cgroup files: remove cpuset_common_file_write()

This patch tweaks the signatures of the update_cpumask() and update_nodemask() functions so that they can be called directly as handlers for the new cgroups write_string() method. This allows cpuset_common_file_write() to be removed. Signed-off-by: Paul Menage <menage@google.com> Cc: Paul Jackson <pj@sgi.com> Cc: Pavel Emelyanov <xemul@openvz.org> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Serge Hallyn <serue@us.ibm.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroup files: remove cpuset_common_file_write()
This patch tweaks the signatures of the update_cpumask() and update_nodemask() functions so that they can be called directly as handlers for the new cgroups write_string() method. This allows cpuset_common_file_write() to be removed. Signed-off-by: Paul Menage <menage@google.com> Cc: Paul Jackson <pj@sgi.com> Cc: Pavel Emelyanov <xemul@openvz.org> Cc: Balbir Singh <balbir@in.ibm.com> Cc: Serge Hallyn <serue@us.ibm.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Paul Menage · Linus Torvalds
1 parent af351026aa
Showing 1 changed file with 35 additions and 74 deletions Side-by-side Diff
kernel/cpuset.c
@@ -227,10 +227,6 @@
  * The task_struct fields mems_allowed and mems_generation may only
  * be accessed in the context of that task, so require no locks.
  *
- * The cpuset_common_file_write handler for operations that modify
- * the cpuset hierarchy holds cgroup_mutex across the entire operation,
- * single threading all such cpuset modifications across the system.
- *
  * The cpuset_common_file_read() handlers only hold callback_mutex across
  * small pieces of code, such as when reading out possibly multi-word
  * cpumasks and nodemasks.
@@ -772,7 +768,7 @@
  * @cs: the cpuset to consider
  * @buf: buffer of cpu numbers written to this cpuset
  */
-static int update_cpumask(struct cpuset *cs, char *buf)
+static int update_cpumask(struct cpuset *cs, const char *buf)
 {
 	struct cpuset trialcs;
 	struct cgroup_scanner scan;
@@ -792,7 +788,6 @@
 	 * that parsing.  The validate_change() call ensures that cpusets
 	 * with tasks have cpus.
 	 */
-	buf = strstrip(buf);
 	if (!*buf) {
 		cpus_clear(trialcs.cpus_allowed);
 	} else {
@@ -902,7 +897,7 @@
  
 static void *cpuset_being_rebound;
  
-static int update_nodemask(struct cpuset *cs, char *buf)
+static int update_nodemask(struct cpuset *cs, const char *buf)
 {
 	struct cpuset trialcs;
 	nodemask_t oldmem;
@@ -929,7 +924,6 @@
 	 * that parsing.  The validate_change() call ensures that cpusets
 	 * with tasks have memory.
 	 */
-	buf = strstrip(buf);
 	if (!*buf) {
 		nodes_clear(trialcs.mems_allowed);
 	} else {
  
  
@@ -1256,72 +1250,14 @@
 	FILE_SPREAD_SLAB,
 } cpuset_filetype_t;
  
-static ssize_t cpuset_common_file_write(struct cgroup *cont,
-					struct cftype *cft,
-					struct file *file,
-					const char __user *userbuf,
-					size_t nbytes, loff_t *unused_ppos)
-{
-	struct cpuset *cs = cgroup_cs(cont);
-	cpuset_filetype_t type = cft->private;
-	char *buffer;
-	int retval = 0;
-
-	/* Crude upper limit on largest legitimate cpulist user might write. */
-	if (nbytes > 100U + 6 * max(NR_CPUS, MAX_NUMNODES))
-		return -E2BIG;
-
-	/* +1 for nul-terminator */
-	buffer = kmalloc(nbytes + 1, GFP_KERNEL);
-	if (!buffer)
-		return -ENOMEM;
-
-	if (copy_from_user(buffer, userbuf, nbytes)) {
-		retval = -EFAULT;
-		goto out1;
-	}
-	buffer[nbytes] = 0;	/* nul-terminate */
-
-	cgroup_lock();
-
-	if (cgroup_is_removed(cont)) {
-		retval = -ENODEV;
-		goto out2;
-	}
-
-	switch (type) {
-	case FILE_CPULIST:
-		retval = update_cpumask(cs, buffer);
-		break;
-	case FILE_MEMLIST:
-		retval = update_nodemask(cs, buffer);
-		break;
-	default:
-		retval = -EINVAL;
-		goto out2;
-	}
-
-	if (retval == 0)
-		retval = nbytes;
-out2:
-	cgroup_unlock();
-out1:
-	kfree(buffer);
-	return retval;
-}
-
 static int cpuset_write_u64(struct cgroup *cgrp, struct cftype *cft, u64 val)
 {
 	int retval = 0;
 	struct cpuset *cs = cgroup_cs(cgrp);
 	cpuset_filetype_t type = cft->private;
  
