nvme: Consolidate chunk_sectors settings
Subsystems (2)
cover nvme: Consolidate chunk_sectors settings
Backport commit 38adf94e166e3cb4eb89683458ca578051e8218d and it's dependencies to linux-stable 5.4.y.
Dependent commits:314d48dd224897e35ddcaf5a1d7d133b5adddeb7e08f2ae850929d40e66268ee47e443e7ea56eeb7
When running test cases to stress NVMe device, a race condition / deadlocks is seen every couple of days or so where multiple threads are trying to acquire ctrl->subsystem->lock or ctrl->scan_lock.
The test cases send a lot nvme-cli requests to do Sanitize, Format, FW Download, FW Activate, Flush, Get Log, Identify, and reset requests to two controllers that share a namespace. Some of those commands target a namespace, some target a controller. The commands are sent in random order and random mix to the two controllers.
The test cases does not wait for nvme-cli requests to finish before sending more. So for example, there could be multiple reset requests, multiple format requests, outstanding at the same time as a sanitize, on both paths at the same time, etc. Many of these test cases include combos that don't really make sense in the context of NVMe, however it is used to create as much stress as possible.
This patchset fixes this issue.
Similar issue with a detailed call trace/log was discussed in the LKML
Revanth Rajashekar (3):
nvme: Cleanup and rename nvme_block_nr()
nvme: Introduce nvme_lba_to_sect()
nvme: consolidate chunk_sectors settings
drivers/nvme/host/core.c | 40 +++++++++++++++++++--------------------- drivers/nvme/host/nvme.h | 16 +++++++++++++--- 2 files changed, 32 insertions(+), 24 deletions(-)
--
2.17.1
PATCH 1/3 nvme: Cleanup and rename nvme_block_nr()
From: Damien Le Moal <[email protected]>
commit 314d48dd224897e35ddcaf5a1d7d133b5adddeb7
Rename nvme_block_nr() to nvme_sect_to_lba() and use SECTOR_SHIFT instead of its hard coded value 9. Also add a comment to decribe this helper.
--- drivers/nvme/host/core.c | 6 +++--- drivers/nvme/host/nvme.h | 7 +++++-- 2 files changed, 8 insertions(+), 5 deletions(-)
drivers/nvme/host/core.c +3 -3
@@ -633,7 +633,7 @@ static blk_status_t nvme_setup_discard(struct nvme_ns *ns, struct request *req, }__rq_for_each_bio (bio, req) { -u64 slba =nvme_block_nr (ns, bio->bi_iter.bi_sector); +u64 slba =nvme_sect_to_lba (ns, bio->bi_iter.bi_sector);u32 nlb = bio->bi_iter.bi_size >> ns->lba_shift;if (n < segments) {
@@ -674,7 +674,7 @@ static inline blk_status_t nvme_setup_write_zeroes(struct nvme_ns *ns, cmnd->write_zeroes.opcode = nvme_cmd_write_zeroes; cmnd->write_zeroes.nsid =cpu_to_le32 (ns->head->ns_id); cmnd->write_zeroes.slba = -cpu_to_le64 (nvme_block_nr (ns,blk_rq_pos (req))); +cpu_to_le64 (nvme_sect_to_lba (ns,blk_rq_pos (req))); cmnd->write_zeroes.length =cpu_to_le16 ((blk_rq_bytes (req) >> ns->lba_shift) -1 ); cmnd->write_zeroes.control =0 ;
@@ -698,7 +698,7 @@ static inline blk_status_t nvme_setup_rw(struct nvme_ns *ns, cmnd->rw.opcode = (rq_data_dir (req) ? nvme_cmd_write : nvme_cmd_read); cmnd->rw.nsid =cpu_to_le32 (ns->head->ns_id); - cmnd->rw.slba =cpu_to_le64 (nvme_block_nr (ns,blk_rq_pos (req))); + cmnd->rw.slba =cpu_to_le64 (nvme_sect_to_lba (ns,blk_rq_pos (req))); cmnd->rw.length =cpu_to_le16 ((blk_rq_bytes (req) >> ns->lba_shift) -1 );if (req_op (req) ==REQ_OP_WRITE && ctrl->nr_streams)
drivers/nvme/host/nvme.h +5 -2
