Skip to contentSkip to conversation

cgroup: make cgroups info more readable

PATCH In review Huan Yang 3 participants 5 messages Apr 9, 2024 – Apr 22, 2024 cgroup

lore
Share & tools
lore

Huan Yang

author
#0 lore raw

PATCH 1/1 cgroup: make cgroups info more readable

The current cgroups output format is based on tabs, which may cause misalignment of output information.

Using placeholder formatting can make the output information more readable.

---
 kernel/cgroup/cgroup-v1.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
kernel/cgroup/cgroup-v1.c +4 -3
index 520a11cb12f4..c082a78f4c22 100644
@@ -669,15 +669,16 @@ int proc_cgroupstats_show(struct seq_file *m, void *v)
 	struct cgroup_subsys *ss;
 	int i;
 
-	seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n");
+	seq_printf(m, "%16s %16s %16s %16s\n", "#subsys_name", "hierarchy",
+		   "num_cgroups", "enabled");
 	/*
 	 * Grab the subsystems state racily. No need to add avenue to
 	 * cgroup_mutex contention.
 	 */
 
 	for_each_subsys(ss, i)
-		seq_printf(m, "%s\t%d\t%d\t%d\n",
-			   ss->legacy_name, ss->root->hierarchy_id,
+		seq_printf(m, "%16s %16d %16d %16d\n", ss->legacy_name,
+			   ss->root->hierarchy_id,
 			   atomic_read(&ss->root->nr_cgrps),
 			   cgroup_ssid_enabled(i));
 
Signature · 1 line
2.34.1
#1 lore raw

Hi,

I found a discussion about this change in the email thread bellow, and hope it helps you.

https://lore.kernel.org/all/YwMwlMv%[email protected]/#t

Quoted 35 lines · Huan Yang

The current cgroups output format is based on tabs, which may cause misalignment of output information.

Using placeholder formatting can make the output information more readable.

---
 kernel/cgroup/cgroup-v1.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
kernel/cgroup/cgroup-v1.c +4 -3
index 520a11cb12f4..c082a78f4c22 100644
@@ -669,15 +669,16 @@ int proc_cgroupstats_show(struct seq_file *m, void *v)
 	struct cgroup_subsys *ss;
 	int i;
 
-	seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n");
+	seq_printf(m, "%16s %16s %16s %16s\n", "#subsys_name", "hierarchy",
+		   "num_cgroups", "enabled");
 	/*
 	 * Grab the subsystems state racily. No need to add avenue to
 	 * cgroup_mutex contention.
 	 */
 
 	for_each_subsys(ss, i)
-		seq_printf(m, "%s\t%d\t%d\t%d\n",
-			   ss->legacy_name, ss->root->hierarchy_id,
+		seq_printf(m, "%16s %16d %16d %16d\n", ss->legacy_name,
+			   ss->root->hierarchy_id,
 			   atomic_read(&ss->root->nr_cgrps),
 			   cgroup_ssid_enabled(i));
 
#2 lore raw

HI jianfeng

Quoted 6 lines · xiujianfeng

[Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

Hi,

I found a discussion about this change in the email thread bellow, and hope it helps you.

It's helpful to know why this patch not need, thank you.

Quoted 2 lines

https://lore.kernel.org/all/YwMwlMv%[email protected]/#t

I have a question, that, now that only for cgroup1, when I running qemu ubuntu, I got this:

Quoted 2 lines

mount | grep cgroup
cgroup2 on /sys/fs/cgroup type cgroup2 (rw,nosuid,nodev,noexec,relatime,nsdelegate,memory_recursiveprot)

Only cgroup2 mount in my system, but /proc/cgroup also worked, maybe better to disable this when only cgroup2 mounted?

Quoted 36 lines · Huan Yang
Quoted 35 lines

The current cgroups output format is based on tabs, which may cause misalignment of output information.

Using placeholder formatting can make the output information more readable.

