* DAMON VA regions don't split on an large Android APP
@ 2022-04-26 23:19 Barry Song
2022-04-27 0:21 ` sj
2022-04-27 6:56 ` Rongwei Wang
0 siblings, 2 replies; 23+ messages in thread
From: Barry Song @ 2022-04-26 23:19 UTC (permalink / raw)
To: sj, Andrew Morton, Linux-MM, LKML
Cc: Matthew Wilcox, shuah, brendanhiggins, foersleo, sieberf,
Shakeel Butt, sjpark, tuhailong, Song Jiang,
张诗明(Simon Zhang),
李培锋(wink)
Hi SeongJae & Andrew,
(also Cc-ed main damon developers)
On an Android phone, I tried to use the DAMON vaddr monitor and found
that vaddr regions don't split well on large Android Apps though
everything works well on native Apps.
I have tried the below two cases on an Android phone with 12GB memory
and snapdragon 888 CPU.
1. a native program with small memory working set as below,
#define size (1024*1024*100)
main()
{
volatile int *p = malloc(size);
memset(p, 0x55, size);
while(1) {
int i;
for (i = 0; i < size / 4; i++)
(void)*(p + i);
usleep(1000);
for (i = 0; i < size / 16; i++)
(void)*(p + i);
usleep(1000);
}
}
For this application, the Damon vaddr monitor works very well.
I have modified monitor.py in the damo userspace tool a little bit to
show the raw data getting from the kernel.
Regions can split decently on this kind of applications, a typical raw
data is as below,
monitoring_start: 2.224 s
monitoring_end: 2.329 s
monitoring_duration: 104.336 ms
target_id: 0
nr_regions: 24
005fb37b2000-005fb734a000( 59.594 MiB): 0
005fb734a000-005fbaf95000( 60.293 MiB): 0
005fbaf95000-005fbec0b000( 60.461 MiB): 0
005fbec0b000-005fc2910000( 61.020 MiB): 0
005fc2910000-005fc6769000( 62.348 MiB): 0
005fc6769000-005fca33f000( 59.836 MiB): 0
005fca33f000-005fcdc8b000( 57.297 MiB): 0
005fcdc8b000-005fd115a000( 52.809 MiB): 0
005fd115a000-005fd45bd000( 52.387 MiB): 0
007661c59000-007661ee4000( 2.543 MiB): 2
007661ee4000-0076623e4000( 5.000 MiB): 3
0076623e4000-007662837000( 4.324 MiB): 2
007662837000-0076630f1000( 8.727 MiB): 3
0076630f1000-007663494000( 3.637 MiB): 2
007663494000-007663753000( 2.746 MiB): 1
007663753000-007664251000( 10.992 MiB): 3
007664251000-0076666fd000( 36.672 MiB): 2
0076666fd000-007666e73000( 7.461 MiB): 1
007666e73000-007667c89000( 14.086 MiB): 2
007667c89000-007667f97000( 3.055 MiB): 0
007667f97000-007668112000( 1.480 MiB): 1
007668112000-00766820f000(1012.000 KiB): 0
007ff27b7000-007ff27d6000( 124.000 KiB): 0
007ff27d6000-007ff27d8000( 8.000 KiB): 8
2. a large Android app like Asphalt 9
For this case, basically regions can't split very well, but monitor
works on small vma:
monitoring_start: 2.220 s
monitoring_end: 2.318 s
monitoring_duration: 98.576 ms
target_id: 0
nr_regions: 15
000012c00000-0001c301e000( 6.754 GiB): 0
0001c301e000-000371b6c000( 6.730 GiB): 0
000371b6c000-000400000000( 2.223 GiB): 0
005c6759d000-005c675a2000( 20.000 KiB): 0
005c675a2000-005c675a3000( 4.000 KiB): 3
005c675a3000-005c675a7000( 16.000 KiB): 0
0072f1e14000-0074928d4000( 6.510 GiB): 0
0074928d4000-00763c71f000( 6.655 GiB): 0
00763c71f000-0077e863e000( 6.687 GiB): 0
0077e863e000-00798e214000( 6.590 GiB): 0
00798e214000-007b0e48a000( 6.002 GiB): 0
007b0e48a000-007c62f00000( 5.323 GiB): 0
007c62f00000-007defb19000( 6.199 GiB): 0
007defb19000-007f794ef000( 6.150 GiB): 0
007f794ef000-007fe8f53000( 1.745 GiB): 0
As you can see, we have some regions which are very very big and they
are losing the chance to be splitted. But
Damon can still monitor memory access for those small VMA areas very well like:
005c675a2000-005c675a3000( 4.000 KiB): 3
Typical characteristics of a large Android app is that it has
thousands of vma and very large virtual address spaces:
~/damo # pmap 2550 | wc -l
8522
~/damo # pmap 2550
...
0000007992bbe000 4K r---- [ anon ]
0000007992bbf000 24K rw--- [ anon ]
0000007fe8753000 4K ----- [ anon ]
0000007fe8754000 8188K rw--- [ stack ]
total 36742112K
Because the whole vma list is too long, I have put the list here for
you to download:
wget http://www.linuxep.com/patches/android-app-vmas
I can reproduce this problem on other Apps like youtube as well.
I suppose we need to boost the algorithm of splitting regions for this
kind of application.
Any thoughts?
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-26 23:19 DAMON VA regions don't split on an large Android APP Barry Song
@ 2022-04-27 0:21 ` sj
2022-04-27 2:08 ` Barry Song
2022-04-27 6:56 ` Rongwei Wang
1 sibling, 1 reply; 23+ messages in thread
From: sj @ 2022-04-27 0:21 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
linux-damon
Hello Barry,
Thank you so much for sharing your great findings! :)
On Wed, 27 Apr 2022 11:19:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
> Hi SeongJae & Andrew,
> (also Cc-ed main damon developers)
> On an Android phone, I tried to use the DAMON vaddr monitor and found
> that vaddr regions don't split well on large Android Apps though
> everything works well on native Apps.
>
> I have tried the below two cases on an Android phone with 12GB memory
> and snapdragon 888 CPU.
> 1. a native program with small memory working set as below,
> #define size (1024*1024*100)
> main()
> {
> volatile int *p = malloc(size);
> memset(p, 0x55, size);
>
> while(1) {
> int i;
> for (i = 0; i < size / 4; i++)
> (void)*(p + i);
> usleep(1000);
>
> for (i = 0; i < size / 16; i++)
> (void)*(p + i);
> usleep(1000);
>
> }
> }
> For this application, the Damon vaddr monitor works very well.
> I have modified monitor.py in the damo userspace tool a little bit to
> show the raw data getting from the kernel.
> Regions can split decently on this kind of applications, a typical raw
> data is as below,
>
> monitoring_start: 2.224 s
> monitoring_end: 2.329 s
> monitoring_duration: 104.336 ms
> target_id: 0
> nr_regions: 24
> 005fb37b2000-005fb734a000( 59.594 MiB): 0
> 005fb734a000-005fbaf95000( 60.293 MiB): 0
> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> 005fbec0b000-005fc2910000( 61.020 MiB): 0
> 005fc2910000-005fc6769000( 62.348 MiB): 0
> 005fc6769000-005fca33f000( 59.836 MiB): 0
> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> 005fd115a000-005fd45bd000( 52.387 MiB): 0
> 007661c59000-007661ee4000( 2.543 MiB): 2
> 007661ee4000-0076623e4000( 5.000 MiB): 3
> 0076623e4000-007662837000( 4.324 MiB): 2
> 007662837000-0076630f1000( 8.727 MiB): 3
> 0076630f1000-007663494000( 3.637 MiB): 2
> 007663494000-007663753000( 2.746 MiB): 1
> 007663753000-007664251000( 10.992 MiB): 3
> 007664251000-0076666fd000( 36.672 MiB): 2
> 0076666fd000-007666e73000( 7.461 MiB): 1
> 007666e73000-007667c89000( 14.086 MiB): 2
> 007667c89000-007667f97000( 3.055 MiB): 0
> 007667f97000-007668112000( 1.480 MiB): 1
> 007668112000-00766820f000(1012.000 KiB): 0
> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>
> 2. a large Android app like Asphalt 9
> For this case, basically regions can't split very well, but monitor
> works on small vma:
>
> monitoring_start: 2.220 s
> monitoring_end: 2.318 s
> monitoring_duration: 98.576 ms
> target_id: 0
> nr_regions: 15
> 000012c00000-0001c301e000( 6.754 GiB): 0
> 0001c301e000-000371b6c000( 6.730 GiB): 0
> 000371b6c000-000400000000( 2.223 GiB): 0
> 005c6759d000-005c675a2000( 20.000 KiB): 0
> 005c675a2000-005c675a3000( 4.000 KiB): 3
> 005c675a3000-005c675a7000( 16.000 KiB): 0
> 0072f1e14000-0074928d4000( 6.510 GiB): 0
> 0074928d4000-00763c71f000( 6.655 GiB): 0
> 00763c71f000-0077e863e000( 6.687 GiB): 0
> 0077e863e000-00798e214000( 6.590 GiB): 0
> 00798e214000-007b0e48a000( 6.002 GiB): 0
> 007b0e48a000-007c62f00000( 5.323 GiB): 0
> 007c62f00000-007defb19000( 6.199 GiB): 0
> 007defb19000-007f794ef000( 6.150 GiB): 0
> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>
> As you can see, we have some regions which are very very big and they
> are losing the chance to be splitted. But
> Damon can still monitor memory access for those small VMA areas very well like:
> 005c675a2000-005c675a3000( 4.000 KiB): 3
In short, DAMON doesn't set regions based on VMA but access pattern, and
therefore this looks not a problem.
DAMON allows users set min/max monitoring overhead limit and provides a best
accuracy under the condition. In detail, users are allowed to set the min/max
monitoring regions as DAMON's monitoring overhead is proportional to the number
of regions. DAMON provides best effort accuracy under the condition by
splitting and merging regions so that pages in each region has different access
frequency.
The default min number of regions is 10. I believe that's why there are many 6
GiB regions.
If we don't see small regions having some non-zero access frequency, we would
be better to be worried. However, it is finding the small 4 KiB regions having
higher access frequency successfully. The 4 KiB region is not because the
region is having 4 KiB VMA, but the address region shows high access frequency.
>
> Typical characteristics of a large Android app is that it has
> thousands of vma and very large virtual address spaces:
> ~/damo # pmap 2550 | wc -l
> 8522
>
> ~/damo # pmap 2550
> ...
> 0000007992bbe000 4K r---- [ anon ]
> 0000007992bbf000 24K rw--- [ anon ]
> 0000007fe8753000 4K ----- [ anon ]
> 0000007fe8754000 8188K rw--- [ stack ]
> total 36742112K
>
> Because the whole vma list is too long, I have put the list here for
> you to download:
> wget http://www.linuxep.com/patches/android-app-vmas
>
> I can reproduce this problem on other Apps like youtube as well.
> I suppose we need to boost the algorithm of splitting regions for this
> kind of application.
> Any thoughts?
As mentioned above, this looks not a problem, as DAMON's monitoring regions is
not constructed based on VMAs but access patterns.
Nevertheless, I believe there are many rooms for improvement of DAMON's access
frequency. I want to implement fixed-gran monitoring feature first, and
develop some accuracy optimizations using the fixed-gran monitoring as
comparison target.
If I'm missing something or the explanation was not enough, please feel free to
let me know.
Thank,
SJ
>
> Thanks
> Barry
>
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 0:21 ` sj
@ 2022-04-27 2:08 ` Barry Song
2022-04-27 17:39 ` sj
0 siblings, 1 reply; 23+ messages in thread
From: Barry Song @ 2022-04-27 2:08 UTC (permalink / raw)
To: sj
Cc: Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
linux-damon
On Wed, Apr 27, 2022 at 12:21 PM <sj@kernel.org> wrote:
>
> Hello Barry,
>
>
> Thank you so much for sharing your great findings! :)
Thanks for your quick response.
>
> On Wed, 27 Apr 2022 11:19:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
>
> > Hi SeongJae & Andrew,
> > (also Cc-ed main damon developers)
> > On an Android phone, I tried to use the DAMON vaddr monitor and found
> > that vaddr regions don't split well on large Android Apps though
> > everything works well on native Apps.
> >
> > I have tried the below two cases on an Android phone with 12GB memory
> > and snapdragon 888 CPU.
> > 1. a native program with small memory working set as below,
> > #define size (1024*1024*100)
> > main()
> > {
> > volatile int *p = malloc(size);
> > memset(p, 0x55, size);
> >
> > while(1) {
> > int i;
> > for (i = 0; i < size / 4; i++)
> > (void)*(p + i);
> > usleep(1000);
> >
> > for (i = 0; i < size / 16; i++)
> > (void)*(p + i);
> > usleep(1000);
> >
> > }
> > }
> > For this application, the Damon vaddr monitor works very well.
> > I have modified monitor.py in the damo userspace tool a little bit to
> > show the raw data getting from the kernel.
> > Regions can split decently on this kind of applications, a typical raw
> > data is as below,
> >
> > monitoring_start: 2.224 s
> > monitoring_end: 2.329 s
> > monitoring_duration: 104.336 ms
> > target_id: 0
> > nr_regions: 24
> > 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > 005fc2910000-005fc6769000( 62.348 MiB): 0
> > 005fc6769000-005fca33f000( 59.836 MiB): 0
> > 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > 007661c59000-007661ee4000( 2.543 MiB): 2
> > 007661ee4000-0076623e4000( 5.000 MiB): 3
> > 0076623e4000-007662837000( 4.324 MiB): 2
> > 007662837000-0076630f1000( 8.727 MiB): 3
> > 0076630f1000-007663494000( 3.637 MiB): 2
> > 007663494000-007663753000( 2.746 MiB): 1
> > 007663753000-007664251000( 10.992 MiB): 3
> > 007664251000-0076666fd000( 36.672 MiB): 2
> > 0076666fd000-007666e73000( 7.461 MiB): 1
> > 007666e73000-007667c89000( 14.086 MiB): 2
> > 007667c89000-007667f97000( 3.055 MiB): 0
> > 007667f97000-007668112000( 1.480 MiB): 1
> > 007668112000-00766820f000(1012.000 KiB): 0
> > 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> >
> > 2. a large Android app like Asphalt 9
> > For this case, basically regions can't split very well, but monitor
> > works on small vma:
> >
> > monitoring_start: 2.220 s
> > monitoring_end: 2.318 s
> > monitoring_duration: 98.576 ms
> > target_id: 0
> > nr_regions: 15
> > 000012c00000-0001c301e000( 6.754 GiB): 0
> > 0001c301e000-000371b6c000( 6.730 GiB): 0
> > 000371b6c000-000400000000( 2.223 GiB): 0
> > 005c6759d000-005c675a2000( 20.000 KiB): 0
> > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > 005c675a3000-005c675a7000( 16.000 KiB): 0
> > 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > 0074928d4000-00763c71f000( 6.655 GiB): 0
> > 00763c71f000-0077e863e000( 6.687 GiB): 0
> > 0077e863e000-00798e214000( 6.590 GiB): 0
> > 00798e214000-007b0e48a000( 6.002 GiB): 0
> > 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > 007c62f00000-007defb19000( 6.199 GiB): 0
> > 007defb19000-007f794ef000( 6.150 GiB): 0
> > 007f794ef000-007fe8f53000( 1.745 GiB): 0
> >
> > As you can see, we have some regions which are very very big and they
> > are losing the chance to be splitted. But
> > Damon can still monitor memory access for those small VMA areas very well like:
> > 005c675a2000-005c675a3000( 4.000 KiB): 3
>
> In short, DAMON doesn't set regions based on VMA but access pattern, and
> therefore this looks not a problem.
>
> DAMON allows users set min/max monitoring overhead limit and provides a best
> accuracy under the condition. In detail, users are allowed to set the min/max
> monitoring regions as DAMON's monitoring overhead is proportional to the number
> of regions. DAMON provides best effort accuracy under the condition by
> splitting and merging regions so that pages in each region has different access
> frequency.
>
> The default min number of regions is 10. I believe that's why there are many 6
> GiB regions.
>
i had actually tried to set min regions to 100 as below:
/sys/kernel/debug/damon # echo 5000 100000 60000000 100 1000 > attrs
/sys/kernel/debug/damon # cat attrs
5000 100000 60000000 100 1000
but it seems i am still only getting regions like 16:
monitoring_start: 805.192 ms
monitoring_end: 905.385 ms
monitoring_duration: 100.193 ms
target_id: 0
nr_regions: 16
000012c00000-0001fc021000( 7.645 GiB): 0
0001fc021000-0003e6b9a000( 7.667 GiB): 0
0003e6b9a000-000400000000( 404.398 MiB): 0
005e95645000-005e9564a000( 20.000 KiB): 0
005e9564a000-005e9564b000( 4.000 KiB): 8
005e9564b000-005e9564c000( 4.000 KiB): 1
005e9564c000-005e9564f000( 12.000 KiB): 0
006ffffff000-0071fa4f8000( 7.911 GiB): 0
0071fa4f8000-0073dacd1000( 7.508 GiB): 0
0073dacd1000-00759533c000( 6.913 GiB): 0
00759533c000-0076cc999000( 4.866 GiB): 0
0076cc999000-00788bea1000( 6.989 GiB): 0
00788bea1000-007a6fe14000( 7.562 GiB): 0
007a6fe14000-007c5c4a3000( 7.694 GiB): 0
007c5c4a3000-007df9bd8000( 6.460 GiB): 0
007df9bd8000-007fe6d7b000( 7.705 GiB): 0
so it seems the 100 set in min regions doesn't mean we will have at
least 100 regions?
> If we don't see small regions having some non-zero access frequency, we would
> be better to be worried. However, it is finding the small 4 KiB regions having
> higher access frequency successfully. The 4 KiB region is not because the
> region is having 4 KiB VMA, but the address region shows high access frequency.
>
> >
> > Typical characteristics of a large Android app is that it has
> > thousands of vma and very large virtual address spaces:
> > ~/damo # pmap 2550 | wc -l
> > 8522
> >
> > ~/damo # pmap 2550
> > ...
> > 0000007992bbe000 4K r---- [ anon ]
> > 0000007992bbf000 24K rw--- [ anon ]
> > 0000007fe8753000 4K ----- [ anon ]
> > 0000007fe8754000 8188K rw--- [ stack ]
> > total 36742112K
> >
> > Because the whole vma list is too long, I have put the list here for
> > you to download:
> > wget http://www.linuxep.com/patches/android-app-vmas
> >
> > I can reproduce this problem on other Apps like youtube as well.
> > I suppose we need to boost the algorithm of splitting regions for this
> > kind of application.
> > Any thoughts?
>
> As mentioned above, this looks not a problem, as DAMON's monitoring regions is
> not constructed based on VMAs but access patterns.
What makes me believe it is a problem is that it seems we are getting
an incorrect
wss based on the data reported from kernel:
~/damo # ./damo monitor --report_type wss --count 20 2561
# <percentile> <wss>
# target_id 0
# avr: 443.854 MiB
0 0 B |
|
25 4.000 KiB |
|
50 12.000 KiB |
|
75 20.000 KiB |
|
100 5.851 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 666.112 MiB
0 0 B |
|
25 4.000 KiB |
|
50 8.000 KiB |
|
75 20.000 KiB |
|
100 6.159 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 566.706 MiB
0 0 B |
|
25 4.000 KiB |
|
50 20.000 KiB |
|
75 36.000 KiB |
|
100 4.654 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 977.249 MiB
0 4.000 KiB |
|
25 12.000 KiB |
|
50 20.000 KiB |
|
75 20.000 KiB |
|
100 8.798 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 837.221 MiB
0 0 B |
|
25 0 B |
|
50 20.000 KiB |
|
75 422.656 MiB |****
|
100 5.113 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 456.690 MiB
0 4.000 KiB |
|
25 5.180 MiB |
|
50 10.023 MiB |
|
75 84.137 MiB |*
|
100 4.503 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 179.652 MiB
0 0 B |
|
25 4.000 KiB |
|
50 4.000 KiB |
|
75 20.000 KiB |
|
100 3.166 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 878.505 MiB
0 4.000 KiB |
|
25 4.000 KiB |
|
50 12.000 KiB |
|
75 20.000 KiB |
|
100 6.741 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 1.000 GiB
0 4.000 KiB |
|
25 4.000 KiB |
|
50 12.000 KiB |
|
75 1004.711 MiB |*******
|
100 7.706 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 1.187 GiB
0 0 B |
|
25 8.000 KiB |
|
50 16.000 KiB |
|
75 757.449 MiB |*******
|
100 5.964 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 991.889 MiB
0 0 B |
|
25 0 B |
|
50 36.000 KiB |
|
75 1.824 GiB |************************
|
100 4.399 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 545.213 MiB
0 0 B |
|
25 4.000 KiB |
|
50 8.000 KiB |
|
75 24.000 KiB |
|
100 6.270 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 523.257 MiB
0 0 B |
|
25 0 B |
|
50 8.000 KiB |
|
75 28.000 KiB |
|
100 3.789 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 1.395 GiB
0 0 B |
|
25 8.000 KiB |
|
50 40.000 KiB |
|
75 1.555 GiB |**************
|
100 6.220 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 685.373 MiB
0 0 B |
|
25 0 B |
|
50 8.000 KiB |
|
75 28.000 KiB |
|
100 5.984 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 411.626 MiB
0 0 B |
|
25 0 B |
|
50 8.000 KiB |
|
75 20.000 KiB |
|
100 3.958 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 402.370 MiB
0 0 B |
|
25 0 B |
|
50 0 B |
|
75 12.000 KiB |
|
100 6.263 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 656.361 MiB
0 0 B |
|
25 0 B |
|
50 0 B |
|
75 28.000 KiB |
|
100 5.275 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 189.350 MiB
0 0 B |
|
25 0 B |
|
50 8.000 KiB |
|
75 20.000 KiB |
|
100 3.698 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 489.517 MiB
0 0 B |
|
25 0 B |
|
50 12.000 KiB |
|
75 32.000 KiB |
|
100 4.663 GiB
|***********************************************************|
And I have a question, what do percentile 0,25,50,75 mean here?
Why are they so different with percentile 100?
For example, 0,25,50,75 has only KiB but 100 has GiB.
I guess Asphalt 9 should be normally accessing hundreds of megabytes of memory.
>
> Nevertheless, I believe there are many rooms for improvement of DAMON's access
> frequency. I want to implement fixed-gran monitoring feature first, and
> develop some accuracy optimizations using the fixed-gran monitoring as
> comparison target.
Does fixed-gran mean splitting VA to some regions equally with the same size?
for example, if we have 1GB VA, we split it into 512 regions in 2MB
size, something like that?
>
> If I'm missing something or the explanation was not enough, please feel free to
> let me know.
>
>
> Thank,
> SJ
>
> >
> > Thanks
> > Barry
> >
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-26 23:19 DAMON VA regions don't split on an large Android APP Barry Song
2022-04-27 0:21 ` sj
@ 2022-04-27 6:56 ` Rongwei Wang
2022-04-27 7:44 ` Barry Song
1 sibling, 1 reply; 23+ messages in thread
From: Rongwei Wang @ 2022-04-27 6:56 UTC (permalink / raw)
To: Barry Song, sj, Andrew Morton, Linux-MM, LKML
Cc: Matthew Wilcox, shuah, brendanhiggins, foersleo, sieberf,
Shakeel Butt, sjpark, tuhailong, Song Jiang,
张诗明(Simon Zhang),
李培锋(wink),
xhao
On 4/27/22 7:19 AM, Barry Song wrote:
> Hi SeongJae & Andrew,
> (also Cc-ed main damon developers)
> On an Android phone, I tried to use the DAMON vaddr monitor and found
> that vaddr regions don't split well on large Android Apps though
> everything works well on native Apps.
>
> I have tried the below two cases on an Android phone with 12GB memory
> and snapdragon 888 CPU.
> 1. a native program with small memory working set as below,
> #define size (1024*1024*100)
> main()
> {
> volatile int *p = malloc(size);
> memset(p, 0x55, size);
>
> while(1) {
> int i;
> for (i = 0; i < size / 4; i++)
> (void)*(p + i);
> usleep(1000);
>
> for (i = 0; i < size / 16; i++)
> (void)*(p + i);
> usleep(1000);
>
> }
> }
> For this application, the Damon vaddr monitor works very well.
> I have modified monitor.py in the damo userspace tool a little bit to
> show the raw data getting from the kernel.
