exec: add high-level controls for blkio cgroup attributes

4 files changed, 228 insertions, 7 deletions

diff --git a/man/systemd.exec.xml b/man/systemd.exec.xml
index 89e3369d3c..6bc8bf3e79 100644
--- a/man/systemd.exec.xml
+++ b/man/systemd.exec.xml
@@ -797,11 +797,13 @@
                                 <term><varname>CPUShares=</varname></term>
 
                                 <listitem><para>Assign the specified
-                                overall CPU time shares to the processes executed. Takes
-                                an integer value. This controls the
+                                overall CPU time shares to the
+                                processes executed. Takes an integer
+                                value. This controls the
                                 <literal>cpu.shares</literal> control
-                                group attribute. For details about
-                                this control group attribute see <ulink
+                                group attribute, which defaults to
+                                1024. For details about this control
+                                group attribute see <ulink
                                 url="http://www.kernel.org/doc/Documentation/scheduler/sched-design-CFS.txt">sched-design-CFS.txt</ulink>.</para></listitem>
                         </varlistentry>
 
@@ -814,7 +816,7 @@
                                 size. Takes a memory size in bytes. If
                                 the value is suffixed with K, M, G or
                                 T the specified memory size is parsed
-                                as Kilobytes, Megabytes, Gigabytes
+                                as Kilobytes, Megabytes, Gigabytes,
                                 resp. Terabytes (to the base
                                 1024). This controls the
                                 <literal>memory.limit_in_bytes</literal>
@@ -849,6 +851,57 @@
                         </varlistentry>
 
                         <varlistentry>
+                                <term><varname>BlockIOWeight=</varname></term>
+
+                                <listitem><para>Set the default or
+                                per-device overall block IO weight
+                                value for the executed
+                                processes. Takes either a single
+                                weight value (between 10 and 1000) to
+                                set the default block IO weight, or a
+                                space separated pair of a device node
+                                path and a weight value to specify the
+                                device specific weight value (Example:
+                                "/dev/sda 500"). This controls the
+                                <literal>blkio.weight</literal> and
+                                <literal>blkio.weight_device</literal>
+                                control group attributes, which
+                                default to 1000. Use this option
+                                multiple times to set weights for
+                                multiple devices. For details about
+                                these control group attributes see
+                                <ulink
+                                url="http://www.kernel.org/doc/Documentation/cgroups/blkio-controller.txt">blkio-controller.txt</ulink>.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
+                                <term><varname>BlockIOReadBandwidth=</varname></term>
+                                <term><varname>BlockIOWriteBandwidth=</varname></term>
+
+                                <listitem><para>Set the per-device
+                                overall block IO bandwith limit for the
+                                executed processes. Takes a space
+                                separated pair of a device node path
+                                and a bandwith value (in bytes per
+                                second) to specify the device specific
+                                bandwidth. If the bandwith is suffixed
+                                with K, M, G, or T the specified
+                                bandwith is parsed as Kilobytes,
+                                Megabytes, Gigabytes, resp. Terabytes
+                                (Example: "/dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0 5M"). This
+                                controls the
+                                <literal>blkio.read_bps_device</literal>
+                                and
+                                <literal>blkio.write_bps_device</literal>
+                                control group attributes. Use this
+                                option multiple times to set bandwith
+                                limits for multiple devices. For
+                                details about these control group
+                                attributes see <ulink
+                                url="http://www.kernel.org/doc/Documentation/cgroups/blkio-controller.txt">blkio-controller.txt</ulink>.</para></listitem>
+                        </varlistentry>
+
+                        <varlistentry>
                                 <term><varname>ReadWriteDirectories=</varname></term>
                                 <term><varname>ReadOnlyDirectories=</varname></term>
                                 <term><varname>InaccessibleDirectories=</varname></term>
diff --git a/src/load-fragment-gperf.gperf.m4 b/src/load-fragment-gperf.gperf.m4
index bfa1e3d1b8..08223c5c1b 100644
--- a/src/load-fragment-gperf.gperf.m4
+++ b/src/load-fragment-gperf.gperf.m4
@@ -71,6 +71,9 @@ $1.MemoryLimit,                  config_parse_unit_memory_limit,     0,
 $1.MemorySoftLimit,              config_parse_unit_memory_limit,     0,                             0
 $1.DeviceAllow,                  config_parse_unit_device_allow,     0,                             0
 $1.DeviceDeny,                   config_parse_unit_device_allow,     0,                             0
+$1.BlockIOWeight,                config_parse_unit_blkio_weight,     0,                             0
+$1.BlockIOReadBandwidth,         config_parse_unit_blkio_bandwidth,  0,                             0
+$1.BlockIOWriteBandwidth,        config_parse_unit_blkio_bandwidth,  0,                             0
 $1.ReadWriteDirectories,         config_parse_path_strv,             0,                             offsetof($1, exec_context.read_write_dirs)
 $1.ReadOnlyDirectories,          config_parse_path_strv,             0,                             offsetof($1, exec_context.read_only_dirs)
 $1.InaccessibleDirectories,      config_parse_path_strv,             0,                             offsetof($1, exec_context.inaccessible_dirs)
diff --git a/src/load-fragment.c b/src/load-fragment.c
index 28439d9b20..b122ea419c 100644
--- a/src/load-fragment.c
+++ b/src/load-fragment.c
@@ -1743,9 +1743,13 @@ int config_parse_unit_memory_limit(const char *filename, unsigned line, const ch
 }
 