---
  kernel/cgroup/cgroup-v1.c | 7 ++++---
  1 file changed, 4 insertions(+), 3 deletions(-)
kernel/cgroup/cgroup-v1.c +4 -3
index 520a11cb12f4..c082a78f4c22 100644
@@ -669,15 +669,16 @@ int proc_cgroupstats_show(struct seq_file *m, void *v)
       struct cgroup_subsys *ss;
       int i;
 
-     seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n");
+     seq_printf(m, "%16s %16s %16s %16s\n", "#subsys_name", "hierarchy",
+                "num_cgroups", "enabled");
       /*
        * Grab the subsystems state racily. No need to add avenue to
        * cgroup_mutex contention.
        */
 
       for_each_subsys(ss, i)
-             seq_printf(m, "%s\t%d\t%d\t%d\n",
-                        ss->legacy_name, ss->root->hierarchy_id,
+             seq_printf(m, "%16s %16d %16d %16d\n", ss->legacy_name,
+                        ss->root->hierarchy_id,
                          atomic_read(&ss->root->nr_cgrps),
                          cgroup_ssid_enabled(i));
 
#3 lore raw

Hi,

Quoted 18 lines · Huan Yang

HI jianfeng

Quoted 8 lines · xiujianfeng

[Some people who received this message don't often get email from [email protected]. Learn why this is important at
https://aka.ms/LearnAboutSenderIdentification ]

Hi,

I found a discussion about this change in the email thread bellow, and hope it helps you.

It's helpful to know why this patch not need, thank you.

Quoted 2 lines

https://lore.kernel.org/all/YwMwlMv%[email protected]/#t

I have a question, that, now that only for cgroup1, when I running qemu ubuntu, I got this:

Quoted 6 lines
Quoted 2 lines

mount | grep cgroup
cgroup2 on /sys/fs/cgroup type cgroup2

(rw,nosuid,nodev,noexec,relatime,nsdelegate,memory_recursiveprot)

Only cgroup2 mount in my system, but /proc/cgroup also worked, maybe better to disable this when only cgroup2 mounted?

I’m not the maintainer, so the official answer to this question should be left to them:).

However, I don’t think this is the right way. Even though the information shown by /proc/cgroups doesn’t seem as useful for cgroup2 as for cgroup1 due to cgroup2 has only single hierarchy, it’s not entirely useless, IMHO.

Quoted 39 lines
Quoted 39 lines · Huan Yang
Quoted 37 lines

The current cgroups output format is based on tabs, which may cause misalignment of output information.

Using placeholder formatting can make the output information more readable.

---
  kernel/cgroup/cgroup-v1.c | 7 ++++---

  1 file changed, 4 insertions(+), 3 deletions(-)
kernel/cgroup/cgroup-v1.c +4 -3
index 520a11cb12f4..c082a78f4c22 100644
@@ -669,15 +669,16 @@ int proc_cgroupstats_show(struct seq_file *m,
 void *v)
        struct cgroup_subsys *ss;
        int i;
 
-     seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n");
+     seq_printf(m, "%16s %16s %16s %16s\n", "#subsys_name",
 "hierarchy",
+                "num_cgroups", "enabled");
        /*
         * Grab the subsystems state racily. No need to add avenue to
         * cgroup_mutex contention.
         */
 
        for_each_subsys(ss, i)
-             seq_printf(m, "%s\t%d\t%d\t%d\n",
-                        ss->legacy_name, ss->root->hierarchy_id,
+             seq_printf(m, "%16s %16d %16d %16d\n", ss->legacy_name,
+                        ss->root->hierarchy_id,
                           atomic_read(&ss->root->nr_cgrps),

                          cgroup_ssid_enabled(i));

#4 lore raw
Quoted 3 lines · xiujianfeng

However, I don’t think this is the right way. Even though the information shown by /proc/cgroups doesn’t seem as useful for cgroup2 as for cgroup1 due to cgroup2 has only single hierarchy, it’s not entirely

OK, thanks for your reply

Quoted 1 line

useless, IMHO.