bcache: always record start time of a sample

euleros/rtos inclusion
category: bugfix
bugzilla: https://gitee.com/openeuler/kernel/issues/I4LOJ6
CVE: NA

--------------------------------

Before this patch bypassed I/O request will not record start time,
therefore when calculating latency data, start time will be calculated
wrongly. This patch makes start time always be recorded to fix this.
Signed-off-by: NLi Ruilin <liruilin4@huawei.com>
Reviewed-by: NSong Chao <chao.song@huawei.com>
Acked-by: NXie XiuQi <xiexiuqi@huawei.com>
Signed-off-by: NCheng Jian <cj.chengjian@huawei.com>
Reviewed-by: NGuangxing Deng <dengguangxing@huawei.com>
Reviewed-by: Nchao song <chao.song@huawei.com>
Signed-off-by: NZheng Zengkai <zhengzengkai@huawei.com>

drivers/md/bcache/request.c

@@ -1252,14 +1252,13 @@ blk_qc_t cached_dev_submit_bio(struct bio *bio)
 		} else {
 			s->iop.bypass = check_should_bypass(dc, bio);
 
-			if (!s->iop.bypass && bio->bi_iter.bi_size && !rw) {
-				s->smp.offset = bio->bi_iter.bi_sector - dc->sb.data_offset;
-				s->smp.length = bio->bi_iter.bi_size;
-				s->smp.type = rw;
-				s->smp.dev = dc->bdev->bd_dev;
-				s->smp.start_time = ktime_get_ns();
+			s->smp.offset = bio->bi_iter.bi_sector - dc->sb.data_offset;
+			s->smp.length = bio->bi_iter.bi_size;
+			s->smp.type = rw;
+			s->smp.dev = dc->bdev->bd_dev;
+			s->smp.start_time = ktime_get_ns();
+			if (!s->iop.bypass && bio->bi_iter.bi_size && !rw)
 				save_circ_item(&s->smp);
-			}
 
 			if (rw)
 				cached_dev_write(dc, s);