@@ -421,9 +421,12 @@ static inline int nvme_reset_subsystem(struct nvme_ctrl *ctrl)return ctrl->ops->reg_write32 (ctrl,NVME_REG_NSSR ,0x4E564D65 ); } -static inline u64 nvme_block_nr (struct nvme_ns *ns,sector_t sector) +/* +* Convert a 512B sector number to a device logical block number. +*/ +static inline u64 nvme_sect_to_lba (struct nvme_ns *ns,sector_t sector) { -return (sector >> (ns->lba_shift -9 )); +return sector >> (ns->lba_shift -SECTOR_SHIFT ); } static inline void nvme_end_request(struct request *req, __le16 status,
--
2.17.1
PATCH 2/3 nvme: Introduce nvme_lba_to_sect()
In reply to Revanth RajashekarFrom: Damien Le Moal <[email protected]>
commit e08f2ae850929d40e66268ee47e443e7ea56eeb7
Introduce the new helper function nvme_lba_to_sect() to convert a device logical block number to a 512B sector number. Use this new helper in obvious places, cleaning up the code.
--- drivers/nvme/host/core.c | 14 +++++++------- drivers/nvme/host/nvme.h | 8 ++++++++ 2 files changed, 15 insertions(+), 7 deletions(-)
drivers/nvme/host/core.c +7 -7
@@ -1694,7 +1694,7 @@ static void nvme_init_integrity(struct gendisk *disk, u16 ms, u8 pi_type)static void nvme_set_chunk_size (struct nvme_ns *ns) { -u32 chunk_size = (((u32 )ns->noiob) << (ns->lba_shift -9 )); +u32 chunk_size =nvme_lba_to_sect (ns, ns->noiob);blk_queue_chunk_sectors (ns->queue,rounddown_pow_of_two (chunk_size)); }
@@ -1731,8 +1731,7 @@ static void nvme_config_discard(struct gendisk *disk, struct nvme_ns *ns)static void nvme_config_write_zeroes (struct gendisk *disk,struct nvme_ns *ns) { -u32 max_sectors; -unsigned short bs =1 << ns->lba_shift; +u64 max_blocks;if (!(ns->ctrl->oncs &NVME_CTRL_ONCS_WRITE_ZEROES ) || (ns->ctrl->quirks &NVME_QUIRK_DISABLE_WRITE_ZEROES ))
@@ -1748,11 +1747,12 @@ static void nvme_config_write_zeroes(struct gendisk *disk, struct nvme_ns *ns)* nvme_init_identify() if available. */ if (ns->ctrl->max_hw_sectors ==UINT_MAX ) - max_sectors = ((u32)(USHRT_MAX +1 ) * bs) >>9 ; + max_blocks = (u64 )USHRT_MAX +1 ;else - max_sectors = ((u32)(ns->ctrl->max_hw_sectors +1 ) * bs) >>9 ; + max_blocks = ns->ctrl->max_hw_sectors +1 ; -blk_queue_max_write_zeroes_sectors (disk->queue, max_sectors); +blk_queue_max_write_zeroes_sectors (disk->queue, +nvme_lba_to_sect (ns, max_blocks)); } static int nvme_report_ns_ids(struct nvme_ctrl *ctrl, unsigned int nsid,
@@ -1786,7 +1786,7 @@ static bool nvme_ns_ids_equal(struct nvme_ns_ids *a, struct nvme_ns_ids *b)static void nvme_update_disk_info (struct gendisk *disk,struct nvme_ns *ns,struct nvme_id_ns *id) { -sector_t capacity =le64_to_cpu (id->nsze) << (ns->lba_shift -9 ); +sector_t capacity =nvme_lba_to_sect (ns,le64_to_cpu (id->nsze));unsigned short bs =1 << ns->lba_shift;u32 atomic_bs, phys_bs, io_opt;
drivers/nvme/host/nvme.h +8 -0
@@ -429,6 +429,14 @@ static inline u64 nvme_sect_to_lba(struct nvme_ns *ns, sector_t sector)return sector >> (ns->lba_shift -SECTOR_SHIFT ); } +/* +* Convert a device logical block number to a 512B sector number. +*/ +static inline sector_t nvme_lba_to_sect (struct nvme_ns *ns,u64 lba) +{ +return lba << (ns->lba_shift -SECTOR_SHIFT ); +} +static inline void nvme_end_request (struct request *req,__le16 status,union nvme_result result) {
--
2.17.1
PATCH 3/3 nvme: consolidate chunk_sectors settings
In reply to Revanth RajashekarFrom: Keith Busch <[email protected]>
commit 38adf94e166e3cb4eb89683458ca578051e8218d
Move the quirked chunk_sectors setting to the same location as noiob so one place registers this setting. And since the noiob value is only used locally, remove the member from struct nvme_ns.
