Statistics might be compounds of several results, or buckets
(see histogram). For any mode of data aggregation, we use one
line per bucket in the output format - with the exception of
a "utilisation indicator", which squeezes a handful of numbers
into a single line (number of samples, minimim, average, maximum,
variance).
This patch splits this compound up into separate and appropriately
named output lines, ridding us of a nasty exception, and improving
readability.
Signed-off-by: Martin Peschke <[email protected]>
---
Documentation/statistics.txt | 22 +++++++++++++++++-----
lib/statistic.c | 14 +++++++++++---
2 files changed, 28 insertions(+), 8 deletions(-)
--- a/lib/statistic.c 2006-10-12 20:37:12.000000000 +0200
+++ b/lib/statistic.c 2006-10-19 19:53:03.000000000 +0200
@@ -952,14 +952,22 @@ static int statistic_fdata_util(struct s
signed long long min = num ? util->min : 0,
max = num ? util->max : 0;
- seg = sgrb_seg_find(&fpriv->read_seg_lh, 128);
+ seg = sgrb_seg_find(&fpriv->read_seg_lh, 512);
if (unlikely(!seg))
return -ENOMEM;
statistic_div(&mean_w, &mean_d, acc, num, 3);
statistic_div(&var_w, &var_d, sqr - mean_w * mean_w, num, 3);
seg->offset += sprintf(seg->address + seg->offset,
- "%s %Lu %Ld %Ld.%03Ld %Ld %Ld.%03Ld\n", name,
- num, min, mean_w, mean_d, max, var_w, var_d);
+ "%s samples %Lu\n"
+ "%s minimum %Ld\n"
+ "%s average %Ld.%03Ld\n"
+ "%s maximum %Ld\n"
+ "%s variance %Ld.%03Ld\n",
+ name, num,
+ name, min,
+ name, mean_w, mean_d,
+ name, max,
+ name, var_w, var_d);
return 0;
}
--- a/Documentation/statistics.txt 2006-10-08 23:03:51.000000000 +0200
+++ b/Documentation/statistics.txt 2006-10-19 20:13:12.000000000 +0200
@@ -199,7 +199,11 @@ has been implemented:
size_write 0x14000 12 |
... |
size_write 0x9000 1 /
- queue_used_depth 970 1 18.122 32 > num min avg max for a queue
+ queue_used_depth samples 970 \
+ queue_used_depth minimum 1 |
+ queue_used_depth average 18.122 > utilisation of a queue
+ queue_used_depth maximum 32 |
+ queue_used_depth variance 53.324 /
Such output can grow as needed in debugfs files. It is human-readable and
could be parsed and postprocessed by simple scripts that are aware of what the
@@ -543,18 +547,26 @@ this:
foo 0x1000 4
foo 0x2000 1
foo 0x5000 2
- bar 961 1 42.000 128
+ bar samples 961
+ bar minimum 1
+ bar average 42.000
+ bar maximum 128
+ bar variance 149.254
Output formats of different statistic types
Statistic Type Output Format Number of Lines
- counter_inc <name> <total of Y> 1
+ counter_inc <name> <total of Y> 1
- counter_prod <name> <total of Xi*Yi> 1
+ counter_prod <name> <total of Xi*Yi> 1
- utilisation <name> <total of Y> <min X> <avg X> <max X> 1
+ utilisation <name> "samples" <total of Y> 5
+ <name> "minimum" <minimum X>
+ <name> "average" <average X>
+ <name> "maximum" <maximum X>
+ <name> "variance" <variance of X>
sparse <name> <Xn> <total of Y for Xn> <= entries
...
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
