme_init_ms (ctrl, head, id); -if (ret) -return ret; +nvme_init_ms (ctrl, head, id, nvm); head->features &= ~(NVME_NS_METADATA_SUPPORTED |NVME_NS_EXT_LBAS );if (!head->ms || !(ctrl->ops->flags &NVME_F_METADATA_SUPPORTED )) -return 0 ; +return ;if (ctrl->ops->flags &NVME_F_FABRICS ) {/*
@@ -1930,7 +1917,7 @@ static int nvme_configure_metadata(struct nvme_ctrl *ctrl,* remap the separate metadata buffer from the block layer. */ if (WARN_ON_ONCE (!(id->flbas &NVME_NS_FLBAS_META_EXT ))) -return 0 ; +return ; head->features |=NVME_NS_EXT_LBAS ;
@@ -1957,7 +1944,6 @@ static int nvme_configure_metadata(struct nvme_ctrl *ctrl,else head->features |=NVME_NS_METADATA_SUPPORTED ; } -return 0 ; }static u32 nvme_max_drv_segments (struct nvme_ctrl *ctrl)
@@ -2092,6 +2078,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, struct nvme_ns_info *info) {bool vwc = ns->ctrl->vwc &NVME_CTRL_VWC_PRESENT ; +struct nvme_id_ns_nvm *nvm =NULL ;struct nvme_id_ns *id;sector_t capacity;unsigned lbaf;
@@ -2108,6 +2095,12 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, goto out; } +if (ns->ctrl->ctratt &NVME_CTRL_ATTR_ELBAS ) { + ret =nvme_identify_ns_nvm (ns->ctrl, info->nsid, &nvm); +if (ret <0 ) + goto out; + } +blk_mq_freeze_queue (ns->disk->queue); lbaf =nvme_lbaf_index (id->flbas); ns->head->lba_shift = id->lbaf[lbaf].ds;
@@ -2115,12 +2108,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, capacity =nvme_lba_to_sect (ns->head,le64_to_cpu (id->nsze));nvme_set_queue_limits (ns->ctrl, ns->queue); - - ret =nvme_configure_metadata (ns->ctrl, ns->head, id); -if (ret <0 ) { -blk_mq_unfreeze_queue (ns->disk->queue); - goto out; - } +nvme_configure_metadata (ns->ctrl, ns->head, id, nvm);nvme_set_chunk_sectors (ns, id);if (!nvme_update_disk_info (ns, id)) capacity =0 ;
@@ -2165,6 +2153,7 @@ static int nvme_update_ns_info_block(struct nvme_ns *ns, ret =0 ; out: +kfree (nvm);kfree (id);return ret; }
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 15/21 nvme: cleanup nvme_configure_metadata
In reply to Christoph HellwigFold nvme_init_ms into nvme_configure_metadata after splitting up a little helper to deal with the extended LBA formats.
--- drivers/nvme/host/core.c | 47 ++++++++++++++++------------------------ 1 file changed, 19 insertions(+), 28 deletions(-)
drivers/nvme/host/core.c +19 -28
@@ -1855,26 +1855,14 @@ static int nvme_identify_ns_nvm(struct nvme_ctrl *ctrl, unsigned int nsid,return ret; } -static void nvme_init_ms (struct nvme_ctrl *ctrl,struct nvme_ns_head *head, +static void nvme_configure_pi_elbas (struct nvme_ns_head *head,struct nvme_id_ns *id,struct nvme_id_ns_nvm *nvm) { -bool first = id->dps &NVME_NS_DPS_PI_FIRST ; -unsigned lbaf =nvme_lbaf_index (id->flbas); -u32 elbaf; - - head->pi_size =0 ; - head->ms =le16_to_cpu (id->lbaf[lbaf].ms); -if (!nvm || !(ctrl->ctratt &NVME_CTRL_ATTR_ELBAS )) { - head->pi_size =sizeof (struct t10_pi_tuple ); - head->guard_type =NVME_NVM_NS_16B_GUARD ; - goto set_pi; - } - - elbaf =le32_to_cpu (nvm->elbaf[lbaf]); +u32 elbaf =le32_to_cpu (nvm->elbaf[nvme_lbaf_index (id->flbas)]);/* no support for storage tag formats right now */ if (nvme_elbaf_sts (elbaf)) - goto set_pi; +return ; head->guard_type =nvme_elbaf_guard_type (elbaf);switch (head->guard_type) {