--- drivers/nvme/host/core.c | 22 ++++++++++------------ drivers/nvme/host/nvme.h | 1 - 2 files changed, 10 insertions(+), 13 deletions(-)
drivers/nvme/host/core.c +10 -12
@@ -1692,12 +1692,6 @@ static void nvme_init_integrity(struct gendisk *disk, u16 ms, u8 pi_type) }#endif /* CONFIG_BLK_DEV_INTEGRITY */ -static void nvme_set_chunk_size (struct nvme_ns *ns) -{ -u32 chunk_size =nvme_lba_to_sect (ns, ns->noiob); -blk_queue_chunk_sectors (ns->queue,rounddown_pow_of_two (chunk_size)); -} -static void nvme_config_discard (struct gendisk *disk,struct nvme_ns *ns) {struct nvme_ctrl *ctrl = ns->ctrl;
@@ -1852,6 +1846,7 @@ static void nvme_update_disk_info(struct gendisk *disk,static void __nvme_revalidate_disk (struct gendisk *disk,struct nvme_id_ns *id) {struct nvme_ns *ns = disk->private_data; +u32 iob;/* * If identify namespace failed, use default 512 byte block size so
@@ -1860,7 +1855,13 @@ static void __nvme_revalidate_disk(struct gendisk *disk, struct nvme_id_ns *id) ns->lba_shift = id->lbaf[id->flbas &NVME_NS_FLBAS_LBA_MASK ].ds;if (ns->lba_shift ==0 ) ns->lba_shift =9 ; - ns->noiob =le16_to_cpu (id->noiob); + +if ((ns->ctrl->quirks &NVME_QUIRK_STRIPE_SIZE ) && +is_power_of_2 (ns->ctrl->max_hw_sectors)) + iob = ns->ctrl->max_hw_sectors; +else + iob =nvme_lba_to_sect (ns,le16_to_cpu (id->noiob)); + ns->ms =le16_to_cpu (id->lbaf[id->flbas &NVME_NS_FLBAS_LBA_MASK ].ms); ns->ext = ns->ms && (id->flbas &NVME_NS_FLBAS_META_EXT );/* the PI implementation requires metadata equal t10 pi tuple size */
@@ -1869,8 +1870,8 @@ static void __nvme_revalidate_disk(struct gendisk *disk, struct nvme_id_ns *id)else ns->pi_type =0 ; -if (ns->noiob) -nvme_set_chunk_size (ns); +if (iob) +blk_queue_chunk_sectors (ns->queue,rounddown_pow_of_two (iob));nvme_update_disk_info (disk, ns, id);#ifdef CONFIG_NVME_MULTIPATH if (ns->head->disk) {
@@ -2221,9 +2222,6 @@ 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, max_segments,USHRT_MAX )); } -if ((ctrl->quirks &NVME_QUIRK_STRIPE_SIZE ) && -is_power_of_2 (ctrl->max_hw_sectors)) -blk_queue_chunk_sectors (q, ctrl->max_hw_sectors);blk_queue_virt_boundary (q, ctrl->page_size -1 );if (ctrl->vwc &NVME_CTRL_VWC_PRESENT ) vwc = true;
drivers/nvme/host/nvme.h +0 -1
@@ -376,7 +376,6 @@ struct nvme_ns {#define NVME_NS_REMOVING 0#define NVME_NS_DEAD 1#define NVME_NS_ANA_PENDING 2 -u16 noiob;struct nvme_fault_inject fault_inject;
--
2.17.1
Quoted 28 lines · Revanth Rajashekar
Backport commit
38adf94e166e3cb4eb89683458ca578051e8218dand it's dependencies to linux-stable 5.4.y.Dependent commits:
314d48dd224897e35ddcaf5a1d7d133b5adddeb7e08f2ae850929d40e66268ee47e443e7ea56eeb7When running test cases to stress NVMe device, a race condition / deadlocks is seen every couple of days or so where multiple threads are trying to acquire
ctrl->subsystem->lockorctrl->scan_lock.The test cases send a lot nvme-cli requests to do Sanitize, Format, FW Download, FW Activate, Flush, Get Log, Identify, and reset requests to two controllers that share a namespace. Some of those commands target a namespace, some target a controller. The commands are sent in random order and random mix to the two controllers.