> Regions can split decently on this kind of applications, a typical raw
> data is as below,
>
> monitoring_start: 2.224 s
> monitoring_end: 2.329 s
> monitoring_duration: 104.336 ms
> target_id: 0
> nr_regions: 24
> 005fb37b2000-005fb734a000( 59.594 MiB): 0
> 005fb734a000-005fbaf95000( 60.293 MiB): 0
> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> 005fbec0b000-005fc2910000( 61.020 MiB): 0
> 005fc2910000-005fc6769000( 62.348 MiB): 0
> 005fc6769000-005fca33f000( 59.836 MiB): 0
> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> 005fd115a000-005fd45bd000( 52.387 MiB): 0
> 007661c59000-007661ee4000( 2.543 MiB): 2
> 007661ee4000-0076623e4000( 5.000 MiB): 3
> 0076623e4000-007662837000( 4.324 MiB): 2
> 007662837000-0076630f1000( 8.727 MiB): 3
> 0076630f1000-007663494000( 3.637 MiB): 2
> 007663494000-007663753000( 2.746 MiB): 1
> 007663753000-007664251000( 10.992 MiB): 3
> 007664251000-0076666fd000( 36.672 MiB): 2
> 0076666fd000-007666e73000( 7.461 MiB): 1
> 007666e73000-007667c89000( 14.086 MiB): 2
> 007667c89000-007667f97000( 3.055 MiB): 0
> 007667f97000-007668112000( 1.480 MiB): 1
> 007668112000-00766820f000(1012.000 KiB): 0
> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>
> 2. a large Android app like Asphalt 9
> For this case, basically regions can't split very well, but monitor
> works on small vma:
>
> monitoring_start: 2.220 s
> monitoring_end: 2.318 s
> monitoring_duration: 98.576 ms
> target_id: 0
> nr_regions: 15
> 000012c00000-0001c301e000( 6.754 GiB): 0
> 0001c301e000-000371b6c000( 6.730 GiB): 0
> 000371b6c000-000400000000( 2.223 GiB): 0
> 005c6759d000-005c675a2000( 20.000 KiB): 0
> 005c675a2000-005c675a3000( 4.000 KiB): 3
> 005c675a3000-005c675a7000( 16.000 KiB): 0
> 0072f1e14000-0074928d4000( 6.510 GiB): 0
> 0074928d4000-00763c71f000( 6.655 GiB): 0
> 00763c71f000-0077e863e000( 6.687 GiB): 0
> 0077e863e000-00798e214000( 6.590 GiB): 0
> 00798e214000-007b0e48a000( 6.002 GiB): 0
> 007b0e48a000-007c62f00000( 5.323 GiB): 0
> 007c62f00000-007defb19000( 6.199 GiB): 0
> 007defb19000-007f794ef000( 6.150 GiB): 0
> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>
> As you can see, we have some regions which are very very big and they
> are losing the chance to be splitted. But
> Damon can still monitor memory access for those small VMA areas very well like:
> 005c675a2000-005c675a3000( 4.000 KiB): 3
Hi, Barry
Actually, we also had found the same problem in redis by ourselves
tool[1]. The DAMON can not split the large anon VMA well, and the anon
VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
hot areas to been monitored or found by DAMON, likes one or more address
choose by DAMON not been accessed during sample period.
I'm not sure whether sets init_regions can deal with the above problem,
or dynamic choose one or limited number VMA to monitor.
I'm not sure, just share my idea.
[1] https://github.com/aliyun/data-profile-tools.git
>
> Typical characteristics of a large Android app is that it has
> thousands of vma and very large virtual address spaces:
> ~/damo # pmap 2550 | wc -l
> 8522
>
> ~/damo # pmap 2550
> ...
> 0000007992bbe000 4K r---- [ anon ]
> 0000007992bbf000 24K rw--- [ anon ]
> 0000007fe8753000 4K ----- [ anon ]
> 0000007fe8754000 8188K rw--- [ stack ]
> total 36742112K
>
> Because the whole vma list is too long, I have put the list here for
> you to download:
> wget http://www.linuxep.com/patches/android-app-vmas
>
> I can reproduce this problem on other Apps like youtube as well.
> I suppose we need to boost the algorithm of splitting regions for this
> kind of application.
> Any thoughts?
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 6:56 ` Rongwei Wang
@ 2022-04-27 7:44 ` Barry Song
2022-04-27 9:22 ` Barry Song
` (2 more replies)
0 siblings, 3 replies; 23+ messages in thread
From: Barry Song @ 2022-04-27 7:44 UTC (permalink / raw)
To: Rongwei Wang
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
<rongwei.wang@linux.alibaba.com> wrote:
>
>
>
> On 4/27/22 7:19 AM, Barry Song wrote:
> > Hi SeongJae & Andrew,
> > (also Cc-ed main damon developers)
> > On an Android phone, I tried to use the DAMON vaddr monitor and found
> > that vaddr regions don't split well on large Android Apps though
> > everything works well on native Apps.
> >
> > I have tried the below two cases on an Android phone with 12GB memory
> > and snapdragon 888 CPU.
> > 1. a native program with small memory working set as below,
> > #define size (1024*1024*100)
> > main()
> > {
> > volatile int *p = malloc(size);
> > memset(p, 0x55, size);
> >
> > while(1) {
> > int i;
> > for (i = 0; i < size / 4; i++)
> > (void)*(p + i);
> > usleep(1000);
> >
> > for (i = 0; i < size / 16; i++)
> > (void)*(p + i);
> > usleep(1000);
> >
> > }
> > }
> > For this application, the Damon vaddr monitor works very well.
> > I have modified monitor.py in the damo userspace tool a little bit to
> > show the raw data getting from the kernel.
> > Regions can split decently on this kind of applications, a typical raw
> > data is as below,
> >
> > monitoring_start: 2.224 s
> > monitoring_end: 2.329 s
> > monitoring_duration: 104.336 ms
> > target_id: 0
> > nr_regions: 24
> > 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > 005fc2910000-005fc6769000( 62.348 MiB): 0
> > 005fc6769000-005fca33f000( 59.836 MiB): 0
> > 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > 007661c59000-007661ee4000( 2.543 MiB): 2
> > 007661ee4000-0076623e4000( 5.000 MiB): 3
> > 0076623e4000-007662837000( 4.324 MiB): 2
> > 007662837000-0076630f1000( 8.727 MiB): 3
> > 0076630f1000-007663494000( 3.637 MiB): 2
> > 007663494000-007663753000( 2.746 MiB): 1
> > 007663753000-007664251000( 10.992 MiB): 3
> > 007664251000-0076666fd000( 36.672 MiB): 2
> > 0076666fd000-007666e73000( 7.461 MiB): 1
> > 007666e73000-007667c89000( 14.086 MiB): 2
> > 007667c89000-007667f97000( 3.055 MiB): 0
> > 007667f97000-007668112000( 1.480 MiB): 1
> > 007668112000-00766820f000(1012.000 KiB): 0
> > 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> >
> > 2. a large Android app like Asphalt 9
> > For this case, basically regions can't split very well, but monitor
> > works on small vma:
> >
> > monitoring_start: 2.220 s
> > monitoring_end: 2.318 s
> > monitoring_duration: 98.576 ms
> > target_id: 0
> > nr_regions: 15
> > 000012c00000-0001c301e000( 6.754 GiB): 0
> > 0001c301e000-000371b6c000( 6.730 GiB): 0
> > 000371b6c000-000400000000( 2.223 GiB): 0
> > 005c6759d000-005c675a2000( 20.000 KiB): 0
> > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > 005c675a3000-005c675a7000( 16.000 KiB): 0
> > 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > 0074928d4000-00763c71f000( 6.655 GiB): 0
> > 00763c71f000-0077e863e000( 6.687 GiB): 0
> > 0077e863e000-00798e214000( 6.590 GiB): 0
> > 00798e214000-007b0e48a000( 6.002 GiB): 0
> > 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > 007c62f00000-007defb19000( 6.199 GiB): 0
> > 007defb19000-007f794ef000( 6.150 GiB): 0
> > 007f794ef000-007fe8f53000( 1.745 GiB): 0
> >
> > As you can see, we have some regions which are very very big and they
> > are losing the chance to be splitted. But
> > Damon can still monitor memory access for those small VMA areas very well like:
> > 005c675a2000-005c675a3000( 4.000 KiB): 3
> Hi, Barry
>
> Actually, we also had found the same problem in redis by ourselves
> tool[1]. The DAMON can not split the large anon VMA well, and the anon
> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> hot areas to been monitored or found by DAMON, likes one or more address
> choose by DAMON not been accessed during sample period.
Hi Rongwei,
Thanks for your comments and thanks for sharing your tools.
I guess the cause might be:
in case a region is very big like 10GiB, we have only 1MiB hot pages
in this large region.
damon will randomly pick one page to sample, but the page has only
1MiB/10GiB, thus
less than 1/10000 chance to hit the hot 1MiB. so probably we need
10000 sample periods
to hit the hot 1MiB in order to split this large region?
@SeongJae, please correct me if I am wrong.
>
> I'm not sure whether sets init_regions can deal with the above problem,
> or dynamic choose one or limited number VMA to monitor.
>
I won't set a limited number of VMA as this will make the damon too hard to use
as nobody wants to make such complex operations, especially an Android
app might have more than 8000 VMAs.
I agree init_regions might be the right place to enhance the situation.
> I'm not sure, just share my idea.
>
> [1] https://github.com/aliyun/data-profile-tools.git
I suppose this tool is based on damon? How do you finally resolve the problem
that large anon VMAs can't be splitted?
Anyway, I will give your tool a try.
> >
> > Typical characteristics of a large Android app is that it has
> > thousands of vma and very large virtual address spaces:
> > ~/damo # pmap 2550 | wc -l
> > 8522
> >
> > ~/damo # pmap 2550
> > ...
> > 0000007992bbe000 4K r---- [ anon ]
> > 0000007992bbf000 24K rw--- [ anon ]
> > 0000007fe8753000 4K ----- [ anon ]
> > 0000007fe8754000 8188K rw--- [ stack ]
> > total 36742112K
> >
> > Because the whole vma list is too long, I have put the list here for
> > you to download:
> > wget http://www.linuxep.com/patches/android-app-vmas
> >
> > I can reproduce this problem on other Apps like youtube as well.
> > I suppose we need to boost the algorithm of splitting regions for this
> > kind of application.
> > Any thoughts?
> >
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 7:44 ` Barry Song
@ 2022-04-27 9:22 ` Barry Song
2022-04-27 11:55 ` Rongwei Wang
2022-04-28 2:04 ` Rongwei Wang
2022-04-27 12:06 ` Rongwei Wang
2022-04-27 17:50 ` sj
2 siblings, 2 replies; 23+ messages in thread
From: Barry Song @ 2022-04-27 9:22 UTC (permalink / raw)
To: Rongwei Wang
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>
> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
> >
> >
> >
> > On 4/27/22 7:19 AM, Barry Song wrote:
> > > Hi SeongJae & Andrew,
> > > (also Cc-ed main damon developers)
> > > On an Android phone, I tried to use the DAMON vaddr monitor and found
> > > that vaddr regions don't split well on large Android Apps though
> > > everything works well on native Apps.
> > >
> > > I have tried the below two cases on an Android phone with 12GB memory
> > > and snapdragon 888 CPU.
> > > 1. a native program with small memory working set as below,
> > > #define size (1024*1024*100)
> > > main()
> > > {
> > > volatile int *p = malloc(size);
> > > memset(p, 0x55, size);
> > >
> > > while(1) {
> > > int i;
> > > for (i = 0; i < size / 4; i++)
> > > (void)*(p + i);
> > > usleep(1000);
> > >
> > > for (i = 0; i < size / 16; i++)
> > > (void)*(p + i);
> > > usleep(1000);
> > >
> > > }
> > > }
> > > For this application, the Damon vaddr monitor works very well.
> > > I have modified monitor.py in the damo userspace tool a little bit to
> > > show the raw data getting from the kernel.
> > > Regions can split decently on this kind of applications, a typical raw
> > > data is as below,
> > >
> > > monitoring_start: 2.224 s
> > > monitoring_end: 2.329 s
> > > monitoring_duration: 104.336 ms
> > > target_id: 0
> > > nr_regions: 24
> > > 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > > 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > > 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > > 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > > 005fc2910000-005fc6769000( 62.348 MiB): 0
> > > 005fc6769000-005fca33f000( 59.836 MiB): 0
> > > 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > > 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > > 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > > 007661c59000-007661ee4000( 2.543 MiB): 2
> > > 007661ee4000-0076623e4000( 5.000 MiB): 3
> > > 0076623e4000-007662837000( 4.324 MiB): 2
> > > 007662837000-0076630f1000( 8.727 MiB): 3
> > > 0076630f1000-007663494000( 3.637 MiB): 2
> > > 007663494000-007663753000( 2.746 MiB): 1
> > > 007663753000-007664251000( 10.992 MiB): 3
> > > 007664251000-0076666fd000( 36.672 MiB): 2
> > > 0076666fd000-007666e73000( 7.461 MiB): 1
> > > 007666e73000-007667c89000( 14.086 MiB): 2
> > > 007667c89000-007667f97000( 3.055 MiB): 0
> > > 007667f97000-007668112000( 1.480 MiB): 1
> > > 007668112000-00766820f000(1012.000 KiB): 0
> > > 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > > 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> > >
> > > 2. a large Android app like Asphalt 9
> > > For this case, basically regions can't split very well, but monitor
> > > works on small vma:
> > >
> > > monitoring_start: 2.220 s
> > > monitoring_end: 2.318 s
> > > monitoring_duration: 98.576 ms
> > > target_id: 0
> > > nr_regions: 15
> > > 000012c00000-0001c301e000( 6.754 GiB): 0
> > > 0001c301e000-000371b6c000( 6.730 GiB): 0
> > > 000371b6c000-000400000000( 2.223 GiB): 0
> > > 005c6759d000-005c675a2000( 20.000 KiB): 0
> > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > > 005c675a3000-005c675a7000( 16.000 KiB): 0
> > > 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > > 0074928d4000-00763c71f000( 6.655 GiB): 0
> > > 00763c71f000-0077e863e000( 6.687 GiB): 0
> > > 0077e863e000-00798e214000( 6.590 GiB): 0
> > > 00798e214000-007b0e48a000( 6.002 GiB): 0
> > > 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > > 007c62f00000-007defb19000( 6.199 GiB): 0
> > > 007defb19000-007f794ef000( 6.150 GiB): 0
> > > 007f794ef000-007fe8f53000( 1.745 GiB): 0
> > >
> > > As you can see, we have some regions which are very very big and they
> > > are losing the chance to be splitted. But
> > > Damon can still monitor memory access for those small VMA areas very well like:
> > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > Hi, Barry
> >
> > Actually, we also had found the same problem in redis by ourselves
> > tool[1]. The DAMON can not split the large anon VMA well, and the anon
> > VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> > hot areas to been monitored or found by DAMON, likes one or more address
> > choose by DAMON not been accessed during sample period.
>
> Hi Rongwei,
> Thanks for your comments and thanks for sharing your tools.
>
> I guess the cause might be:
> in case a region is very big like 10GiB, we have only 1MiB hot pages
> in this large region.
> damon will randomly pick one page to sample, but the page has only
> 1MiB/10GiB, thus
> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> 10000 sample periods
> to hit the hot 1MiB in order to split this large region?
>
> @SeongJae, please correct me if I am wrong.
>
> >
> > I'm not sure whether sets init_regions can deal with the above problem,
> > or dynamic choose one or limited number VMA to monitor.
> >
>
> I won't set a limited number of VMA as this will make the damon too hard to use
> as nobody wants to make such complex operations, especially an Android
> app might have more than 8000 VMAs.
>
> I agree init_regions might be the right place to enhance the situation.
>
> > I'm not sure, just share my idea.
> >
> > [1] https://github.com/aliyun/data-profile-tools.git
>
> I suppose this tool is based on damon? How do you finally resolve the problem
> that large anon VMAs can't be splitted?
> Anyway, I will give your tool a try.
Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
though autogen.sh
runs successfully.
/usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
/root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
reference to `wgetch'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
/root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
reference to `subwin'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
/root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
/root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
/root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
/root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
/root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
/root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
reference to `wattr_off'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
reference to `wattr_on'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
/root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
reference to `wattr_off'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
reference to `mvwprintw'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
reference to `wattr_off'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
/root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
reference to `wclear'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
reference to `wrefresh'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
reference to `endwin'
/usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
/root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
reference to `initscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
reference to `wrefresh'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
reference to `use_default_colors'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
reference to `start_color'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
reference to `keypad'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
reference to `nonl'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
reference to `cbreak'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
reference to `noecho'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
reference to `curs_set'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
reference to `stdscr'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
reference to `mvwprintw'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
reference to `mvwprintw'
/usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
reference to `wrefresh'
collect2: error: ld returned 1 exit status
make[1]: *** [Makefile:592: datop] Error 1
make[1]: Leaving directory '/root/data-profile-tools'
make: *** [Makefile:438: all] Error 2
>
> > >
> > > Typical characteristics of a large Android app is that it has
> > > thousands of vma and very large virtual address spaces:
> > > ~/damo # pmap 2550 | wc -l
> > > 8522
> > >
> > > ~/damo # pmap 2550
> > > ...
> > > 0000007992bbe000 4K r---- [ anon ]
> > > 0000007992bbf000 24K rw--- [ anon ]
> > > 0000007fe8753000 4K ----- [ anon ]
> > > 0000007fe8754000 8188K rw--- [ stack ]
> > > total 36742112K
> > >
> > > Because the whole vma list is too long, I have put the list here for
> > > you to download:
> > > wget http://www.linuxep.com/patches/android-app-vmas
> > >
> > > I can reproduce this problem on other Apps like youtube as well.
> > > I suppose we need to boost the algorithm of splitting regions for this
> > > kind of application.
> > > Any thoughts?
> > >
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 9:22 ` Barry Song
@ 2022-04-27 11:55 ` Rongwei Wang
2022-04-28 2:04 ` Rongwei Wang
1 sibling, 0 replies; 23+ messages in thread
From: Rongwei Wang @ 2022-04-27 11:55 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On 4/27/22 5:22 PM, Barry Song wrote:
> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>>
>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
>> <rongwei.wang@linux.alibaba.com> wrote:
>>>
>>>
>>>
>>> On 4/27/22 7:19 AM, Barry Song wrote:
>>>> Hi SeongJae & Andrew,
>>>> (also Cc-ed main damon developers)
>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>>> that vaddr regions don't split well on large Android Apps though
>>>> everything works well on native Apps.
>>>>
>>>> I have tried the below two cases on an Android phone with 12GB memory
>>>> and snapdragon 888 CPU.
>>>> 1. a native program with small memory working set as below,
>>>> #define size (1024*1024*100)
>>>> main()
>>>> {
>>>> volatile int *p = malloc(size);
>>>> memset(p, 0x55, size);
>>>>
>>>> while(1) {
>>>> int i;
>>>> for (i = 0; i < size / 4; i++)
>>>> (void)*(p + i);
>>>> usleep(1000);
>>>>
>>>> for (i = 0; i < size / 16; i++)
>>>> (void)*(p + i);
>>>> usleep(1000);
>>>>
>>>> }
>>>> }
>>>> For this application, the Damon vaddr monitor works very well.
>>>> I have modified monitor.py in the damo userspace tool a little bit to
>>>> show the raw data getting from the kernel.
>>>> Regions can split decently on this kind of applications, a typical raw
>>>> data is as below,
>>>>
>>>> monitoring_start: 2.224 s
>>>> monitoring_end: 2.329 s
>>>> monitoring_duration: 104.336 ms
>>>> target_id: 0
>>>> nr_regions: 24
>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>>> 007663494000-007663753000( 2.746 MiB): 1
>>>> 007663753000-007664251000( 10.992 MiB): 3
>>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>>> 007667f97000-007668112000( 1.480 MiB): 1
>>>> 007668112000-00766820f000(1012.000 KiB): 0
>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>>
>>>> 2. a large Android app like Asphalt 9
>>>> For this case, basically regions can't split very well, but monitor
>>>> works on small vma:
>>>>
>>>> monitoring_start: 2.220 s
>>>> monitoring_end: 2.318 s
>>>> monitoring_duration: 98.576 ms
>>>> target_id: 0
>>>> nr_regions: 15
>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>>
>>>> As you can see, we have some regions which are very very big and they
>>>> are losing the chance to be splitted. But
>>>> Damon can still monitor memory access for those small VMA areas very well like:
>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>> Hi, Barry
>>>
>>> Actually, we also had found the same problem in redis by ourselves
>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>>> hot areas to been monitored or found by DAMON, likes one or more address
>>> choose by DAMON not been accessed during sample period.
>>
>> Hi Rongwei,
>> Thanks for your comments and thanks for sharing your tools.
>>
>> I guess the cause might be:
>> in case a region is very big like 10GiB, we have only 1MiB hot pages
>> in this large region.
>> damon will randomly pick one page to sample, but the page has only
>> 1MiB/10GiB, thus
>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
>> 10000 sample periods
>> to hit the hot 1MiB in order to split this large region?
>>
>> @SeongJae, please correct me if I am wrong.
>>
>>>
>>> I'm not sure whether sets init_regions can deal with the above problem,
>>> or dynamic choose one or limited number VMA to monitor.
>>>
>>
>> I won't set a limited number of VMA as this will make the damon too hard to use
>> as nobody wants to make such complex operations, especially an Android
>> app might have more than 8000 VMAs.
>>
>> I agree init_regions might be the right place to enhance the situation.
>>
>>> I'm not sure, just share my idea.
>>>
>>> [1] https://github.com/aliyun/data-profile-tools.git
>>
>> I suppose this tool is based on damon? How do you finally resolve the problem
>> that large anon VMAs can't be splitted?
>> Anyway, I will give your tool a try.
>
> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
> though autogen.sh
> runs successfully.
>
> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> reference to `wgetch'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> reference to `subwin'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
> reference to `wattr_off'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
> reference to `wattr_on'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> reference to `wattr_off'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
> reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> reference to `wattr_off'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> reference to `wclear'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
> reference to `wrefresh'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
> reference to `endwin'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
> reference to `initscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> reference to `wrefresh'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
> reference to `use_default_colors'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
> reference to `start_color'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
> reference to `keypad'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
> reference to `nonl'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
> reference to `cbreak'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
> reference to `noecho'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
> reference to `curs_set'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
> reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
> reference to `wrefresh'
> collect2: error: ld returned 1 exit status
> make[1]: *** [Makefile:592: datop] Error 1
> make[1]: Leaving directory '/root/data-profile-tools'
> make: *** [Makefile:438: all] Error 2
Hi, Barry
Thank you for this bug report.
It seems this tool had not supported with ubuntu. And we just support
for CentOS or AnolisOS. I am trying to fix this bug.
I see all these errors reported by you are extensions to the course
library? I am not familiar with Ubuntu and It looks that these errors
can be fixed if course relevant library installed.
Anyway, I will try to fix it next.
Thanks.
>
>>
>>>>
>>>> Typical characteristics of a large Android app is that it has
>>>> thousands of vma and very large virtual address spaces:
>>>> ~/damo # pmap 2550 | wc -l
>>>> 8522
>>>>
>>>> ~/damo # pmap 2550
>>>> ...
>>>> 0000007992bbe000 4K r---- [ anon ]
>>>> 0000007992bbf000 24K rw--- [ anon ]
>>>> 0000007fe8753000 4K ----- [ anon ]
>>>> 0000007fe8754000 8188K rw--- [ stack ]
>>>> total 36742112K
>>>>
>>>> Because the whole vma list is too long, I have put the list here for
>>>> you to download:
>>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>>
>>>> I can reproduce this problem on other Apps like youtube as well.
>>>> I suppose we need to boost the algorithm of splitting regions for this
>>>> kind of application.
>>>> Any thoughts?
>>>>
>>
>> Thanks
>> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 7:44 ` Barry Song
2022-04-27 9:22 ` Barry Song
@ 2022-04-27 12:06 ` Rongwei Wang
2022-04-27 17:50 ` sj
2 siblings, 0 replies; 23+ messages in thread
From: Rongwei Wang @ 2022-04-27 12:06 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On 4/27/22 3:44 PM, Barry Song wrote:
> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
>>
>>
>>
>> On 4/27/22 7:19 AM, Barry Song wrote:
>>> Hi SeongJae & Andrew,
>>> (also Cc-ed main damon developers)
>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>> that vaddr regions don't split well on large Android Apps though
>>> everything works well on native Apps.
>>>
>>> I have tried the below two cases on an Android phone with 12GB memory
>>> and snapdragon 888 CPU.
>>> 1. a native program with small memory working set as below,
>>> #define size (1024*1024*100)
>>> main()
>>> {
>>> volatile int *p = malloc(size);
>>> memset(p, 0x55, size);
>>>
>>> while(1) {
>>> int i;
>>> for (i = 0; i < size / 4; i++)
>>> (void)*(p + i);
>>> usleep(1000);
>>>
>>> for (i = 0; i < size / 16; i++)
>>> (void)*(p + i);
>>> usleep(1000);
>>>
>>> }
>>> }
>>> For this application, the Damon vaddr monitor works very well.
>>> I have modified monitor.py in the damo userspace tool a little bit to
>>> show the raw data getting from the kernel.
