Re: [PATCH 1/1] mm/page_alloc: add scheduling point to free_unref_page_list

[wangjianxing]

On 03/03/2022 07:34 AM, Andrew Morton wrote:
> On Tue,  1 Mar 2022 20:38:25 -0500 wangjianxing <wangjianxing@xxxxxxxxxxx> wrote:
>
> > free a large list of pages maybe cause rcu_sched starved on
> > non-preemptible kernels
> >
> > rcu: rcu_sched kthread starved for 5359 jiffies! g454793 f0x0
> > RCU_GP_WAIT_FQS(5) ->state=0x0 ->cpu=19
> > [...]
> > Call Trace:
> >    free_unref_page_list+0x19c/0x270
> >    release_pages+0x3cc/0x498
> >    tlb_flush_mmu_free+0x44/0x70
> >    zap_pte_range+0x450/0x738
> >    unmap_page_range+0x108/0x240
> >    unmap_vmas+0x74/0xf0
> >    unmap_region+0xb0/0x120
> >    do_munmap+0x264/0x438
> >    vm_munmap+0x58/0xa0
> >    sys_munmap+0x10/0x20
> >    syscall_common+0x24/0x38
> Thanks.
>
> How did this large list of pages come about?
>
> Will people be seeing this message in upstream kernels, or is it
> specific to some caller code which you have added?
>
> Please always include details such as this so that others can determine
> whether the fix should be backported into -stable kernels.
Thanks.

I try to increase the overcommit ratio of cpu to 1:2~1:3 in KVM 
hypervisor, per-vm has the same number of vcpu with host cpu, then setup 
2 or 3 vm.
Run ltpstress test in per vm, both host and guest is non-preemptiable 
kernel, vm dmesg will throw some rcu_sched warning.

ltp version is 20180926, but until now I didn't analysis ltpstress code 
deeply.