RE: Re: Re: [PATCH] mm: Add nr_free_highatomimic to fix incorrect watermatk routine

RE: Re: Re: [PATCH] mm: Add nr_free_highatomimic to fix incorrect watermatk routine

[Park Sangwoo]

>On Mon 02-09-19 13:34:54, Sangwoo� wrote:
>>>On Fri 30-08-19 18:25:53, Sangwoo wrote:
>>>> The highatomic migrate block can be increased to 1% of Total memory.
>>>> And, this is for only highorder ( > 0 order). So, this block size is
>>>> excepted during check watermark if allocation type isn't alloc_harder.
>>>>
>>>> It has problem. The usage of highatomic is already calculated at
>> NR_FREE_PAGES.
>>>> So, if we except total block size of highatomic, it's twice minus size of
>>allocated
>>>> highatomic.
>>>> It's cause allocation fail although free pages enough.
>>>>
>>>> We checked this by random test on my target(8GB RAM).
>>>>
>>>>  Binder:6218_2: page allocation failure: order:0, mode:0x14200ca
>> (GFP_HIGHUSER_MOVABLE), nodemask=(null)
>>>>  Binder:6218_2 cpuset=background mems_allowed=0
>>>
>>>How come this order-0 sleepable allocation fails? The upstream kernel
>>>doesn't fail those allocations unless the process context is killed by
>>>the oom killer.
>> 
>> Most calltacks are zsmalloc, as shown below.
>
>What makes those allocations special so that they fail unlike any other
>normal order-0 requests? Also do you see the same problem with the
>current upstream kernel? Is it possible this is an Android specific
>issue?

There is the other case of fail order-0 fail.
----
hvdcp_opti: page allocation failure: order:0, mode:0x1004000(GFP_NOWAIT|__GFP_COMP), nodemask=(null)
hvdcp_opti cpuset=/ mems_allowed=0
CPU: 0 PID: 1882 Comm: hvdcp_opti Tainted: P S      W  O    4.14.83-perf+ #1
Hardware name: Qualcomm Technologies, Inc. SM6150 PM6150 LG Electronics, mh3_lao_kr, rev-C (DT)
Call trace:
dump_backtrace+0x0/0x1f0
show_stack+0x18/0x20
dump_stack+0xc4/0x100
warn_alloc+0x100/0x198
__alloc_pages_nodemask+0x116c/0x1188
new_slab+0x130/0x5e0
___slab_alloc+0x490/0x610
kmem_cache_alloc+0x2a8/0x2c8
avc_alloc_node+0x34/0x268
avc_compute_av+0xb8/0x1f8
avc_has_perm_noaudit+0xcc/0x100
selinux_inode_permission+0x100/0x1b0
security_inode_permission+0x58/0x78
__inode_permission2+0x40/0xe8
may_open+0x78/0x118
path_openat+0x8f8/0x14d0
do_filp_open+0x74/0x120
do_sys_open+0x13c/0x260
SyS_openat+0x10/0x18
el0_svc_naked+0x34/0x38
snipped...
DMA free:11320kB min:3440kB low:46092kB high:47812kB active_anon:143344kB inactive_anon:145812kB active_file:171900kB inactive_file:146976kB u
lowmem_reserve[]: 0 1901 1901
Normal free:3928kB min:3940kB low:52748kB high:54716kB active_anon:85100kB inactive_anon:81772kB active_file:103312kB inactive_file:114732kB u
lowmem_reserve[]: 0 0 0
DMA: 343*4kB (UMECH) 947*8kB (UMCH) 26*16kB (UH) 23*32kB (UH) 11*64kB (H) 6*128kB (H) 3*256kB (H) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 12340kB
Normal: 798*4kB (UMH) 104*8kB (UMH) 20*16kB (U) 3*32kB (UH) 11*64kB (H) 1*128kB (H) 1*256kB (H) 0*512kB 0*1024kB 0*2048kB 0*4096kB = 5528kB
----
In my test, most case are using camera. So, memory usage is increased momentarily,
it cause free page go to under low value of watermark.
If free page is under low and 0-order fail is occured, its normal operation.
But, although free page is higher than min, fail is occurred.
After fix routin for checking highatomic size, it's not reproduced.

I now develop smartphone is applied kernel-4.14. And I didn't checked current upstream kernel.
I thinks this symptom can be occurred the any platform that have use-case of memory intensive moment.

>
>>  Call trace:
>>   dump_backtrace+0x0/0x1f0
>>   show_stack+0x18/0x20
>>   dump_stack+0xc4/0x100
>>   warn_alloc+0x100/0x198
>>   __alloc_pages_nodemask+0x116c/0x1188
>>   do_swap_page+0x10c/0x6f0
>>   handle_pte_fault+0x12c/0xfe0
>>   handle_mm_fault+0x1d0/0x328
>>   do_page_fault+0x2a0/0x3e0
>>   do_translation_fault+0x44/0xa8
>>   do_mem_abort+0x4c/0xd0
>>   el1_da+0x24/0x84
>>   __arch_copy_to_user+0x5c/0x220
>>   binder_ioctl+0x20c/0x740
>>   compat_SyS_ioctl+0x128/0x248
>>   __sys_trace_return+0x0/0x4
>

Re: Re: Re: [PATCH] mm: Add nr_free_highatomimic to fix incorrect watermatk routine

[Michal Hocko]

On Tue 03-09-19 18:59:59, Park Sangwoo wrote:
> >On Mon 02-09-19 13:34:54, Sangwoo� wrote:
> >>>On Fri 30-08-19 18:25:53, Sangwoo wrote:
> >>>> The highatomic migrate block can be increased to 1% of Total memory.
> >>>> And, this is for only highorder ( > 0 order). So, this block size is
> >>>> excepted during check watermark if allocation type isn't alloc_harder.
> >>>>
> >>>> It has problem. The usage of highatomic is already calculated at
> >> NR_FREE_PAGES.
> >>>> So, if we except total block size of highatomic, it's twice minus size of
> >>allocated
> >>>> highatomic.
> >>>> It's cause allocation fail although free pages enough.
> >>>>
> >>>> We checked this by random test on my target(8GB RAM).
> >>>>
> >>>>  Binder:6218_2: page allocation failure: order:0, mode:0x14200ca
> >> (GFP_HIGHUSER_MOVABLE), nodemask=(null)
> >>>>  Binder:6218_2 cpuset=background mems_allowed=0
> >>>
> >>>How come this order-0 sleepable allocation fails? The upstream kernel
> >>>doesn't fail those allocations unless the process context is killed by
> >>>the oom killer.
> >> 
> >> Most calltacks are zsmalloc, as shown below.
> >
> >What makes those allocations special so that they fail unlike any other
> >normal order-0 requests? Also do you see the same problem with the
> >current upstream kernel? Is it possible this is an Android specific
> >issue?
> 
> There is the other case of fail order-0 fail.
> ----
> hvdcp_opti: page allocation failure: order:0, mode:0x1004000(GFP_NOWAIT|__GFP_COMP), nodemask=(null)

This is an atomic allocation and failing that one is not a problem
usually. High atomic reservations might prevent GFP_NOWAIT allocation
from suceeding but I do not see that as a problem. This is the primary
purpose of the reservation. 
[...]
> In my test, most case are using camera. So, memory usage is increased momentarily,
> it cause free page go to under low value of watermark.
> If free page is under low and 0-order fail is occured, its normal operation.
> But, although free page is higher than min, fail is occurred.
> After fix routin for checking highatomic size, it's not reproduced.

But you are stealing from the atomic reserves and thus defeating the
purpose of it.
-- 
Michal Hocko
SUSE Labs