Re: [Patch v4] storvsc: setup 1:1 mapping between hardware queue and CPU queue
Long, > storvsc doesn't use a dedicated hardware queue for a given CPU > queue. When issuing I/O, it selects returning CPU (hardware queue) > dynamically based on vmbus channel usage across all channels. Applied to 5.4/scsi-fixes. Thanks! -- Martin K. Petersen Oracle Linux Engineering
RE: [Patch v4] storvsc: setup 1:1 mapping between hardware queue and CPU queue
>Subject: RE: [Patch v4] storvsc: setup 1:1 mapping between hardware queue >and CPU queue > >>Subject: Re: [Patch v4] storvsc: setup 1:1 mapping between hardware >>queue and CPU queue >> >>On Fri, Sep 06, 2019 at 10:24:20AM -0700, lon...@linuxonhyperv.com wrote: >>>From: Long Li >>> >>>storvsc doesn't use a dedicated hardware queue for a given CPU queue. >>>When issuing I/O, it selects returning CPU (hardware queue) >>>dynamically based on vmbus channel usage across all channels. >>> >>>This patch advertises num_present_cpus() as number of hardware queues. >>>This will have upper layer setup 1:1 mapping between hardware queue >>>and CPU queue and avoid unnecessary locking when issuing I/O. >>> >>>Signed-off-by: Long Li Hi Martin, I have addressed all comments on this patch. Can you merge it to SCSI? If there is anything else I need to change, please let me know. Thanks Long >>>--- >>> >>>Changes: >>>v2: rely on default upper layer function to map queues. (suggested by >>>Ming Lei >>>) >>>v3: use num_present_cpus() instead of num_online_cpus(). Hyper-v >>>doesn't support hot-add CPUs. (suggested by Michael Kelley >>>) >>>v4: move change logs to after Signed-of-by >>> >>> drivers/scsi/storvsc_drv.c | 3 +-- >>> 1 file changed, 1 insertion(+), 2 deletions(-) >>> >>>diff --git a/drivers/scsi/storvsc_drv.c b/drivers/scsi/storvsc_drv.c >>>index b89269120a2d..cf987712041a 100644 >>>--- a/drivers/scsi/storvsc_drv.c >>>+++ b/drivers/scsi/storvsc_drv.c >>>@@ -1836,8 +1836,7 @@ static int storvsc_probe(struct hv_device *device, >>> /* >>> * Set the number of HW queues we are supporting. >>> */ >>>-if (stor_device->num_sc != 0) >>>-host->nr_hw_queues = stor_device->num_sc + 1; >>>+host->nr_hw_queues = num_present_cpus(); >> >>Just looking at the change notes for v3: why isn't this >>num_active_cpus() then? One can still isolate CPUs on hyper-v, no? > >The isolated CPU can be made online at run time. For example, even >maxcpus=x is put on the boot line, individual CPUs can still be made >online/offline. > >> >>-- >>Thanks, >>Sasha
RE: [Patch v4] storvsc: setup 1:1 mapping between hardware queue and CPU queue
>Subject: Re: [Patch v4] storvsc: setup 1:1 mapping between hardware queue >and CPU queue > >On Fri, Sep 06, 2019 at 10:24:20AM -0700, lon...@linuxonhyperv.com wrote: >>From: Long Li >> >>storvsc doesn't use a dedicated hardware queue for a given CPU queue. >>When issuing I/O, it selects returning CPU (hardware queue) dynamically >>based on vmbus channel usage across all channels. >> >>This patch advertises num_present_cpus() as number of hardware queues. >>This will have upper layer setup 1:1 mapping between hardware queue and >>CPU queue and avoid unnecessary locking when issuing I/O. >> >>Signed-off-by: Long Li >>--- >> >>Changes: >>v2: rely on default upper layer function to map queues. (suggested by >>Ming Lei >>) >>v3: use num_present_cpus() instead of num_online_cpus(). Hyper-v >>doesn't support hot-add CPUs. (suggested by Michael Kelley >>) >>v4: move change logs to after Signed-of-by >> >> drivers/scsi/storvsc_drv.c | 3 +-- >> 1 file changed, 1 insertion(+), 2 deletions(-) >> >>diff --git a/drivers/scsi/storvsc_drv.c b/drivers/scsi/storvsc_drv.c >>index b89269120a2d..cf987712041a 100644 >>--- a/drivers/scsi/storvsc_drv.c >>+++ b/drivers/scsi/storvsc_drv.c >>@@ -1836,8 +1836,7 @@ static int storvsc_probe(struct hv_device *device, >> /* >> * Set the number of HW queues we are supporting. >> */ >>- if (stor_device->num_sc != 0) >>- host->nr_hw_queues = stor_device->num_sc + 1; >>+ host->nr_hw_queues = num_present_cpus(); > >Just looking at the change notes for v3: why isn't this >num_active_cpus() then? One can still isolate CPUs on hyper-v, no? The isolated CPU can be made online at run time. For example, even maxcpus=x is put on the boot line, individual CPUs can still be made online/offline. > >-- >Thanks, >Sasha
Re: [Patch v4] storvsc: setup 1:1 mapping between hardware queue and CPU queue
On Fri, Sep 06, 2019 at 10:24:20AM -0700, lon...@linuxonhyperv.com wrote: From: Long Li storvsc doesn't use a dedicated hardware queue for a given CPU queue. When issuing I/O, it selects returning CPU (hardware queue) dynamically based on vmbus channel usage across all channels. This patch advertises num_present_cpus() as number of hardware queues. This will have upper layer setup 1:1 mapping between hardware queue and CPU queue and avoid unnecessary locking when issuing I/O. Signed-off-by: Long Li --- Changes: v2: rely on default upper layer function to map queues. (suggested by Ming Lei ) v3: use num_present_cpus() instead of num_online_cpus(). Hyper-v doesn't support hot-add CPUs. (suggested by Michael Kelley ) v4: move change logs to after Signed-of-by drivers/scsi/storvsc_drv.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/scsi/storvsc_drv.c b/drivers/scsi/storvsc_drv.c index b89269120a2d..cf987712041a 100644 --- a/drivers/scsi/storvsc_drv.c +++ b/drivers/scsi/storvsc_drv.c @@ -1836,8 +1836,7 @@ static int storvsc_probe(struct hv_device *device, /* * Set the number of HW queues we are supporting. */ - if (stor_device->num_sc != 0) - host->nr_hw_queues = stor_device->num_sc + 1; + host->nr_hw_queues = num_present_cpus(); Just looking at the change notes for v3: why isn't this num_active_cpus() then? One can still isolate CPUs on hyper-v, no? -- Thanks, Sasha
RE: [Patch v4] storvsc: setup 1:1 mapping between hardware queue and CPU queue
From: Long Li Sent: Friday, September 6, 2019 10:24 AM > > storvsc doesn't use a dedicated hardware queue for a given CPU queue. When > issuing I/O, it selects returning CPU (hardware queue) dynamically based on > vmbus channel usage across all channels. > > This patch advertises num_present_cpus() as number of hardware queues. This > will have upper layer setup 1:1 mapping between hardware queue and CPU queue > and avoid unnecessary locking when issuing I/O. > > Signed-off-by: Long Li > --- > > Changes: > v2: rely on default upper layer function to map queues. (suggested by Ming Lei > ) > v3: use num_present_cpus() instead of num_online_cpus(). Hyper-v doesn't > support hot-add CPUs. (suggested by Michael Kelley ) > v4: move change logs to after Signed-of-by > > drivers/scsi/storvsc_drv.c | 3 +-- > 1 file changed, 1 insertion(+), 2 deletions(-) > > diff --git a/drivers/scsi/storvsc_drv.c b/drivers/scsi/storvsc_drv.c > index b89269120a2d..cf987712041a 100644 > --- a/drivers/scsi/storvsc_drv.c > +++ b/drivers/scsi/storvsc_drv.c > @@ -1836,8 +1836,7 @@ static int storvsc_probe(struct hv_device *device, > /* >* Set the number of HW queues we are supporting. >*/ > - if (stor_device->num_sc != 0) > - host->nr_hw_queues = stor_device->num_sc + 1; > + host->nr_hw_queues = num_present_cpus(); > > /* >* Set the error handler work queue. > -- > 2.17.1 Reviewed-by: Michael Kelley