Jul. 13th, 2018

izard: (Default)
Yesterday I posted about a strange bug. I am still debugging, but I found the root cause of "linux kernel has memory between reboots".

The benchmark simply leaves /tmp/benchmarks.pid file with its PID. It works differently if same PID is already used or not. When rebooting with same kernel, PID is likely to stay the same, and when rebooting with different kernel, PID is more likely to change, because of different timings in the scheduler.

My bad, everything is simple. I wish the original bug was so simple to resolve too :(

Profile

izard: (Default)
izard

September 2025

S M T W T F S
 1 23456
78910111213
14151617181920
21222324252627
282930    

Most Popular Tags

Page Summary

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 9th, 2025 03:15 am
Powered by Dreamwidth Studios