>>> Regions can split decently on this kind of applications, a typical raw
>>> data is as below,
>>>
>>> monitoring_start: 2.224 s
>>> monitoring_end: 2.329 s
>>> monitoring_duration: 104.336 ms
>>> target_id: 0
>>> nr_regions: 24
>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>> 007663494000-007663753000( 2.746 MiB): 1
>>> 007663753000-007664251000( 10.992 MiB): 3
>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>> 007667f97000-007668112000( 1.480 MiB): 1
>>> 007668112000-00766820f000(1012.000 KiB): 0
>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>
>>> 2. a large Android app like Asphalt 9
>>> For this case, basically regions can't split very well, but monitor
>>> works on small vma:
>>>
>>> monitoring_start: 2.220 s
>>> monitoring_end: 2.318 s
>>> monitoring_duration: 98.576 ms
>>> target_id: 0
>>> nr_regions: 15
>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>
>>> As you can see, we have some regions which are very very big and they
>>> are losing the chance to be splitted. But
>>> Damon can still monitor memory access for those small VMA areas very well like:
>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>> Hi, Barry
>>
>> Actually, we also had found the same problem in redis by ourselves
>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>> hot areas to been monitored or found by DAMON, likes one or more address
>> choose by DAMON not been accessed during sample period.
>
> Hi Rongwei,
> Thanks for your comments and thanks for sharing your tools.
>
> I guess the cause might be:
> in case a region is very big like 10GiB, we have only 1MiB hot pages
> in this large region.
> damon will randomly pick one page to sample, but the page has only
> 1MiB/10GiB, thus
> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> 10000 sample periods
> to hit the hot 1MiB in order to split this large region?
>
> @SeongJae, please correct me if I am wrong.
>
>>
>> I'm not sure whether sets init_regions can deal with the above problem,
>> or dynamic choose one or limited number VMA to monitor.
>>
>
> I won't set a limited number of VMA as this will make the damon too hard to use
> as nobody wants to make such complex operations, especially an Android
> app might have more than 8000 VMAs.
>
> I agree init_regions might be the right place to enhance the situation.
>
>> I'm not sure, just share my idea.
>>
>> [1] https://github.com/aliyun/data-profile-tools.git
>
> I suppose this tool is based on damon? How do you finally resolve the problem
Yes, and we plan to design it to be a user agent.
> that large anon VMAs can't be splitted?
I see your have different environment with mine. Finally, I just monitor
the anon VMA and set a large regions number.
There is one or two large anon VMAs in my environment. It seems
different with your.
> Anyway, I will give your tool a try.
>
>>>
>>> Typical characteristics of a large Android app is that it has
>>> thousands of vma and very large virtual address spaces:
>>> ~/damo # pmap 2550 | wc -l
>>> 8522
>>>
>>> ~/damo # pmap 2550
>>> ...
>>> 0000007992bbe000 4K r---- [ anon ]
>>> 0000007992bbf000 24K rw--- [ anon ]
>>> 0000007fe8753000 4K ----- [ anon ]
>>> 0000007fe8754000 8188K rw--- [ stack ]
>>> total 36742112K
>>>
>>> Because the whole vma list is too long, I have put the list here for
>>> you to download:
>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>
>>> I can reproduce this problem on other Apps like youtube as well.
>>> I suppose we need to boost the algorithm of splitting regions for this
>>> kind of application.
>>> Any thoughts?
>>>
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 2:08 ` Barry Song
@ 2022-04-27 17:39 ` sj
2022-04-28 1:27 ` Barry Song
0 siblings, 1 reply; 23+ messages in thread
From: sj @ 2022-04-27 17:39 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
linux-damon
On Wed, 27 Apr 2022 14:08:04 +1200 Barry Song <21cnbao@gmail.com> wrote:
> On Wed, Apr 27, 2022 at 12:21 PM <sj@kernel.org> wrote:
> >
> > Hello Barry,
> >
> >
> > Thank you so much for sharing your great findings! :)
>
> Thanks for your quick response.
My pleasure :)
>
> >
> > On Wed, 27 Apr 2022 11:19:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
> >
> > > Hi SeongJae & Andrew,
> > > (also Cc-ed main damon developers)
> > > On an Android phone, I tried to use the DAMON vaddr monitor and found
> > > that vaddr regions don't split well on large Android Apps though
> > > everything works well on native Apps.
> > >
> > > I have tried the below two cases on an Android phone with 12GB memory
> > > and snapdragon 888 CPU.
> > > 1. a native program with small memory working set as below,
> > > #define size (1024*1024*100)
> > > main()
> > > {
> > > volatile int *p = malloc(size);
> > > memset(p, 0x55, size);
> > >
> > > while(1) {
> > > int i;
> > > for (i = 0; i < size / 4; i++)
> > > (void)*(p + i);
> > > usleep(1000);
> > >
> > > for (i = 0; i < size / 16; i++)
> > > (void)*(p + i);
> > > usleep(1000);
> > >
> > > }
> > > }
> > > For this application, the Damon vaddr monitor works very well.
> > > I have modified monitor.py in the damo userspace tool a little bit to
> > > show the raw data getting from the kernel.
> > > Regions can split decently on this kind of applications, a typical raw
> > > data is as below,
> > >
> > > monitoring_start: 2.224 s
> > > monitoring_end: 2.329 s
> > > monitoring_duration: 104.336 ms
> > > target_id: 0
> > > nr_regions: 24
> > > 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > > 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > > 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > > 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > > 005fc2910000-005fc6769000( 62.348 MiB): 0
> > > 005fc6769000-005fca33f000( 59.836 MiB): 0
> > > 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > > 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > > 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > > 007661c59000-007661ee4000( 2.543 MiB): 2
> > > 007661ee4000-0076623e4000( 5.000 MiB): 3
> > > 0076623e4000-007662837000( 4.324 MiB): 2
> > > 007662837000-0076630f1000( 8.727 MiB): 3
> > > 0076630f1000-007663494000( 3.637 MiB): 2
> > > 007663494000-007663753000( 2.746 MiB): 1
> > > 007663753000-007664251000( 10.992 MiB): 3
> > > 007664251000-0076666fd000( 36.672 MiB): 2
> > > 0076666fd000-007666e73000( 7.461 MiB): 1
> > > 007666e73000-007667c89000( 14.086 MiB): 2
> > > 007667c89000-007667f97000( 3.055 MiB): 0
> > > 007667f97000-007668112000( 1.480 MiB): 1
> > > 007668112000-00766820f000(1012.000 KiB): 0
> > > 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > > 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> > >
> > > 2. a large Android app like Asphalt 9
> > > For this case, basically regions can't split very well, but monitor
> > > works on small vma:
> > >
> > > monitoring_start: 2.220 s
> > > monitoring_end: 2.318 s
> > > monitoring_duration: 98.576 ms
> > > target_id: 0
> > > nr_regions: 15
> > > 000012c00000-0001c301e000( 6.754 GiB): 0
> > > 0001c301e000-000371b6c000( 6.730 GiB): 0
> > > 000371b6c000-000400000000( 2.223 GiB): 0
> > > 005c6759d000-005c675a2000( 20.000 KiB): 0
> > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > > 005c675a3000-005c675a7000( 16.000 KiB): 0
> > > 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > > 0074928d4000-00763c71f000( 6.655 GiB): 0
> > > 00763c71f000-0077e863e000( 6.687 GiB): 0
> > > 0077e863e000-00798e214000( 6.590 GiB): 0
> > > 00798e214000-007b0e48a000( 6.002 GiB): 0
> > > 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > > 007c62f00000-007defb19000( 6.199 GiB): 0
> > > 007defb19000-007f794ef000( 6.150 GiB): 0
> > > 007f794ef000-007fe8f53000( 1.745 GiB): 0
> > >
> > > As you can see, we have some regions which are very very big and they
> > > are losing the chance to be splitted. But
> > > Damon can still monitor memory access for those small VMA areas very well like:
> > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> >
> > In short, DAMON doesn't set regions based on VMA but access pattern, and
> > therefore this looks not a problem.
> >
> > DAMON allows users set min/max monitoring overhead limit and provides a best
> > accuracy under the condition. In detail, users are allowed to set the min/max
> > monitoring regions as DAMON's monitoring overhead is proportional to the number
> > of regions. DAMON provides best effort accuracy under the condition by
> > splitting and merging regions so that pages in each region has different access
> > frequency.
> >
> > The default min number of regions is 10. I believe that's why there are many 6
> > GiB regions.
> >
>
> i had actually tried to set min regions to 100 as below:
> /sys/kernel/debug/damon # echo 5000 100000 60000000 100 1000 > attrs
> /sys/kernel/debug/damon # cat attrs
> 5000 100000 60000000 100 1000
>
> but it seems i am still only getting regions like 16:
>
> monitoring_start: 805.192 ms
> monitoring_end: 905.385 ms
> monitoring_duration: 100.193 ms
> target_id: 0
> nr_regions: 16
> 000012c00000-0001fc021000( 7.645 GiB): 0
> 0001fc021000-0003e6b9a000( 7.667 GiB): 0
> 0003e6b9a000-000400000000( 404.398 MiB): 0
> 005e95645000-005e9564a000( 20.000 KiB): 0
> 005e9564a000-005e9564b000( 4.000 KiB): 8
> 005e9564b000-005e9564c000( 4.000 KiB): 1
> 005e9564c000-005e9564f000( 12.000 KiB): 0
> 006ffffff000-0071fa4f8000( 7.911 GiB): 0
> 0071fa4f8000-0073dacd1000( 7.508 GiB): 0
> 0073dacd1000-00759533c000( 6.913 GiB): 0
> 00759533c000-0076cc999000( 4.866 GiB): 0
> 0076cc999000-00788bea1000( 6.989 GiB): 0
> 00788bea1000-007a6fe14000( 7.562 GiB): 0
> 007a6fe14000-007c5c4a3000( 7.694 GiB): 0
> 007c5c4a3000-007df9bd8000( 6.460 GiB): 0
> 007df9bd8000-007fe6d7b000( 7.705 GiB): 0
>
> so it seems the 100 set in min regions doesn't mean we will have at
> least 100 regions?
Have you used the debugfs interface for turning DAMON on for the above case?
If not, and if you have used 'damo monitor' again, 'damo monitor' should have
reset the 'min_nr_regions' to 10.
On my x86 test machine, the 'min_nr_regions' was respected:
$ sudo damo record --minr 100 paddr
[...]
$ sudo damo report nr_regions
# <percentile> <# regions>
# target_id 0
# avr: 111
0 109
25 111
50 111
75 113
100 116
Unfortunately, 'damo monitor' has no option to set the min_nr_regions. You
could modify the default value of the option of damo-record on 'damo_record.py'
for the case. Sorry for your inconvenient.
>
> > If we don't see small regions having some non-zero access frequency, we would
> > be better to be worried. However, it is finding the small 4 KiB regions having
> > higher access frequency successfully. The 4 KiB region is not because the
> > region is having 4 KiB VMA, but the address region shows high access frequency.
> >
> > >
> > > Typical characteristics of a large Android app is that it has
> > > thousands of vma and very large virtual address spaces:
> > > ~/damo # pmap 2550 | wc -l
> > > 8522
> > >
> > > ~/damo # pmap 2550
> > > ...
> > > 0000007992bbe000 4K r---- [ anon ]
> > > 0000007992bbf000 24K rw--- [ anon ]
> > > 0000007fe8753000 4K ----- [ anon ]
> > > 0000007fe8754000 8188K rw--- [ stack ]
> > > total 36742112K
> > >
> > > Because the whole vma list is too long, I have put the list here for
> > > you to download:
> > > wget http://www.linuxep.com/patches/android-app-vmas
> > >
> > > I can reproduce this problem on other Apps like youtube as well.
> > > I suppose we need to boost the algorithm of splitting regions for this
> > > kind of application.
> > > Any thoughts?
> >
> > As mentioned above, this looks not a problem, as DAMON's monitoring regions is
> > not constructed based on VMAs but access patterns.
>
> What makes me believe it is a problem is that it seems we are getting
> an incorrect
> wss based on the data reported from kernel:
> ~/damo # ./damo monitor --report_type wss --count 20 2561
> # <percentile> <wss>
> # target_id 0
> # avr: 443.854 MiB
> 0 0 B |
> |
> 25 4.000 KiB |
> |
> 50 12.000 KiB |
> |
> 75 20.000 KiB |
> |
> 100 5.851 GiB
> |***********************************************************|
[...]
>
> And I have a question, what do percentile 0,25,50,75 mean here?
> Why are they so different with percentile 100?
> For example, 0,25,50,75 has only KiB but 100 has GiB.
For each aggregation interval, we get one snapshot. So, if we have a
monitoring results that recorded for, say, 100 aggregation interval, we have
100 snapshots. 'damo' calculates working set size of each snapshot by summing
size of regions assumed to be accessed at least once. So, in this example, we
get 100 wss values. Then, 'damo' sorts the values and provides the smallest
one as 0-th percentile, 25th small value as 25-th percentile, and so on.
100-th percentile wss is usually noisy, as DAMON regions shouldn't be converged
well at the beginning of the record. I believe that could be the reason why
the 100-th percentile wss is so unexpectedly big.
I personally use 50-th percentile as reliable value.
>
> I guess Asphalt 9 should be normally accessing hundreds of megabytes of memory.
Each wss value is calculated for each aggregation interval. That is, the wss
value is amount of memory that accessed within an aggregation interval. As the
default aggregation interval is 100ms, the working set size could be smaller
than your expectation.
For such cases, 'damo report wss' has an option called '--work_time' to
simulate higher aggregation intervals. That is, you can get the total size of
the memory regions that DAMON shown access to those within arbitrary amount of
time using the option. 'damo monitor' doesn't support the option at the
moment, though.
>
> >
> > Nevertheless, I believe there are many rooms for improvement of DAMON's access
> > frequency. I want to implement fixed-gran monitoring feature first, and
> > develop some accuracy optimizations using the fixed-gran monitoring as
> > comparison target.
>
> Does fixed-gran mean splitting VA to some regions equally with the same size?
> for example, if we have 1GB VA, we split it into 512 regions in 2MB
> size, something like that?
Yes, exactly. That's the idea. It would incur much higher, and unbounded
overhead, but could be a good comparison target for accuracy improvement of
DAMON. Also, for some environments that doesn't care about the monitoring
overhead, it could be used well.
Thanks,
SJ
> >
> > If I'm missing something or the explanation was not enough, please feel free to
> > let me know.
> >
> >
> > Thank,
> > SJ
> >
> > >
> > > Thanks
> > > Barry
> > >
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 7:44 ` Barry Song
2022-04-27 9:22 ` Barry Song
2022-04-27 12:06 ` Rongwei Wang
@ 2022-04-27 17:50 ` sj
2022-05-29 19:54 ` SeongJae Park
2 siblings, 1 reply; 23+ messages in thread
From: sj @ 2022-04-27 17:50 UTC (permalink / raw)
To: Barry Song
Cc: Rongwei Wang, sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox,
shuah, brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
Hello Rongwei and Barry,
On Wed, 27 Apr 2022 19:44:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
> >
> >
> >
> > On 4/27/22 7:19 AM, Barry Song wrote:
> > > Hi SeongJae & Andrew,
> > > (also Cc-ed main damon developers)
> > > On an Android phone, I tried to use the DAMON vaddr monitor and found
> > > that vaddr regions don't split well on large Android Apps though
> > > everything works well on native Apps.
> > >
> > > I have tried the below two cases on an Android phone with 12GB memory
> > > and snapdragon 888 CPU.
> > > 1. a native program with small memory working set as below,
> > > #define size (1024*1024*100)
> > > main()
> > > {
> > > volatile int *p = malloc(size);
> > > memset(p, 0x55, size);
> > >
> > > while(1) {
> > > int i;
> > > for (i = 0; i < size / 4; i++)
> > > (void)*(p + i);
> > > usleep(1000);
> > >
> > > for (i = 0; i < size / 16; i++)
> > > (void)*(p + i);
> > > usleep(1000);
> > >
> > > }
> > > }
> > > For this application, the Damon vaddr monitor works very well.
> > > I have modified monitor.py in the damo userspace tool a little bit to
> > > show the raw data getting from the kernel.
> > > Regions can split decently on this kind of applications, a typical raw
> > > data is as below,
> > >
> > > monitoring_start: 2.224 s
> > > monitoring_end: 2.329 s
> > > monitoring_duration: 104.336 ms
> > > target_id: 0
> > > nr_regions: 24
> > > 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > > 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > > 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > > 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > > 005fc2910000-005fc6769000( 62.348 MiB): 0
> > > 005fc6769000-005fca33f000( 59.836 MiB): 0
> > > 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > > 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > > 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > > 007661c59000-007661ee4000( 2.543 MiB): 2
> > > 007661ee4000-0076623e4000( 5.000 MiB): 3
> > > 0076623e4000-007662837000( 4.324 MiB): 2
> > > 007662837000-0076630f1000( 8.727 MiB): 3
> > > 0076630f1000-007663494000( 3.637 MiB): 2
> > > 007663494000-007663753000( 2.746 MiB): 1
> > > 007663753000-007664251000( 10.992 MiB): 3
> > > 007664251000-0076666fd000( 36.672 MiB): 2
> > > 0076666fd000-007666e73000( 7.461 MiB): 1
> > > 007666e73000-007667c89000( 14.086 MiB): 2
> > > 007667c89000-007667f97000( 3.055 MiB): 0
> > > 007667f97000-007668112000( 1.480 MiB): 1
> > > 007668112000-00766820f000(1012.000 KiB): 0
> > > 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > > 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> > >
> > > 2. a large Android app like Asphalt 9
> > > For this case, basically regions can't split very well, but monitor
> > > works on small vma:
> > >
> > > monitoring_start: 2.220 s
> > > monitoring_end: 2.318 s
> > > monitoring_duration: 98.576 ms
> > > target_id: 0
> > > nr_regions: 15
> > > 000012c00000-0001c301e000( 6.754 GiB): 0
> > > 0001c301e000-000371b6c000( 6.730 GiB): 0
> > > 000371b6c000-000400000000( 2.223 GiB): 0
> > > 005c6759d000-005c675a2000( 20.000 KiB): 0
> > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > > 005c675a3000-005c675a7000( 16.000 KiB): 0
> > > 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > > 0074928d4000-00763c71f000( 6.655 GiB): 0
> > > 00763c71f000-0077e863e000( 6.687 GiB): 0
> > > 0077e863e000-00798e214000( 6.590 GiB): 0
> > > 00798e214000-007b0e48a000( 6.002 GiB): 0
> > > 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > > 007c62f00000-007defb19000( 6.199 GiB): 0
> > > 007defb19000-007f794ef000( 6.150 GiB): 0
> > > 007f794ef000-007fe8f53000( 1.745 GiB): 0
> > >
> > > As you can see, we have some regions which are very very big and they
> > > are losing the chance to be splitted. But
> > > Damon can still monitor memory access for those small VMA areas very well like:
> > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > Hi, Barry
> >
> > Actually, we also had found the same problem in redis by ourselves
> > tool[1]. The DAMON can not split the large anon VMA well, and the anon
> > VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> > hot areas to been monitored or found by DAMON, likes one or more address
> > choose by DAMON not been accessed during sample period.
>
> Hi Rongwei,
> Thanks for your comments and thanks for sharing your tools.
I also appreciate for sharing the great tool, Rongwei!
>
> I guess the cause might be:
> in case a region is very big like 10GiB, we have only 1MiB hot pages
> in this large region.
> damon will randomly pick one page to sample, but the page has only
> 1MiB/10GiB, thus
> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> 10000 sample periods
> to hit the hot 1MiB in order to split this large region?
>
> @SeongJae, please correct me if I am wrong.
I think your theory makes sense. There was a similar concern, so we made DAMON
to split regions into 3 sub-regions when we don't see advance[1]. My current
rough idea for improving DAMON accuracy is making it more aggressive while
keeping the monitoring overhead low.
[1] https://git.kernel.org/pub/scm/linux/kernel/git/sj/linux.git/tree/mm/damon/core.c?h=damon/next-2022-04-21-08-31-on-v5.18-rc3-mmots-2022-04-20-17-37#n1053
>
> >
> > I'm not sure whether sets init_regions can deal with the above problem,
> > or dynamic choose one or limited number VMA to monitor.
> >
>
> I won't set a limited number of VMA as this will make the damon too hard to use
> as nobody wants to make such complex operations, especially an Android
> app might have more than 8000 VMAs.
>
> I agree init_regions might be the right place to enhance the situation.
'init_regions' has developed for the purpose, where user space knows some good
information for starting point of the regions adjustment, and thus want to hint
DAMON. Nevertheless, it might not work as expected, because DAMON
automatically updates the target regions to cover all VMAs as much as it can.
I have posted a patchset for the use case yesterday[1].
[1] https://lore.kernel.org/linux-mm/20220426231750.48822-1-sj@kernel.org/
>
> > I'm not sure, just share my idea.
> >
> > [1] https://github.com/aliyun/data-profile-tools.git
>
> I suppose this tool is based on damon? How do you finally resolve the problem
> that large anon VMAs can't be splitted?
> Anyway, I will give your tool a try.
>
> > >
> > > Typical characteristics of a large Android app is that it has
> > > thousands of vma and very large virtual address spaces:
> > > ~/damo # pmap 2550 | wc -l
> > > 8522
> > >
> > > ~/damo # pmap 2550
> > > ...
> > > 0000007992bbe000 4K r---- [ anon ]
> > > 0000007992bbf000 24K rw--- [ anon ]
> > > 0000007fe8753000 4K ----- [ anon ]
> > > 0000007fe8754000 8188K rw--- [ stack ]
> > > total 36742112K
> > >
> > > Because the whole vma list is too long, I have put the list here for
> > > you to download:
> > > wget http://www.linuxep.com/patches/android-app-vmas
> > >
> > > I can reproduce this problem on other Apps like youtube as well.
> > > I suppose we need to boost the algorithm of splitting regions for this
> > > kind of application.
> > > Any thoughts?
> > >
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 17:39 ` sj
@ 2022-04-28 1:27 ` Barry Song
2022-04-28 16:16 ` sj
0 siblings, 1 reply; 23+ messages in thread
From: Barry Song @ 2022-04-28 1:27 UTC (permalink / raw)
To: sj
Cc: Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
linux-damon
On Thu, Apr 28, 2022 at 5:39 AM <sj@kernel.org> wrote:
>
> On Wed, 27 Apr 2022 14:08:04 +1200 Barry Song <21cnbao@gmail.com> wrote:
>
> > On Wed, Apr 27, 2022 at 12:21 PM <sj@kernel.org> wrote:
> > >
> > > Hello Barry,
> > >
> > >
> > > Thank you so much for sharing your great findings! :)
> >
> > Thanks for your quick response.
>
> My pleasure :)
>
> >
> > >
> > > On Wed, 27 Apr 2022 11:19:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
> > >
> > > > Hi SeongJae & Andrew,
> > > > (also Cc-ed main damon developers)
> > > > On an Android phone, I tried to use the DAMON vaddr monitor and found
> > > > that vaddr regions don't split well on large Android Apps though
> > > > everything works well on native Apps.
> > > >
> > > > I have tried the below two cases on an Android phone with 12GB memory
> > > > and snapdragon 888 CPU.
> > > > 1. a native program with small memory working set as below,
> > > > #define size (1024*1024*100)
> > > > main()
> > > > {
> > > > volatile int *p = malloc(size);
> > > > memset(p, 0x55, size);
> > > >
> > > > while(1) {
> > > > int i;
> > > > for (i = 0; i < size / 4; i++)
> > > > (void)*(p + i);
> > > > usleep(1000);
> > > >
> > > > for (i = 0; i < size / 16; i++)
> > > > (void)*(p + i);
> > > > usleep(1000);
> > > >
> > > > }
> > > > }
> > > > For this application, the Damon vaddr monitor works very well.
> > > > I have modified monitor.py in the damo userspace tool a little bit to
> > > > show the raw data getting from the kernel.