 static int device_map(const char *controller, const char *name, const char *value, char **ret) {
-        struct stat st;
         char **l;
 
+        assert(controller);
+        assert(name);
+        assert(value);
+        assert(ret);
+
         l = strv_split_quoted(value);
         if (!l)
                 return -ENOMEM;
@@ -1761,7 +1765,9 @@ static int device_map(const char *controller, const char *name, const char *valu
                 }
 
         } else {
-                if (lstat(l[0], &st) < 0) {
+                struct stat st;
+
+                if (stat(l[0], &st) < 0) {
                         log_warning("Couldn't stat device %s", l[0]);
                         strv_free(l);
                         return -errno;
@@ -1834,6 +1840,163 @@ int config_parse_unit_device_allow(const char *filename, unsigned line, const ch
         return 0;
 }
 
+static int blkio_map(const char *controller, const char *name, const char *value, char **ret) {
+        struct stat st;
+        char **l;
+
+        assert(controller);
+        assert(name);
+        assert(value);
+        assert(ret);
+
+        l = strv_split_quoted(value);
+        if (!l)
+                return -ENOMEM;
+
+        assert(strv_length(l) == 2);
+
+        if (stat(l[0], &st) < 0) {
+                log_warning("Couldn't stat device %s", l[0]);
+                strv_free(l);
+                return -errno;
+        }
+
+        if (!S_ISBLK(st.st_mode)) {
+                log_warning("%s is not a block device.", l[0]);
+                strv_free(l);
+                return -ENODEV;
+        }
+
+        if (asprintf(ret, "%u:%u %s", major(st.st_rdev), minor(st.st_rdev), l[1]) < 0) {
+                strv_free(l);
+                return -ENOMEM;
+        }
+
+        strv_free(l);
+        return 0;
+}
+
+int config_parse_unit_blkio_weight(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata) {
+        Unit *u = data;
+        int r;
+        unsigned long ul;
+        const char *device = NULL, *weight;
+        unsigned k;
+        char *t, **l;
+
+        assert(filename);
+        assert(lvalue);
+        assert(rvalue);
+        assert(data);
+
+        l = strv_split_quoted(rvalue);
+        if (!l)
+                return -ENOMEM;
+
+        k = strv_length(l);
+        if (k < 1 || k > 2) {
+                log_error("[%s:%u] Failed to parse weight value, ignoring: %s", filename, line, rvalue);
+                strv_free(l);
+                return 0;
+        }
+
+        if (k == 1)
+                weight = l[0];
+        else {
+                device = l[0];
+                weight = l[1];
+        }
+
+        if (device && !path_startswith(device, "/dev/")) {
+                log_error("[%s:%u] Failed to parse block device node value, ignoring: %s", filename, line, rvalue);
+                strv_free(l);
+                return 0;
+        }
+
+        if (safe_atolu(weight, &ul) < 0 || ul < 10 || ul > 1000) {
+                log_error("[%s:%u] Failed to parse block IO weight value, ignoring: %s", filename, line, rvalue);
+                strv_free(l);
+                return 0;
+        }
+
+        if (device)
+                r = asprintf(&t, "%s %lu", device, ul);
+        else