@@ -1887,29 +1875,32 @@ static void nvme_init_ms(struct nvme_ctrl *ctrl, struct nvme_ns_head *head,default :break ; } - -set_pi: -if (head->pi_size && head->ms >= head->pi_size) - head->pi_type = id->dps &NVME_NS_DPS_PI_MASK ; -else - head->pi_type =0 ; - -if (first) - head->pi_offset =0 ; -else - head->pi_offset = head->ms - head->pi_size; }static void nvme_configure_metadata (struct nvme_ctrl *ctrl,struct nvme_ns_head *head,struct nvme_id_ns *id,struct nvme_id_ns_nvm *nvm) { -nvme_init_ms (ctrl, head, id, nvm); - head->features &= ~(NVME_NS_METADATA_SUPPORTED |NVME_NS_EXT_LBAS ); + head->pi_type =0 ; + head->pi_size =0 ; + head->pi_offset =0 ; + head->ms =le16_to_cpu (id->lbaf[nvme_lbaf_index (id->flbas)].ms);if (!head->ms || !(ctrl->ops->flags &NVME_F_METADATA_SUPPORTED ))return ; +if (nvm && (ctrl->ctratt &NVME_CTRL_ATTR_ELBAS )) { +nvme_configure_pi_elbas (head, id, nvm); + }else { + head->pi_size =sizeof (struct t10_pi_tuple ); + head->guard_type =NVME_NVM_NS_16B_GUARD ; + } + +if (head->pi_size && head->ms >= head->pi_size) + head->pi_type = id->dps &NVME_NS_DPS_PI_MASK ; +if (!(id->dps &NVME_NS_DPS_PI_FIRST )) + head->pi_offset = head->ms - head->pi_size; +if (ctrl->ops->flags &NVME_F_FABRICS ) {/* * The NVMe over Fabrics specification only supports metadata as
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 16/21 nvme-rdma: initialize max_hw_sectors earlier
In reply to Christoph HellwigInitialize max_fr_pages and the values depending on it a little earlier so that nvme_alloc_admin_tag_set can rely on it to set the initial queue limits.
--- drivers/nvme/host/rdma.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-)
drivers/nvme/host/rdma.c +6 -7
@@ -797,6 +797,12 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl, ctrl->max_fr_pages =nvme_rdma_get_max_fr_pages (ctrl->device->dev, pi_capable); + ctrl->ctrl.max_segments = ctrl->max_fr_pages; + ctrl->ctrl.max_hw_sectors = ctrl->max_fr_pages << (ilog2 (SZ_4K ) -9 ); +if (pi_capable) + ctrl->ctrl.max_integrity_segments = ctrl->max_fr_pages; +else + ctrl->ctrl.max_integrity_segments =0 ;/* * Bind the async event SQE DMA mapping to the admin queue lifetime.
@@ -826,13 +832,6 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl,if (error) goto out_stop_queue; - ctrl->ctrl.max_segments = ctrl->max_fr_pages; - ctrl->ctrl.max_hw_sectors = ctrl->max_fr_pages << (ilog2 (SZ_4K ) -9 ); -if (pi_capable) - ctrl->ctrl.max_integrity_segments = ctrl->max_fr_pages; -else - ctrl->ctrl.max_integrity_segments =0 ; -nvme_unquiesce_admin_queue (&ctrl->ctrl); error =nvme_init_ctrl_finish (&ctrl->ctrl, false);
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 17/21 nvme-loop: initialize max_hw_sectors earlier
In reply to Christoph HellwigInitialize max_hw_sectors a little earlier so that nvme_alloc_admin_tag_set can rely on it to set the initial queue limits.