> > > > Regions can split decently on this kind of applications, a typical raw
> > > > data is as below,
> > > >
> > > > monitoring_start: 2.224 s
> > > > monitoring_end: 2.329 s
> > > > monitoring_duration: 104.336 ms
> > > > target_id: 0
> > > > nr_regions: 24
> > > > 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > > > 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > > > 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > > > 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > > > 005fc2910000-005fc6769000( 62.348 MiB): 0
> > > > 005fc6769000-005fca33f000( 59.836 MiB): 0
> > > > 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > > > 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > > > 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > > > 007661c59000-007661ee4000( 2.543 MiB): 2
> > > > 007661ee4000-0076623e4000( 5.000 MiB): 3
> > > > 0076623e4000-007662837000( 4.324 MiB): 2
> > > > 007662837000-0076630f1000( 8.727 MiB): 3
> > > > 0076630f1000-007663494000( 3.637 MiB): 2
> > > > 007663494000-007663753000( 2.746 MiB): 1
> > > > 007663753000-007664251000( 10.992 MiB): 3
> > > > 007664251000-0076666fd000( 36.672 MiB): 2
> > > > 0076666fd000-007666e73000( 7.461 MiB): 1
> > > > 007666e73000-007667c89000( 14.086 MiB): 2
> > > > 007667c89000-007667f97000( 3.055 MiB): 0
> > > > 007667f97000-007668112000( 1.480 MiB): 1
> > > > 007668112000-00766820f000(1012.000 KiB): 0
> > > > 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > > > 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> > > >
> > > > 2. a large Android app like Asphalt 9
> > > > For this case, basically regions can't split very well, but monitor
> > > > works on small vma:
> > > >
> > > > monitoring_start: 2.220 s
> > > > monitoring_end: 2.318 s
> > > > monitoring_duration: 98.576 ms
> > > > target_id: 0
> > > > nr_regions: 15
> > > > 000012c00000-0001c301e000( 6.754 GiB): 0
> > > > 0001c301e000-000371b6c000( 6.730 GiB): 0
> > > > 000371b6c000-000400000000( 2.223 GiB): 0
> > > > 005c6759d000-005c675a2000( 20.000 KiB): 0
> > > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > > > 005c675a3000-005c675a7000( 16.000 KiB): 0
> > > > 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > > > 0074928d4000-00763c71f000( 6.655 GiB): 0
> > > > 00763c71f000-0077e863e000( 6.687 GiB): 0
> > > > 0077e863e000-00798e214000( 6.590 GiB): 0
> > > > 00798e214000-007b0e48a000( 6.002 GiB): 0
> > > > 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > > > 007c62f00000-007defb19000( 6.199 GiB): 0
> > > > 007defb19000-007f794ef000( 6.150 GiB): 0
> > > > 007f794ef000-007fe8f53000( 1.745 GiB): 0
> > > >
> > > > As you can see, we have some regions which are very very big and they
> > > > are losing the chance to be splitted. But
> > > > Damon can still monitor memory access for those small VMA areas very well like:
> > > > 005c675a2000-005c675a3000( 4.000 KiB): 3
> > >
> > > In short, DAMON doesn't set regions based on VMA but access pattern, and
> > > therefore this looks not a problem.
> > >
> > > DAMON allows users set min/max monitoring overhead limit and provides a best
> > > accuracy under the condition. In detail, users are allowed to set the min/max
> > > monitoring regions as DAMON's monitoring overhead is proportional to the number
> > > of regions. DAMON provides best effort accuracy under the condition by
> > > splitting and merging regions so that pages in each region has different access
> > > frequency.
> > >
> > > The default min number of regions is 10. I believe that's why there are many 6
> > > GiB regions.
> > >
> >
> > i had actually tried to set min regions to 100 as below:
> > /sys/kernel/debug/damon # echo 5000 100000 60000000 100 1000 > attrs
> > /sys/kernel/debug/damon # cat attrs
> > 5000 100000 60000000 100 1000
> >
> > but it seems i am still only getting regions like 16:
> >
> > monitoring_start: 805.192 ms
> > monitoring_end: 905.385 ms
> > monitoring_duration: 100.193 ms
> > target_id: 0
> > nr_regions: 16
> > 000012c00000-0001fc021000( 7.645 GiB): 0
> > 0001fc021000-0003e6b9a000( 7.667 GiB): 0
> > 0003e6b9a000-000400000000( 404.398 MiB): 0
> > 005e95645000-005e9564a000( 20.000 KiB): 0
> > 005e9564a000-005e9564b000( 4.000 KiB): 8
> > 005e9564b000-005e9564c000( 4.000 KiB): 1
> > 005e9564c000-005e9564f000( 12.000 KiB): 0
> > 006ffffff000-0071fa4f8000( 7.911 GiB): 0
> > 0071fa4f8000-0073dacd1000( 7.508 GiB): 0
> > 0073dacd1000-00759533c000( 6.913 GiB): 0
> > 00759533c000-0076cc999000( 4.866 GiB): 0
> > 0076cc999000-00788bea1000( 6.989 GiB): 0
> > 00788bea1000-007a6fe14000( 7.562 GiB): 0
> > 007a6fe14000-007c5c4a3000( 7.694 GiB): 0
> > 007c5c4a3000-007df9bd8000( 6.460 GiB): 0
> > 007df9bd8000-007fe6d7b000( 7.705 GiB): 0
> >
> > so it seems the 100 set in min regions doesn't mean we will have at
> > least 100 regions?
>
> Have you used the debugfs interface for turning DAMON on for the above case?
> If not, and if you have used 'damo monitor' again, 'damo monitor' should have
> reset the 'min_nr_regions' to 10.
>
Thanks for the clarification.
i hardcoded min_nr_regions to 200 by:
diff --git a/_damon.py b/_damon.py
index 1306ea1..82342a5 100644
--- a/_damon.py
+++ b/_damon.py
@@ -344,7 +344,7 @@ def set_attrs_argparser(parser):
parser.add_argument('-u', '--updr', metavar='<interval>', type=int,
default=1000000, help='regions update interval (us)')
parser.add_argument('-n', '--minr', metavar='<# regions>', type=int,
- default=10, help='minimal number of regions')
+ default=200, help='minimal number of regions')
parser.add_argument('-m', '--maxr', metavar='<# regions>', type=int,
default=1000, help='maximum number of regions')
Now wss seems to make more senses:
~/damo # ./damo monitor --report_type=wss --count=20 2551
# <percentile> <wss>
# target_id 0
# avr: 235.394 MiB
0 0 B |
|
25 2.164 MiB |
|
50 129.875 MiB |*********
|
75 430.547 MiB |******************************
|
100 844.238 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 352.501 MiB
0 8.781 MiB |
|
25 100.793 MiB |*****
|
50 309.883 MiB |***************
|
75 487.027 MiB |************************
|
100 1.146 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 161.014 MiB
0 2.398 MiB |
|
25 9.902 MiB |*
|
50 48.539 MiB |*****
|
75 326.391 MiB |************************************
|
100 528.809 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 226.624 MiB
0 3.336 MiB |
|
25 4.078 MiB |
|
50 248.410 MiB |***************************
|
75 374.551 MiB |*****************************************
|
100 536.539 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 312.534 MiB
0 0 B |
|
25 123.539 MiB |********
|
50 248.914 MiB |*****************
|
75 461.578 MiB |********************************
|
100 839.188 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 134.619 MiB
0 0 B |
|
25 0 B |
|
50 32.883 MiB |**
|
75 240.508 MiB |********************
|
100 705.434 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 126.954 MiB
0 2.371 MiB |
|
25 4.195 MiB |
|
50 4.500 MiB |
|
75 228.879 MiB |***************
|
100 851.113 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 217.697 MiB
0 12.391 MiB |
|
25 14.273 MiB |
|
50 17.473 MiB |
|
75 377.797 MiB |********************
|
100 1.043 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 147.215 MiB
0 1.660 MiB |
|
25 2.078 MiB |
|
50 42.012 MiB |**
|
75 136.105 MiB |*******
|
100 1.096 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 185.126 MiB
0 2.164 MiB |
|
25 4.203 MiB |
|
50 34.738 MiB |**
|
75 270.617 MiB |********************
|
100 790.008 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 324.269 MiB
0 14.824 MiB |
|
25 25.648 MiB |
|
50 45.023 MiB |*
|
75 450.941 MiB |************
|
100 2.019 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 165.244 MiB
0 9.664 MiB |
|
25 10.316 MiB |
|
50 31.832 MiB |**
|
75 330.348 MiB |****************************
|
100 679.520 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 127.422 MiB
0 13.043 MiB |*
|
25 15.020 MiB |**
|
50 16.145 MiB |**
|
75 300.398 MiB |****************************************
|
100 432.977 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 256.081 MiB
0 13.387 MiB |
|
25 18.477 MiB |*
|
50 48.340 MiB |**
|
75 497.965 MiB |****************************
|
100 1.014 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 378.882 MiB
0 2.832 MiB |
|
25 56.801 MiB |***
|
50 383.211 MiB |**********************
|
75 658.094 MiB |*************************************
|
100 1022.305 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 180.306 MiB
0 3.137 MiB |
|
25 12.395 MiB |
|
50 68.301 MiB |****
|
75 287.266 MiB |****************
|
100 1001.176 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 212.827 MiB
0 10.887 MiB |
|
25 20.914 MiB |*
|
50 38.023 MiB |**
|
75 458.492 MiB |*******************************
|
100 866.816 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 108.935 MiB
0 1.840 MiB |
|
25 2.219 MiB |
|
50 5.582 MiB |
|
75 119.770 MiB |******
|
100 1.004 GiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 164.604 MiB
0 3.973 MiB |
|
25 4.559 MiB |
|
50 36.355 MiB |***
|
75 279.863 MiB |*************************
|
100 650.730 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 226.909 MiB
0 9.691 MiB |
|
25 11.453 MiB |*
|
50 25.648 MiB |**
|
75 560.414 MiB |*************************************************
|
100 664.480 MiB
|***********************************************************|
Regions are like:
monitoring_start: 2.250 s
monitoring_end: 2.350 s
monitoring_duration: 100.425 ms
target_id: 0
nr_regions: 488
000012c00000-00002c14a000( 405.289 MiB): 0
00002c14a000-000044f05000( 397.730 MiB): 0
000044f05000-00005d106000( 386.004 MiB): 0
00005d106000-0000765f9000( 404.949 MiB): 0
0000765f9000-0000867b8000( 257.746 MiB): 0
0000867b8000-00009fb18000( 403.375 MiB): 0
00009fb18000-0000b932d000( 408.082 MiB): 0
0000b932d000-0000d1376000( 384.285 MiB): 0
0000d1376000-0000e765d000( 354.902 MiB): 0
0000e765d000-0000fccab000( 342.305 MiB): 0
0000fccab000-0001132b2000( 358.027 MiB): 0
0001132b2000-00012bc48000( 393.586 MiB): 0
00012bc48000-0001445c8000( 393.500 MiB): 0
0001445c8000-00015cc7c000( 390.703 MiB): 0
00015cc7c000-000172a92000( 350.086 MiB): 0
000172a92000-00018b19b000( 391.035 MiB): 0
00018b19b000-0001a0f74000( 349.848 MiB): 0
0001a0f74000-0001b707b000( 353.027 MiB): 0
0001b707b000-0001d05fc000( 405.504 MiB): 0
0001d05fc000-0001e10f9000( 266.988 MiB): 0
0001e10f9000-0001f75c6000( 356.801 MiB): 0
0001f75c6000-00020fc0e000( 390.281 MiB): 0
00020fc0e000-00022923d000( 406.184 MiB): 0
00022923d000-000240acb000( 376.555 MiB): 0
000240acb000-000250b48000( 256.488 MiB): 0
000250b48000-000266cc5000( 353.488 MiB): 0
000266cc5000-00027bf32000( 338.426 MiB): 0
00027bf32000-0002944fe000( 389.797 MiB): 0
0002944fe000-0002ac8c6000( 387.781 MiB): 0
0002ac8c6000-0002c52e8000( 394.133 MiB): 0
0002c52e8000-0002db30d000( 352.145 MiB): 0
0002db30d000-0002eedd0000( 314.762 MiB): 0
0002eedd0000-0003075f9000( 392.160 MiB): 0
0003075f9000-000316ff8000( 249.996 MiB): 0
000316ff8000-00032f8ec000( 392.953 MiB): 0
00032f8ec000-000342f72000( 310.523 MiB): 0
000342f72000-00035be55000( 398.887 MiB): 0
00035be55000-00037299d000( 363.281 MiB): 0
00037299d000-00038a649000( 380.672 MiB): 0
00038a649000-0003a2dbd000( 391.453 MiB): 0
0003a2dbd000-0003b2b96000( 253.848 MiB): 0
0003b2b96000-0003c8902000( 349.422 MiB): 0
0003c8902000-0003e06c2000( 381.750 MiB): 0
0003e06c2000-0003f92d9000( 396.090 MiB): 0
0003f92d9000-000400000000( 109.152 MiB): 0
005858822000-00585882c000( 40.000 KiB): 0
006fd5edd000-006fef5c4000( 406.902 MiB): 0
006fef5c4000-0070089ba000( 403.961 MiB): 0
0070089ba000-007021339000( 393.496 MiB): 0
007021339000-007039f08000( 395.809 MiB): 0
007039f08000-00704fb11000( 348.035 MiB): 0
00704fb11000-007067cfb000( 385.914 MiB): 0
007067cfb000-00707e1a2000( 356.652 MiB): 0
00707e1a2000-0070940cd000( 351.168 MiB): 0
0070940cd000-0070ad473000( 403.648 MiB): 0
0070ad473000-0070c65d6000( 401.387 MiB): 0
0070c65d6000-0070dbf07000( 345.191 MiB): 0
0070dbf07000-0070f5077000( 401.438 MiB): 0
0070f5077000-007107593000( 293.109 MiB): 0
007107593000-00711a860000( 306.801 MiB): 0
00711a860000-007131800000( 367.625 MiB): 0
007131800000-007143aff000( 290.996 MiB): 0
007143aff000-00715ce42000( 403.262 MiB): 0
00715ce42000-0071756e3000( 392.629 MiB): 0
0071756e3000-00718d7f1000( 385.055 MiB): 0
00718d7f1000-0071a5bf2000( 388.004 MiB): 0
0071a5bf2000-0071be018000( 388.148 MiB): 0
0071be018000-0071ceefa000( 270.883 MiB): 0
0071ceefa000-0071e3f2d000( 336.199 MiB): 0
0071e3f2d000-0071faa8c000( 363.371 MiB): 0
0071faa8c000-00721257e000( 378.945 MiB): 0
00721257e000-00722ace6000( 391.406 MiB): 0
00722ace6000-00724308f000( 387.660 MiB): 0
00724308f000-0072597ab000( 359.109 MiB): 0
0072597ab000-007272a9e000( 402.949 MiB): 0
007272a9e000-00728bb31000( 400.574 MiB): 0
00728bb31000-00729fa37000( 319.023 MiB): 0
00729fa37000-0072b3486000( 314.309 MiB): 0
0072b3486000-0072cc684000( 401.992 MiB): 0
0072cc684000-0072e51af000( 395.168 MiB): 0
0072e51af000-0072fb3c6000( 354.090 MiB): 0
0072fb3c6000-007305891000( 164.793 MiB): 0
007305891000-007305893000( 8.000 KiB): 11
007305893000-007305897000( 16.000 KiB): 14
007305897000-00730589a000( 12.000 KiB): 11
00730589a000-0073058a0000( 24.000 KiB): 0
0073058a0000-0073058a2000( 8.000 KiB): 11
0073058a2000-0073058ab000( 36.000 KiB): 8
0073058ab000-0073058ad000( 8.000 KiB): 0
0073058ad000-0073058ae000( 4.000 KiB): 13
0073058ae000-0073058b3000( 20.000 KiB): 10
0073058b3000-0073058b4000( 4.000 KiB): 5
0073058b4000-0073058b8000( 16.000 KiB): 13
0073058b8000-0073058ca000( 72.000 KiB): 11
0073058ca000-0073058cf000( 20.000 KiB): 0
0073058cf000-0073058d1000( 8.000 KiB): 13
0073058d1000-0073058d4000( 12.000 KiB): 10
0073058d4000-0073058d5000( 4.000 KiB): 0
0073058d5000-0073058d7000( 8.000 KiB): 13
0073058d7000-0073058da000( 12.000 KiB): 10
0073058da000-0073058db000( 4.000 KiB): 6
0073058db000-0073058dd000( 8.000 KiB): 11
0073058dd000-0073058df000( 8.000 KiB): 6
0073058df000-0073058e0000( 4.000 KiB): 0
0073058e0000-0073058e2000( 8.000 KiB): 11
0073058e2000-0073058e9000( 28.000 KiB): 14
0073058e9000-0073058ef000( 24.000 KiB): 6
0073058ef000-0073058f1000( 8.000 KiB): 12
0073058f1000-007305997000( 664.000 KiB): 5
007305997000-007305999000( 8.000 KiB): 11
007305999000-00730599d000( 16.000 KiB): 7
00730599d000-00730599e000( 4.000 KiB): 9
00730599e000-00730599f000( 4.000 KiB): 0
00730599f000-0073059a1000( 8.000 KiB): 2
0073059a1000-0073059a6000( 20.000 KiB): 0
0073059a6000-0073059ae000( 32.000 KiB): 9
0073059ae000-0073059c1000( 76.000 KiB): 0
0073059c1000-0073059c2000( 4.000 KiB): 6
0073059c2000-0073059c4000( 8.000 KiB): 3
0073059c4000-0073059c6000( 8.000 KiB): 12
0073059c6000-0073059c8000( 8.000 KiB): 0
0073059c8000-0073059ca000( 8.000 KiB): 5
0073059ca000-0073059d4000( 40.000 KiB): 0
0073059d4000-0073059d5000( 4.000 KiB): 13
0073059d5000-0073059d6000( 4.000 KiB): 4
0073059d6000-0073059d8000( 8.000 KiB): 13
0073059d8000-0073059db000( 12.000 KiB): 1
0073059db000-0073059dd000( 8.000 KiB): 12
0073059dd000-0073059e1000( 16.000 KiB): 7
0073059e1000-0073059e3000( 8.000 KiB): 12
0073059e3000-0073059e6000( 12.000 KiB): 6
0073059e6000-0073059e8000( 8.000 KiB): 10
0073059e8000-0073059e9000( 4.000 KiB): 13
0073059e9000-0073059ea000( 4.000 KiB): 9
0073059ea000-0073059ec000( 8.000 KiB): 11
0073059ec000-0073059ee000( 8.000 KiB): 3
0073059ee000-0073059f0000( 8.000 KiB): 13
0073059f0000-0073059f2000( 8.000 KiB): 0
0073059f2000-0073059f3000( 4.000 KiB): 13
0073059f3000-007305a1b000( 160.000 KiB): 3
007305a1b000-007305a20000( 20.000 KiB): 5
007305a20000-007305a22000( 8.000 KiB): 13
007305a22000-007305a23000( 4.000 KiB): 6
007305a23000-007305a25000( 8.000 KiB): 11
007305a25000-007305a28000( 12.000 KiB): 7
007305a28000-007305a2f000( 28.000 KiB): 1
007305a2f000-007305a39000( 40.000 KiB): 4
007305a39000-007305a3b000( 8.000 KiB): 9
007305a3b000-007305a3d000( 8.000 KiB): 11
007305a3d000-007305a3f000( 8.000 KiB): 5
007305a3f000-007305a43000( 16.000 KiB): 0
007305a43000-007305a44000( 4.000 KiB): 9
007305a44000-007305a49000( 20.000 KiB): 13
007305a49000-00731d952000( 383.035 MiB): 0
00731d952000-0073371b4000( 408.383 MiB): 0
0073371b4000-00735039e000( 401.914 MiB): 0
00735039e000-007368d88000( 393.914 MiB): 0
007368d88000-007381dd7000( 400.309 MiB): 0
007381dd7000-00739b19a000( 403.762 MiB): 0
00739b19a000-0073a76ea000( 197.312 MiB): 0
0073a76ea000-0073a76ec000( 8.000 KiB): 8
0073a76ec000-0073a76f0000( 16.000 KiB): 0
0073a76f0000-0073a76f1000( 4.000 KiB): 7
0073a76f1000-0073a76f4000( 12.000 KiB): 3
0073a76f4000-0073a76f5000( 4.000 KiB): 0
0073a76f5000-0073a76f7000( 8.000 KiB): 10
0073a76f7000-0073a76fa000( 12.000 KiB): 4
0073a76fa000-0073a76fb000( 4.000 KiB): 11
0073a76fb000-0073a76fe000( 12.000 KiB): 5
0073a76fe000-0073a7700000( 8.000 KiB): 13
0073a7700000-0073a7711000( 68.000 KiB): 9
0073a7711000-0073a772e000( 116.000 KiB): 5
0073a772e000-0073a7730000( 8.000 KiB): 10
0073a7730000-0073a7733000( 12.000 KiB): 0
0073a7733000-0073a7734000( 4.000 KiB): 4
0073a7734000-0073a7738000( 16.000 KiB): 2
0073a7738000-0073a773b000( 12.000 KiB): 4
0073a773b000-0073a773d000( 8.000 KiB): 0
0073a773d000-0073a7745000( 32.000 KiB): 4
0073a7745000-0073a778c000( 284.000 KiB): 0
0073a778c000-0073a77a0000( 80.000 KiB): 5
0073a77a0000-0073a77a2000( 8.000 KiB): 11
0073a77a2000-0073a77a5000( 12.000 KiB): 8
0073a77a5000-0073a77fc000( 348.000 KiB): 0
0073a77fc000-0073a7805000( 36.000 KiB): 5
0073a7805000-0073a7809000( 16.000 KiB): 7
0073a7809000-0073a780b000( 8.000 KiB): 10
0073a780b000-0073a780f000( 16.000 KiB): 6
0073a780f000-0073a7811000( 8.000 KiB): 8
0073a7811000-0073a7815000( 16.000 KiB): 3
0073a7815000-0073a7817000( 8.000 KiB): 6
0073a7817000-0073a781c000( 20.000 KiB): 10
0073a781c000-0073a781f000( 12.000 KiB): 12
0073a781f000-0073a7827000( 32.000 KiB): 10
0073a7827000-0073a782a000( 12.000 KiB): 6
0073a782a000-0073a784c000( 136.000 KiB): 7
0073a784c000-0073a784d000( 4.000 KiB): 9
0073a784d000-0073a7855000( 32.000 KiB): 6
0073a7855000-0073a7860000( 44.000 KiB): 11
0073a7860000-0073a7864000( 16.000 KiB): 9
0073a7864000-0073a7865000( 4.000 KiB): 12
0073a7865000-0073a78f6000( 580.000 KiB): 7
0073a78f6000-0073a78f7000( 4.000 KiB): 0
0073a78f7000-0073a78fb000( 16.000 KiB): 7
0073a78fb000-0073a78fc000( 4.000 KiB): 0
0073a78fc000-0073a793a000( 248.000 KiB): 8
0073a793a000-0073a7941000( 28.000 KiB): 6
0073a7941000-0073a7948000( 28.000 KiB): 8
0073a7948000-0073a7953000( 44.000 KiB): 4
0073a7953000-0073a795b000( 32.000 KiB): 10
0073a795b000-0073a7961000( 24.000 KiB): 7
0073a7961000-0073a7964000( 12.000 KiB): 10
0073a7964000-0073a7967000( 12.000 KiB): 7
0073a7967000-0073a7969000( 8.000 KiB): 2
0073a7969000-0073a796b000( 8.000 KiB): 7
0073a796b000-0073a796d000( 8.000 KiB): 1
0073a796d000-0073a796f000( 8.000 KiB): 5
0073a796f000-0073a7974000( 20.000 KiB): 0
0073a7974000-0073a7975000( 4.000 KiB): 6
0073a7975000-0073a7977000( 8.000 KiB): 2
0073a7977000-0073a7978000( 4.000 KiB): 13
0073a7978000-0073a797a000( 8.000 KiB): 7
0073a797a000-0073a797d000( 12.000 KiB): 3
0073a797d000-0073a7983000( 24.000 KiB): 1
0073a7983000-0073a7985000( 8.000 KiB): 12
0073a7985000-0073a7987000( 8.000 KiB): 0
0073a7987000-0073a7994000( 52.000 KiB): 11
0073a7994000-0073a7998000( 16.000 KiB): 8
0073a7998000-0073a799d000( 20.000 KiB): 10
0073a799d000-0073a799f000( 8.000 KiB): 8
0073a799f000-0073a79a1000( 8.000 KiB): 13
0073a79a1000-0073a79a6000( 20.000 KiB): 10
0073a79a6000-0073a79ad000( 28.000 KiB): 6
0073a79ad000-0073a79b1000( 16.000 KiB): 8
0073a79b1000-0073a79b2000( 4.000 KiB): 4
0073a79b2000-0073a79b3000( 4.000 KiB): 11
0073a79b3000-0073a79b6000( 12.000 KiB): 5
0073a79b6000-0073a79b7000( 4.000 KiB): 1
0073a79b7000-0073a79d2000( 108.000 KiB): 12
0073a79d2000-0073a79f7000( 148.000 KiB): 8
0073a79f7000-0073a79f8000( 4.000 KiB): 1
0073a79f8000-0073a7a13000( 108.000 KiB): 6
0073a7a13000-0073a7a19000( 24.000 KiB): 9
0073a7a19000-0073a7a1a000( 4.000 KiB): 0
0073a7a1a000-0073a7a1b000( 4.000 KiB): 7
0073a7a1b000-0073a7a2e000( 76.000 KiB): 0
0073a7a2e000-0073a7a30000( 8.000 KiB): 3