+                r = asprintf(&t, "%lu", ul);
+        strv_free(l);
+
+        if (r < 0)
+                return -ENOMEM;
+
+        if (device)
+                r = unit_add_cgroup_attribute(u, "blkio", "blkio.weight_device", t, blkio_map);
+        else
+                r = unit_add_cgroup_attribute(u, "blkio", "blkio.weight", t, NULL);
+        free(t);
+
+        if (r < 0) {
+                log_error("[%s:%u] Failed to add cgroup attribute value, ignoring: %s", filename, line, rvalue);
+                return 0;
+        }
+
+        return 0;
+}
+
+int config_parse_unit_blkio_bandwidth(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata) {
+        Unit *u = data;
+        int r;
+        off_t bytes;
+        unsigned k;
+        char *t, **l;
+
+        assert(filename);
+        assert(lvalue);
+        assert(rvalue);
+        assert(data);
+
+        l = strv_split_quoted(rvalue);
+        if (!l)
+                return -ENOMEM;
+
+        k = strv_length(l);
+        if (k != 2) {
+                log_error("[%s:%u] Failed to parse bandwidth value, ignoring: %s", filename, line, rvalue);
+                strv_free(l);
+                return 0;
+        }
+
+        if (!path_startswith(l[0], "/dev/")) {
+                log_error("[%s:%u] Failed to parse block device node value, ignoring: %s", filename, line, rvalue);
+                strv_free(l);
+                return 0;
+        }
+
+        if (parse_bytes(l[1], &bytes) < 0 || bytes <= 0) {
+                log_error("[%s:%u] Failed to parse block IO bandwith value, ignoring: %s", filename, line, rvalue);
+                strv_free(l);
+                return 0;
+        }
+
+        r = asprintf(&t, "%s %llu", l[0], (unsigned long long) bytes);
+        strv_free(l);
+
+        if (r < 0)
+                return -ENOMEM;
+
+        r = unit_add_cgroup_attribute(u, "blkio",
+                                      streq(lvalue, "BlockIOReadBandwidth") ? "blkio.read_bps_device" : "blkio.write_bps_device",
+                                      t, blkio_map);
+        free(t);
+
+        if (r < 0) {
+                log_error("[%s:%u] Failed to add cgroup attribute value, ignoring: %s", filename, line, rvalue);
+                return 0;
+        }
+
+        return 0;
+}
+
+
 #define FOLLOW_MAX 8
 
 static int open_follow(char **filename, FILE **_f, Set *names, char **_final) {
diff --git a/src/load-fragment.h b/src/load-fragment.h
index 8521ca0918..fbb31f9b9a 100644
--- a/src/load-fragment.h
+++ b/src/load-fragment.h
@@ -80,6 +80,8 @@ int config_parse_unit_cgroup_attr(const char *filename, unsigned line, const cha
 int config_parse_unit_cpu_shares(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
 int config_parse_unit_memory_limit(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
 int config_parse_unit_device_allow(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
+int config_parse_unit_blkio_weight(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
+int config_parse_unit_blkio_bandwidth(const char *filename, unsigned line, const char *section, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
 
 /* gperf prototypes */
 const struct ConfigPerfItem* load_fragment_gperf_lookup(const char *key, unsigned length);