cmdprio: Add support for per I/O priority hint

Introduce the new option cmdprio_hint to allow specifying I/O priority
hints per IO with the io_uring and libaio IO engines. A third acceptable
format for the cmdprio_bssplit option is also introduced to allow
specifying an I/O hint in addition to a priority class and level.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Niklas Cassel <niklas.cassel@wdc.com>
Link: https://lore.kernel.org/r/20230721110510.44772-6-dlemoal@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>

HOWTO.rst

@@ -2287,6 +2287,16 @@ with the caveat that when used on the command line, they must come after the
 	reads and writes. See :manpage:`ionice(1)`. See also the
 	:option:`prioclass` option.
 
+.. option:: cmdprio_hint=int[,int] : [io_uring] [libaio]
+
+	Set the I/O priority hint to use for I/Os that must be issued with
+	a priority when :option:`cmdprio_percentage` or
+	:option:`cmdprio_bssplit` is set. If not specified when
+	:option:`cmdprio_percentage` or :option:`cmdprio_bssplit` is set,
+	this defaults to 0 (no hint). A single value applies to reads and
+	writes. Comma-separated values may be specified for reads and writes.
+	See also the :option:`priohint` option.
+
 .. option:: cmdprio=int[,int] : [io_uring] [libaio]
 
 	Set the I/O priority value to use for I/Os that must be issued with
@@ -2313,9 +2323,9 @@ with the caveat that when used on the command line, they must come after the
 
 		cmdprio_bssplit=blocksize/percentage:blocksize/percentage
 
-	In this case, each entry will use the priority class and priority
-	level defined by the options :option:`cmdprio_class` and
-	:option:`cmdprio` respectively.
+	In this case, each entry will use the priority class, priority hint
+	and priority level defined by the options :option:`cmdprio_class`,
+        :option:`cmdprio` and :option:`cmdprio_hint` respectively.
 
 	The second accepted format for this option is:
 
@@ -2326,7 +2336,14 @@ with the caveat that when used on the command line, they must come after the
 	accepted format does not restrict all entries to have the same priority
 	class and priority level.
 
-	For both formats, only the read and write data directions are supported,
+	The third accepted format for this option is:
+
+		cmdprio_bssplit=blocksize/percentage/class/level/hint:...
+
+	This is an extension of the second accepted format that allows to also
+	specify a priority hint.
+
+	For all formats, only the read and write data directions are supported,
 	values for trim IOs are ignored. This option is mutually exclusive with
 	the :option:`cmdprio_percentage` option.
 
@@ -3445,6 +3462,9 @@ Threads, processes and job synchronization
 	of I/Os so that the device can optimize its internal command scheduling
 	according to the latency limits indicated by the user.
 
+	For per-I/O priority hint setting, see the I/O engine specific
+	:option:`cmdprio_hint` option.
+
 .. option:: cpus_allowed=str
 
 	Controls the same options as :option:`cpumask`, but accepts a textual

engines/cmdprio.c

@@ -267,7 +267,8 @@ static int fio_cmdprio_percentage(struct cmdprio *cmdprio, struct io_u *io_u,
  * to be set. If the random percentage value is within the user specified
  * percentage of I/Os that should use a cmdprio priority value (rather than
  * the default priority), then this function updates the io_u with an ioprio
- * value as defined by the cmdprio/cmdprio_class or cmdprio_bssplit options.
+ * value as defined by the cmdprio/cmdprio_hint/cmdprio_class or
+ * cmdprio_bssplit options.
  *
  * Return true if the io_u ioprio was changed and false otherwise.
  */
@@ -342,7 +343,8 @@ static int fio_cmdprio_gen_perc(struct thread_data *td, struct cmdprio *cmdprio)
 		prio = &cmdprio->perc_entry[ddir];
 		prio->perc = options->percentage[ddir];
 		prio->prio = ioprio_value(options->class[ddir],
-					  options->level[ddir], 0);
+					  options->level[ddir],
+					  options->hint[ddir]);
 		assign_clat_prio_index(prio, &values[ddir]);
 
