linux-scsi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Don Brace <don.brace@pmcs.com>
To: scott.teel@pmcs.com, Kevin.Barnett@pmcs.com,
	james.bottomley@parallels.com, hch@infradead.org,
	Justin.Lindley@pmcs.com, brace@pmcs.com
Cc: linux-scsi@vger.kernel.org
Subject: [PATCH v2 30/48] hpsa: optimize cmd_alloc function by remembering last allocation
Date: Fri, 23 Jan 2015 16:43:41 -0600	[thread overview]
Message-ID: <20150123224340.14919.95076.stgit@brunhilda> (raw)
In-Reply-To: <20150123224020.14919.29458.stgit@brunhilda>

From: Robert Elliott <elliott@hp.com>

Empirically, this improves performance slightly (~2% max IOPS) by
allowing cmd_alloc to remember where it left off searching for
free commands between calls instead of always starting its search
at command 0.

Reviewed-by: Scott Teel <scott.teel@pmcs.com>
Signed-off-by: Robert Elliott <elliott@hp.com>
Signed-off-by: Don Brace <don.brace@pmcs.com>
---
 drivers/scsi/hpsa.c |    7 +++++--
 drivers/scsi/hpsa.h |    1 +
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/scsi/hpsa.c b/drivers/scsi/hpsa.c
index c95a20c..72abcf3 100644
--- a/drivers/scsi/hpsa.c
+++ b/drivers/scsi/hpsa.c
@@ -4649,9 +4649,10 @@ static struct CommandList *cmd_alloc(struct ctlr_info *h)
 	union u64bit temp64;
 	dma_addr_t cmd_dma_handle, err_dma_handle;
 	int refcount;
-	unsigned long offset = 0;
+	unsigned long offset;
 
-	/* There is some *extremely* small but non-zero chance that that
+	/*
+	 * There is some *extremely* small but non-zero chance that that
 	 * multiple threads could get in here, and one thread could
 	 * be scanning through the list of bits looking for a free
 	 * one, but the free ones are always behind him, and other
@@ -4662,6 +4663,7 @@ static struct CommandList *cmd_alloc(struct ctlr_info *h)
 	 * infrequently as to be indistinguishable from never.
 	 */
 
+	offset = h->last_allocation; /* benignly racy */
 	for (;;) {
 		i = find_next_zero_bit(h->cmd_pool_bits, h->nr_cmds, offset);
 		if (unlikely(i == h->nr_cmds)) {
@@ -4679,6 +4681,7 @@ static struct CommandList *cmd_alloc(struct ctlr_info *h)
 			h->cmd_pool_bits + (i / BITS_PER_LONG));
 		break; /* it's ours now. */
 	}
+	h->last_allocation = i; /* benignly racy */
 
 	/* Zero out all of commandlist except the last field, refcount */
 	memset(c, 0, offsetof(struct CommandList, refcount));
diff --git a/drivers/scsi/hpsa.h b/drivers/scsi/hpsa.h
index 679e4d2..981479a 100644
--- a/drivers/scsi/hpsa.h
+++ b/drivers/scsi/hpsa.h
@@ -133,6 +133,7 @@ struct ctlr_info {
 	struct CfgTable __iomem *cfgtable;
 	int	interrupts_enabled;
 	int 	max_commands;
+	int last_allocation;
 	atomic_t commands_outstanding;
 #	define PERF_MODE_INT	0
 #	define DOORBELL_INT	1


  parent reply	other threads:[~2015-01-23 22:44 UTC|newest]