0073a7a30000-0073a7a33000( 12.000 KiB): 0
0073a7a33000-0073a7a35000( 8.000 KiB): 4
0073a7a35000-0073a7a36000( 4.000 KiB): 9
0073a7a36000-0073a7a37000( 4.000 KiB): 12
0073a7a37000-0073a7a50000( 100.000 KiB): 4
0073a7a50000-0073a7a58000( 32.000 KiB): 6
0073a7a58000-0073a7aa1000( 292.000 KiB): 4
0073a7aa1000-0073a7b23000( 520.000 KiB): 9
0073a7b23000-0073a7b32000( 60.000 KiB): 7
0073a7b32000-0073a7b71000( 252.000 KiB): 10
0073a7b71000-0073c0c49000( 400.844 MiB): 0
0073c0c49000-0073d9ea9000( 402.375 MiB): 0
0073d9ea9000-0073f31cb000( 403.133 MiB): 0
0073f31cb000-00740c9b8000( 407.926 MiB): 0
00740c9b8000-0074179b1000( 175.973 MiB): 0
0074179b1000-0074179b2000( 4.000 KiB): 5
0074179b2000-0074179b3000( 4.000 KiB): 10
0074179b3000-0074179b7000( 16.000 KiB): 4
0074179b7000-0074179ba000( 12.000 KiB): 11
0074179ba000-0074179bb000( 4.000 KiB): 6
0074179bb000-0074179bc000( 4.000 KiB): 9
0074179bc000-0074179be000( 8.000 KiB): 5
0074179be000-0074179bf000( 4.000 KiB): 0
0074179bf000-0074179c0000( 4.000 KiB): 5
0074179c0000-0074179ca000( 40.000 KiB): 10
0074179ca000-0074179cd000( 12.000 KiB): 3
0074179cd000-0074179d0000( 12.000 KiB): 0
0074179d0000-0074179e6000( 88.000 KiB): 5
0074179e6000-00742f340000( 377.352 MiB): 0
00742f340000-0074479a6000( 390.398 MiB): 0
0074479a6000-007457a3d000( 256.590 MiB): 0
007457a3d000-007457a43000( 24.000 KiB): 7
007457a43000-007457a44000( 4.000 KiB): 12
007457a44000-007457a5c000( 96.000 KiB): 6
007457a5c000-007457a5e000( 8.000 KiB): 11
007457a5e000-007457a60000( 8.000 KiB): 6
007457a60000-007457a69000( 36.000 KiB): 0
007457a69000-007457a7f000( 88.000 KiB): 2
007457a7f000-007457a82000( 12.000 KiB): 7
007457a82000-007457a8b000( 36.000 KiB): 0
007457a8b000-007457a90000( 20.000 KiB): 2
007457a90000-007457a91000( 4.000 KiB): 8
007457a91000-007457a98000( 28.000 KiB): 5
007457a98000-007457a9a000( 8.000 KiB): 8
007457a9a000-007457aa9000( 60.000 KiB): 5
007457aa9000-007457abc000( 76.000 KiB): 2
007457abc000-007457ad8000( 112.000 KiB): 5
007457ad8000-007457ae9000( 68.000 KiB): 2
007457ae9000-007457aef000( 24.000 KiB): 0
007457aef000-007457af2000( 12.000 KiB): 4
007457af2000-007457af8000( 24.000 KiB): 7
007457af8000-007457b0a000( 72.000 KiB): 5
007457b0a000-007457b17000( 52.000 KiB): 0
007457b17000-007457b21000( 40.000 KiB): 7
007457b21000-007457b23000( 8.000 KiB): 0
007457b23000-007457b25000( 8.000 KiB): 3
007457b25000-007457b2c000( 28.000 KiB): 6
007457b2c000-007457b2e000( 8.000 KiB): 11
007457b2e000-007457b30000( 8.000 KiB): 0
007457b30000-007457b34000( 16.000 KiB): 3
007457b34000-007457b3c000( 32.000 KiB): 7
007457b3c000-007457b3e000( 8.000 KiB): 5
007457b3e000-007457b3f000( 4.000 KiB): 9
007457b3f000-007457b5b000( 112.000 KiB): 0
007457b5b000-007457b5c000( 4.000 KiB): 9
007457b5c000-007457b5e000( 8.000 KiB): 4
007457b5e000-007457b61000( 12.000 KiB): 7
007457b61000-007457b62000( 4.000 KiB): 0
007457b62000-007457b63000( 4.000 KiB): 8
007457b63000-007457b65000( 8.000 KiB): 3
007457b65000-007457b6d000( 32.000 KiB): 0
007457b6d000-007457b6f000( 8.000 KiB): 7
007457b6f000-007457b73000( 16.000 KiB): 1
007457b73000-007457b77000( 16.000 KiB): 4
007457b77000-007457b78000( 4.000 KiB): 10
007457b78000-007457b88000( 64.000 KiB): 0
007457b88000-007457b8b000( 12.000 KiB): 4
007457b8b000-007457b8d000( 8.000 KiB): 1
007457b8d000-007457b92000( 20.000 KiB): 4
007457b92000-007457baa000( 96.000 KiB): 6
007457baa000-007457bac000( 8.000 KiB): 0
007457bac000-007457bae000( 8.000 KiB): 6
007457bae000-007457bb2000( 16.000 KiB): 0
007457bb2000-007457bb5000( 12.000 KiB): 4
007457bb5000-007457bb7000( 8.000 KiB): 9
007457bb7000-007457bb8000( 4.000 KiB): 0
007457bb8000-007457bb9000( 4.000 KiB): 7
007457bb9000-007457bba000( 4.000 KiB): 0
007457bba000-007457bbc000( 8.000 KiB): 5
007457bbc000-007457bbd000( 4.000 KiB): 9
007457bbd000-007457bbe000( 4.000 KiB): 7
007457bbe000-007457bc0000( 8.000 KiB): 4
007457bc0000-007457bc1000( 4.000 KiB): 6
007457bc1000-007457bc3000( 8.000 KiB): 0
007457bc3000-007457bc6000( 12.000 KiB): 8
007457bc6000-007457bc8000( 8.000 KiB): 10
007457bc8000-007457bca000( 8.000 KiB): 6
007457bca000-007457bcf000( 20.000 KiB): 8
007457bcf000-007457bd0000( 4.000 KiB): 0
007457bd0000-007457bd1000( 4.000 KiB): 7
007457bd1000-007457bd3000( 8.000 KiB): 2
007457bd3000-007457bd5000( 8.000 KiB): 0
007457bd5000-007457bd7000( 8.000 KiB): 5
007457bd7000-007457c8c000( 724.000 KiB): 0
007457c8c000-007457cbd000( 196.000 KiB): 6
007457cbd000-007457cd0000( 76.000 KiB): 0
007457cd0000-007457d50000( 512.000 KiB): 3
007457d50000-007457dfe000( 696.000 KiB): 7
007457dfe000-007457e12000( 80.000 KiB): 5
007457e12000-0074687b1000( 265.621 MiB): 0
0074687b1000-007481a01000( 402.312 MiB): 0
007481a01000-00749997a000( 383.473 MiB): 0
00749997a000-0074b130f000( 377.582 MiB): 0
0074b130f000-0074ca3dd000( 400.805 MiB): 0
0074ca3dd000-0074e1ba0000( 375.762 MiB): 0
0074e1ba0000-0074fa5d3000( 394.199 MiB): 0
0074fa5d3000-007513006000( 394.199 MiB): 0
007513006000-007525068000( 288.383 MiB): 0
007525068000-007534c34000( 251.797 MiB): 0
007534c34000-00754ca08000( 381.828 MiB): 0
00754ca08000-007560ddc000( 323.828 MiB): 0
007560ddc000-007579e5e000( 400.508 MiB): 0
007579e5e000-007591e6d000( 384.059 MiB): 0
007591e6d000-0075a86ab000( 360.242 MiB): 0
0075a86ab000-0075bddd3000( 343.156 MiB): 0
0075bddd3000-0075d562c000( 376.348 MiB): 0
0075d562c000-0075ed81b000( 385.934 MiB): 0
0075ed81b000-00760536d000( 379.320 MiB): 0
00760536d000-00761b6c3000( 355.336 MiB): 0
00761b6c3000-007631df5000( 359.195 MiB): 0
007631df5000-00764ac4a000( 398.332 MiB): 0
00764ac4a000-007663774000( 395.164 MiB): 0
007663774000-00767a2f4000( 363.500 MiB): 0
00767a2f4000-00769004e000( 349.352 MiB): 0
00769004e000-0076a32e5000( 306.590 MiB): 0
0076a32e5000-0076ba7dc000( 372.965 MiB): 0
0076ba7dc000-0076d104c000( 360.438 MiB): 0
0076d104c000-0076e9e04000( 397.719 MiB): 0
0076e9e04000-0077005cd000( 359.785 MiB): 0
0077005cd000-0077163ad000( 349.875 MiB): 0
0077163ad000-00772f3b6000( 400.035 MiB): 0
00772f3b6000-007747429000( 384.449 MiB): 0
007747429000-00775ccc7000( 344.617 MiB): 0
00775ccc7000-007775b1d000( 398.336 MiB): 0
007775b1d000-00778d5fa000( 378.863 MiB): 0
00778d5fa000-0077a0b88000( 309.555 MiB): 0
0077a0b88000-0077b415b000( 309.824 MiB): 0
0077b415b000-0077cca1e000( 392.762 MiB): 0
0077cca1e000-0077de313000( 280.957 MiB): 0
0077de313000-0077f420d000( 350.977 MiB): 0
0077f420d000-00780d701000( 404.953 MiB): 0
00780d701000-007826047000( 393.273 MiB): 0
007826047000-007836cda000( 268.574 MiB): 0
007836cda000-00785010d000( 404.199 MiB): 0
00785010d000-0078695e1000( 404.828 MiB): 0
0078695e1000-00788210a000( 395.160 MiB): 0
00788210a000-00789b0e9000( 399.871 MiB): 0
00789b0e9000-0078b13f8000( 355.059 MiB): 0
0078b13f8000-0078c8052000( 364.352 MiB): 0
0078c8052000-0078d8b7e000( 267.172 MiB): 0
0078d8b7e000-0078eb5e6000( 298.406 MiB): 0
0078eb5e6000-007904c7e000( 406.594 MiB): 0
007904c7e000-00791bbef000( 367.441 MiB): 0
00791bbef000-00792887d000( 204.555 MiB): 0
00792887d000-007941070000( 391.949 MiB): 0
007941070000-007955283000( 322.074 MiB): 0
007955283000-00796e01c000( 397.598 MiB): 0
00796e01c000-0079870e8000( 400.797 MiB): 0
0079870e8000-0079972d5000( 257.926 MiB): 0
0079972d5000-0079ada20000( 359.293 MiB): 0
0079ada20000-0079c66f9000( 396.848 MiB): 0
0079c66f9000-0079df2f4000( 395.980 MiB): 0
0079df2f4000-0079f4943000( 342.309 MiB): 0
0079f4943000-007a0918c000( 328.285 MiB): 0
007a0918c000-007a21cd9000( 395.301 MiB): 0
007a21cd9000-007a3a8e6000( 396.051 MiB): 0
007a3a8e6000-007a51841000( 367.355 MiB): 0
007a51841000-007a66e4a000( 342.035 MiB): 0
007a66e4a000-007a72270000( 180.148 MiB): 0
007a72270000-007a8a370000( 385.000 MiB): 0
007a8a370000-007a9b42e000( 272.742 MiB): 0
007a9b42e000-007aafd69000( 329.230 MiB): 0
007aafd69000-007ac7e3a000( 384.816 MiB): 0
007ac7e3a000-007adfbbc000( 381.508 MiB): 0
007adfbbc000-007af6579000( 361.738 MiB): 0
007af6579000-007b0c320000( 349.652 MiB): 0
007b0c320000-007b1f4b2000( 305.570 MiB): 0
007b1f4b2000-007b380d9000( 396.152 MiB): 0
007b380d9000-007b516be000( 405.895 MiB): 0
007b516be000-007b69d9b000( 390.863 MiB): 0
007b69d9b000-007b8206a000( 386.809 MiB): 0
007b8206a000-007b92bf0000( 267.523 MiB): 0
007b92bf0000-007ba1d2a000( 241.227 MiB): 0
007ba1d2a000-007bb716c000( 340.258 MiB): 0
007bb716c000-007bd0252000( 400.898 MiB): 0
007bd0252000-007be5ad8000( 344.523 MiB): 0
007be5ad8000-007bfc6f6000( 364.117 MiB): 0
007bfc6f6000-007c148e4000( 385.930 MiB): 0
007c148e4000-007c2bf85000( 374.629 MiB): 0
007c2bf85000-007c401ce000( 322.285 MiB): 0
007c401ce000-007c57cd1000( 379.012 MiB): 0
007c57cd1000-007c70b1c000( 398.293 MiB): 0
007c70b1c000-007c8895e000( 382.258 MiB): 0
007c8895e000-007c9fca1000( 371.262 MiB): 0
007c9fca1000-007cb6566000( 360.770 MiB): 0
007cb6566000-007cce54f000( 383.910 MiB): 0
007cce54f000-007ce6ef9000( 393.664 MiB): 0
007ce6ef9000-007cff709000( 392.062 MiB): 0
007cff709000-007d158ba000( 353.691 MiB): 0
007d158ba000-007d2e725000( 398.418 MiB): 0
007d2e725000-007d46614000( 382.934 MiB): 0
007d46614000-007d5e963000( 387.309 MiB): 0
007d5e963000-007d6b9d0000( 208.426 MiB): 0
007d6b9d0000-007d843c4000( 393.953 MiB): 0
007d843c4000-007d9aa8a000( 358.773 MiB): 0
007d9aa8a000-007db1101000( 358.465 MiB): 0
007db1101000-007dc8fb0000( 382.684 MiB): 0
007dc8fb0000-007de1fc3000( 400.074 MiB): 0
007de1fc3000-007df9408000( 372.270 MiB): 0
007df9408000-007e09ae4000( 262.859 MiB): 0
007e09ae4000-007e21d3b000( 386.340 MiB): 0
007e21d3b000-007e3921e000( 372.887 MiB): 0
007e3921e000-007e519f0000( 391.820 MiB): 0
007e519f0000-007e63a63000( 288.449 MiB): 0
007e63a63000-007e7a69c000( 364.223 MiB): 0
007e7a69c000-007e8c3b9000( 285.113 MiB): 0
007e8c3b9000-007ea3ed2000( 379.098 MiB): 0
007ea3ed2000-007eb6e90000( 303.742 MiB): 0
007eb6e90000-007ecbdeb000( 335.355 MiB): 0
007ecbdeb000-007ee2f8e000( 369.637 MiB): 0
007ee2f8e000-007ef9ec8000( 367.227 MiB): 0
007ef9ec8000-007f12ac7000( 395.996 MiB): 0
007f12ac7000-007f2c18f000( 406.781 MiB): 0
007f2c18f000-007f43c22000( 378.574 MiB): 0
007f43c22000-007f5baee000( 382.797 MiB): 0
007f5baee000-007f74a66000( 399.469 MiB): 0
007f74a66000-007f8caaf000( 384.285 MiB): 0
007f8caaf000-007fa423b000( 375.547 MiB): 0
007fa423b000-007fb9fb6000( 349.480 MiB): 0
007fb9fb6000-007fd29ae000( 393.969 MiB): 0
007fd29ae000-007fdbd6e000( 147.750 MiB): 0
Though I am not quite sure if it is accurate enough :-) so fixed-gran would be
a nice feature.
> On my x86 test machine, the 'min_nr_regions' was respected:
>
> $ sudo damo record --minr 100 paddr
> [...]
> $ sudo damo report nr_regions
> # <percentile> <# regions>
> # target_id 0
> # avr: 111
> 0 109
> 25 111
> 50 111
> 75 113
> 100 116
>
> Unfortunately, 'damo monitor' has no option to set the min_nr_regions. You
> could modify the default value of the option of damo-record on 'damo_record.py'
> for the case. Sorry for your inconvenient.
>
> >
> > > If we don't see small regions having some non-zero access frequency, we would
> > > be better to be worried. However, it is finding the small 4 KiB regions having
> > > higher access frequency successfully. The 4 KiB region is not because the
> > > region is having 4 KiB VMA, but the address region shows high access frequency.
> > >
> > > >
> > > > Typical characteristics of a large Android app is that it has
> > > > thousands of vma and very large virtual address spaces:
> > > > ~/damo # pmap 2550 | wc -l
> > > > 8522
> > > >
> > > > ~/damo # pmap 2550
> > > > ...
> > > > 0000007992bbe000 4K r---- [ anon ]
> > > > 0000007992bbf000 24K rw--- [ anon ]
> > > > 0000007fe8753000 4K ----- [ anon ]
> > > > 0000007fe8754000 8188K rw--- [ stack ]
> > > > total 36742112K
> > > >
> > > > Because the whole vma list is too long, I have put the list here for
> > > > you to download:
> > > > wget http://www.linuxep.com/patches/android-app-vmas
> > > >
> > > > I can reproduce this problem on other Apps like youtube as well.
> > > > I suppose we need to boost the algorithm of splitting regions for this
> > > > kind of application.
> > > > Any thoughts?
> > >
> > > As mentioned above, this looks not a problem, as DAMON's monitoring regions is
> > > not constructed based on VMAs but access patterns.
> >
> > What makes me believe it is a problem is that it seems we are getting
> > an incorrect
> > wss based on the data reported from kernel:
> > ~/damo # ./damo monitor --report_type wss --count 20 2561
> > # <percentile> <wss>
> > # target_id 0
> > # avr: 443.854 MiB
> > 0 0 B |
> > |
> > 25 4.000 KiB |
> > |
> > 50 12.000 KiB |
> > |
> > 75 20.000 KiB |
> > |
> > 100 5.851 GiB
> > |***********************************************************|
> [...]
> >
> > And I have a question, what do percentile 0,25,50,75 mean here?
> > Why are they so different with percentile 100?
> > For example, 0,25,50,75 has only KiB but 100 has GiB.
>
> For each aggregation interval, we get one snapshot. So, if we have a
> monitoring results that recorded for, say, 100 aggregation interval, we have
> 100 snapshots. 'damo' calculates working set size of each snapshot by summing
> size of regions assumed to be accessed at least once. So, in this example, we
> get 100 wss values. Then, 'damo' sorts the values and provides the smallest
> one as 0-th percentile, 25th small value as 25-th percentile, and so on.
>
> 100-th percentile wss is usually noisy, as DAMON regions shouldn't be converged
> well at the beginning of the record. I believe that could be the reason why
> the 100-th percentile wss is so unexpectedly big.
>
> I personally use 50-th percentile as reliable value.
Thanks, it seems you mean if we get 100 snapshots with values exactly as
2, 4, 6, 8, 10..... , 198, 200 (just an example)
for 25%, we will get 50; for 50%, we will get 100; for 75%, we will
get 150, for 100%,
we will get 200. Right?
I am not quite sure I understand "as DAMON regions shouldn't be converged well
at the beginning of the record", in case we are monitoring with
--count=2000, I suppose
only at the beginning, regions are not splitted very well? When we
have run monitor
for a while, regions should have been relatively stable? I mean I
don't quite understand
why 100% is noise and 50% is more reliable.
>
> >
> > I guess Asphalt 9 should be normally accessing hundreds of megabytes of memory.
>
> Each wss value is calculated for each aggregation interval. That is, the wss
> value is amount of memory that accessed within an aggregation interval. As the
> default aggregation interval is 100ms, the working set size could be smaller
> than your expectation.
>
> For such cases, 'damo report wss' has an option called '--work_time' to
> simulate higher aggregation intervals. That is, you can get the total size of
> the memory regions that DAMON shown access to those within arbitrary amount of
> time using the option. 'damo monitor' doesn't support the option at the
> moment, though.
>
> >
> > >
> > > Nevertheless, I believe there are many rooms for improvement of DAMON's access
> > > frequency. I want to implement fixed-gran monitoring feature first, and
> > > develop some accuracy optimizations using the fixed-gran monitoring as
> > > comparison target.
> >
> > Does fixed-gran mean splitting VA to some regions equally with the same size?
> > for example, if we have 1GB VA, we split it into 512 regions in 2MB
> > size, something like that?
>
> Yes, exactly. That's the idea. It would incur much higher, and unbounded
> overhead, but could be a good comparison target for accuracy improvement of
> DAMON. Also, for some environments that doesn't care about the monitoring
> overhead, it could be used well.
>
>
> Thanks,
> SJ
>
> > >
> > > If I'm missing something or the explanation was not enough, please feel free to
> > > let me know.
> > >
> > >
> > > Thank,
> > > SJ
> > >
Thanks
Barry
^ permalink raw reply related [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 9:22 ` Barry Song
2022-04-27 11:55 ` Rongwei Wang
@ 2022-04-28 2:04 ` Rongwei Wang
2022-04-28 7:37 ` Barry Song
1 sibling, 1 reply; 23+ messages in thread
From: Rongwei Wang @ 2022-04-28 2:04 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On 4/27/22 5:22 PM, Barry Song wrote:
> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>>
>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
>> <rongwei.wang@linux.alibaba.com> wrote:
>>>
>>>
>>>
>>> On 4/27/22 7:19 AM, Barry Song wrote:
>>>> Hi SeongJae & Andrew,
>>>> (also Cc-ed main damon developers)
>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>>> that vaddr regions don't split well on large Android Apps though
>>>> everything works well on native Apps.
>>>>
>>>> I have tried the below two cases on an Android phone with 12GB memory
>>>> and snapdragon 888 CPU.
>>>> 1. a native program with small memory working set as below,
>>>> #define size (1024*1024*100)
>>>> main()
>>>> {
>>>> volatile int *p = malloc(size);
>>>> memset(p, 0x55, size);
>>>>
>>>> while(1) {
>>>> int i;
>>>> for (i = 0; i < size / 4; i++)
>>>> (void)*(p + i);
>>>> usleep(1000);
>>>>
>>>> for (i = 0; i < size / 16; i++)
>>>> (void)*(p + i);
>>>> usleep(1000);
>>>>
>>>> }
>>>> }
>>>> For this application, the Damon vaddr monitor works very well.
>>>> I have modified monitor.py in the damo userspace tool a little bit to
>>>> show the raw data getting from the kernel.
>>>> Regions can split decently on this kind of applications, a typical raw
>>>> data is as below,
>>>>
>>>> monitoring_start: 2.224 s
>>>> monitoring_end: 2.329 s
>>>> monitoring_duration: 104.336 ms
>>>> target_id: 0
>>>> nr_regions: 24
>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>>> 007663494000-007663753000( 2.746 MiB): 1
>>>> 007663753000-007664251000( 10.992 MiB): 3
>>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>>> 007667f97000-007668112000( 1.480 MiB): 1
>>>> 007668112000-00766820f000(1012.000 KiB): 0
>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>>
>>>> 2. a large Android app like Asphalt 9
>>>> For this case, basically regions can't split very well, but monitor
>>>> works on small vma:
>>>>
>>>> monitoring_start: 2.220 s
>>>> monitoring_end: 2.318 s
>>>> monitoring_duration: 98.576 ms
>>>> target_id: 0
>>>> nr_regions: 15
>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>>
>>>> As you can see, we have some regions which are very very big and they
>>>> are losing the chance to be splitted. But
>>>> Damon can still monitor memory access for those small VMA areas very well like:
>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>> Hi, Barry
>>>
>>> Actually, we also had found the same problem in redis by ourselves
>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>>> hot areas to been monitored or found by DAMON, likes one or more address
>>> choose by DAMON not been accessed during sample period.
>>
>> Hi Rongwei,
>> Thanks for your comments and thanks for sharing your tools.
>>
>> I guess the cause might be:
>> in case a region is very big like 10GiB, we have only 1MiB hot pages
>> in this large region.
>> damon will randomly pick one page to sample, but the page has only
>> 1MiB/10GiB, thus
>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
>> 10000 sample periods
>> to hit the hot 1MiB in order to split this large region?
>>
>> @SeongJae, please correct me if I am wrong.
>>
>>>
>>> I'm not sure whether sets init_regions can deal with the above problem,
>>> or dynamic choose one or limited number VMA to monitor.
>>>
>>
>> I won't set a limited number of VMA as this will make the damon too hard to use
>> as nobody wants to make such complex operations, especially an Android
>> app might have more than 8000 VMAs.
>>
>> I agree init_regions might be the right place to enhance the situation.
>>
>>> I'm not sure, just share my idea.
>>>
>>> [1] https://github.com/aliyun/data-profile-tools.git
>>
>> I suppose this tool is based on damon? How do you finally resolve the problem
>> that large anon VMAs can't be splitted?
>> Anyway, I will give your tool a try.
>
> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
> though autogen.sh
> runs successfully.