--- drivers/nvme/target/loop.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/nvme/target/loop.c +2 -3
@@ -351,6 +351,8 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)if (error)return error; ctrl->ctrl.queue_count =1 ; + ctrl->ctrl.max_hw_sectors = + (NVME_LOOP_MAX_SEGMENTS -1 ) <<PAGE_SECTORS_SHIFT ; error = nvme_alloc_admin_tag_set(&ctrl->ctrl, &ctrl->admin_tag_set, &nvme_loop_admin_mq_ops,
@@ -372,9 +374,6 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)if (error) goto out_cleanup_tagset; - ctrl->ctrl.max_hw_sectors = - (NVME_LOOP_MAX_SEGMENTS -1 ) <<PAGE_SECTORS_SHIFT ; -nvme_unquiesce_admin_queue (&ctrl->ctrl); error =nvme_init_ctrl_finish (&ctrl->ctrl, false);
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 18/21 nvme-fc: initialize max_hw_sectors earlier
In reply to Christoph HellwigInitialize max_hw_sectors and max_sectors a little earlier so that nvme_alloc_admin_tag_set can rely on it to set the initial queue limits.
--- drivers/nvme/host/fc.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/nvme/host/fc.c +4 -4
@@ -3113,10 +3113,6 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)if (ret) goto out_disconnect_admin_queue; - ctrl->ctrl.max_segments = ctrl->lport->ops->max_sgl_segments; - ctrl->ctrl.max_hw_sectors = ctrl->ctrl.max_segments << - (ilog2 (SZ_4K ) -9 ); -nvme_unquiesce_admin_queue (&ctrl->ctrl); ret =nvme_init_ctrl_finish (&ctrl->ctrl, false);
@@ -3542,6 +3538,10 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,/* at this point, teardown path changes to ref counting on nvme ctrl */ + ctrl->ctrl.max_segments = ctrl->lport->ops->max_sgl_segments; + ctrl->ctrl.max_hw_sectors = + ctrl->ctrl.max_segments << (ilog2 (SZ_4K ) -9 ); + ret = nvme_alloc_admin_tag_set(&ctrl->ctrl, &ctrl->admin_tag_set, &nvme_fc_admin_mq_ops, struct_size_t(struct nvme_fcp_op_w_sgl, priv,
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 19/21 nvme-apple: initialize max_hw_sectors earlier
In reply to Christoph HellwigInitialize max_hw_sectors and max_sectors a little earlier so they can be passed directly to blk_mq_alloc_queue.
--- drivers/nvme/host/apple.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/nvme/host/apple.c +8 -8
@@ -1046,14 +1046,6 @@ static void apple_nvme_reset_work(struct work_struct *work)dev_dbg (anv->dev,"ANS booted successfully." ); -/* -* Limit the max command size to prevent iod->sg allocations going -* over a single page. -*/ - anv->ctrl.max_hw_sectors =min_t (u32,NVME_MAX_KB_SZ <<1 , -dma_max_mapping_size (anv->dev) >>9 ); - anv->ctrl.max_segments =NVME_MAX_SEGS ; -dma_set_max_seg_size (anv->dev,0xffffffff );/*
@@ -1516,6 +1508,14 @@ static int apple_nvme_probe(struct platform_device *pdev) goto put_dev; } +/* +* Limit the max command size to prevent iod->sg allocations going +* over a single page. +*/ + anv->ctrl.max_hw_sectors =min_t (u32,NVME_MAX_KB_SZ <<1 , +dma_max_mapping_size (anv->dev) >>9 ); + anv->ctrl.max_segments =NVME_MAX_SEGS ; + anv->ctrl.admin_q =blk_mq_alloc_queue (&anv->admin_tagset,NULL ,NULL );if (IS_ERR (anv->ctrl.admin_q)) { ret = -ENOMEM ;
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 20/21 nvme: use the atomic queue limits update API
In reply to Christoph HellwigChanges the callchains that update queue_limits to build an on-stack queue_limits and update it atomically. Note that for now only the admin queue actually passes it to the queue allocation function. Doing the same for the gendisks used for the namespaces will require a little more work.
