| Message ID | cover.1663961449.git.tom@talpey.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Reduce SMBDirect RDMA SGE counts and sizes | expand |
2022-09-24 6:53 GMT+09:00, Tom Talpey <tom@talpey.com>: > Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect > implementations. > > The current maximum values (16 SGEs and 8192 bytes) cause failures on the > SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and > 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, > and for debug sanity, reformat client-side logging to more clearly show > addresses, lengths and flags in the appropriate base. > > Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and > general. > > v2: correct an uninitialized value issue found by Coverity > > Tom Talpey (6): > Decrease the number of SMB3 smbdirect client SGEs > Decrease the number of SMB3 smbdirect server SGEs > Reduce client smbdirect max receive segment size > Reduce server smbdirect max send/receive segment sizes > Handle variable number of SGEs in client smbdirect send. > Fix formatting of client smbdirect RDMA logging You are missing adding prefix(cifs or ksmbd:) to each patch. > > fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- > fs/cifs/smbdirect.h | 14 ++- > fs/ksmbd/transport_rdma.c | 6 +- > 3 files changed, 109 insertions(+), 138 deletions(-) > > -- > 2.34.1 > >
On 9/24/2022 11:45 PM, Namjae Jeon wrote: > 2022-09-24 6:53 GMT+09:00, Tom Talpey <tom@talpey.com>: >> Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect >> implementations. >> >> The current maximum values (16 SGEs and 8192 bytes) cause failures on the >> SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and >> 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, >> and for debug sanity, reformat client-side logging to more clearly show >> addresses, lengths and flags in the appropriate base. >> >> Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and >> general. >> >> v2: correct an uninitialized value issue found by Coverity >> >> Tom Talpey (6): >> Decrease the number of SMB3 smbdirect client SGEs >> Decrease the number of SMB3 smbdirect server SGEs >> Reduce client smbdirect max receive segment size >> Reduce server smbdirect max send/receive segment sizes >> Handle variable number of SGEs in client smbdirect send. >> Fix formatting of client smbdirect RDMA logging > You are missing adding prefix(cifs or ksmbd:) to each patch. Ok, sure I'll add those. Tom. > >> >> fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- >> fs/cifs/smbdirect.h | 14 ++- >> fs/ksmbd/transport_rdma.c | 6 +- >> 3 files changed, 109 insertions(+), 138 deletions(-) >> >> -- >> 2.34.1 >> >> >
merged patches 1, 3, 5, 6 of this series into cifs-2.6.git for-next (will let Namjae test/try the server patches, 2 and 4) pending additional testing. Let me know if any Reviewed-by to add On Fri, Sep 23, 2022 at 4:54 PM Tom Talpey <tom@talpey.com> wrote: > > Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect > implementations. > > The current maximum values (16 SGEs and 8192 bytes) cause failures on the > SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and > 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, > and for debug sanity, reformat client-side logging to more clearly show > addresses, lengths and flags in the appropriate base. > > Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and general. > > v2: correct an uninitialized value issue found by Coverity > > Tom Talpey (6): > Decrease the number of SMB3 smbdirect client SGEs > Decrease the number of SMB3 smbdirect server SGEs > Reduce client smbdirect max receive segment size > Reduce server smbdirect max send/receive segment sizes > Handle variable number of SGEs in client smbdirect send. > Fix formatting of client smbdirect RDMA logging > > fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- > fs/cifs/smbdirect.h | 14 ++- > fs/ksmbd/transport_rdma.c | 6 +- > 3 files changed, 109 insertions(+), 138 deletions(-) > > -- > 2.34.1 >
I need to add the "cifs" and "ksmbd" prefixes, and a couple of Acked-by's. I'm still pretty ill so not getting much done just now though. I'll try to get on it later today. On 9/29/2022 1:02 AM, Steve French wrote: > merged patches 1, 3, 5, 6 of this series into cifs-2.6.git for-next > (will let Namjae test/try the server patches, 2 and 4) pending > additional testing. > > Let me know if any Reviewed-by to add > > On Fri, Sep 23, 2022 at 4:54 PM Tom Talpey <tom@talpey.com> wrote: >> >> Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect >> implementations. >> >> The current maximum values (16 SGEs and 8192 bytes) cause failures on the >> SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and >> 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, >> and for debug sanity, reformat client-side logging to more clearly show >> addresses, lengths and flags in the appropriate base. >> >> Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and general. >> >> v2: correct an uninitialized value issue found by Coverity >> >> Tom Talpey (6): >> Decrease the number of SMB3 smbdirect client SGEs >> Decrease the number of SMB3 smbdirect server SGEs >> Reduce client smbdirect max receive segment size >> Reduce server smbdirect max send/receive segment sizes >> Handle variable number of SGEs in client smbdirect send. >> Fix formatting of client smbdirect RDMA logging >> >> fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- >> fs/cifs/smbdirect.h | 14 ++- >> fs/ksmbd/transport_rdma.c | 6 +- >> 3 files changed, 109 insertions(+), 138 deletions(-) >> >> -- >> 2.34.1 >> > >