 		ret = init_ts_clat_prio(ts, ddir, &values[ddir]);
@@ -400,7 +402,8 @@ static int fio_cmdprio_parse_and_gen_bssplit(struct thread_data *td,
 			goto err;
 
 		implicit_cmdprio = ioprio_value(options->class[ddir],
-						options->level[ddir], 0);
+						options->level[ddir],
+						options->hint[ddir]);
 
 		ret = fio_cmdprio_generate_bsprio_desc(&cmdprio->bsprio_desc[ddir],
 						       &parse_res[ddir],

engines/cmdprio.h

@@ -40,6 +40,7 @@ struct cmdprio_options {
 	unsigned int percentage[CMDPRIO_RWDIR_CNT];
 	unsigned int class[CMDPRIO_RWDIR_CNT];
 	unsigned int level[CMDPRIO_RWDIR_CNT];
+	unsigned int hint[CMDPRIO_RWDIR_CNT];
 	char *bssplit_str;
 };
 
@@ -74,6 +75,21 @@ struct cmdprio_options {
 		.category = FIO_OPT_C_ENGINE,					\
 		.group	= opt_group,						\
 	},									\
+	{									\
+		.name	= "cmdprio_hint",					\
+		.lname	= "Asynchronous I/O priority hint",			\
+		.type	= FIO_OPT_INT,						\
+		.off1	= offsetof(opt_struct,					\
+				   cmdprio_options.hint[DDIR_READ]),		\
+		.off2	= offsetof(opt_struct,					\
+				   cmdprio_options.hint[DDIR_WRITE]),		\
+		.help	= "Set asynchronous IO priority hint",			\
+		.minval	= IOPRIO_MIN_PRIO_HINT,					\
+		.maxval	= IOPRIO_MAX_PRIO_HINT,					\
+		.interval = 1,							\
+		.category = FIO_OPT_C_ENGINE,					\
+		.group	= opt_group,						\
+	},									\
 	{									\
 		.name	= "cmdprio",						\
 		.lname	= "Asynchronous I/O priority level",			\
@@ -112,6 +128,12 @@ struct cmdprio_options {
 		.type	= FIO_OPT_UNSUPPORTED,					\
 		.help	= "Platform does not support I/O priority classes",	\
 	},									\
+	{									\
+		.name	= "cmdprio_hint",					\
+		.lname	= "Asynchronous I/O priority hint",			\
+		.type	= FIO_OPT_UNSUPPORTED,					\
+		.help	= "Platform does not support I/O priority classes",	\
+	},									\
 	{									\
 		.name	= "cmdprio",						\
 		.lname	= "Asynchronous I/O priority level",			\

engines/io_uring.c

@@ -285,8 +285,8 @@ static int fio_ioring_prep(struct thread_data *td, struct io_u *io_u)
 		/*
 		 * Since io_uring can have a submission context (sqthread_poll)
 		 * that is different from the process context, we cannot rely on
-		 * the IO priority set by ioprio_set() (option prio/prioclass)
-		 * to be inherited.
+		 * the IO priority set by ioprio_set() (options prio, prioclass,
+		 * and priohint) to be inherited.
 		 * td->ioprio will have the value of the "default prio", so set
 		 * this unconditionally. This value might get overridden by
 		 * fio_ioring_cmdprio_prep() if the option cmdprio_percentage or