>
> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> reference to `wgetch'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> reference to `subwin'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
> reference to `wattr_off'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
> reference to `wattr_on'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> reference to `wattr_off'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
> reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> reference to `wattr_off'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> reference to `wclear'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
> reference to `wrefresh'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
> reference to `endwin'
> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
> reference to `initscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> reference to `wrefresh'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
> reference to `use_default_colors'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
> reference to `start_color'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
> reference to `keypad'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
> reference to `nonl'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
> reference to `cbreak'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
> reference to `noecho'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
> reference to `curs_set'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> reference to `stdscr'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
> reference to `mvwprintw'
> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
> reference to `wrefresh'
> collect2: error: ld returned 1 exit status
> make[1]: *** [Makefile:592: datop] Error 1
> make[1]: Leaving directory '/root/data-profile-tools'
> make: *** [Makefile:438: all] Error 2
Hi, Barry
Now, the question made me realize that the compatibility of this tool is
very poor. I built a ubuntu environment at yesterday, and fixed above
errors by:
diff --git a/configure.ac b/configure.ac
index 7922f27..1ed823c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -21,13 +21,9 @@ AC_PROG_INSTALL
AC_CHECK_LIB([numa], [numa_free])
AC_CHECK_LIB([pthread], [pthread_create])
-PKG_CHECK_MODULES([CHECK], [check])
-
-PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
$ncurses_LIBS"], [
- AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
- AC_MSG_ERROR([ncurses is required but was not found])
- ], [])
-])
+AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
+ AC_MSG_ERROR([required library libncurses or ncurses not found])
+ ])
It works. But I found an another thing will hinder you using this tool.
We had developed other patches about DAMON base on upstream. This tool
only works well in ourselves kernel(anolis kernel, already open source).
Of course, I think it's unnecessary for you to change kernel, just let
you know this tool still has this problem.
Anyway, the question that you reported was valuable, made me realize
what we need to improve next.
Thanks,
Rongwei Wang
>
>>
>>>>
>>>> Typical characteristics of a large Android app is that it has
>>>> thousands of vma and very large virtual address spaces:
>>>> ~/damo # pmap 2550 | wc -l
>>>> 8522
>>>>
>>>> ~/damo # pmap 2550
>>>> ...
>>>> 0000007992bbe000 4K r---- [ anon ]
>>>> 0000007992bbf000 24K rw--- [ anon ]
>>>> 0000007fe8753000 4K ----- [ anon ]
>>>> 0000007fe8754000 8188K rw--- [ stack ]
>>>> total 36742112K
>>>>
>>>> Because the whole vma list is too long, I have put the list here for
>>>> you to download:
>>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>>
>>>> I can reproduce this problem on other Apps like youtube as well.
>>>> I suppose we need to boost the algorithm of splitting regions for this
>>>> kind of application.
>>>> Any thoughts?
>>>>
>>
>> Thanks
>> Barry
^ permalink raw reply related [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-28 2:04 ` Rongwei Wang
@ 2022-04-28 7:37 ` Barry Song
2022-04-28 11:19 ` Rongwei Wang
2022-05-16 7:03 ` Barry Song
0 siblings, 2 replies; 23+ messages in thread
From: Barry Song @ 2022-04-28 7:37 UTC (permalink / raw)
To: Rongwei Wang
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
<rongwei.wang@linux.alibaba.com> wrote:
>
>
>
> On 4/27/22 5:22 PM, Barry Song wrote:
> > On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
> >>
> >> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> >> <rongwei.wang@linux.alibaba.com> wrote:
> >>>
> >>>
> >>>
> >>> On 4/27/22 7:19 AM, Barry Song wrote:
> >>>> Hi SeongJae & Andrew,
> >>>> (also Cc-ed main damon developers)
> >>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
> >>>> that vaddr regions don't split well on large Android Apps though
> >>>> everything works well on native Apps.
> >>>>
> >>>> I have tried the below two cases on an Android phone with 12GB memory
> >>>> and snapdragon 888 CPU.
> >>>> 1. a native program with small memory working set as below,
> >>>> #define size (1024*1024*100)
> >>>> main()
> >>>> {
> >>>> volatile int *p = malloc(size);
> >>>> memset(p, 0x55, size);
> >>>>
> >>>> while(1) {
> >>>> int i;
> >>>> for (i = 0; i < size / 4; i++)
> >>>> (void)*(p + i);
> >>>> usleep(1000);
> >>>>
> >>>> for (i = 0; i < size / 16; i++)
> >>>> (void)*(p + i);
> >>>> usleep(1000);
> >>>>
> >>>> }
> >>>> }
> >>>> For this application, the Damon vaddr monitor works very well.
> >>>> I have modified monitor.py in the damo userspace tool a little bit to
> >>>> show the raw data getting from the kernel.
> >>>> Regions can split decently on this kind of applications, a typical raw
> >>>> data is as below,
> >>>>
> >>>> monitoring_start: 2.224 s
> >>>> monitoring_end: 2.329 s
> >>>> monitoring_duration: 104.336 ms
> >>>> target_id: 0
> >>>> nr_regions: 24
> >>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
> >>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
> >>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> >>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
> >>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
> >>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
> >>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> >>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> >>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
> >>>> 007661c59000-007661ee4000( 2.543 MiB): 2
> >>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
> >>>> 0076623e4000-007662837000( 4.324 MiB): 2
> >>>> 007662837000-0076630f1000( 8.727 MiB): 3
> >>>> 0076630f1000-007663494000( 3.637 MiB): 2
> >>>> 007663494000-007663753000( 2.746 MiB): 1
> >>>> 007663753000-007664251000( 10.992 MiB): 3
> >>>> 007664251000-0076666fd000( 36.672 MiB): 2
> >>>> 0076666fd000-007666e73000( 7.461 MiB): 1
> >>>> 007666e73000-007667c89000( 14.086 MiB): 2
> >>>> 007667c89000-007667f97000( 3.055 MiB): 0
> >>>> 007667f97000-007668112000( 1.480 MiB): 1
> >>>> 007668112000-00766820f000(1012.000 KiB): 0
> >>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> >>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> >>>>
> >>>> 2. a large Android app like Asphalt 9
> >>>> For this case, basically regions can't split very well, but monitor
> >>>> works on small vma:
> >>>>
> >>>> monitoring_start: 2.220 s
> >>>> monitoring_end: 2.318 s
> >>>> monitoring_duration: 98.576 ms
> >>>> target_id: 0
> >>>> nr_regions: 15
> >>>> 000012c00000-0001c301e000( 6.754 GiB): 0
> >>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
> >>>> 000371b6c000-000400000000( 2.223 GiB): 0
> >>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
> >>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> >>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
> >>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
> >>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
> >>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
> >>>> 0077e863e000-00798e214000( 6.590 GiB): 0
> >>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
> >>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
> >>>> 007c62f00000-007defb19000( 6.199 GiB): 0
> >>>> 007defb19000-007f794ef000( 6.150 GiB): 0
> >>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
> >>>>
> >>>> As you can see, we have some regions which are very very big and they
> >>>> are losing the chance to be splitted. But
> >>>> Damon can still monitor memory access for those small VMA areas very well like:
> >>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> >>> Hi, Barry
> >>>
> >>> Actually, we also had found the same problem in redis by ourselves
> >>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
> >>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> >>> hot areas to been monitored or found by DAMON, likes one or more address
> >>> choose by DAMON not been accessed during sample period.
> >>
> >> Hi Rongwei,
> >> Thanks for your comments and thanks for sharing your tools.
> >>
> >> I guess the cause might be:
> >> in case a region is very big like 10GiB, we have only 1MiB hot pages
> >> in this large region.
> >> damon will randomly pick one page to sample, but the page has only
> >> 1MiB/10GiB, thus
> >> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> >> 10000 sample periods
> >> to hit the hot 1MiB in order to split this large region?
> >>
> >> @SeongJae, please correct me if I am wrong.
> >>
> >>>
> >>> I'm not sure whether sets init_regions can deal with the above problem,
> >>> or dynamic choose one or limited number VMA to monitor.
> >>>
> >>
> >> I won't set a limited number of VMA as this will make the damon too hard to use
> >> as nobody wants to make such complex operations, especially an Android
> >> app might have more than 8000 VMAs.
> >>
> >> I agree init_regions might be the right place to enhance the situation.
> >>
> >>> I'm not sure, just share my idea.
> >>>
> >>> [1] https://github.com/aliyun/data-profile-tools.git
> >>
> >> I suppose this tool is based on damon? How do you finally resolve the problem
> >> that large anon VMAs can't be splitted?
> >> Anyway, I will give your tool a try.
> >
> > Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
> > though autogen.sh
> > runs successfully.
> >
> > /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
> > /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> > reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> > reference to `wgetch'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
> > /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> > reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> > reference to `subwin'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
> > /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
> > /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
> > /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
> > /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
> > /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
> > /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
> > reference to `wattr_off'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
> > reference to `wattr_on'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
> > /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> > reference to `wattr_off'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
> > reference to `mvwprintw'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> > reference to `wattr_off'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
> > /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> > reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> > reference to `wclear'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
> > reference to `wrefresh'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
> > reference to `endwin'
> > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
> > /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
> > reference to `initscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> > reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> > reference to `wrefresh'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
> > reference to `use_default_colors'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
> > reference to `start_color'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
> > reference to `keypad'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
> > reference to `nonl'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
> > reference to `cbreak'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
> > reference to `noecho'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
> > reference to `curs_set'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> > reference to `stdscr'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> > reference to `mvwprintw'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
> > reference to `mvwprintw'
> > /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
> > reference to `wrefresh'
> > collect2: error: ld returned 1 exit status
> > make[1]: *** [Makefile:592: datop] Error 1
> > make[1]: Leaving directory '/root/data-profile-tools'
> > make: *** [Makefile:438: all] Error 2
> Hi, Barry
>
> Now, the question made me realize that the compatibility of this tool is
> very poor. I built a ubuntu environment at yesterday, and fixed above
> errors by:
>
> diff --git a/configure.ac b/configure.ac
> index 7922f27..1ed823c 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
> AC_CHECK_LIB([numa], [numa_free])
> AC_CHECK_LIB([pthread], [pthread_create])
>
> -PKG_CHECK_MODULES([CHECK], [check])
> -
> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
> $ncurses_LIBS"], [
> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
> - AC_MSG_ERROR([ncurses is required but was not found])
> - ], [])
> -])
> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
> + AC_MSG_ERROR([required library libncurses or ncurses not found])
> + ])
>
I can confirm the patch fixed the issue I reported yesterday, thanks!
> It works. But I found an another thing will hinder you using this tool.
> We had developed other patches about DAMON base on upstream. This tool
> only works well in ourselves kernel(anolis kernel, already open source).
> Of course, I think it's unnecessary for you to change kernel, just let
> you know this tool still has this problem.
>
Although I can't use this tool directly as I am not a NUMA right now,
~/data-profile-tools # ./datop --help
Not support NUMA fault stat (DAMON)!
I am still quite interested in your design and the purpose of this project.
Unfortunately the project seems to be lacking some design doc.
And would you like to send patches to lkml regarding what you
have changed atop DAMON?
> Anyway, the question that you reported was valuable, made me realize
> what we need to improve next.
>
> Thanks,
> Rongwei Wang
> >
> >>
> >>>>
> >>>> Typical characteristics of a large Android app is that it has
> >>>> thousands of vma and very large virtual address spaces:
> >>>> ~/damo # pmap 2550 | wc -l
> >>>> 8522
> >>>>
> >>>> ~/damo # pmap 2550
> >>>> ...
> >>>> 0000007992bbe000 4K r---- [ anon ]
> >>>> 0000007992bbf000 24K rw--- [ anon ]
> >>>> 0000007fe8753000 4K ----- [ anon ]
> >>>> 0000007fe8754000 8188K rw--- [ stack ]
> >>>> total 36742112K
> >>>>
> >>>> Because the whole vma list is too long, I have put the list here for
> >>>> you to download:
> >>>> wget http://www.linuxep.com/patches/android-app-vmas
> >>>>
> >>>> I can reproduce this problem on other Apps like youtube as well.
> >>>> I suppose we need to boost the algorithm of splitting regions for this
> >>>> kind of application.
> >>>> Any thoughts?
> >>>>
> >>
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-28 7:37 ` Barry Song
@ 2022-04-28 11:19 ` Rongwei Wang
2022-05-16 7:03 ` Barry Song
1 sibling, 0 replies; 23+ messages in thread
From: Rongwei Wang @ 2022-04-28 11:19 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On 4/28/22 3:37 PM, Barry Song wrote:
> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
>>
>>
>>
>> On 4/27/22 5:22 PM, Barry Song wrote:
>>> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>>>>
>>>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
>>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On 4/27/22 7:19 AM, Barry Song wrote:
>>>>>> Hi SeongJae & Andrew,
>>>>>> (also Cc-ed main damon developers)
>>>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>>>>> that vaddr regions don't split well on large Android Apps though
>>>>>> everything works well on native Apps.
>>>>>>
>>>>>> I have tried the below two cases on an Android phone with 12GB memory
>>>>>> and snapdragon 888 CPU.
>>>>>> 1. a native program with small memory working set as below,
>>>>>> #define size (1024*1024*100)
>>>>>> main()
>>>>>> {
>>>>>> volatile int *p = malloc(size);
>>>>>> memset(p, 0x55, size);
>>>>>>
>>>>>> while(1) {
>>>>>> int i;
>>>>>> for (i = 0; i < size / 4; i++)
>>>>>> (void)*(p + i);
>>>>>> usleep(1000);
>>>>>>
>>>>>> for (i = 0; i < size / 16; i++)
>>>>>> (void)*(p + i);
>>>>>> usleep(1000);
>>>>>>
>>>>>> }
>>>>>> }
>>>>>> For this application, the Damon vaddr monitor works very well.
>>>>>> I have modified monitor.py in the damo userspace tool a little bit to
>>>>>> show the raw data getting from the kernel.
>>>>>> Regions can split decently on this kind of applications, a typical raw
>>>>>> data is as below,
>>>>>>
>>>>>> monitoring_start: 2.224 s
>>>>>> monitoring_end: 2.329 s
>>>>>> monitoring_duration: 104.336 ms
>>>>>> target_id: 0
>>>>>> nr_regions: 24
>>>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>>>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>>>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>>>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>>>>> 007663494000-007663753000( 2.746 MiB): 1
>>>>>> 007663753000-007664251000( 10.992 MiB): 3
>>>>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>>>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>>>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>>>>> 007667f97000-007668112000( 1.480 MiB): 1
>>>>>> 007668112000-00766820f000(1012.000 KiB): 0
>>>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>>>>
>>>>>> 2. a large Android app like Asphalt 9
>>>>>> For this case, basically regions can't split very well, but monitor
>>>>>> works on small vma:
>>>>>>
>>>>>> monitoring_start: 2.220 s
>>>>>> monitoring_end: 2.318 s
>>>>>> monitoring_duration: 98.576 ms
>>>>>> target_id: 0
>>>>>> nr_regions: 15
>>>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>>>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>>>>
>>>>>> As you can see, we have some regions which are very very big and they
>>>>>> are losing the chance to be splitted. But
>>>>>> Damon can still monitor memory access for those small VMA areas very well like:
>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>> Hi, Barry
>>>>>
>>>>> Actually, we also had found the same problem in redis by ourselves
>>>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>>>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>>>>> hot areas to been monitored or found by DAMON, likes one or more address
>>>>> choose by DAMON not been accessed during sample period.
>>>>
>>>> Hi Rongwei,
>>>> Thanks for your comments and thanks for sharing your tools.
>>>>
>>>> I guess the cause might be:
>>>> in case a region is very big like 10GiB, we have only 1MiB hot pages
>>>> in this large region.
>>>> damon will randomly pick one page to sample, but the page has only
>>>> 1MiB/10GiB, thus
>>>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
>>>> 10000 sample periods
>>>> to hit the hot 1MiB in order to split this large region?
>>>>
>>>> @SeongJae, please correct me if I am wrong.
>>>>
>>>>>
>>>>> I'm not sure whether sets init_regions can deal with the above problem,
>>>>> or dynamic choose one or limited number VMA to monitor.
>>>>>
>>>>
>>>> I won't set a limited number of VMA as this will make the damon too hard to use
>>>> as nobody wants to make such complex operations, especially an Android
>>>> app might have more than 8000 VMAs.
>>>>
>>>> I agree init_regions might be the right place to enhance the situation.
>>>>
>>>>> I'm not sure, just share my idea.
>>>>>
>>>>> [1] https://github.com/aliyun/data-profile-tools.git
>>>>
>>>> I suppose this tool is based on damon? How do you finally resolve the problem
>>>> that large anon VMAs can't be splitted?
>>>> Anyway, I will give your tool a try.
>>>
>>> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
>>> though autogen.sh
>>> runs successfully.
>>>
>>> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
>>> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>> reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>> reference to `wgetch'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
>>> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>> reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>> reference to `subwin'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
>>> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
>>> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
>>> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
>>> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
>>> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
>>> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
>>> reference to `wattr_off'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
>>> reference to `wattr_on'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
>>> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>> reference to `wattr_off'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
>>> reference to `mvwprintw'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>> reference to `wattr_off'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
>>> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>> reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>> reference to `wclear'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
>>> reference to `wrefresh'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
>>> reference to `endwin'
>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
>>> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
>>> reference to `initscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>> reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>> reference to `wrefresh'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
>>> reference to `use_default_colors'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
>>> reference to `start_color'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
>>> reference to `keypad'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
>>> reference to `nonl'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
>>> reference to `cbreak'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
>>> reference to `noecho'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
>>> reference to `curs_set'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>> reference to `stdscr'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>> reference to `mvwprintw'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
>>> reference to `mvwprintw'
>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
>>> reference to `wrefresh'
>>> collect2: error: ld returned 1 exit status
>>> make[1]: *** [Makefile:592: datop] Error 1
>>> make[1]: Leaving directory '/root/data-profile-tools'
>>> make: *** [Makefile:438: all] Error 2
>> Hi, Barry
>>
>> Now, the question made me realize that the compatibility of this tool is
>> very poor. I built a ubuntu environment at yesterday, and fixed above
>> errors by:
>>
>> diff --git a/configure.ac b/configure.ac
>> index 7922f27..1ed823c 100644
>> --- a/configure.ac
>> +++ b/configure.ac
>> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
>> AC_CHECK_LIB([numa], [numa_free])
>> AC_CHECK_LIB([pthread], [pthread_create])
>>
>> -PKG_CHECK_MODULES([CHECK], [check])
>> -
>> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
>> $ncurses_LIBS"], [
>> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
>> - AC_MSG_ERROR([ncurses is required but was not found])
>> - ], [])
>> -])
>> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
>> + AC_MSG_ERROR([required library libncurses or ncurses not found])
>> + ])
>>
>
> I can confirm the patch fixed the issue I reported yesterday, thanks!
>
>> It works. But I found an another thing will hinder you using this tool.
>> We had developed other patches about DAMON base on upstream. This tool
>> only works well in ourselves kernel(anolis kernel, already open source).
>> Of course, I think it's unnecessary for you to change kernel, just let
>> you know this tool still has this problem.
>>
>
> Although I can't use this tool directly as I am not a NUMA right now,
> ~/data-profile-tools # ./datop --help
> Not support NUMA fault stat (DAMON)!
>
> I am still quite interested in your design and the purpose of this project.
> Unfortunately the project seems to be lacking some design doc.
Thanks for your feedback. I am planning to write these design doc.
>
> And would you like to send patches to lkml regarding what you
> have changed atop DAMON?
Actually, Hao Xin had tried to submit these relevant patchset[1]. you
can find them in the following link. It seems the latest DAMON has
changed too much, makes these patchset can not merge directly. We will
send next version for reviewing after adjusting base on latest upstream.
[1] https://lore.kernel.org/lkml/20220217082939.2850-1-sj@kernel.org/T/
>
>> Anyway, the question that you reported was valuable, made me realize
>> what we need to improve next.
>>
>> Thanks,
>> Rongwei Wang
>>>
>>>>
>>>>>>
>>>>>> Typical characteristics of a large Android app is that it has
>>>>>> thousands of vma and very large virtual address spaces:
>>>>>> ~/damo # pmap 2550 | wc -l
>>>>>> 8522
>>>>>>
>>>>>> ~/damo # pmap 2550
>>>>>> ...
>>>>>> 0000007992bbe000 4K r---- [ anon ]
>>>>>> 0000007992bbf000 24K rw--- [ anon ]
>>>>>> 0000007fe8753000 4K ----- [ anon ]
>>>>>> 0000007fe8754000 8188K rw--- [ stack ]
>>>>>> total 36742112K
>>>>>>
>>>>>> Because the whole vma list is too long, I have put the list here for
>>>>>> you to download:
>>>>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>>>>
>>>>>> I can reproduce this problem on other Apps like youtube as well.
>>>>>> I suppose we need to boost the algorithm of splitting regions for this
>>>>>> kind of application.
>>>>>> Any thoughts?
>>>>>>
>>>>
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-28 1:27 ` Barry Song
@ 2022-04-28 16:16 ` sj
0 siblings, 0 replies; 23+ messages in thread
From: sj @ 2022-04-28 16:16 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
brendanhiggins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
linux-damon
On Thu, 28 Apr 2022 13:27:59 +1200 Barry Song <21cnbao@gmail.com> wrote:
[...]
>
> Thanks for the clarification.
>
> i hardcoded min_nr_regions to 200 by:
> diff --git a/_damon.py b/_damon.py
> index 1306ea1..82342a5 100644
> --- a/_damon.py
> +++ b/_damon.py
> @@ -344,7 +344,7 @@ def set_attrs_argparser(parser):
> parser.add_argument('-u', '--updr', metavar='<interval>', type=int,
> default=1000000, help='regions update interval (us)')
> parser.add_argument('-n', '--minr', metavar='<# regions>', type=int,
> - default=10, help='minimal number of regions')
> + default=200, help='minimal number of regions')
> parser.add_argument('-m', '--maxr', metavar='<# regions>', type=int,
> default=1000, help='maximum number of regions')
>
>
> Now wss seems to make more senses:
>
> ~/damo # ./damo monitor --report_type=wss --count=20 2551
> # <percentile> <wss>
> # target_id 0
> # avr: 235.394 MiB
> 0 0 B |
> |
> 25 2.164 MiB |
> |
> 50 129.875 MiB |*********
> |
> 75 430.547 MiB |******************************
> |
> 100 844.238 MiB
> |***********************************************************|
>
> # <percentile> <wss>
> # target_id 0
> # avr: 352.501 MiB
> 0 8.781 MiB |
[...]
> |
> 100 664.480 MiB
> |***********************************************************|
>
> Regions are like:
>
> monitoring_start: 2.250 s
> monitoring_end: 2.350 s
> monitoring_duration: 100.425 ms
> target_id: 0
> nr_regions: 488
> 000012c00000-00002c14a000( 405.289 MiB): 0
> 00002c14a000-000044f05000( 397.730 MiB): 0
> 000044f05000-00005d106000( 386.004 MiB): 0
> 00005d106000-0000765f9000( 404.949 MiB): 0
> 0000765f9000-0000867b8000( 257.746 MiB): 0
> 0000867b8000-00009fb18000( 403.375 MiB): 0
[...]
> 007f74a66000-007f8caaf000( 384.285 MiB): 0
> 007f8caaf000-007fa423b000( 375.547 MiB): 0
> 007fa423b000-007fb9fb6000( 349.480 MiB): 0
> 007fb9fb6000-007fd29ae000( 393.969 MiB): 0
> 007fd29ae000-007fdbd6e000( 147.750 MiB): 0
>
> Though I am not quite sure if it is accurate enough :-) so fixed-gran would be
> a nice feature.
Totally agreed. Thank you for making your voice! I will use this for
re-prioritizing my TODO list items.
[...]
> > >
> > > And I have a question, what do percentile 0,25,50,75 mean here?
> > > Why are they so different with percentile 100?
> > > For example, 0,25,50,75 has only KiB but 100 has GiB.
> >
> > For each aggregation interval, we get one snapshot. So, if we have a
> > monitoring results that recorded for, say, 100 aggregation interval, we have
> > 100 snapshots. 'damo' calculates working set size of each snapshot by summing
> > size of regions assumed to be accessed at least once. So, in this example, we
> > get 100 wss values. Then, 'damo' sorts the values and provides the smallest
> > one as 0-th percentile, 25th small value as 25-th percentile, and so on.
> >
> > 100-th percentile wss is usually noisy, as DAMON regions shouldn't be converged
> > well at the beginning of the record. I believe that could be the reason why
> > the 100-th percentile wss is so unexpectedly big.
> >
> > I personally use 50-th percentile as reliable value.
>
> Thanks, it seems you mean if we get 100 snapshots with values exactly as
> 2, 4, 6, 8, 10..... , 198, 200 (just an example)
>
> for 25%, we will get 50; for 50%, we will get 100; for 75%, we will
> get 150, for 100%,
> we will get 200. Right?