Thread overview: 51+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-01-23 22:41 [PATCH v2 00/48] hpsa driver updates Don Brace
2015-01-23 22:41 ` [PATCH v2 01/48] hpsa: correct endian sparse warnings Don Brace
2015-01-23 22:41 ` [PATCH v2 02/48] hpsa: fix memory leak in kdump hard reset Don Brace
2015-01-23 22:41 ` [PATCH v2 03/48] hpsa: turn off interrupts when kdump starts Don Brace
2015-01-23 22:41 ` [PATCH v2 04/48] hpsa: change how SA controllers are reset Don Brace
2015-01-23 22:41 ` [PATCH v2 05/48] hpsa: correct change_queue_depth Don Brace
2015-01-23 22:41 ` [PATCH v2 06/48] hpsa: adjust RAID-1, RAID-1ADM, and RAID-6 names Don Brace
2015-01-23 22:41 ` [PATCH v2 07/48] hpsa: rename free_irqs to hpsa_free_irqs Don Brace
2015-01-23 22:41 ` [PATCH v2 08/48] hpsa: Fix -Wunused-but-set-variable warning Don Brace
2015-01-23 22:41 ` [PATCH v2 09/48] hpsa: notice all request_irq errors Don Brace
2015-01-23 22:41 ` [PATCH v2 10/48] hpsa: remove 0x from queue depth print which is in decimal Don Brace
2015-01-23 22:42 ` [PATCH v2 11/48] hpsa: propagate hard_reset failures in reset_devices mode Don Brace
2015-01-23 22:42 ` [PATCH v2 12/48] hpsa: propagate return value from board ID lookup Don Brace
2015-01-23 22:42 ` [PATCH v2 13/48] hpsa: downgrade the Waiting for no-op print to dev_info Don Brace
2015-01-23 22:42 ` [PATCH v2 14/48] hpsa: refactor hpsa_find_board_params() to encapsulate legacy test Don Brace
2015-01-23 22:42 ` [PATCH v2 15/48] hpsa: trivial message and comment clean ups Don Brace
2015-01-23 22:42 ` [PATCH v2 16/48] hpsa: report failure to ioremap config table Don Brace
2015-01-23 22:42 ` [PATCH v2 17/48] hpsa: rename hpsa_request_irq to hpsa_request_irqs Don Brace
2015-01-23 22:42 ` [PATCH v2 18/48] hpsa: pass error from pci_set_consistent_dma_mask from hpsa_message Don Brace
2015-01-23 22:42 ` [PATCH v2 19/48] hpsa: report allocation failures while allocating SG chain blocks Don Brace
2015-01-23 22:42 ` [PATCH v2 20/48] hpsa: fix memory leak in hpsa_alloc_cmd_pool Don Brace
2015-01-23 22:42 ` [PATCH v2 21/48] hpsa: avoid unneccesary calls to resource freeing functions Don Brace
2015-01-23 22:42 ` [PATCH v2 22/48] hpsa: reserve some commands for use by driver Don Brace
2015-01-23 22:43 ` [PATCH v2 23/48] hpsa: get rid of cmd_special_alloc and cmd_special_free Don Brace
2015-01-23 22:43 ` [PATCH v2 24/48] hpsa: do not queue commands internally in driver Don Brace
2015-01-23 22:43 ` [PATCH v2 25/48] hpsa: do not request device rescan on every ioaccel path error Don Brace
2015-01-23 22:43 ` [PATCH v2 26/48] hpsa: factor out hpsa_ciss_submit function Don Brace
2015-01-23 22:43 ` [PATCH v2 27/48] hpsa: use workqueue to resubmit failed ioaccel commands Don Brace
2015-01-23 22:43 ` [PATCH v2 28/48] hpsa: honor queue depth of physical devices Don Brace
2015-01-23 22:43 ` [PATCH v2 29/48] hpsa: fix race between abort handler and main i/o path Don Brace
2015-01-23 22:43 ` Don Brace [this message]
2015-01-23 22:43 ` [PATCH v2 31/48] hpsa: count passthru cmds with atomics, not a spin locked int Don Brace
2015-01-23 22:43 ` [PATCH v2 32/48] hpsa: slightly optimize SA5_performant_completed Don Brace
2015-01-23 22:43 ` [PATCH v2 33/48] hpsa: do not check for msi(x) in interrupt_pending Don Brace
2015-01-23 22:44 ` [PATCH v2 34/48] hpsa: remove incorrect BUG_ONs checking for raid offload enable Don Brace
2015-01-23 22:44 ` [PATCH v2 35/48] hpsa: do not ack controller events on controllers that do not support it Don Brace
2015-01-23 22:44 ` [PATCH v2 36/48] hpsa: guard against overflowing raid map array Don Brace
2015-01-23 22:44 ` [PATCH v2 37/48] hpsa: check for ctlr lockup after command allocation in main io path Don Brace
2015-01-23 22:44 ` [PATCH v2 38/48] hpsa: return failed from device reset/abort handlers Don Brace
2015-01-23 22:44 ` [PATCH v2 39/48] hpsa: do not use a void pointer for scsi_cmd field of struct CommandList Don Brace
2015-01-23 22:44 ` [PATCH v2 40/48] hpsa: print CDBs instead of kernel virtual addresses for uncommon errors Don Brace
2015-01-23 22:44 ` [PATCH v2 41/48] hpsa: do not use function pointers in fast path command submission Don Brace
2015-01-23 22:44 ` [PATCH v2 42/48] hpsa: move SG descriptor set-up out of hpsa_scatter_gather() Don Brace
2015-01-23 22:44 ` [PATCH v2 43/48] hpsa: refactor duplicated scan completion code into a new routine Don Brace
2015-01-23 22:44 ` [PATCH v2 44/48] hpsa: shorten the wait for the CISS doorbell mode change ack Don Brace
2015-01-23 22:45 ` [PATCH v2 45/48] hpsa: detect and report failures changing controller transport modes Don Brace
2015-01-23 22:45 ` [PATCH v2 46/48] hpsa: add in gen9 controller model names Don Brace
2015-01-23 22:45 ` [PATCH v2 47/48] hpsa: add in P840ar controller model name Don Brace
2015-01-23 22:45 ` [PATCH v2 48/48] hpsa: Use local workqueues instead of system workqueues Don Brace
2015-01-27 16:55   ` Tomas Henzl
2015-01-27 17:13     ` Tomas Henzl

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20150123224340.14919.95076.stgit@brunhilda \
    --to=don.brace@pmcs.com \
    --cc=Justin.Lindley@pmcs.com \
    --cc=Kevin.Barnett@pmcs.com \
    --cc=brace@pmcs.com \
    --cc=hch@infradead.org \
    --cc=james.bottomley@parallels.com \
    --cc=linux-scsi@vger.kernel.org \
    --cc=scott.teel@pmcs.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).