fio.1

@@ -2084,6 +2084,14 @@ is set, this defaults to the highest priority class. A single value applies
 to reads and writes. Comma-separated values may be specified for reads and
 writes. See man \fBionice\fR\|(1). See also the \fBprioclass\fR option.
 .TP
+.BI (io_uring,libaio)cmdprio_hint \fR=\fPint[,int]
+Set the I/O priority hint to use for I/Os that must be issued with a
+priority when \fBcmdprio_percentage\fR or \fBcmdprio_bssplit\fR is set.
+If not specified when \fBcmdprio_percentage\fR or \fBcmdprio_bssplit\fR
+is set, this defaults to 0 (no hint). A single value applies to reads and
+writes. Comma-separated values may be specified for reads and writes.
+See also the \fBpriohint\fR option.
+.TP
 .BI (io_uring,libaio)cmdprio \fR=\fPint[,int]
 Set the I/O priority value to use for I/Os that must be issued with a
 priority when \fBcmdprio_percentage\fR or \fBcmdprio_bssplit\fR is set.
@@ -2109,8 +2117,9 @@ The first accepted format for this option is the same as the format of the
 cmdprio_bssplit=blocksize/percentage:blocksize/percentage
 .RE
 .P
-In this case, each entry will use the priority class and priority level defined
-by the options \fBcmdprio_class\fR and \fBcmdprio\fR respectively.
+In this case, each entry will use the priority class, priority hint and
+priority level defined by the options \fBcmdprio_class\fR, \fBcmdprio\fR
+and \fBcmdprio_hint\fR respectively.
 .P
 The second accepted format for this option is:
 .RS
@@ -2123,7 +2132,16 @@ entry. In comparison with the first accepted format, the second accepted format
 does not restrict all entries to have the same priority class and priority
 level.
 .P
-For both formats, only the read and write data directions are supported, values
+The third accepted format for this option is:
+.RS
+.P
+cmdprio_bssplit=blocksize/percentage/class/level/hint:...
+.RE
+.P
+This is an extension of the second accepted format that allows to also
+specify a priority hint.
+.P
+For all formats, only the read and write data directions are supported, values
 for trim IOs are ignored. This option is mutually exclusive with the
 \fBcmdprio_percentage\fR option.
 .RE
@@ -3150,7 +3168,8 @@ I/O priority classes and to devices with features controlled through priority
 hints, e.g. block devices supporting command duration limits, or CDL. CDL is a
 way to indicate the desired maximum latency of I/Os so that the device can
 optimize its internal command scheduling according to the latency limits
-indicated by the user.
+indicated by the user. For per-I/O priority hint setting, see the I/O engine
+specific \fBcmdprio_hint\fB option.
 .TP
 .BI cpus_allowed \fR=\fPstr
 Controls the same options as \fBcpumask\fR, but accepts a textual

options.c

@@ -313,15 +313,17 @@ static int parse_cmdprio_bssplit_entry(struct thread_options *o,
 	int matches = 0;
 	char *bs_str = NULL;
 	long long bs_val;
-	unsigned int perc = 0, class, level;
+	unsigned int perc = 0, class, level, hint;
 
 	/*
 	 * valid entry formats:
 	 * bs/ - %s/ - set perc to 0, prio to -1.
 	 * bs/perc - %s/%u - set prio to -1.
 	 * bs/perc/class/level - %s/%u/%u/%u
+	 * bs/perc/class/level/hint - %s/%u/%u/%u/%u
 	 */
-	matches = sscanf(str, "%m[^/]/%u/%u/%u", &bs_str, &perc, &class, &level);
+	matches = sscanf(str, "%m[^/]/%u/%u/%u/%u",
+			 &bs_str, &perc, &class, &level, &hint);
 	if (matches < 1) {
 		log_err("fio: invalid cmdprio_bssplit format\n");
 		return 1;
@@ -342,9 +344,14 @@ static int parse_cmdprio_bssplit_entry(struct thread_options *o,
 	case 2: /* bs/perc case */
 		break;
 	case 4: /* bs/perc/class/level case */
+	case 5: /* bs/perc/class/level/hint case */
 		class = min(class, (unsigned int) IOPRIO_MAX_PRIO_CLASS);
 		level = min(level, (unsigned int) IOPRIO_MAX_PRIO);
-		entry->prio = ioprio_value(class, level, 0);
+		if (matches == 5)
+			hint = min(hint, (unsigned int) IOPRIO_MAX_PRIO_HINT);
+		else
+			hint = 0;
+		entry->prio = ioprio_value(class, level, hint);
 		break;
 	default:
 		log_err("fio: invalid cmdprio_bssplit format\n");