This might be a scenario.
In those cases one thread detects a low mem condition that has not been
handled [happens when much memory is allocated fast and kswapd has too
low prio to get run] by kswapd and tries to free some pages itself.
It succeeds. But waits/reschedules to a higher prio process.
During that run the higher prio process steels the pages.
Back to the original trying to get hold of the pages - no success...
Note: since this process is both __GFP_IO and __GFP_WAIT it could
happen during waiting for IO and not only due to the rescheduling in
do_try_to_free_pages()
/RogerL
Mike Kravetz wrote:
>
> Hello,
>
> While running a workload on a 2.4.0-test4 kernel, the message:
>
> VM: killing process db2sysc
>
> is displayed on the console and the process s killed. This message
> is coming from the file fault.c:
>
> /*
> * We ran out of memory, or some other thing happened to us that made
> * us unable to handle the page fault gracefully.
> */
> out_of_memory:
> up(&mm->mmap_sem);
> printk("VM: killing process %s\n", tsk->comm);
>
> Does this mean that the kernel ran out of memory? This system has
> 512MB of mem and 1GB of swap. There is plenty of swap left when
> the process is killed.
>
> This workload runs on the same system with a 2.2.14 kernel. The
> kernels are configured the same (or as much the same as they can be).
>
> I suppose it is possible that with the 2.2.14 kernel my workload
> put kernel memory usage just under the limit, while the 2.4-test4
> pushed it over the limit.
>
> Can anyone provide suggestions on how to further analyze this
> situation? I can start instrumenting the kernel, but am looking for
> better/more established options.
>
> Thanks,
> --
> Mike Kravetz mkravetz@us.ibm.com
> IBM Linux Technology Center
> 15450 SW Koll Parkway T/L 775-3494
> Beaverton, OR 97006-6063 (503)578-3494
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
-- Home page: http://www.norran.net/nra02596/- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Jul 23 2000 - 21:00:13 EST