-	cgroup_lock();
-
-	if (cgroup_is_removed(cgrp)) {
-		cgroup_unlock();
+	if (!cgroup_lock_live_group(cgrp))
 		return -ENODEV;
-	}
  
 	switch (type) {
 	case FILE_CPU_EXCLUSIVE:
  
@@ -1367,12 +1303,9 @@
 	struct cpuset *cs = cgroup_cs(cgrp);
 	cpuset_filetype_t type = cft->private;
  
-	cgroup_lock();
-
-	if (cgroup_is_removed(cgrp)) {
-		cgroup_unlock();
+	if (!cgroup_lock_live_group(cgrp))
 		return -ENODEV;
-	}
+
 	switch (type) {
 	case FILE_SCHED_RELAX_DOMAIN_LEVEL:
 		retval = update_relax_domain_level(cs, val);
@@ -1386,6 +1319,32 @@
 }
  
 /*
+ * Common handling for a write to a "cpus" or "mems" file.
+ */
+static int cpuset_write_resmask(struct cgroup *cgrp, struct cftype *cft,
+				const char *buf)
+{
+	int retval = 0;
+
+	if (!cgroup_lock_live_group(cgrp))
+		return -ENODEV;
+
+	switch (cft->private) {
+	case FILE_CPULIST:
+		retval = update_cpumask(cgroup_cs(cgrp), buf);
+		break;
+	case FILE_MEMLIST:
+		retval = update_nodemask(cgroup_cs(cgrp), buf);
+		break;
+	default:
+		retval = -EINVAL;
+		break;
+	}
+	cgroup_unlock();
+	return retval;
+}
+
+/*
  * These ascii lists should be read in a single call, by using a user
  * buffer large enough to hold the entire map.  If read in smaller
  * chunks, there is no guarantee of atomicity.  Since the display format
  
@@ -1504,14 +1463,16 @@
 	{
 		.name = "cpus",
 		.read = cpuset_common_file_read,
-		.write = cpuset_common_file_write,
+		.write_string = cpuset_write_resmask,
+		.max_write_len = (100U + 6 * NR_CPUS),
 		.private = FILE_CPULIST,
 	},
  