I can add the Acked-bys if you send them to me (for the cifs.ko ones) The client for server (cifs vs ksmbd prefix) in the title is more of an issue for email threads and patch review. On Thu, Sep 29, 2022 at 10:15 AM Tom Talpey <tom@talpey.com> wrote: > > I need to add the "cifs" and "ksmbd" prefixes, and a couple of > Acked-by's. I'm still pretty ill so not getting much done just > now though. I'll try to get on it later today. > > On 9/29/2022 1:02 AM, Steve French wrote: > > merged patches 1, 3, 5, 6 of this series into cifs-2.6.git for-next > > (will let Namjae test/try the server patches, 2 and 4) pending > > additional testing. > > > > Let me know if any Reviewed-by to add > > > > On Fri, Sep 23, 2022 at 4:54 PM Tom Talpey <tom@talpey.com> wrote: > >> > >> Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect > >> implementations. > >> > >> The current maximum values (16 SGEs and 8192 bytes) cause failures on the > >> SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and > >> 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, > >> and for debug sanity, reformat client-side logging to more clearly show > >> addresses, lengths and flags in the appropriate base. > >> > >> Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and general. > >> > >> v2: correct an uninitialized value issue found by Coverity > >> > >> Tom Talpey (6): > >> Decrease the number of SMB3 smbdirect client SGEs > >> Decrease the number of SMB3 smbdirect server SGEs > >> Reduce client smbdirect max receive segment size > >> Reduce server smbdirect max send/receive segment sizes > >> Handle variable number of SGEs in client smbdirect send. > >> Fix formatting of client smbdirect RDMA logging > >> > >> fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- > >> fs/cifs/smbdirect.h | 14 ++- > >> fs/ksmbd/transport_rdma.c | 6 +- > >> 3 files changed, 109 insertions(+), 138 deletions(-) > >> > >> -- > >> 2.34.1 > >> > > > >
On 9/29/2022 11:27 AM, Steve French wrote: > I can add the Acked-bys if you send them to me (for the cifs.ko ones) That would be a big help! Patch 1: Add "cifs:" Patch 2: Add "ksmbd:" and Acked-by: Namjae Jeon <linkinjeon@kernel.org> Patch 3: Add "cifs:" Patch 4: Add "ksmbd:" and Acked-by: Namjae Jeon <linkinjeon@kernel.org> Patch 5" Add "cifs:" Patch 6: Add "cifs:" No R-B's received, and no code changes. Tom. > > The client for server (cifs vs ksmbd prefix) in the title is more of > an issue for email threads and patch review. > > On Thu, Sep 29, 2022 at 10:15 AM Tom Talpey <tom@talpey.com> wrote: >> >> I need to add the "cifs" and "ksmbd" prefixes, and a couple of >> Acked-by's. I'm still pretty ill so not getting much done just >> now though. I'll try to get on it later today. >> >> On 9/29/2022 1:02 AM, Steve French wrote: >>> merged patches 1, 3, 5, 6 of this series into cifs-2.6.git for-next >>> (will let Namjae test/try the server patches, 2 and 4) pending >>> additional testing. >>> >>> Let me know if any Reviewed-by to add >>> >>> On Fri, Sep 23, 2022 at 4:54 PM Tom Talpey <tom@talpey.com> wrote: >>>> >>>> Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect >>>> implementations. >>>> >>>> The current maximum values (16 SGEs and 8192 bytes) cause failures on the >>>> SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and >>>> 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, >>>> and for debug sanity, reformat client-side logging to more clearly show >>>> addresses, lengths and flags in the appropriate base. >>>> >>>> Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and general. >>>> >>>> v2: correct an uninitialized value issue found by Coverity >>>> >>>> Tom Talpey (6): >>>> Decrease the number of SMB3 smbdirect client SGEs >>>> Decrease the number of SMB3 smbdirect server SGEs >>>> Reduce client smbdirect max receive segment size >>>> Reduce server smbdirect max send/receive segment sizes >>>> Handle variable number of SGEs in client smbdirect send. >>>> Fix formatting of client smbdirect RDMA logging >>>> >>>> fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- >>>> fs/cifs/smbdirect.h | 14 ++- >>>> fs/ksmbd/transport_rdma.c | 6 +- >>>> 3 files changed, 109 insertions(+), 138 deletions(-) >>>> >>>> -- >>>> 2.34.1 >>>> >>> >>> > > >
Tom Talpey <tom@talpey.com> writes: > Allocate fewer SGEs and standard packet sizes in both kernel SMBDirect > implementations. > > The current maximum values (16 SGEs and 8192 bytes) cause failures on the > SoftiWARP provider, and are suboptimal on others. Reduce these to 6 and > 1364. Additionally, recode smbd_send() to work with as few as 2 SGEs, > and for debug sanity, reformat client-side logging to more clearly show > addresses, lengths and flags in the appropriate base. > > Tested over SoftiWARP and SoftRoCE with shell, Connectathon basic and general. > > v2: correct an uninitialized value issue found by Coverity > > Tom Talpey (6): > Decrease the number of SMB3 smbdirect client SGEs > Decrease the number of SMB3 smbdirect server SGEs > Reduce client smbdirect max receive segment size > Reduce server smbdirect max send/receive segment sizes > Handle variable number of SGEs in client smbdirect send. > Fix formatting of client smbdirect RDMA logging > > fs/cifs/smbdirect.c | 227 ++++++++++++++++---------------------- > fs/cifs/smbdirect.h | 14 ++- > fs/ksmbd/transport_rdma.c | 6 +- > 3 files changed, 109 insertions(+), 138 deletions(-) Acked-by: Paulo Alcantara (SUSE) <pc@cjr.nz>