emo/procps
1
0
Fork 0
Code Issues Pull Requests Packages Releases Activity

top: limit task cpu % based on total number of threads

Browse Source 
Everyone has either accepted this potential distortion
or patched top to eliminate it. Now, the time has come
to regain some consistency when calculating that %CPU.

We'll now limit such values to: 100.0 * total threads.
And, it took way too long to address this little flaw.

(now that we know a '.' + 2 spaces is squeezed to one)
(everything's perfectly justified, but it's just luck)

Reference(s):
http://www.freelists.org/post/procps/top-regression-reports
http://www.freelists.org/post/procps/top-regression-reports,1

Signed-off-by: Jim Warner <james.warner@comcast.net>
This commit is contained in:
Jim Warner 2012-12-18 00:00:00 -06:00 committed by Craig Small
parent 27c2ef1e6d
commit ced5009f75
2 changed files with 12 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
top/top.1
View File

@ -454,9 +454,15 @@ Such fields can, however, be scaled differently via the 'e' \*(CI.
1.\fB %CPU \*(Em \*(PU Usage \fR
The task's share of the elapsed \*(PU time since the last screen update,
expressed as a percentage of total \*(PU time.
In a true SMP environment, if 'Irix mode' is \*F, \*(We will operate
in 'Solaris mode' where a task's \*(Pu usage will be divided by the total
number of \*(PUs.
In a true SMP environment, if a process is multi-threaded and \*(We is
\fInot\fR operating in Threads mode, amounts greater than 100% may be
reported.
You toggle Threads mode with the 'H' \*(CI.
Also for multi-processor environments, if 'Irix mode' is \*F, \*(We
will operate in 'Solaris mode' where a task's \*(Pu usage will be
divided by the total number of \*(PUs.
You toggle 'Irix/Solaris' modes with the 'I' \*(CI.
.TP 4

3
top/top.c
View File

@ -4631,6 +4631,9 @@ static const char *task_show (const WIN_t *q, const proc_t *p) {
break;
case P_CPU:
{ float u = (float)p->pcpu * Frame_etscale;
/* process can't use more %cpu than number of threads it has
( thanks Jaromir Capik <jcapik@redhat.com> ) */
if (u > 100.0 * p->nlwp) u = 100.0 * p->nlwp;
if (u > Cpu_pmax) u = Cpu_pmax;
cp = scale_pcnt(u, W, Jn);
}