mm: add ratio in slabinfo print

PATCH In review Xishi Qiu 3 participants 6 messages Jan 11, 2016 – Jan 12, 2016

Subsystems (2)

Memory management Slab allocator

lore

Share & tools

lore

Xishi Qiu

author

2016-01-11 03:58 UTC #0 lore raw

PATCH 1/1 mm: add ratio in slabinfo print

Add ratio(active_objs/num_objs) in /proc/slabinfo, it is used to show the availability factor in each slab. Also adjustment format because some slabs' name is too long.

before applied
...

ext4_inode_cache    1591   3008   1008   32    8 : tunables    0    0    0 : slabdata     94     94      0
ext4_free_data       640    640     64   64    1 : tunables    0    0    0 : slabdata     10     10      0
ext4_allocation_context    480    480    128   32    1 : tunables    0    0    0 : slabdata     15     15      0
ext4_io_end          616    616     72   56    1 : tunables    0    0    0 : slabdata     11     11      0
ext4_extent_status   3979   4794     40  102    1 : tunables    0    0    0 : slabdata     47     47      0

jbd2_journal_handle 1360 1360 48 85 1 : tunables 0 0 0 : slabdata 16 16 0
jbd2_journal_head 510 510 120 34 1 : tunables 0 0 0 : slabdata 15 15 0
jbd2_revoke_table_s 768 768 16 256 1 : tunables 0 0 0 : slabdata 3 3 0
jbd2_revoke_record_s 384 384 32 128 1 : tunables 0 0 0 : slabdata 3 3 0

scsi_data_buffer       0      0     24  170    1 : tunables    0    0    0 : slabdata      0      0      0
cfq_queue            560    560    232   35    2 : tunables    0    0    0 : slabdata     16     16      0
bsg_cmd                0      0    312   26    2 : tunables    0    0    0 : slabdata      0      0      0
mqueue_inode_cache     36     36    896   36    8 : tunables    0    0    0 : slabdata      1      1      0
isofs_inode_cache      0      0    600   27    4 : tunables    0    0    0 : slabdata      0      0      0

hugetlbfs_inode_cache 28 28 568 28 4 : tunables 0 0 0 : slabdata 1 1 0
dquot 448 448 256 32 2 : tunables 0 0 0 : slabdata 14 14 0

after applied
...

ext4_inode_cache            1287   2400     53%   1008   32    8 : tunables    0    0    0 : slabdata     75     75      0
ext4_free_data               640    640    100%     64   64    1 : tunables    0    0    0 : slabdata     10     10      0
ext4_allocation_context      512    512    100%    128   32    1 : tunables    0    0    0 : slabdata     16     16      0
ext4_io_end                  560    560    100%     72   56    1 : tunables    0    0    0 : slabdata     10     10      0
ext4_extent_status          3775   4692     80%     40  102    1 : tunables    0    0    0 : slabdata     46     46      0
jbd2_journal_handle         1360   1360    100%     48   85    1 : tunables    0    0    0 : slabdata     16     16      0
jbd2_journal_head            544    544    100%    120   34    1 : tunables    0    0    0 : slabdata     16     16      0
jbd2_revoke_table_s          768    768    100%     16  256    1 : tunables    0    0    0 : slabdata      3      3      0
jbd2_revoke_record_s         512    512    100%     32  128    1 : tunables    0    0    0 : slabdata      4      4      0
scsi_data_buffer               0      0      0%     24  170    1 : tunables    0    0    0 : slabdata      0      0      0
cfq_queue                    560    560    100%    232   35    2 : tunables    0    0    0 : slabdata     16     16      0
bsg_cmd                        0      0      0%    312   26    2 : tunables    0    0    0 : slabdata      0      0      0
mqueue_inode_cache            36     36    100%    896   36    8 : tunables    0    0    0 : slabdata      1      1      0
isofs_inode_cache              0      0      0%    600   27    4 : tunables    0    0    0 : slabdata      0      0      0
hugetlbfs_inode_cache         28     28    100%    568   28    4 : tunables    0    0    0 : slabdata      1      1      0
dquot                        448    448    100%    256   32    2 : tunables    0    0    0 : slabdata     14     14      0

Signed-off-by Xishi Qiu <[email protected]>

---
 mm/slab_common.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

mm/slab_common.c +8 -5

index 3c6a86b..6f1e130 100644
@@ -1041,8 +1041,8 @@ static void print_slabinfo_header(struct seq_file *m)
 #else
 	seq_puts(m, "slabinfo - version: 2.1\n");
 #endif
-	seq_puts(m, "# name            <active_objs> <num_objs> <objsize> "
-		 "<objperslab> <pagesperslab>");
+	seq_puts(m, "# name                   <active_objs> <num_objs> <ratio> "
+		 "<objsize> <objperslab> <pagesperslab>");
 	seq_puts(m, " : tunables <limit> <batchcount> <sharedfactor>");
 	seq_puts(m, " : slabdata <active_slabs> <num_slabs> <sharedavail>");
 #ifdef CONFIG_DEBUG_SLAB