 	{
 		.name = "mems",
 		.read = cpuset_common_file_read,
-		.write = cpuset_common_file_write,
+		.write_string = cpuset_write_resmask,
+		.max_write_len = (100U + 6 * MAX_NUMNODES),
 		.private = FILE_MEMLIST,
 	},
...	...	@@ -227,10 +227,6 @@
227	227	* The task_struct fields mems_allowed and mems_generation may only
228	228	* be accessed in the context of that task, so require no locks.
229	229	*
230		- * The cpuset_common_file_write handler for operations that modify
231		- * the cpuset hierarchy holds cgroup_mutex across the entire operation,
232		- * single threading all such cpuset modifications across the system.
233		- *
234	230	* The cpuset_common_file_read() handlers only hold callback_mutex across
235	231	* small pieces of code, such as when reading out possibly multi-word
236	232	* cpumasks and nodemasks.
...	...	@@ -772,7 +768,7 @@
772	768	* @cs: the cpuset to consider
773	769	* @buf: buffer of cpu numbers written to this cpuset
774	770	*/
775		-static int update_cpumask(struct cpuset cs, char buf)
	771	+static int update_cpumask(struct cpuset cs, const char buf)
776	772	{
777	773	struct cpuset trialcs;
778	774	struct cgroup_scanner scan;
...	...	@@ -792,7 +788,6 @@
792	788	* that parsing. The validate_change() call ensures that cpusets
793	789	* with tasks have cpus.
794	790	*/
795		- buf = strstrip(buf);
796	791	if (!*buf) {
797	792	cpus_clear(trialcs.cpus_allowed);
798	793	} else {
...	...	@@ -902,7 +897,7 @@
902	897
903	898	static void *cpuset_being_rebound;
904	899
905		-static int update_nodemask(struct cpuset cs, char buf)
	900	+static int update_nodemask(struct cpuset cs, const char buf)
906	901	{
907	902	struct cpuset trialcs;
908	903	nodemask_t oldmem;
...	...	@@ -929,7 +924,6 @@
929	924	* that parsing. The validate_change() call ensures that cpusets
930	925	* with tasks have memory.
931	926	*/
932		- buf = strstrip(buf);
933	927	if (!*buf) {
934	928	nodes_clear(trialcs.mems_allowed);
935	929	} else {
936	930
937	931
...	...	@@ -1256,72 +1250,14 @@
1256	1250	FILE_SPREAD_SLAB,
1257	1251	} cpuset_filetype_t;
1258	1252
1259		-static ssize_t cpuset_common_file_write(struct cgroup *cont,
1260		- struct cftype *cft,
1261		- struct file *file,
1262		- const char __user *userbuf,
1263		- size_t nbytes, loff_t *unused_ppos)
1264		-{
1265		- struct cpuset *cs = cgroup_cs(cont);
1266		- cpuset_filetype_t type = cft->private;
1267		- char *buffer;
1268		- int retval = 0;
1269		-
1270		- /* Crude upper limit on largest legitimate cpulist user might write. */
1271		- if (nbytes > 100U + 6 * max(NR_CPUS, MAX_NUMNODES))
1272		- return -E2BIG;
1273		-
1274		- /* +1 for nul-terminator */
1275		- buffer = kmalloc(nbytes + 1, GFP_KERNEL);
1276		- if (!buffer)
1277		- return -ENOMEM;
1278		-
1279		- if (copy_from_user(buffer, userbuf, nbytes)) {
1280		- retval = -EFAULT;
1281		- goto out1;
1282		- }
1283		- buffer[nbytes] = 0; /* nul-terminate */
1284		-
1285		- cgroup_lock();
1286		-
1287		- if (cgroup_is_removed(cont)) {
1288		- retval = -ENODEV;
1289		- goto out2;
1290		- }
1291		-
1292		- switch (type) {
1293		- case FILE_CPULIST:
1294		- retval = update_cpumask(cs, buffer);
1295		- break;
1296		- case FILE_MEMLIST:
1297		- retval = update_nodemask(cs, buffer);
1298		- break;
1299		- default:
1300		- retval = -EINVAL;
1301		- goto out2;
1302		- }
1303		-
1304		- if (retval == 0)
1305		- retval = nbytes;
1306		-out2:
1307		- cgroup_unlock();
1308		-out1:
1309		- kfree(buffer);
1310		- return retval;
1311		-}
1312		-
1313	1253	static int cpuset_write_u64(struct cgroup cgrp, struct cftype cft, u64 val)
1314	1254	{
1315	1255	int retval = 0;
1316	1256	struct cpuset *cs = cgroup_cs(cgrp);
1317	1257	cpuset_filetype_t type = cft->private;
1318	1258
1319		- cgroup_lock();
1320		-
1321		- if (cgroup_is_removed(cgrp)) {
1322		- cgroup_unlock();
	1259	+ if (!cgroup_lock_live_group(cgrp))
1323	1260	return -ENODEV;
1324		- }
1325	1261
1326	1262	switch (type) {
1327	1263	case FILE_CPU_EXCLUSIVE:
1328	1264
...	...	@@ -1367,12 +1303,9 @@
1367	1303	struct cpuset *cs = cgroup_cs(cgrp);
1368	1304	cpuset_filetype_t type = cft->private;
1369	1305
1370		- cgroup_lock();
1371		-
1372		- if (cgroup_is_removed(cgrp)) {
1373		- cgroup_unlock();
	1306	+ if (!cgroup_lock_live_group(cgrp))
1374	1307	return -ENODEV;
1375		- }
	1308	+
1376	1309	switch (type) {
1377	1310	case FILE_SCHED_RELAX_DOMAIN_LEVEL:
1378	1311	retval = update_relax_domain_level(cs, val);
...	...	@@ -1386,6 +1319,32 @@
1386	1319	}
1387	1320
1388	1321	/*
	1322	+ * Common handling for a write to a "cpus" or "mems" file.
	1323	+ */
	1324	+static int cpuset_write_resmask(struct cgroup cgrp, struct cftype cft,
	1325	+ const char *buf)
	1326	+{
	1327	+ int retval = 0;
	1328	+
	1329	+ if (!cgroup_lock_live_group(cgrp))
	1330	+ return -ENODEV;
	1331	+
	1332	+ switch (cft->private) {
	1333	+ case FILE_CPULIST:
	1334	+ retval = update_cpumask(cgroup_cs(cgrp), buf);
	1335	+ break;
	1336	+ case FILE_MEMLIST:
	1337	+ retval = update_nodemask(cgroup_cs(cgrp), buf);
	1338	+ break;
	1339	+ default:
	1340	+ retval = -EINVAL;
	1341	+ break;
	1342	+ }
	1343	+ cgroup_unlock();
	1344	+ return retval;
	1345	+}
	1346	+
	1347	+/*
1389	1348	* These ascii lists should be read in a single call, by using a user
1390	1349	* buffer large enough to hold the entire map. If read in smaller
1391	1350	* chunks, there is no guarantee of atomicity. Since the display format
1392	1351
...	...	@@ -1504,14 +1463,16 @@
1504	1463	{
1505	1464	.name = "cpus",
1506	1465	.read = cpuset_common_file_read,
1507		- .write = cpuset_common_file_write,
	1466	+ .write_string = cpuset_write_resmask,
	1467	+ .max_write_len = (100U + 6 * NR_CPUS),
1508	1468	.private = FILE_CPULIST,
1509	1469	},
1510	1470
1511	1471	{
1512	1472	.name = "mems",
1513	1473	.read = cpuset_common_file_read,
1514		- .write = cpuset_common_file_write,
	1474	+ .write_string = cpuset_write_resmask,
	1475	+ .max_write_len = (100U + 6 * MAX_NUMNODES),
1515	1476	.private = FILE_MEMLIST,
1516	1477	},
1517	1478