| Age | Commit message (Collapse) | Author |
|---|
|
Print function names and registered names with ~tw.
Open dump file with encoding utf8.
|
|
|
|
Originally, etop would only wait for 1 second after starting the
collection of process info from the observed node. If data was not
received in time, etop would exit with reason 'connection_lost'.
In some cases, however, this timer would expire due to the amount of
processes on the observed node. To avoid this, the timer is now
extended to the same as the update interval, which means that if the
problem occurs, it can be helped by extending the interval time.
|
|
Since commit 265998a7e412cd382315632964846569c370a5a2, etop reports
the CPU (scheduler) utilization since the tool was started instead of
for the last interval only. This is now corrected.
|
|
|
|
|
|
Use observer for the gui part
|
|
Now using scheduler_wall_time instead of runtime/wall_clock to get cpu
utilization. The old version could give CPU utilization far beyond 100%.
Also, a bug which sometimes gave a badarith when calculating the CPU
utilization is corrected.
|
|
|
|
|
|
|