--- drivers/nvme/host/apple.c | 2 + drivers/nvme/host/core.c | 126 +++++++++++++++++++------------------- drivers/nvme/host/nvme.h | 11 +--- drivers/nvme/host/zns.c | 16 ++--- 4 files changed, 76 insertions(+), 79 deletions(-)
drivers/nvme/host/apple.c +2 -0
@@ -1382,6 +1382,7 @@ static void devm_apple_nvme_mempool_destroy(void *data)static int apple_nvme_probe (struct platform_device *pdev) { +struct queue_limits lim = { };struct device *dev = &pdev->dev;struct apple_nvme *anv;int ret;
@@ -1516,6 +1517,7 @@ static int apple_nvme_probe(struct platform_device *pdev) dma_max_mapping_size(anv->dev) >> 9); anv->ctrl.max_segments =NVME_MAX_SEGS ; +nvme_set_ctrl_limits (&anv->ctrl, &lim); anv->ctrl.admin_q =blk_mq_alloc_queue (&anv->admin_tagset,NULL ,NULL );if (IS_ERR (anv->ctrl.admin_q)) { ret = -ENOMEM ;
drivers/nvme/host/core.c +63 -63
223 lines not on this page · raw message
drivers/nvme/host/nvme.h +3 -8
@@ -756,6 +756,7 @@ static __always_inline void nvme_complete_batch(struct io_comp_batch *iob,blk_mq_end_request_batch (iob); } +void nvme_set_ctrl_limits (struct nvme_ctrl *ctrl,struct queue_limits *lim);blk_status_t nvme_host_path_error (struct request *req);bool nvme_cancel_request (struct request *req,void *data);void nvme_cancel_tagset (struct nvme_ctrl *ctrl);
@@ -1038,8 +1039,9 @@ static inline bool nvme_disk_is_ns_head(struct gendisk *disk)int nvme_ns_report_zones (struct nvme_ns *ns,sector_t sector,unsigned int nr_zones,report_zones_cb cb,void *data); +int nvme_update_zone_info (struct nvme_ns *ns,unsigned lbaf, +struct queue_limits *lim);#ifdef CONFIG_BLK_DEV_ZONED -int nvme_update_zone_info (struct nvme_ns *ns,unsigned lbaf);blk_status_t nvme_setup_zone_mgmt_send (struct nvme_ns *ns,struct request *req,struct nvme_command *cmnd,enum nvme_zone_mgmt_action action);
@@ -1050,13 +1052,6 @@ static inline blk_status_t nvme_setup_zone_mgmt_send(struct nvme_ns *ns, {return BLK_STS_NOTSUPP ; } - -static inline int nvme_update_zone_info (struct nvme_ns *ns,unsigned lbaf) -{ -dev_warn (ns->ctrl->device, -"Please enable CONFIG_BLK_DEV_ZONED to support ZNS devices\n" ); -return -EPROTONOSUPPORT ; -}#endif static inline struct nvme_ns *nvme_get_ns_from_dev (struct device *dev)
drivers/nvme/host/zns.c +8 -8
@@ -35,10 +35,10 @@ static int nvme_set_max_append(struct nvme_ctrl *ctrl)return 0 ; } -int nvme_update_zone_info (struct nvme_ns *ns,unsigned lbaf) +int nvme_update_zone_info (struct nvme_ns *ns,unsigned lbaf, +struct queue_limits *lim) {struct nvme_effects_log *log = ns->head->effects; -struct request_queue *q = ns->queue;struct nvme_command c = { };struct nvme_id_ns_zns *id;int status;
@@ -99,12 +99,12 @@ int nvme_update_zone_info(struct nvme_ns *ns, unsigned lbaf) goto free_data; } -disk_set_zoned (ns->disk); -blk_queue_flag_set (QUEUE_FLAG_ZONE_RESETALL , q); -disk_set_max_open_zones (ns->disk,le32_to_cpu (id->mor) +1 ); -disk_set_max_active_zones (ns->disk,le32_to_cpu (id->mar) +1 ); -blk_queue_chunk_sectors (ns->queue, ns->head->zsze); -blk_queue_max_zone_append_sectors (ns->queue, ns->ctrl->max_zone_append); +blk_queue_flag_set (QUEUE_FLAG_ZONE_RESETALL , ns->queue); + lim->zoned =1 ; + lim->max_open_zones =le32_to_cpu (id->mor) +1 ; + lim->max_active_zones =le32_to_cpu (id->mar) +1 ; + lim->chunk_sectors = ns->head->zsze; + lim->max_zone_append_sectors = ns->ctrl->max_zone_append; free_data:kfree (id);return status;
Signature · 1 line
2.39.2
Christoph Hellwig
maintainerPATCH 21/21 nvme-multipath: pass queue_limits to blk_alloc_disk
In reply to Christoph HellwigThe multipath disk starts out with the stacking default limits. The one interesting part here is that blk_set_stacking_limits sets the max_zone_append_sectorts to UINT_MAX, which fails the validation for non-zoned devices. With the old one call per limit scheme this was fine because no one verified this weird mismatch and it was fixed by blk_stack_limits a little later before I/O could be issued.
