| Message ID | 20210331011526epcms2p37684869a9781d1eb45bfcbfe9babd217@epcms2p3 (mailing list archive) |
|---|---|
| Headers | show |
| Series | scsi: ufs: Add Host Performance Booster Support | expand |
Hi Martin I don't know when/how do you plan to accept this patch. I think the Mobile vendors and chipset vendors are all looking forward to this UFS HPB feature that can be mainlined in the upstream Linux. Since the first version HPB driver submitted in the community, it is now V32, and we have been working on this feature for two years. Would you please take a look at this? thanks. Kind regards, Bean On Wed, 2021-03-31 at 10:15 +0900, Daejun Park wrote: > Changelog: > > v31 -> v32 > Delete unused parameter of unmap API. > > v30 -> v31 > Delete unnecessary debug message. > > v29 -> v30 > 1. Add support to reuse bio of pre-request. > 2. Delete unreached code in the ufshpb_issue_map_req. > > v28 -> v29 > 1. Remove unused variable that reported by kernel test robot. > > v27 -> v28 > 1. Fix wrong return value of ufshpb_prep. > > v26 -> v27 > 1. Fix wrong refernce of sense buffer in pre_req complete function. > 2. Fix read_id error. > 3. Fix chunk size checking for HPB 1.0. > 4. Mute unnecessary messages before HPB initialization. > > v25 -> v26 > 1. Fix wrong chunk size checking for HPB 1.0. > 2. Fix wrong max data size for HPB single command. > 3. Fix typo error. > > v24 -> v25 > 1. Change write buffer API for unmap region. > 2. Add checking hpb_enable for avoiding unnecessary memory > allocation. > 3. Change pr_info to dev_info. > 4. Change default requeue timeout value for HPB read. > 5. Fix wrong offset manipulation on ufshpb_prep_entry. > > v23 -> v24 > 1. Fix build error reported by kernel test robot. > > v22 -> v23 > 1. Add support compatibility of HPB 1.0. > 2. Fix read id for single HPB read command. > 3. Fix number of pre-allocated requests for write buffer. > 4. Add fast path for response UPIU that has same LUN in sense data. > 5. Remove WARN_ON for preventing kernel crash. > 7. Fix wrong argument for read buffer command. > > v21 -> v22 > 1. Add support processing response UPIU in suspend state. > 2. Add support HPB hint from other LU. > 3. Add sending write buffer with 0x03 after HPB init. > > v20 -> v21 > 1. Add bMAX_DATA_SIZE_FOR_HPB_SINGLE_CMD attr. and fHPBen flag > support. > > v19 -> v20 > 1. Add documentation for sysfs entries of hpb->stat. > 2. Fix read buffer command for under-sized sub-region. > 3. Fix wrong condition checking for kick map work. > 4. Delete redundant response UPIU checking. > 5. Add LUN checking in response UPIU. > 6. Fix possible deadlock problem due to runtime PM. > 7. Add instant changing of sub-region state from response UPIU. > 8. Fix endian problem in prefetched PPN. > 9. Add JESD220-3A (HPB v2.0) support. > > v18 -> 19 > 1. Fix null pointer error when printing sysfs from non-HPB LU. > 2. Apply HPB read opcode in lrbp->cmd->cmnd (from Can Guo's review). > 3. Rebase the patch on 5.12/scsi-queue. > > v17 -> v18 > Fix build error which reported by kernel test robot. > > v16 -> v17 > 1. Rename hpb_state_lock to rgn_state_lock and move it to > corresponding > patch. > 2. Remove redundant information messages. > > v15 -> v16 > 1. Add missed sysfs ABI documentation. > > v14 -> v15 > 1. Remove duplicated sysfs ABI entries in documentation. > 2. Add experiment result of HPB performance testing with iozone. > > v13 -> v14 > 1. Cleanup codes by commentted in Greg's review. > 2. Add documentation for sysfs entries (from Greg's review). > 3. Add experiment result of HPB performance testing. > > v12 -> v13 > 1. Cleanup codes by comments from Can Guo. > 2. Add HPB related descriptor/flag/attributes in sysfs. > 3. Change base commit from 5.10/scsi-queue to 5.11/scsi-queue. > > v11 -> v12 > 1. Fixed to return error value when HPB fails to initialize pinned > active > region. > 2. Fixed to disable HPB feature if HPB fails to allocate essential > memory > and workqueue. > 3. Fixed to change proper sub-region state when region is already > evicted. > > v10 -> v11 > Add a newline at end the last line on Kconfig file. > > v9 -> v10 > 1. Fixed 64-bit division error > 2. Fixed problems commentted in Bart's review. > > v8 -> v9 > 1. Change sysfs initialization. > 2. Change reading descriptor during HPB initialization > 3. Fixed problems commentted in Bart's review. > 4. Change base commit from 5.9/scsi-queue to 5.10/scsi-queue. > > v7 -> v8 > Remove wrongly added tags. > > v6 -> v7 > 1. Remove UFS feature layer. > 2. Cleanup for sparse error. > > v5 -> v6 > Change base commit to b53293fa662e28ae0cdd40828dc641c09f133405 > > v4 -> v5 > Delete unused macro define. > > v3 -> v4 > 1. Cleanup. > > v2 -> v3 > 1. Add checking input module parameter value. > 2. Change base commit from 5.8/scsi-queue to 5.9/scsi-queue. > 3. Cleanup for unused variables and label. > > v1 -> v2 > 1. Change the full boilerplate text to SPDX style. > 2. Adopt dynamic allocation for sub-region data structure. > 3. Cleanup. > > NAND flash memory-based storage devices use Flash Translation Layer > (FTL) > to translate logical addresses of I/O requests to corresponding flash > memory addresses. Mobile storage devices typically have RAM with > constrained size, thus lack in memory to keep the whole mapping > table. > Therefore, mapping tables are partially retrieved from NAND flash on > demand, causing random-read performance degradation. > > To improve random read performance, JESD220-3 (HPB v1.0) proposes HPB > (Host Performance Booster) which uses host system memory as a cache > for the > FTL mapping table. By using HPB, FTL data can be read from host > memory > faster than from NAND flash memory. > > The current version only supports the DCM (device control mode). > This patch consists of 3 parts to support HPB feature. > > 1) HPB probe and initialization process > 2) READ -> HPB READ using cached map information > 3) L2P (logical to physical) map management > > In the HPB probe and init process, the device information of the UFS > is > queried. After checking supported features, the data structure for > the HPB > is initialized according to the device information. > > A read I/O in the active sub-region where the map is cached is > changed to > HPB READ by the HPB. > > The HPB manages the L2P map using information received from the > device. For active sub-region, the HPB caches through ufshpb_map > request. For the in-active region, the HPB discards the L2P map. > When a write I/O occurs in an active sub-region area, associated > dirty > bitmap checked as dirty for preventing stale read. > > HPB is shown to have a performance improvement of 58 - 67% for random > read > workload. [1] > > [1]: > https://www.usenix.org/conference/hotstorage17/program/presentation/jeong > > Daejun Park (4): > scsi: ufs: Introduce HPB feature > scsi: ufs: L2P map management for HPB read > scsi: ufs: Prepare HPB read for cached sub-region > scsi: ufs: Add HPB 2.0 support > > Documentation/ABI/testing/sysfs-driver-ufs | 162 ++ > drivers/scsi/ufs/Kconfig | 9 + > drivers/scsi/ufs/Makefile | 1 + > drivers/scsi/ufs/ufs-sysfs.c | 22 + > drivers/scsi/ufs/ufs.h | 54 +- > drivers/scsi/ufs/ufshcd.c | 74 +- > drivers/scsi/ufs/ufshcd.h | 29 + > drivers/scsi/ufs/ufshpb.c | 2387 > ++++++++++++++++++++ > drivers/scsi/ufs/ufshpb.h | 277 +++ > 9 files changed, 3013 insertions(+), 2 deletions(-) > create mode 100644 drivers/scsi/ufs/ufshpb.c > create mode 100644 drivers/scsi/ufs/ufshpb.h >
On 31.03.2021 21:40, Bean Huo wrote: >Hi Martin > >I don't know when/how do you plan to accept this patch. I think the >Mobile vendors and chipset vendors are all looking forward to this UFS >HPB feature that can be mainlined in the upstream Linux. Since the >first version HPB driver submitted in the community, it is now V32, and >we have been working on this feature for two years. Would you please >take a look at this? thanks. Hi Bean, I believe it would help if you and others are willing to review this patchset thoroughly and give a reviewed-by / tested-by tag. This will give Martin and others more confidence that there is vendor alignment and that there is a group of people is willing to maintain HPB going forward. What do you think? Javier