The test cases does not wait for nvme-cli requests to finish before sending more. So for example, there could be multiple reset requests, multiple format requests, outstanding at the same time as a sanitize, on both paths at the same time, etc. Many of these test cases include combos that don't really make sense in the context of NVMe, however it is used to create as much stress as possible.
This patchset fixes this issue.
Similar issue with a detailed call trace/log was discussed in the LKML
Quoted 8 lines
Revanth Rajashekar (3):
nvme: Cleanup and renamenvme_block_nr()
nvme: Introducenvme_lba_to_sect()
nvme: consolidatechunk_sectorssettingsdrivers/nvme/host/core.c | 40 +++++++++++++++++++--------------------- drivers/nvme/host/nvme.h | 16 +++++++++++++--- 2 files changed, 32 insertions(+), 24 deletions(-)
For some reason you forgot to credit, and cc: all of the people on the original patches, which isn't very nice, don't you think?
Next time please be more careful.
greg k-h
Quoted 37 lines · Greg KH
Quoted 28 lines · Revanth Rajashekar
Backport commit
38adf94e166e3cb4eb89683458ca578051e8218dand it's dependencies to linux-stable 5.4.y.Dependent commits:
314d48dd224897e35ddcaf5a1d7d133b5adddeb7e08f2ae850929d40e66268ee47e443e7ea56eeb7When running test cases to stress NVMe device, a race condition / deadlocks is seen every couple of days or so where multiple threads are trying to acquire
ctrl->subsystem->lockorctrl->scan_lock.The test cases send a lot nvme-cli requests to do Sanitize, Format, FW Download, FW Activate, Flush, Get Log, Identify, and reset requests to two controllers that share a namespace. Some of those commands target a namespace, some target a controller. The commands are sent in random order and random mix to the two controllers.
The test cases does not wait for nvme-cli requests to finish before sending more. So for example, there could be multiple reset requests, multiple format requests, outstanding at the same time as a sanitize, on both paths at the same time, etc. Many of these test cases include combos that don't really make sense in the context of NVMe, however it is used to create as much stress as possible.
This patchset fixes this issue.
Similar issue with a detailed call trace/log was discussed in the LKML
Quoted 8 lines
Revanth Rajashekar (3):
nvme: Cleanup and renamenvme_block_nr()
nvme: Introducenvme_lba_to_sect()
nvme: consolidatechunk_sectorssettingsdrivers/nvme/host/core.c | 40 +++++++++++++++++++--------------------- drivers/nvme/host/nvme.h | 16 +++++++++++++--- 2 files changed, 32 insertions(+), 24 deletions(-)
Quoted 6 lines
For some reason you forgot to credit, and cc: all of the people on the original patches, which isn't very nice, don't you think?
Next time please be more careful.
greg k-h
I'm really sorry for missing out a few people from the cc list who are on the original patch. Thought only the signed-off names go in there.
Will make sure this won't happen again and thanks for pointing it out :)
Thanks!
Revanth