@@ -1093,15 +1093,18 @@ memcg_accumulate_slabinfo(struct kmem_cache *s, struct slabinfo *info)
 static void cache_show(struct kmem_cache *s, struct seq_file *m)
 {
 	struct slabinfo sinfo;
+	unsigned long ratio;
 
 	memset(&sinfo, 0, sizeof(sinfo));
 	get_slabinfo(s, &sinfo);
 
 	memcg_accumulate_slabinfo(s, &sinfo);
+	ratio = sinfo.num_objs ? sinfo.active_objs * 100 / sinfo.num_objs : 0;
 
-	seq_printf(m, "%-17s %6lu %6lu %6u %4u %4d",
-		   cache_name(s), sinfo.active_objs, sinfo.num_objs, s->size,
-		   sinfo.objects_per_slab, (1 << sinfo.cache_order));
+	seq_printf(m, "%-25s %6lu %6lu %6lu%% %6u %4u %4d",
+		   cache_name(s), sinfo.active_objs, sinfo.num_objs,
+		   ratio, s->size, sinfo.objects_per_slab,
+		   (1 << sinfo.cache_order));
 
 	seq_printf(m, " : tunables %4u %4u %4u",
 		   sinfo.limit, sinfo.batchcount, sinfo.shared);

Signature · 1 line

2.0.0

Michal Hocko

2016-01-11 12:25 UTC #1 lore raw

Quoted 2 lines · Xishi Qiu

Add ratio(active_objs/num_objs) in /proc/slabinfo, it is used to show the availability factor in each slab.

What is the reason to add such a new value when it can be trivially calculated from the userspace?

Besides that such a change would break existing parsers no?

Signature · 8 lines

Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Xishi Qiu

author

2016-01-11 13:40 UTC #2 lore raw

Quoted 8 lines · Michal Hocko

Quoted 2 lines · Xishi Qiu
Add ratio(active_objs/num_objs) in /proc/slabinfo, it is used to show the availability factor in each slab.
What is the reason to add such a new value when it can be trivially calculated from the userspace?
Besides that such a change would break existing parsers no?

Oh, maybe it is.

How about adjustment the format because some names are too long?

Thanks,
Xishi Qiu

Michal Hocko

2016-01-11 13:39 UTC #3 lore raw

Quoted 6 lines · Xishi Qiu

Quoted 4 lines · Michal Hocko
Quoted 2 lines · Xishi Qiu
Add ratio(active_objs/num_objs) in /proc/slabinfo, it is used to show the availability factor in each slab.

Quoted 8 lines

Quoted 4 lines
What is the reason to add such a new value when it can be trivially calculated from the userspace?
Besides that such a change would break existing parsers no?
Oh, maybe it is.
How about adjustment the format because some names are too long?

Parsers should be clever enough to process white spaces properly but there is no guarantee this will be the case. A more important question is whether it really makes sense to change this in the first place. What would be the benefit? Somehow nicer output? Does this justify a potential breakage of tool processing this file? To me this all sounds like such a change is not worth it.

Signature · 8 lines

Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to [email protected].  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

David Rientjes

reviewer

2016-01-12 00:20 UTC #4 lore raw

In reply to Xishi Qiu

Quoted 11 lines · Xishi Qiu

Quoted 8 lines · Xishi Qiu
Quoted 2 lines
Add ratio(active_objs/num_objs) in /proc/slabinfo, it is used to show the availability factor in each slab.
What is the reason to add such a new value when it can be trivially calculated from the userspace?
Besides that such a change would break existing parsers no?
Oh, maybe it is.

If you need the information internally, you could always create a library around slabinfo and export the information for users who are interested for your own use. Doing anything other than appending fields to each line is too dangerous, however, as a general rule.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to [email protected]. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

Xishi Qiu

author

2016-01-12 01:05 UTC #5 lore raw

Quoted 11 lines · David Rientjes

Quoted 9 lines · Xishi Qiu
Quoted 8 lines · Xishi Qiu
Quoted 2 lines
Add ratio(active_objs/num_objs) in /proc/slabinfo, it is used to show the availability factor in each slab.
What is the reason to add such a new value when it can be trivially calculated from the userspace?
Besides that such a change would break existing parsers no?

Quoted 9 lines

Quoted 2 lines
Oh, maybe it is.
If you need the information internally, you could always create a library around slabinfo and export the information for users who are interested for your own use. Doing anything other than appending fields to each line is too dangerous, however, as a general rule.

OK, I know.

Thanks,
Xishi Qiu