On 2021-03-31 09:15, Daejun Park wrote: > Changelog: > > v31 -> v32 > Delete unused parameter of unmap API. > > v30 -> v31 > Delete unnecessary debug message. > > v29 -> v30 > 1. Add support to reuse bio of pre-request. > 2. Delete unreached code in the ufshpb_issue_map_req. > > v28 -> v29 > 1. Remove unused variable that reported by kernel test robot. > > v27 -> v28 > 1. Fix wrong return value of ufshpb_prep. > > v26 -> v27 > 1. Fix wrong refernce of sense buffer in pre_req complete function. > 2. Fix read_id error. > 3. Fix chunk size checking for HPB 1.0. > 4. Mute unnecessary messages before HPB initialization. > > v25 -> v26 > 1. Fix wrong chunk size checking for HPB 1.0. > 2. Fix wrong max data size for HPB single command. > 3. Fix typo error. > > v24 -> v25 > 1. Change write buffer API for unmap region. > 2. Add checking hpb_enable for avoiding unnecessary memory allocation. > 3. Change pr_info to dev_info. > 4. Change default requeue timeout value for HPB read. > 5. Fix wrong offset manipulation on ufshpb_prep_entry. > > v23 -> v24 > 1. Fix build error reported by kernel test robot. > > v22 -> v23 > 1. Add support compatibility of HPB 1.0. > 2. Fix read id for single HPB read command. > 3. Fix number of pre-allocated requests for write buffer. > 4. Add fast path for response UPIU that has same LUN in sense data. > 5. Remove WARN_ON for preventing kernel crash. > 7. Fix wrong argument for read buffer command. > > v21 -> v22 > 1. Add support processing response UPIU in suspend state. > 2. Add support HPB hint from other LU. > 3. Add sending write buffer with 0x03 after HPB init. > > v20 -> v21 > 1. Add bMAX_DATA_SIZE_FOR_HPB_SINGLE_CMD attr. and fHPBen flag support. > > v19 -> v20 > 1. Add documentation for sysfs entries of hpb->stat. > 2. Fix read buffer command for under-sized sub-region. > 3. Fix wrong condition checking for kick map work. > 4. Delete redundant response UPIU checking. > 5. Add LUN checking in response UPIU. > 6. Fix possible deadlock problem due to runtime PM. > 7. Add instant changing of sub-region state from response UPIU. > 8. Fix endian problem in prefetched PPN. > 9. Add JESD220-3A (HPB v2.0) support. > > v18 -> 19 > 1. Fix null pointer error when printing sysfs from non-HPB LU. > 2. Apply HPB read opcode in lrbp->cmd->cmnd (from Can Guo's review). > 3. Rebase the patch on 5.12/scsi-queue. > > v17 -> v18 > Fix build error which reported by kernel test robot. > > v16 -> v17 > 1. Rename hpb_state_lock to rgn_state_lock and move it to corresponding > patch. > 2. Remove redundant information messages. > > v15 -> v16 > 1. Add missed sysfs ABI documentation. > > v14 -> v15 > 1. Remove duplicated sysfs ABI entries in documentation. > 2. Add experiment result of HPB performance testing with iozone. > > v13 -> v14 > 1. Cleanup codes by commentted in Greg's review. > 2. Add documentation for sysfs entries (from Greg's review). > 3. Add experiment result of HPB performance testing. > > v12 -> v13 > 1. Cleanup codes by comments from Can Guo. > 2. Add HPB related descriptor/flag/attributes in sysfs. > 3. Change base commit from 5.10/scsi-queue to 5.11/scsi-queue. > > v11 -> v12 > 1. Fixed to return error value when HPB fails to initialize pinned > active > region. > 2. Fixed to disable HPB feature if HPB fails to allocate essential > memory > and workqueue. > 3. Fixed to change proper sub-region state when region is already > evicted. > > v10 -> v11 > Add a newline at end the last line on Kconfig file. > > v9 -> v10 > 1. Fixed 64-bit division error > 2. Fixed problems commentted in Bart's review. > > v8 -> v9 > 1. Change sysfs initialization. > 2. Change reading descriptor during HPB initialization > 3. Fixed problems commentted in Bart's review. > 4. Change base commit from 5.9/scsi-queue to 5.10/scsi-queue. > > v7 -> v8 > Remove wrongly added tags. > > v6 -> v7 > 1. Remove UFS feature layer. > 