| Message ID | 1501047404-14456-3-git-send-email-anup.patel@broadcom.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Jul 26, 2017 at 11:06:40AM +0530, Anup Patel wrote: > We should peek mbox channels when we are left with no free > sba_requests in sba_alloc_request() and why is the world should we do that, how does that help?? > > Signed-off-by: Anup Patel <anup.patel@broadcom.com> > Reviewed-by: Ray Jui <ray.jui@broadcom.com> > Reviewed-by: Scott Branden <scott.branden@broadcom.com> > --- > drivers/dma/bcm-sba-raid.c | 25 ++++++++++++++++++++----- > 1 file changed, 20 insertions(+), 5 deletions(-) > > diff --git a/drivers/dma/bcm-sba-raid.c b/drivers/dma/bcm-sba-raid.c > index 6d15fed..321420b 100644 > --- a/drivers/dma/bcm-sba-raid.c > +++ b/drivers/dma/bcm-sba-raid.c > @@ -199,6 +199,14 @@ static inline u32 __pure sba_cmd_pq_c_mdata(u32 d, u32 b1, u32 b0) > > /* ====== General helper routines ===== */ > > +static void sba_peek_mchans(struct sba_device *sba) > +{ > + int mchan_idx; > + > + for (mchan_idx = 0; mchan_idx < sba->mchans_count; mchan_idx++) > + mbox_client_peek_data(sba->mchans[mchan_idx]); > +} > + > static struct sba_request *sba_alloc_request(struct sba_device *sba) > { > unsigned long flags; > @@ -210,8 +218,17 @@ static struct sba_request *sba_alloc_request(struct sba_device *sba) > if (req) > list_move_tail(&req->node, &sba->reqs_alloc_list); > spin_unlock_irqrestore(&sba->reqs_lock, flags); > - if (!req) > + > + if (!req) { > + /* > + * We have no more free requests so, we peek > + * mailbox channels hoping few active requests > + * would have completed which will create more > + * room for new requests. > + */ > + sba_peek_mchans(sba); > return NULL; > + } > > req->flags = SBA_REQUEST_STATE_ALLOCED; > req->first = req; > @@ -558,17 +575,15 @@ static enum dma_status sba_tx_status(struct dma_chan *dchan, > dma_cookie_t cookie, > struct dma_tx_state *txstate) > { > - int mchan_idx; > enum dma_status ret; > struct sba_device *sba = to_sba_device(dchan); > > - for (mchan_idx = 0; mchan_idx < sba->mchans_count; mchan_idx++) > - mbox_client_peek_data(sba->mchans[mchan_idx]); > - > ret = dma_cookie_status(dchan, cookie, txstate); > if (ret == DMA_COMPLETE) > return ret; > > + sba_peek_mchans(sba); > + > return dma_cookie_status(dchan, cookie, txstate); > } > > -- > 2.7.4 >
On Wed, Jul 26, 2017 at 10:40 PM, Vinod Koul <vinod.koul@intel.com> wrote: > On Wed, Jul 26, 2017 at 11:06:40AM +0530, Anup Patel wrote: >> We should peek mbox channels when we are left with no free >> sba_requests in sba_alloc_request() > > and why is the world should we do that, how does that help?? When setting up RAID array on several NVMe disk we observed that sba_alloc_request() start failing (due to no free requests left) and RAID array setup becomes very slow. Doing mbox channel peek when we have no free requests left, improves performance of RAID array setup. This change is inspired from mv_chan_alloc_slot() implemented in drivers/dma/mv_xor.c Regards, Anup
On Thu, Jul 27, 2017 at 10:25:25AM +0530, Anup Patel wrote: > On Wed, Jul 26, 2017 at 10:40 PM, Vinod Koul <vinod.koul@intel.com> wrote: > > On Wed, Jul 26, 2017 at 11:06:40AM +0530, Anup Patel wrote: > >> We should peek mbox channels when we are left with no free > >> sba_requests in sba_alloc_request() > > > > and why is the world should we do that, how does that help?? > > When setting up RAID array on several NVMe disk we observed > that sba_alloc_request() start failing (due to no free requests left) > and RAID array setup becomes very slow. > > Doing mbox channel peek when we have no free requests left, > improves performance of RAID array setup. How about documenting this tribal knowledge in the changelog. Changelogs are very useful, 6 months down the line, you will struggle to remember why this was changed.. > > This change is inspired from mv_chan_alloc_slot() implemented > in drivers/dma/mv_xor.c > > Regards, > Anup
On Fri, Jul 28, 2017 at 8:45 AM, Vinod Koul <vinod.koul@intel.com> wrote: > On Thu, Jul 27, 2017 at 10:25:25AM +0530, Anup Patel wrote: >> On Wed, Jul 26, 2017 at 10:40 PM, Vinod Koul <vinod.koul@intel.com> wrote: >> > On Wed, Jul 26, 2017 at 11:06:40AM +0530, Anup Patel wrote: >> >> We should peek mbox channels when we are left with no free >> >> sba_requests in sba_alloc_request() >> > >> > and why is the world should we do that, how does that help?? >> >> When setting up RAID array on several NVMe disk we observed >> that sba_alloc_request() start failing (due to no free requests left) >> and RAID array setup becomes very slow. >> >> Doing mbox channel peek when we have no free requests left, >> improves performance of RAID array setup. > > How about documenting this tribal knowledge in the changelog. Changelogs are > very useful, 6 months down the line, you will struggle to remember why this > was changed.. Sure, I will have detailed commit description for this. Regards, Anup
diff --git a/drivers/dma/bcm-sba-raid.c b/drivers/dma/bcm-sba-raid.c index 6d15fed..321420b 100644 --- a/drivers/dma/bcm-sba-raid.c +++ b/drivers/dma/bcm-sba-raid.c @@ -199,6 +199,14 @@ static inline u32 __pure sba_cmd_pq_c_mdata(u32 d, u32 b1, u32 b0) /* ====== General helper routines ===== */ +static void sba_peek_mchans(struct sba_device *sba) +{ + int mchan_idx; + + for (mchan_idx = 0; mchan_idx < sba->mchans_count; mchan_idx++) + mbox_client_peek_data(sba->mchans[mchan_idx]); +} + static struct sba_request *sba_alloc_request(struct sba_device *sba) { unsigned long flags; @@ -210,8 +218,17 @@ static struct sba_request *sba_alloc_request(struct sba_device *sba) if (req) list_move_tail(&req->node, &sba->reqs_alloc_list); spin_unlock_irqrestore(&sba->reqs_lock, flags); - if (!req) + + if (!req) { + /* + * We have no more free requests so, we peek + * mailbox channels hoping few active requests + * would have completed which will create more + * room for new requests. + */ + sba_peek_mchans(sba); return NULL; + } req->flags = SBA_REQUEST_STATE_ALLOCED; req->first = req; @@ -558,17 +575,15 @@ static enum dma_status sba_tx_status(struct dma_chan *dchan, dma_cookie_t cookie, struct dma_tx_state *txstate) { - int mchan_idx; enum dma_status ret; struct sba_device *sba = to_sba_device(dchan); - for (mchan_idx = 0; mchan_idx < sba->mchans_count; mchan_idx++) - mbox_client_peek_data(sba->mchans[mchan_idx]); - ret = dma_cookie_status(dchan, cookie, txstate); if (ret == DMA_COMPLETE) return ret; + sba_peek_mchans(sba); + return dma_cookie_status(dchan, cookie, txstate); }