#include "blk.h"
#include "blk-mq.h"
+#include <linux/math64.h>
+
EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_remap);
EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_remap);
EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete);
*/
blk_qc_t generic_make_request(struct bio *bio)
{
- struct bio_list bio_list_on_stack;
+ /*
+ * bio_list_on_stack[0] contains bios submitted by the current
+ * make_request_fn.
+ * bio_list_on_stack[1] contains bios that were submitted before
+ * the current make_request_fn, but that haven't been processed
+ * yet.
+ */
+ struct bio_list bio_list_on_stack[2];
blk_qc_t ret = BLK_QC_T_NONE;
if (!generic_make_request_checks(bio))
* should be added at the tail
*/
if (current->bio_list) {
- bio_list_add(current->bio_list, bio);
+ bio_list_add(¤t->bio_list[0], bio);
goto out;
}
* bio_list, and call into ->make_request() again.
*/
BUG_ON(bio->bi_next);
- bio_list_init(&bio_list_on_stack);
- current->bio_list = &bio_list_on_stack;
+ bio_list_init(&bio_list_on_stack[0]);
+ current->bio_list = bio_list_on_stack;
do {
struct request_queue *q = bdev_get_queue(bio->bi_bdev);
if (likely(blk_queue_enter(q, __GFP_DIRECT_RECLAIM) == 0)) {
+ struct bio_list lower, same;
+
+ /* Create a fresh bio_list for all subordinate requests */
+ bio_list_on_stack[1] = bio_list_on_stack[0];
+ bio_list_init(&bio_list_on_stack[0]);
ret = q->make_request_fn(q, bio);
blk_queue_exit(q);
-
- bio = bio_list_pop(current->bio_list);
+ /* sort new bios into those for a lower level
+ * and those for the same level
+ */
+ bio_list_init(&lower);
+ bio_list_init(&same);
+ while ((bio = bio_list_pop(&bio_list_on_stack[0])) != NULL)
+ if (q == bdev_get_queue(bio->bi_bdev))
+ bio_list_add(&same, bio);
+ else
+ bio_list_add(&lower, bio);
+ /* now assemble so we handle the lowest level first */
+ bio_list_merge(&bio_list_on_stack[0], &lower);
+ bio_list_merge(&bio_list_on_stack[0], &same);
+ bio_list_merge(&bio_list_on_stack[0], &bio_list_on_stack[1]);
} else {
- struct bio *bio_next = bio_list_pop(current->bio_list);
-
bio_io_error(bio);
- bio = bio_next;
}
+ bio = bio_list_pop(&bio_list_on_stack[0]);
} while (bio);
current->bio_list = NULL; /* deactivate */
return 0;
}
+
+/*
+ * Blk IO latency support. We want this to be as cheap as possible, so doing
+ * this lockless (and avoiding atomics), a few off by a few errors in this
+ * code is not harmful, and we don't want to do anything that is
+ * perf-impactful.
+ * TODO : If necessary, we can make the histograms per-cpu and aggregate
+ * them when printing them out.
+ */
+void
+blk_zero_latency_hist(struct io_latency_state *s)
+{
+ memset(s->latency_y_axis_read, 0,
+ sizeof(s->latency_y_axis_read));
+ memset(s->latency_y_axis_write, 0,
+ sizeof(s->latency_y_axis_write));
+ s->latency_reads_elems = 0;
+ s->latency_writes_elems = 0;
+}
+EXPORT_SYMBOL(blk_zero_latency_hist);
+
+ssize_t
+blk_latency_hist_show(struct io_latency_state *s, char *buf)
+{
+ int i;
+ int bytes_written = 0;
+ u_int64_t num_elem, elem;
+ int pct;
+
+ num_elem = s->latency_reads_elems;
+ if (num_elem > 0) {
+ bytes_written += scnprintf(buf + bytes_written,
+ PAGE_SIZE - bytes_written,
+ "IO svc_time Read Latency Histogram (n = %llu):\n",
+ num_elem);
+ for (i = 0;
+ i < ARRAY_SIZE(latency_x_axis_us);
+ i++) {
+ elem = s->latency_y_axis_read[i];
+ pct = div64_u64(elem * 100, num_elem);
+ bytes_written += scnprintf(buf + bytes_written,
+ PAGE_SIZE - bytes_written,
+ "\t< %5lluus%15llu%15d%%\n",
+ latency_x_axis_us[i],
+ elem, pct);
+ }
+ /* Last element in y-axis table is overflow */
+ elem = s->latency_y_axis_read[i];
+ pct = div64_u64(elem * 100, num_elem);
+ bytes_written += scnprintf(buf + bytes_written,
+ PAGE_SIZE - bytes_written,
+ "\t> %5dms%15llu%15d%%\n", 10,
+ elem, pct);
+ }
+ num_elem = s->latency_writes_elems;
+ if (num_elem > 0) {
+ bytes_written += scnprintf(buf + bytes_written,
+ PAGE_SIZE - bytes_written,
+ "IO svc_time Write Latency Histogram (n = %llu):\n",
+ num_elem);
+ for (i = 0;
+ i < ARRAY_SIZE(latency_x_axis_us);
+ i++) {
+ elem = s->latency_y_axis_write[i];
+ pct = div64_u64(elem * 100, num_elem);
+ bytes_written += scnprintf(buf + bytes_written,
+ PAGE_SIZE - bytes_written,
+ "\t< %5lluus%15llu%15d%%\n",
+ latency_x_axis_us[i],
+ elem, pct);
+ }
+ /* Last element in y-axis table is overflow */
+ elem = s->latency_y_axis_write[i];
+ pct = div64_u64(elem * 100, num_elem);
+ bytes_written += scnprintf(buf + bytes_written,
+ PAGE_SIZE - bytes_written,
+ "\t> %5dms%15llu%15d%%\n", 10,
+ elem, pct);
+ }
+ return bytes_written;
+}
+EXPORT_SYMBOL(blk_latency_hist_show);