2. Cleanup for sparse error. > > v5 -> v6 > Change base commit to b53293fa662e28ae0cdd40828dc641c09f133405 > > v4 -> v5 > Delete unused macro define. > > v3 -> v4 > 1. Cleanup. > > v2 -> v3 > 1. Add checking input module parameter value. > 2. Change base commit from 5.8/scsi-queue to 5.9/scsi-queue. > 3. Cleanup for unused variables and label. > > v1 -> v2 > 1. Change the full boilerplate text to SPDX style. > 2. Adopt dynamic allocation for sub-region data structure. > 3. Cleanup. > > NAND flash memory-based storage devices use Flash Translation Layer > (FTL) > to translate logical addresses of I/O requests to corresponding flash > memory addresses. Mobile storage devices typically have RAM with > constrained size, thus lack in memory to keep the whole mapping table. > Therefore, mapping tables are partially retrieved from NAND flash on > demand, causing random-read performance degradation. > > To improve random read performance, JESD220-3 (HPB v1.0) proposes HPB > (Host Performance Booster) which uses host system memory as a cache for > the > FTL mapping table. By using HPB, FTL data can be read from host memory > faster than from NAND flash memory. > > The current version only supports the DCM (device control mode). > This patch consists of 3 parts to support HPB feature. > > 1) HPB probe and initialization process > 2) READ -> HPB READ using cached map information > 3) L2P (logical to physical) map management > > In the HPB probe and init process, the device information of the UFS is > queried. After checking supported features, the data structure for the > HPB > is initialized according to the device information. > > A read I/O in the active sub-region where the map is cached is changed > to > HPB READ by the HPB. > > The HPB manages the L2P map using information received from the > device. For active sub-region, the HPB caches through ufshpb_map > request. For the in-active region, the HPB discards the L2P map. > When a write I/O occurs in an active sub-region area, associated dirty > bitmap checked as dirty for preventing stale read. > > HPB is shown to have a performance improvement of 58 - 67% for random > read > workload. [1] > > [1]: > https://www.usenix.org/conference/hotstorage17/program/presentation/jeong > > Daejun Park (4): > scsi: ufs: Introduce HPB feature > scsi: ufs: L2P map management for HPB read > scsi: ufs: Prepare HPB read for cached sub-region > scsi: ufs: Add HPB 2.0 support > > Documentation/ABI/testing/sysfs-driver-ufs | 162 ++ > drivers/scsi/ufs/Kconfig | 9 + > drivers/scsi/ufs/Makefile | 1 + > drivers/scsi/ufs/ufs-sysfs.c | 22 + > drivers/scsi/ufs/ufs.h | 54 +- > drivers/scsi/ufs/ufshcd.c | 74 +- > drivers/scsi/ufs/ufshcd.h | 29 + > drivers/scsi/ufs/ufshpb.c | 2387 ++++++++++++++++++++ > drivers/scsi/ufs/ufshpb.h | 277 +++ > 9 files changed, 3013 insertions(+), 2 deletions(-) > create mode 100644 drivers/scsi/ufs/ufshpb.c > create mode 100644 drivers/scsi/ufs/ufshpb.h To the entire series: Tested-by: Can Guo <cang@codeaurora.org>
Hi Can Guo, >> >> HPB is shown to have a performance improvement of 58 - 67% for random >> read >> workload. [1] >> >> [1]: >> https://www.usenix.org/conference/hotstorage17/program/presentation/jeong >> >> Daejun Park (4): >> scsi: ufs: Introduce HPB feature >> scsi: ufs: L2P map management for HPB read >> scsi: ufs: Prepare HPB read for cached sub-region >> scsi: ufs: Add HPB 2.0 support >> >> Documentation/ABI/testing/sysfs-driver-ufs | 162 ++ >> drivers/scsi/ufs/Kconfig | 9 + >> drivers/scsi/ufs/Makefile | 1 + >> drivers/scsi/ufs/ufs-sysfs.c | 22 + >> drivers/scsi/ufs/ufs.h | 54 +- >> drivers/scsi/ufs/ufshcd.c | 74 +- >> drivers/scsi/ufs/ufshcd.h | 29 + >> drivers/scsi/ufs/ufshpb.c | 2387 ++++++++++++++++++++ >> drivers/scsi/ufs/ufshpb.h | 277 +++ >> 9 files changed, 3013 insertions(+), 2 deletions(-) >> create mode 100644 drivers/scsi/ufs/ufshpb.c >> create mode 100644 drivers/scsi/ufs/ufshpb.h > >To the entire series: > >Tested-by: Can Guo <cang@codeaurora.org> Thanks for testing the patch series. :) Daejun