--- drivers/nvme/host/multipath.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-)
drivers/nvme/host/multipath.c +7 -6
@@ -516,6 +516,7 @@ static void nvme_requeue_work(struct work_struct *work)int nvme_mpath_alloc_disk (struct nvme_ctrl *ctrl,struct nvme_ns_head *head) { +struct queue_limits lim;bool vwc = false;mutex_init (&head->lock);
@@ -532,7 +533,12 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head) !nvme_is_unique_nsid(ctrl, head) || !multipath)return 0 ; - head->disk =blk_alloc_disk (NULL , ctrl->numa_node); +blk_set_stacking_limits (&lim); + lim.dma_alignment =3 ; +if (head->ids.csi !=NVME_CSI_ZNS ) + lim.max_zone_append_sectors =0 ; + + head->disk =blk_alloc_disk (&lim, ctrl->numa_node);if (IS_ERR (head->disk))return PTR_ERR (head->disk); head->disk->fops = &nvme_ns_head_ops;
@@ -553,11 +559,6 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head) ctrl->tagset->map[HCTX_TYPE_POLL].nr_queues)blk_queue_flag_set (QUEUE_FLAG_POLL , head->disk->queue); -/* set to a default value of 512 until the disk is validated */ -blk_queue_logical_block_size (head->disk->queue,512 ); -blk_set_stacking_limits (&head->disk->queue->limits); -blk_queue_dma_alignment (head->disk->queue,3 ); -/* we need to propagate up the VMC settings */ if (ctrl->vwc &NVME_CTRL_VWC_PRESENT ) vwc = true;
Signature · 1 line
2.39.2
Keith Busch
maintainerQuoted 7 lines · Christoph Hellwig
Hi all,
this series converts nvme to the new atomic queue limit updates, and as part of that refactors a lot of the setup code. The first two patches are block layer patches already sent out as part of the md queue limits conversion and will hopefully get merged into Jens' tree soon.
The whole series looks good. I see Jens has got the first two patches, so I'll rebase nvme-6.9 to that and apply the rest of the nvme patches.
Christoph Hellwig
maintainerQuoted 11 lines · Keith Busch
Quoted 7 lines · Christoph Hellwig
Hi all,
this series converts nvme to the new atomic queue limit updates, and as part of that refactors a lot of the setup code. The first two patches are block layer patches already sent out as part of the md queue limits conversion and will hopefully get merged into Jens' tree soon.
The whole series looks good. I see Jens has got the first two patches, so I'll rebase nvme-6.9 to that and apply the rest of the nvme patches.
I had to fix up the little thing Max noticed an also added another patch for multuipath. I'll resend with those updates once you rebased.
Keith Busch
maintainerQuoted 8 lines · Christoph Hellwig
Quoted 3 lines · Keith Busch
The whole series looks good. I see Jens has got the first two patches, so I'll rebase nvme-6.9 to that and apply the rest of the nvme patches.
I had to fix up the little thing Max noticed an also added another patch for multuipath. I'll resend with those updates once you rebased.
Done, nvme-6.9 is rebased to the current for-6.9/block.