MFC r286320: Issue all reads of single XCOPY segment simultaneously.

During vMotion and Clone VMware by default runs multiple sequential 4MB
XCOPY requests same time.  If CTL issues reads sequentially in 1MB chunks
for each XCOPY command, reads from different commands are not detected
as sequential by serseq option code and allowed to execute simultaneously.
Such read pattern confused ZFS prefetcher, causing suboptimal disk access.
Issuing all reads same time make serseq code work properly, serializing
reads both within each XCOPY command and between them.

My tests with ZFS pool of 14 disks in RAID10 shows prefetcher efficiency
improved from 37% to 99.7%, copying speed improved by 10-60%, average
read latency reduced twice on HDD layer and by five times on zvol layer.

1 files changed, 4 insertions, 5 deletions

diff --git a/sys/cam/ctl/ctl_tpc.c b/sys/cam/ctl/ctl_tpc.c
index 490cddd..662ee3d 100644
--- a/sys/cam/ctl/ctl_tpc.c
+++ b/sys/cam/ctl/ctl_tpc.c
@@ -817,7 +817,7 @@ tpc_process_b2b(struct tpc_list *list)
 	struct scsi_ec_segment_b2b *seg;
 	struct scsi_ec_cscd_dtsp *sdstp, *ddstp;
 	struct tpc_io *tior, *tiow;
-	struct runl run, *prun;
+	struct runl run;
 	uint64_t sl, dl;
 	off_t srclba, dstlba, numbytes, donebytes, roundbytes;
 	int numlba;
@@ -889,8 +889,7 @@ tpc_process_b2b(struct tpc_list *list)
 	list->segsectors = numbytes / dstblock;
 	donebytes = 0;
 	TAILQ_INIT(&run);
-	prun = &run;
-	list->tbdio = 1;
+	list->tbdio = 0;
 	while (donebytes < numbytes) {
 		roundbytes = numbytes - donebytes;
 		if (roundbytes > TPC_MAX_IO_SIZE) {
@@ -942,8 +941,8 @@ tpc_process_b2b(struct tpc_list *list)
 		tiow->io->io_hdr.ctl_private[CTL_PRIV_FRONTEND].ptr = tiow;
 
 		TAILQ_INSERT_TAIL(&tior->run, tiow, rlinks);
-		TAILQ_INSERT_TAIL(prun, tior, rlinks);
-		prun = &tior->run;
+		TAILQ_INSERT_TAIL(&run, tior, rlinks);
+		list->tbdio++;
 		donebytes += roundbytes;
 		srclba += roundbytes / srcblock;
 		dstlba += roundbytes / dstblock;