You're perfectly understanding my point.
>
> I am not quite sure I understand "as DAMON regions shouldn't be converged well
> at the beginning of the record", in case we are monitoring with
> --count=2000, I suppose
> only at the beginning, regions are not splitted very well? When we
> have run monitor
> for a while, regions should have been relatively stable? I mean I
> don't quite understand
> why 100% is noise and 50% is more reliable.
'damo monitor' simply repeats 'damo record' and 'damo report'. That is, it
starts recording, stop recording, reporting, and repeat. Therefore every 'damo
moitor' results are fresh ones, not a snapshot of ongoing record. Therefore
regions converge from the beginning for every 'damo monitor' output. Sorry for
the ugly implementation. It should be improved in a near future.
Thanks,
SJ
[...]
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-28 7:37 ` Barry Song
2022-04-28 11:19 ` Rongwei Wang
@ 2022-05-16 7:03 ` Barry Song
2022-05-16 15:00 ` Rongwei Wang
1 sibling, 1 reply; 23+ messages in thread
From: Barry Song @ 2022-05-16 7:03 UTC (permalink / raw)
To: Rongwei Wang
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On Thu, Apr 28, 2022 at 7:37 PM Barry Song <21cnbao@gmail.com> wrote:
>
> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
> >
> >
> >
> > On 4/27/22 5:22 PM, Barry Song wrote:
> > > On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
> > >>
> > >> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> > >> <rongwei.wang@linux.alibaba.com> wrote:
> > >>>
> > >>>
> > >>>
> > >>> On 4/27/22 7:19 AM, Barry Song wrote:
> > >>>> Hi SeongJae & Andrew,
> > >>>> (also Cc-ed main damon developers)
> > >>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
> > >>>> that vaddr regions don't split well on large Android Apps though
> > >>>> everything works well on native Apps.
> > >>>>
> > >>>> I have tried the below two cases on an Android phone with 12GB memory
> > >>>> and snapdragon 888 CPU.
> > >>>> 1. a native program with small memory working set as below,
> > >>>> #define size (1024*1024*100)
> > >>>> main()
> > >>>> {
> > >>>> volatile int *p = malloc(size);
> > >>>> memset(p, 0x55, size);
> > >>>>
> > >>>> while(1) {
> > >>>> int i;
> > >>>> for (i = 0; i < size / 4; i++)
> > >>>> (void)*(p + i);
> > >>>> usleep(1000);
> > >>>>
> > >>>> for (i = 0; i < size / 16; i++)
> > >>>> (void)*(p + i);
> > >>>> usleep(1000);
> > >>>>
> > >>>> }
> > >>>> }
> > >>>> For this application, the Damon vaddr monitor works very well.
> > >>>> I have modified monitor.py in the damo userspace tool a little bit to
> > >>>> show the raw data getting from the kernel.
> > >>>> Regions can split decently on this kind of applications, a typical raw
> > >>>> data is as below,
> > >>>>
> > >>>> monitoring_start: 2.224 s
> > >>>> monitoring_end: 2.329 s
> > >>>> monitoring_duration: 104.336 ms
> > >>>> target_id: 0
> > >>>> nr_regions: 24
> > >>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
> > >>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
> > >>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> > >>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
> > >>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
> > >>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
> > >>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> > >>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> > >>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
> > >>>> 007661c59000-007661ee4000( 2.543 MiB): 2
> > >>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
> > >>>> 0076623e4000-007662837000( 4.324 MiB): 2
> > >>>> 007662837000-0076630f1000( 8.727 MiB): 3
> > >>>> 0076630f1000-007663494000( 3.637 MiB): 2
> > >>>> 007663494000-007663753000( 2.746 MiB): 1
> > >>>> 007663753000-007664251000( 10.992 MiB): 3
> > >>>> 007664251000-0076666fd000( 36.672 MiB): 2
> > >>>> 0076666fd000-007666e73000( 7.461 MiB): 1
> > >>>> 007666e73000-007667c89000( 14.086 MiB): 2
> > >>>> 007667c89000-007667f97000( 3.055 MiB): 0
> > >>>> 007667f97000-007668112000( 1.480 MiB): 1
> > >>>> 007668112000-00766820f000(1012.000 KiB): 0
> > >>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> > >>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> > >>>>
> > >>>> 2. a large Android app like Asphalt 9
> > >>>> For this case, basically regions can't split very well, but monitor
> > >>>> works on small vma:
> > >>>>
> > >>>> monitoring_start: 2.220 s
> > >>>> monitoring_end: 2.318 s
> > >>>> monitoring_duration: 98.576 ms
> > >>>> target_id: 0
> > >>>> nr_regions: 15
> > >>>> 000012c00000-0001c301e000( 6.754 GiB): 0
> > >>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
> > >>>> 000371b6c000-000400000000( 2.223 GiB): 0
> > >>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
> > >>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> > >>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
> > >>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
> > >>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
> > >>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
> > >>>> 0077e863e000-00798e214000( 6.590 GiB): 0
> > >>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
> > >>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
> > >>>> 007c62f00000-007defb19000( 6.199 GiB): 0
> > >>>> 007defb19000-007f794ef000( 6.150 GiB): 0
> > >>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
> > >>>>
> > >>>> As you can see, we have some regions which are very very big and they
> > >>>> are losing the chance to be splitted. But
> > >>>> Damon can still monitor memory access for those small VMA areas very well like:
> > >>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> > >>> Hi, Barry
> > >>>
> > >>> Actually, we also had found the same problem in redis by ourselves
> > >>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
> > >>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> > >>> hot areas to been monitored or found by DAMON, likes one or more address
> > >>> choose by DAMON not been accessed during sample period.
> > >>
> > >> Hi Rongwei,
> > >> Thanks for your comments and thanks for sharing your tools.
> > >>
> > >> I guess the cause might be:
> > >> in case a region is very big like 10GiB, we have only 1MiB hot pages
> > >> in this large region.
> > >> damon will randomly pick one page to sample, but the page has only
> > >> 1MiB/10GiB, thus
> > >> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> > >> 10000 sample periods
> > >> to hit the hot 1MiB in order to split this large region?
> > >>
> > >> @SeongJae, please correct me if I am wrong.
> > >>
> > >>>
> > >>> I'm not sure whether sets init_regions can deal with the above problem,
> > >>> or dynamic choose one or limited number VMA to monitor.
> > >>>
> > >>
> > >> I won't set a limited number of VMA as this will make the damon too hard to use
> > >> as nobody wants to make such complex operations, especially an Android
> > >> app might have more than 8000 VMAs.
> > >>
> > >> I agree init_regions might be the right place to enhance the situation.
> > >>
> > >>> I'm not sure, just share my idea.
> > >>>
> > >>> [1] https://github.com/aliyun/data-profile-tools.git
> > >>
> > >> I suppose this tool is based on damon? How do you finally resolve the problem
> > >> that large anon VMAs can't be splitted?
> > >> Anyway, I will give your tool a try.
> > >
> > > Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
> > > though autogen.sh
> > > runs successfully.
> > >
> > > /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
> > > /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> > > reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> > > reference to `wgetch'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
> > > /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> > > reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> > > reference to `subwin'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
> > > /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
> > > /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
> > > /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
> > > /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
> > > /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
> > > /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
> > > reference to `wattr_off'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
> > > reference to `wattr_on'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
> > > /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> > > reference to `wattr_off'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
> > > reference to `mvwprintw'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> > > reference to `wattr_off'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
> > > /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> > > reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> > > reference to `wclear'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
> > > reference to `wrefresh'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
> > > reference to `endwin'
> > > /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
> > > /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
> > > reference to `initscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> > > reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> > > reference to `wrefresh'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
> > > reference to `use_default_colors'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
> > > reference to `start_color'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
> > > reference to `keypad'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
> > > reference to `nonl'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
> > > reference to `cbreak'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
> > > reference to `noecho'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
> > > reference to `curs_set'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> > > reference to `stdscr'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> > > reference to `mvwprintw'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
> > > reference to `mvwprintw'
> > > /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
> > > reference to `wrefresh'
> > > collect2: error: ld returned 1 exit status
> > > make[1]: *** [Makefile:592: datop] Error 1
> > > make[1]: Leaving directory '/root/data-profile-tools'
> > > make: *** [Makefile:438: all] Error 2
> > Hi, Barry
> >
> > Now, the question made me realize that the compatibility of this tool is
> > very poor. I built a ubuntu environment at yesterday, and fixed above
> > errors by:
> >
> > diff --git a/configure.ac b/configure.ac
> > index 7922f27..1ed823c 100644
> > --- a/configure.ac
> > +++ b/configure.ac
> > @@ -21,13 +21,9 @@ AC_PROG_INSTALL
> > AC_CHECK_LIB([numa], [numa_free])
> > AC_CHECK_LIB([pthread], [pthread_create])
> >
> > -PKG_CHECK_MODULES([CHECK], [check])
> > -
> > -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
> > $ncurses_LIBS"], [
> > - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
> > - AC_MSG_ERROR([ncurses is required but was not found])
> > - ], [])
> > -])
> > +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
> > + AC_MSG_ERROR([required library libncurses or ncurses not found])
> > + ])
> >
>
> I can confirm the patch fixed the issue I reported yesterday, thanks!
>
> > It works. But I found an another thing will hinder you using this tool.
> > We had developed other patches about DAMON base on upstream. This tool
> > only works well in ourselves kernel(anolis kernel, already open source).
> > Of course, I think it's unnecessary for you to change kernel, just let
> > you know this tool still has this problem.
> >
>
> Although I can't use this tool directly as I am not a NUMA right now,
> ~/data-profile-tools # ./datop --help
> Not support NUMA fault stat (DAMON)!
>
I wonder if you can extend it to non-numa by setting "remote" to 0%
and local to "100%" always for non-numa machines rather than death.
as your tools can map regions to .so, which seems to be quite useful.
> I am still quite interested in your design and the purpose of this project.
> Unfortunately the project seems to be lacking some design doc.
>
> And would you like to send patches to lkml regarding what you
> have changed atop DAMON?
>
> > Anyway, the question that you reported was valuable, made me realize
> > what we need to improve next.
> >
> > Thanks,
> > Rongwei Wang
> > >
> > >>
> > >>>>
> > >>>> Typical characteristics of a large Android app is that it has
> > >>>> thousands of vma and very large virtual address spaces:
> > >>>> ~/damo # pmap 2550 | wc -l
> > >>>> 8522
> > >>>>
> > >>>> ~/damo # pmap 2550
> > >>>> ...
> > >>>> 0000007992bbe000 4K r---- [ anon ]
> > >>>> 0000007992bbf000 24K rw--- [ anon ]
> > >>>> 0000007fe8753000 4K ----- [ anon ]
> > >>>> 0000007fe8754000 8188K rw--- [ stack ]
> > >>>> total 36742112K
> > >>>>
> > >>>> Because the whole vma list is too long, I have put the list here for
> > >>>> you to download:
> > >>>> wget http://www.linuxep.com/patches/android-app-vmas
> > >>>>
> > >>>> I can reproduce this problem on other Apps like youtube as well.
> > >>>> I suppose we need to boost the algorithm of splitting regions for this
> > >>>> kind of application.
> > >>>> Any thoughts?
> > >>>>
> > >>
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-05-16 7:03 ` Barry Song
@ 2022-05-16 15:00 ` Rongwei Wang
2022-05-17 11:14 ` Barry Song
0 siblings, 1 reply; 23+ messages in thread
From: Rongwei Wang @ 2022-05-16 15:00 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao
On 5/16/22 3:03 PM, Barry Song wrote:
> On Thu, Apr 28, 2022 at 7:37 PM Barry Song <21cnbao@gmail.com> wrote:
>>
>> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
>> <rongwei.wang@linux.alibaba.com> wrote:
>>>
>>>
>>>
>>> On 4/27/22 5:22 PM, Barry Song wrote:
>>>> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>>>>>
>>>>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
>>>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 4/27/22 7:19 AM, Barry Song wrote:
>>>>>>> Hi SeongJae & Andrew,
>>>>>>> (also Cc-ed main damon developers)
>>>>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>>>>>> that vaddr regions don't split well on large Android Apps though
>>>>>>> everything works well on native Apps.
>>>>>>>
>>>>>>> I have tried the below two cases on an Android phone with 12GB memory
>>>>>>> and snapdragon 888 CPU.
>>>>>>> 1. a native program with small memory working set as below,
>>>>>>> #define size (1024*1024*100)
>>>>>>> main()
>>>>>>> {
>>>>>>> volatile int *p = malloc(size);
>>>>>>> memset(p, 0x55, size);
>>>>>>>
>>>>>>> while(1) {
>>>>>>> int i;
>>>>>>> for (i = 0; i < size / 4; i++)
>>>>>>> (void)*(p + i);
>>>>>>> usleep(1000);
>>>>>>>
>>>>>>> for (i = 0; i < size / 16; i++)
>>>>>>> (void)*(p + i);
>>>>>>> usleep(1000);
>>>>>>>
>>>>>>> }
>>>>>>> }
>>>>>>> For this application, the Damon vaddr monitor works very well.
>>>>>>> I have modified monitor.py in the damo userspace tool a little bit to
>>>>>>> show the raw data getting from the kernel.
>>>>>>> Regions can split decently on this kind of applications, a typical raw
>>>>>>> data is as below,
>>>>>>>
>>>>>>> monitoring_start: 2.224 s
>>>>>>> monitoring_end: 2.329 s
>>>>>>> monitoring_duration: 104.336 ms
>>>>>>> target_id: 0
>>>>>>> nr_regions: 24
>>>>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>>>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>>>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>>>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>>>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>>>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>>>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>>>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>>>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>>>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>>>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>>>>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>>>>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>>>>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>>>>>> 007663494000-007663753000( 2.746 MiB): 1
>>>>>>> 007663753000-007664251000( 10.992 MiB): 3
>>>>>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>>>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>>>>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>>>>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>>>>>> 007667f97000-007668112000( 1.480 MiB): 1
>>>>>>> 007668112000-00766820f000(1012.000 KiB): 0
>>>>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>>>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>>>>>
>>>>>>> 2. a large Android app like Asphalt 9
>>>>>>> For this case, basically regions can't split very well, but monitor
>>>>>>> works on small vma:
>>>>>>>
>>>>>>> monitoring_start: 2.220 s
>>>>>>> monitoring_end: 2.318 s
>>>>>>> monitoring_duration: 98.576 ms
>>>>>>> target_id: 0
>>>>>>> nr_regions: 15
>>>>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>>>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>>>>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>>>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>>>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>>>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>>>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>>>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>>>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>>>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>>>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>>>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>>>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>>>>>
>>>>>>> As you can see, we have some regions which are very very big and they
>>>>>>> are losing the chance to be splitted. But
>>>>>>> Damon can still monitor memory access for those small VMA areas very well like:
>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>> Hi, Barry
>>>>>>
>>>>>> Actually, we also had found the same problem in redis by ourselves
>>>>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>>>>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>>>>>> hot areas to been monitored or found by DAMON, likes one or more address
>>>>>> choose by DAMON not been accessed during sample period.
>>>>>
>>>>> Hi Rongwei,
>>>>> Thanks for your comments and thanks for sharing your tools.
>>>>>
>>>>> I guess the cause might be:
>>>>> in case a region is very big like 10GiB, we have only 1MiB hot pages
>>>>> in this large region.
>>>>> damon will randomly pick one page to sample, but the page has only
>>>>> 1MiB/10GiB, thus
>>>>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
>>>>> 10000 sample periods
>>>>> to hit the hot 1MiB in order to split this large region?
>>>>>
>>>>> @SeongJae, please correct me if I am wrong.
>>>>>
>>>>>>
>>>>>> I'm not sure whether sets init_regions can deal with the above problem,
>>>>>> or dynamic choose one or limited number VMA to monitor.
>>>>>>
>>>>>
>>>>> I won't set a limited number of VMA as this will make the damon too hard to use
>>>>> as nobody wants to make such complex operations, especially an Android
>>>>> app might have more than 8000 VMAs.
>>>>>
>>>>> I agree init_regions might be the right place to enhance the situation.
>>>>>
>>>>>> I'm not sure, just share my idea.
>>>>>>
>>>>>> [1] https://github.com/aliyun/data-profile-tools.git
>>>>>
>>>>> I suppose this tool is based on damon? How do you finally resolve the problem
>>>>> that large anon VMAs can't be splitted?
>>>>> Anyway, I will give your tool a try.
>>>>
>>>> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
>>>> though autogen.sh
>>>> runs successfully.
>>>>
>>>> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
>>>> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>>> reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>>> reference to `wgetch'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
>>>> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>>> reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>>> reference to `subwin'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
>>>> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
>>>> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
>>>> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
>>>> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
>>>> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
>>>> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
>>>> reference to `wattr_off'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
>>>> reference to `wattr_on'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
>>>> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>>> reference to `wattr_off'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
>>>> reference to `mvwprintw'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>>> reference to `wattr_off'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
>>>> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>>> reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>>> reference to `wclear'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
>>>> reference to `wrefresh'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
>>>> reference to `endwin'
>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
>>>> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
>>>> reference to `initscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>>> reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>>> reference to `wrefresh'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
>>>> reference to `use_default_colors'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
>>>> reference to `start_color'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
>>>> reference to `keypad'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
>>>> reference to `nonl'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
>>>> reference to `cbreak'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
>>>> reference to `noecho'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
>>>> reference to `curs_set'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>>> reference to `stdscr'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>>> reference to `mvwprintw'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
>>>> reference to `mvwprintw'
>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
>>>> reference to `wrefresh'
>>>> collect2: error: ld returned 1 exit status
>>>> make[1]: *** [Makefile:592: datop] Error 1
>>>> make[1]: Leaving directory '/root/data-profile-tools'
>>>> make: *** [Makefile:438: all] Error 2
>>> Hi, Barry
>>>
>>> Now, the question made me realize that the compatibility of this tool is
>>> very poor. I built a ubuntu environment at yesterday, and fixed above
>>> errors by:
>>>
>>> diff --git a/configure.ac b/configure.ac
>>> index 7922f27..1ed823c 100644
>>> --- a/configure.ac
>>> +++ b/configure.ac
>>> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
>>> AC_CHECK_LIB([numa], [numa_free])
>>> AC_CHECK_LIB([pthread], [pthread_create])
>>>
>>> -PKG_CHECK_MODULES([CHECK], [check])
>>> -
>>> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
>>> $ncurses_LIBS"], [
>>> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
>>> - AC_MSG_ERROR([ncurses is required but was not found])
>>> - ], [])
>>> -])
>>> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
>>> + AC_MSG_ERROR([required library libncurses or ncurses not found])
>>> + ])
>>>
>>
>> I can confirm the patch fixed the issue I reported yesterday, thanks!
>>
>>> It works. But I found an another thing will hinder you using this tool.
>>> We had developed other patches about DAMON base on upstream. This tool
>>> only works well in ourselves kernel(anolis kernel, already open source).
>>> Of course, I think it's unnecessary for you to change kernel, just let
>>> you know this tool still has this problem.
>>>
>>
>> Although I can't use this tool directly as I am not a NUMA right now,
>> ~/data-profile-tools # ./datop --help
>> Not support NUMA fault stat (DAMON)!
>>
>
> I wonder if you can extend it to non-numa by setting "remote" to 0%
> and local to "100%" always for non-numa machines rather than death.
Hi Barry
That's a great suggestion. Actually, I have removed 'numa_stat' check in
datop. Maybe you can found. It does not enable numa stat when
'numa_stat' sysfs not found in the current system.
What's more, a new hot key 'f' will be introduced which can enable some
features dynamically, such as numa stat. Others features can be used
only in our internal version, likes 'f' in top, and will be open source
when stable.
> as your tools can map regions to .so, which seems to be quite useful.
enen, I'm agree with you. But you know, one region maybe covers one or
more VMAs, hard to map access count of regions to the related .so or
anon. A lazy way used by me now. I still think it's valuable in the future.
Anyway, any idea are welcome.
Thanks,
-wrw
>
>> I am still quite interested in your design and the purpose of this project.
>> Unfortunately the project seems to be lacking some design doc.
>>
>> And would you like to send patches to lkml regarding what you
>> have changed atop DAMON?
>>
>>> Anyway, the question that you reported was valuable, made me realize
>>> what we need to improve next.
>>>
>>> Thanks,
>>> Rongwei Wang
>>>>
>>>>>
>>>>>>>
>>>>>>> Typical characteristics of a large Android app is that it has
>>>>>>> thousands of vma and very large virtual address spaces:
>>>>>>> ~/damo # pmap 2550 | wc -l
>>>>>>> 8522
>>>>>>>
>>>>>>> ~/damo # pmap 2550
>>>>>>> ...
>>>>>>> 0000007992bbe000 4K r---- [ anon ]
>>>>>>> 0000007992bbf000 24K rw--- [ anon ]
>>>>>>> 0000007fe8753000 4K ----- [ anon ]
>>>>>>> 0000007fe8754000 8188K rw--- [ stack ]
>>>>>>> total 36742112K
>>>>>>>
>>>>>>> Because the whole vma list is too long, I have put the list here for
>>>>>>> you to download:
>>>>>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>>>>>
>>>>>>> I can reproduce this problem on other Apps like youtube as well.
>>>>>>> I suppose we need to boost the algorithm of splitting regions for this
>>>>>>> kind of application.
>>>>>>> Any thoughts?
>>>>>>>
>>>>>
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-05-16 15:00 ` Rongwei Wang
@ 2022-05-17 11:14 ` Barry Song
2022-05-18 3:03 ` Rongwei Wang
0 siblings, 1 reply; 23+ messages in thread
From: Barry Song @ 2022-05-17 11:14 UTC (permalink / raw)
To: Rongwei Wang
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao, vxc5208
On Tue, May 17, 2022 at 3:00 AM Rongwei Wang
<rongwei.wang@linux.alibaba.com> wrote:
>
>
>
> On 5/16/22 3:03 PM, Barry Song wrote:
> > On Thu, Apr 28, 2022 at 7:37 PM Barry Song <21cnbao@gmail.com> wrote:
> >>
> >> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
> >> <rongwei.wang@linux.alibaba.com> wrote:
> >>>
> >>>
> >>>
> >>> On 4/27/22 5:22 PM, Barry Song wrote:
> >>>> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
> >>>>>
> >>>>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> >>>>> <rongwei.wang@linux.alibaba.com> wrote:
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> On 4/27/22 7:19 AM, Barry Song wrote:
> >>>>>>> Hi SeongJae & Andrew,
> >>>>>>> (also Cc-ed main damon developers)
> >>>>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
> >>>>>>> that vaddr regions don't split well on large Android Apps though
> >>>>>>> everything works well on native Apps.
> >>>>>>>
> >>>>>>> I have tried the below two cases on an Android phone with 12GB memory
> >>>>>>> and snapdragon 888 CPU.
> >>>>>>> 1. a native program with small memory working set as below,
> >>>>>>> #define size (1024*1024*100)
> >>>>>>> main()
> >>>>>>> {
> >>>>>>> volatile int *p = malloc(size);
> >>>>>>> memset(p, 0x55, size);
> >>>>>>>
> >>>>>>> while(1) {
> >>>>>>> int i;
> >>>>>>> for (i = 0; i < size / 4; i++)
> >>>>>>> (void)*(p + i);
> >>>>>>> usleep(1000);
> >>>>>>>
> >>>>>>> for (i = 0; i < size / 16; i++)
> >>>>>>> (void)*(p + i);
> >>>>>>> usleep(1000);
> >>>>>>>
> >>>>>>> }
> >>>>>>> }
> >>>>>>> For this application, the Damon vaddr monitor works very well.
> >>>>>>> I have modified monitor.py in the damo userspace tool a little bit to
> >>>>>>> show the raw data getting from the kernel.
