On Tue, 4 Mar 2014, He, Bo wrote: > Sorry, the title is confusing. Change it to: mm, oom: normalize the adj to > ensure oom_badness returns a positive number
There's something seriously wrong with your email client, it's not replying correctly to threads so this appears as a completely new thread. Meanwhile, your patch was completely whitespace damaged. Please read Documentation/email-clients.txt. > We are enabling android mobiles. When running stress memory test, there is a > bad issue. Some critical processes such as Healthd and watchdogd are killed, > while some other processes are still alive. > OOM should kill the tasks whose oom_score are biggest. Many processes use a > minus oom_score_adj. oom_badness returns 1 for all of them and their > oom_score are all 0. > No, you're confusing two different things. A negative /proc/pid/oom_score_adj discounts a process's memory usage in comparison to other processes. If this value discounts all memory usage for every process, then your setting of /proc/pid/oom_score_adj is wrong. oom_badness() returns 1 so that the oom killer may still kill these processes, the /proc/pid/oom_score is irrelevant to this (and the only reason it appears as 0 is because it already normalizes by multiplying by 1000 / totalpages). > The patch tries to convert the minus oom_score_adj to a positive number when > calculating oom_score. oom_score can keep right process priority sequence. > Nack, this patch is completely disregarding the value specified by the user for /proc/pid/oom_score_adj. Your userspace should be using /proc/pid/oom_score_adj, not /proc/pid/oom_adj. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