> >>>>>>> Regions can split decently on this kind of applications, a typical raw
> >>>>>>> data is as below,
> >>>>>>>
> >>>>>>> monitoring_start: 2.224 s
> >>>>>>> monitoring_end: 2.329 s
> >>>>>>> monitoring_duration: 104.336 ms
> >>>>>>> target_id: 0
> >>>>>>> nr_regions: 24
> >>>>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
> >>>>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
> >>>>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> >>>>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
> >>>>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
> >>>>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
> >>>>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> >>>>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> >>>>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
> >>>>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
> >>>>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
> >>>>>>> 0076623e4000-007662837000( 4.324 MiB): 2
> >>>>>>> 007662837000-0076630f1000( 8.727 MiB): 3
> >>>>>>> 0076630f1000-007663494000( 3.637 MiB): 2
> >>>>>>> 007663494000-007663753000( 2.746 MiB): 1
> >>>>>>> 007663753000-007664251000( 10.992 MiB): 3
> >>>>>>> 007664251000-0076666fd000( 36.672 MiB): 2
> >>>>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
> >>>>>>> 007666e73000-007667c89000( 14.086 MiB): 2
> >>>>>>> 007667c89000-007667f97000( 3.055 MiB): 0
> >>>>>>> 007667f97000-007668112000( 1.480 MiB): 1
> >>>>>>> 007668112000-00766820f000(1012.000 KiB): 0
> >>>>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> >>>>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> >>>>>>>
> >>>>>>> 2. a large Android app like Asphalt 9
> >>>>>>> For this case, basically regions can't split very well, but monitor
> >>>>>>> works on small vma:
> >>>>>>>
> >>>>>>> monitoring_start: 2.220 s
> >>>>>>> monitoring_end: 2.318 s
> >>>>>>> monitoring_duration: 98.576 ms
> >>>>>>> target_id: 0
> >>>>>>> nr_regions: 15
> >>>>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
> >>>>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
> >>>>>>> 000371b6c000-000400000000( 2.223 GiB): 0
> >>>>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
> >>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> >>>>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
> >>>>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
> >>>>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
> >>>>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
> >>>>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
> >>>>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
> >>>>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
> >>>>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
> >>>>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
> >>>>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
> >>>>>>>
> >>>>>>> As you can see, we have some regions which are very very big and they
> >>>>>>> are losing the chance to be splitted. But
> >>>>>>> Damon can still monitor memory access for those small VMA areas very well like:
> >>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> >>>>>> Hi, Barry
> >>>>>>
> >>>>>> Actually, we also had found the same problem in redis by ourselves
> >>>>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
> >>>>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> >>>>>> hot areas to been monitored or found by DAMON, likes one or more address
> >>>>>> choose by DAMON not been accessed during sample period.
> >>>>>
> >>>>> Hi Rongwei,
> >>>>> Thanks for your comments and thanks for sharing your tools.
> >>>>>
> >>>>> I guess the cause might be:
> >>>>> in case a region is very big like 10GiB, we have only 1MiB hot pages
> >>>>> in this large region.
> >>>>> damon will randomly pick one page to sample, but the page has only
> >>>>> 1MiB/10GiB, thus
> >>>>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> >>>>> 10000 sample periods
> >>>>> to hit the hot 1MiB in order to split this large region?
> >>>>>
> >>>>> @SeongJae, please correct me if I am wrong.
> >>>>>
> >>>>>>
> >>>>>> I'm not sure whether sets init_regions can deal with the above problem,
> >>>>>> or dynamic choose one or limited number VMA to monitor.
> >>>>>>
> >>>>>
> >>>>> I won't set a limited number of VMA as this will make the damon too hard to use
> >>>>> as nobody wants to make such complex operations, especially an Android
> >>>>> app might have more than 8000 VMAs.
> >>>>>
> >>>>> I agree init_regions might be the right place to enhance the situation.
> >>>>>
> >>>>>> I'm not sure, just share my idea.
> >>>>>>
> >>>>>> [1] https://github.com/aliyun/data-profile-tools.git
> >>>>>
> >>>>> I suppose this tool is based on damon? How do you finally resolve the problem
> >>>>> that large anon VMAs can't be splitted?
> >>>>> Anyway, I will give your tool a try.
> >>>>
> >>>> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
> >>>> though autogen.sh
> >>>> runs successfully.
> >>>>
> >>>> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
> >>>> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> >>>> reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> >>>> reference to `wgetch'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
> >>>> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> >>>> reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> >>>> reference to `subwin'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
> >>>> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
> >>>> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
> >>>> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
> >>>> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
> >>>> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
> >>>> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
> >>>> reference to `wattr_off'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
> >>>> reference to `wattr_on'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
> >>>> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> >>>> reference to `wattr_off'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
> >>>> reference to `mvwprintw'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> >>>> reference to `wattr_off'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
> >>>> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> >>>> reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> >>>> reference to `wclear'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
> >>>> reference to `wrefresh'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
> >>>> reference to `endwin'
> >>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
> >>>> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
> >>>> reference to `initscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> >>>> reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> >>>> reference to `wrefresh'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
> >>>> reference to `use_default_colors'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
> >>>> reference to `start_color'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
> >>>> reference to `keypad'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
> >>>> reference to `nonl'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
> >>>> reference to `cbreak'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
> >>>> reference to `noecho'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
> >>>> reference to `curs_set'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> >>>> reference to `stdscr'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> >>>> reference to `mvwprintw'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
> >>>> reference to `mvwprintw'
> >>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
> >>>> reference to `wrefresh'
> >>>> collect2: error: ld returned 1 exit status
> >>>> make[1]: *** [Makefile:592: datop] Error 1
> >>>> make[1]: Leaving directory '/root/data-profile-tools'
> >>>> make: *** [Makefile:438: all] Error 2
> >>> Hi, Barry
> >>>
> >>> Now, the question made me realize that the compatibility of this tool is
> >>> very poor. I built a ubuntu environment at yesterday, and fixed above
> >>> errors by:
> >>>
> >>> diff --git a/configure.ac b/configure.ac
> >>> index 7922f27..1ed823c 100644
> >>> --- a/configure.ac
> >>> +++ b/configure.ac
> >>> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
> >>> AC_CHECK_LIB([numa], [numa_free])
> >>> AC_CHECK_LIB([pthread], [pthread_create])
> >>>
> >>> -PKG_CHECK_MODULES([CHECK], [check])
> >>> -
> >>> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
> >>> $ncurses_LIBS"], [
> >>> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
> >>> - AC_MSG_ERROR([ncurses is required but was not found])
> >>> - ], [])
> >>> -])
> >>> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
> >>> + AC_MSG_ERROR([required library libncurses or ncurses not found])
> >>> + ])
> >>>
> >>
> >> I can confirm the patch fixed the issue I reported yesterday, thanks!
> >>
> >>> It works. But I found an another thing will hinder you using this tool.
> >>> We had developed other patches about DAMON base on upstream. This tool
> >>> only works well in ourselves kernel(anolis kernel, already open source).
> >>> Of course, I think it's unnecessary for you to change kernel, just let
> >>> you know this tool still has this problem.
> >>>
> >>
> >> Although I can't use this tool directly as I am not a NUMA right now,
> >> ~/data-profile-tools # ./datop --help
> >> Not support NUMA fault stat (DAMON)!
> >>
> >
> > I wonder if you can extend it to non-numa by setting "remote" to 0%
> > and local to "100%" always for non-numa machines rather than death.
> Hi Barry
>
> That's a great suggestion. Actually, I have removed 'numa_stat' check in
> datop. Maybe you can found. It does not enable numa stat when
> 'numa_stat' sysfs not found in the current system.
yep. i am able to run it on a non-numa machine, but datop immediately crashes
due to some memory corruption issues:
Monitoring 270 processes (interval: 5.5s)
PID PROC TYPE START END
SIZE(KiB) ACCESS AGE
1693 Binder:1693 ---- 0 0
0 0 0
428 ueventd ---- 0 0
0 0 0
28654 adbd ---- 0 0
0 0 0
971 usb@1.2-ser ---- 0 0
0 0 0
619 logd ---- 0 0
0 0 0
4311 a...
<- Hotkey for sorting: 1(PID), 2(START), 3(SIZE), 4(ACCESS), 5(RMA) ->
CPU% = system CPU utilization
Q: Quit; H: Home; B: Back; R: Refresh; D: DAMON
double free or corruption (!prev)
Aborted
if i move to monitor only one process, datop doesn't crash but it
doesn't show any
data either:
# pgrep youtube
4311
# ./datop -p 4311
Monitoring 1 processes (interval: 5.0s)
PID PROC TYPE START END
SIZE(KiB) *ACCESS AGE
4311 youtube ---- 0 0 0
0 0
>
> What's more, a new hot key 'f' will be introduced which can enable some
> features dynamically, such as numa stat. Others features can be used
> only in our internal version, likes 'f' in top, and will be open source
> when stable.
>
> > as your tools can map regions to .so, which seems to be quite useful.
> enen, I'm agree with you. But you know, one region maybe covers one or
> more VMAs, hard to map access count of regions to the related .so or
> anon. A lazy way used by me now. I still think it's valuable in the future.
>
it seems really an interesting topic worth more investigation. I wonder if
damon vaddr monitor should actually take vmas, or at least the types of
vmas into consideration while splitting.
Different vma types should be inherently different in hotness. for example,
if 1mb text and 1mb data are put in the same region, the monitored data
to reflect the hotness for the whole 2mb seems to be pointless at all.
Hi SeongJae,
what do you think about it?
> Anyway, any idea are welcome.
>
> Thanks,
> -wrw
>
> >
> >> I am still quite interested in your design and the purpose of this project.
> >> Unfortunately the project seems to be lacking some design doc.
> >>
> >> And would you like to send patches to lkml regarding what you
> >> have changed atop DAMON?
> >>
> >>> Anyway, the question that you reported was valuable, made me realize
> >>> what we need to improve next.
> >>>
> >>> Thanks,
> >>> Rongwei Wang
> >>>>
> >>>>>
> >>>>>>>
> >>>>>>> Typical characteristics of a large Android app is that it has
> >>>>>>> thousands of vma and very large virtual address spaces:
> >>>>>>> ~/damo # pmap 2550 | wc -l
> >>>>>>> 8522
> >>>>>>>
> >>>>>>> ~/damo # pmap 2550
> >>>>>>> ...
> >>>>>>> 0000007992bbe000 4K r---- [ anon ]
> >>>>>>> 0000007992bbf000 24K rw--- [ anon ]
> >>>>>>> 0000007fe8753000 4K ----- [ anon ]
> >>>>>>> 0000007fe8754000 8188K rw--- [ stack ]
> >>>>>>> total 36742112K
> >>>>>>>
> >>>>>>> Because the whole vma list is too long, I have put the list here for
> >>>>>>> you to download:
> >>>>>>> wget http://www.linuxep.com/patches/android-app-vmas
> >>>>>>>
> >>>>>>> I can reproduce this problem on other Apps like youtube as well.
> >>>>>>> I suppose we need to boost the algorithm of splitting regions for this
> >>>>>>> kind of application.
> >>>>>>> Any thoughts?
> >>>>>>>
> >>>>>
> >
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-05-17 11:14 ` Barry Song
@ 2022-05-18 3:03 ` Rongwei Wang
2022-05-18 9:51 ` Barry Song
0 siblings, 1 reply; 23+ messages in thread
From: Rongwei Wang @ 2022-05-18 3:03 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao, vxc5208
On 5/17/22 7:14 PM, Barry Song wrote:
> On Tue, May 17, 2022 at 3:00 AM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
>>
>>
>>
>> On 5/16/22 3:03 PM, Barry Song wrote:
>>> On Thu, Apr 28, 2022 at 7:37 PM Barry Song <21cnbao@gmail.com> wrote:
>>>>
>>>> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
>>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>> On 4/27/22 5:22 PM, Barry Song wrote:
>>>>>> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>>>>>>>
>>>>>>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
>>>>>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On 4/27/22 7:19 AM, Barry Song wrote:
>>>>>>>>> Hi SeongJae & Andrew,
>>>>>>>>> (also Cc-ed main damon developers)
>>>>>>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>>>>>>>> that vaddr regions don't split well on large Android Apps though
>>>>>>>>> everything works well on native Apps.
>>>>>>>>>
>>>>>>>>> I have tried the below two cases on an Android phone with 12GB memory
>>>>>>>>> and snapdragon 888 CPU.
>>>>>>>>> 1. a native program with small memory working set as below,
>>>>>>>>> #define size (1024*1024*100)
>>>>>>>>> main()
>>>>>>>>> {
>>>>>>>>> volatile int *p = malloc(size);
>>>>>>>>> memset(p, 0x55, size);
>>>>>>>>>
>>>>>>>>> while(1) {
>>>>>>>>> int i;
>>>>>>>>> for (i = 0; i < size / 4; i++)
>>>>>>>>> (void)*(p + i);
>>>>>>>>> usleep(1000);
>>>>>>>>>
>>>>>>>>> for (i = 0; i < size / 16; i++)
>>>>>>>>> (void)*(p + i);
>>>>>>>>> usleep(1000);
>>>>>>>>>
>>>>>>>>> }
>>>>>>>>> }
>>>>>>>>> For this application, the Damon vaddr monitor works very well.
>>>>>>>>> I have modified monitor.py in the damo userspace tool a little bit to
>>>>>>>>> show the raw data getting from the kernel.
>>>>>>>>> Regions can split decently on this kind of applications, a typical raw
>>>>>>>>> data is as below,
>>>>>>>>>
>>>>>>>>> monitoring_start: 2.224 s
>>>>>>>>> monitoring_end: 2.329 s
>>>>>>>>> monitoring_duration: 104.336 ms
>>>>>>>>> target_id: 0
>>>>>>>>> nr_regions: 24
>>>>>>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>>>>>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>>>>>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>>>>>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>>>>>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>>>>>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>>>>>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>>>>>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>>>>>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>>>>>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>>>>>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>>>>>>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>>>>>>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>>>>>>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>>>>>>>> 007663494000-007663753000( 2.746 MiB): 1
>>>>>>>>> 007663753000-007664251000( 10.992 MiB): 3
>>>>>>>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>>>>>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>>>>>>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>>>>>>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>>>>>>>> 007667f97000-007668112000( 1.480 MiB): 1
>>>>>>>>> 007668112000-00766820f000(1012.000 KiB): 0
>>>>>>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>>>>>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>>>>>>>
>>>>>>>>> 2. a large Android app like Asphalt 9
>>>>>>>>> For this case, basically regions can't split very well, but monitor
>>>>>>>>> works on small vma:
>>>>>>>>>
>>>>>>>>> monitoring_start: 2.220 s
>>>>>>>>> monitoring_end: 2.318 s
>>>>>>>>> monitoring_duration: 98.576 ms
>>>>>>>>> target_id: 0
>>>>>>>>> nr_regions: 15
>>>>>>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>>>>>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>>>>>>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>>>>>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>>>>>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>>>>>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>>>>>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>>>>>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>>>>>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>>>>>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>>>>>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>>>>>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>>>>>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>>>>>>>
>>>>>>>>> As you can see, we have some regions which are very very big and they
>>>>>>>>> are losing the chance to be splitted. But
>>>>>>>>> Damon can still monitor memory access for those small VMA areas very well like:
>>>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>>>> Hi, Barry
>>>>>>>>
>>>>>>>> Actually, we also had found the same problem in redis by ourselves
>>>>>>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>>>>>>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>>>>>>>> hot areas to been monitored or found by DAMON, likes one or more address
>>>>>>>> choose by DAMON not been accessed during sample period.
>>>>>>>
>>>>>>> Hi Rongwei,
>>>>>>> Thanks for your comments and thanks for sharing your tools.
>>>>>>>
>>>>>>> I guess the cause might be:
>>>>>>> in case a region is very big like 10GiB, we have only 1MiB hot pages
>>>>>>> in this large region.
>>>>>>> damon will randomly pick one page to sample, but the page has only
>>>>>>> 1MiB/10GiB, thus
>>>>>>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
>>>>>>> 10000 sample periods
>>>>>>> to hit the hot 1MiB in order to split this large region?
>>>>>>>
>>>>>>> @SeongJae, please correct me if I am wrong.
>>>>>>>
>>>>>>>>
>>>>>>>> I'm not sure whether sets init_regions can deal with the above problem,
>>>>>>>> or dynamic choose one or limited number VMA to monitor.
>>>>>>>>
>>>>>>>
>>>>>>> I won't set a limited number of VMA as this will make the damon too hard to use
>>>>>>> as nobody wants to make such complex operations, especially an Android
>>>>>>> app might have more than 8000 VMAs.
>>>>>>>
>>>>>>> I agree init_regions might be the right place to enhance the situation.
>>>>>>>
>>>>>>>> I'm not sure, just share my idea.
>>>>>>>>
>>>>>>>> [1] https://github.com/aliyun/data-profile-tools.git
>>>>>>>
>>>>>>> I suppose this tool is based on damon? How do you finally resolve the problem
>>>>>>> that large anon VMAs can't be splitted?
>>>>>>> Anyway, I will give your tool a try.
>>>>>>
>>>>>> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
>>>>>> though autogen.sh
>>>>>> runs successfully.
>>>>>>
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
>>>>>> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>>>>> reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>>>>> reference to `wgetch'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
>>>>>> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>>>>> reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>>>>> reference to `subwin'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
>>>>>> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
>>>>>> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
>>>>>> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
>>>>>> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
>>>>>> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
>>>>>> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
>>>>>> reference to `wattr_off'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
>>>>>> reference to `wattr_on'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
>>>>>> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>>>>> reference to `wattr_off'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
>>>>>> reference to `mvwprintw'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>>>>> reference to `wattr_off'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
>>>>>> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>>>>> reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>>>>> reference to `wclear'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
>>>>>> reference to `wrefresh'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
>>>>>> reference to `endwin'
>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
>>>>>> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
>>>>>> reference to `initscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>>>>> reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>>>>> reference to `wrefresh'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
>>>>>> reference to `use_default_colors'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
>>>>>> reference to `start_color'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
>>>>>> reference to `keypad'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
>>>>>> reference to `nonl'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
>>>>>> reference to `cbreak'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
>>>>>> reference to `noecho'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
>>>>>> reference to `curs_set'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>>>>> reference to `stdscr'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>>>>> reference to `mvwprintw'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
>>>>>> reference to `mvwprintw'
>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
>>>>>> reference to `wrefresh'
>>>>>> collect2: error: ld returned 1 exit status
>>>>>> make[1]: *** [Makefile:592: datop] Error 1
>>>>>> make[1]: Leaving directory '/root/data-profile-tools'
>>>>>> make: *** [Makefile:438: all] Error 2
>>>>> Hi, Barry
>>>>>
>>>>> Now, the question made me realize that the compatibility of this tool is
>>>>> very poor. I built a ubuntu environment at yesterday, and fixed above
>>>>> errors by:
>>>>>
>>>>> diff --git a/configure.ac b/configure.ac
>>>>> index 7922f27..1ed823c 100644
>>>>> --- a/configure.ac
>>>>> +++ b/configure.ac
>>>>> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
>>>>> AC_CHECK_LIB([numa], [numa_free])
>>>>> AC_CHECK_LIB([pthread], [pthread_create])
>>>>>
>>>>> -PKG_CHECK_MODULES([CHECK], [check])
>>>>> -
>>>>> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
>>>>> $ncurses_LIBS"], [
>>>>> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
>>>>> - AC_MSG_ERROR([ncurses is required but was not found])
>>>>> - ], [])
>>>>> -])
>>>>> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
>>>>> + AC_MSG_ERROR([required library libncurses or ncurses not found])
>>>>> + ])
>>>>>
>>>>
>>>> I can confirm the patch fixed the issue I reported yesterday, thanks!
>>>>
>>>>> It works. But I found an another thing will hinder you using this tool.
>>>>> We had developed other patches about DAMON base on upstream. This tool
>>>>> only works well in ourselves kernel(anolis kernel, already open source).
>>>>> Of course, I think it's unnecessary for you to change kernel, just let
>>>>> you know this tool still has this problem.
>>>>>
>>>>
>>>> Although I can't use this tool directly as I am not a NUMA right now,
>>>> ~/data-profile-tools # ./datop --help
>>>> Not support NUMA fault stat (DAMON)!
>>>>
>>>
>>> I wonder if you can extend it to non-numa by setting "remote" to 0%
>>> and local to "100%" always for non-numa machines rather than death.
>> Hi Barry
>>
>> That's a great suggestion. Actually, I have removed 'numa_stat' check in
>> datop. Maybe you can found. It does not enable numa stat when
>> 'numa_stat' sysfs not found in the current system.
>
> yep. i am able to run it on a non-numa machine, but datop immediately crashes
> due to some memory corruption issues:
>
> Monitoring 270 processes (interval: 5.5s)
Barry, it's known bug. I remember the maximum number of processes that
is 32 in datop. The reason that setting like this is that I feel
impossible to monitor so many processes at the beginning.
And it seems that the error message should been printed here, instead of
crash. Thank you for reminding me.
>
> PID PROC TYPE START END
> SIZE(KiB) ACCESS AGE
> 1693 Binder:1693 ---- 0 0
> 0 0 0
> 428 ueventd ---- 0 0
> 0 0 0
> 28654 adbd ---- 0 0
> 0 0 0
> 971 usb@1.2-ser ---- 0 0
> 0 0 0
> 619 logd ---- 0 0
> 0 0 0
> 4311 a...
>
> <- Hotkey for sorting: 1(PID), 2(START), 3(SIZE), 4(ACCESS), 5(RMA) ->
> CPU% = system CPU utilization
>
> Q: Quit; H: Home; B: Back; R: Refresh; D: DAMON
> double free or corruption (!prev)
>
> Aborted
>
> if i move to monitor only one process, datop doesn't crash but it
> doesn't show any
> data either:
>
> # pgrep youtube
> 4311
> # ./datop -p 4311
>
> Monitoring 1 processes (interval: 5.0s)
Oh, it's ever happen to me. Does It always show like this when
monitoring one process in your environment?
>
> PID PROC TYPE START END
> SIZE(KiB) *ACCESS AGE
> 4311 youtube ---- 0 0 0
> 0 0
>
>
>>
>> What's more, a new hot key 'f' will be introduced which can enable some
>> features dynamically, such as numa stat. Others features can be used
>> only in our internal version, likes 'f' in top, and will be open source
>> when stable.
>>
>>> as your tools can map regions to .so, which seems to be quite useful.
>> enen, I'm agree with you. But you know, one region maybe covers one or
>> more VMAs, hard to map access count of regions to the related .so or
>> anon. A lazy way used by me now. I still think it's valuable in the future.
>>
>
> it seems really an interesting topic worth more investigation. I wonder if
> damon vaddr monitor should actually take vmas, or at least the types of
> vmas into consideration while splitting.
>
> Different vma types should be inherently different in hotness. for example,
> if 1mb text and 1mb data are put in the same region, the monitored data
> to reflect the hotness for the whole 2mb seems to be pointless at all.
>
> Hi SeongJae,
> what do you think about it?
>
>> Anyway, any idea are welcome.
>>
>> Thanks,
>> -wrw
>>
>>>
>>>> I am still quite interested in your design and the purpose of this project.
>>>> Unfortunately the project seems to be lacking some design doc.
>>>>
>>>> And would you like to send patches to lkml regarding what you
>>>> have changed atop DAMON?
>>>>
>>>>> Anyway, the question that you reported was valuable, made me realize
>>>>> what we need to improve next.
>>>>>
>>>>> Thanks,
>>>>> Rongwei Wang
>>>>>>
>>>>>>>
>>>>>>>>>
>>>>>>>>> Typical characteristics of a large Android app is that it has
>>>>>>>>> thousands of vma and very large virtual address spaces:
>>>>>>>>> ~/damo # pmap 2550 | wc -l
>>>>>>>>> 8522
>>>>>>>>>
>>>>>>>>> ~/damo # pmap 2550
>>>>>>>>> ...
>>>>>>>>> 0000007992bbe000 4K r---- [ anon ]
>>>>>>>>> 0000007992bbf000 24K rw--- [ anon ]
>>>>>>>>> 0000007fe8753000 4K ----- [ anon ]
>>>>>>>>> 0000007fe8754000 8188K rw--- [ stack ]
>>>>>>>>> total 36742112K
>>>>>>>>>
>>>>>>>>> Because the whole vma list is too long, I have put the list here for
>>>>>>>>> you to download:
>>>>>>>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>>>>>>>
>>>>>>>>> I can reproduce this problem on other Apps like youtube as well.
>>>>>>>>> I suppose we need to boost the algorithm of splitting regions for this
>>>>>>>>> kind of application.
>>>>>>>>> Any thoughts?
>>>>>>>>>
>>>>>>>
>>>
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-05-18 3:03 ` Rongwei Wang
@ 2022-05-18 9:51 ` Barry Song
2022-05-19 3:09 ` Rongwei Wang
0 siblings, 1 reply; 23+ messages in thread
From: Barry Song @ 2022-05-18 9:51 UTC (permalink / raw)
To: Rongwei Wang
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao, vxc5208
On Wed, May 18, 2022 at 3:03 PM Rongwei Wang
<rongwei.wang@linux.alibaba.com> wrote:
>
>
>
> On 5/17/22 7:14 PM, Barry Song wrote:
> > On Tue, May 17, 2022 at 3:00 AM Rongwei Wang
> > <rongwei.wang@linux.alibaba.com> wrote:
> >>
> >>
> >>
> >> On 5/16/22 3:03 PM, Barry Song wrote:
> >>> On Thu, Apr 28, 2022 at 7:37 PM Barry Song <21cnbao@gmail.com> wrote:
> >>>>
> >>>> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
> >>>> <rongwei.wang@linux.alibaba.com> wrote:
> >>>>>
> >>>>>
> >>>>>
> >>>>> On 4/27/22 5:22 PM, Barry Song wrote:
> >>>>>> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
> >>>>>>>
> >>>>>>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> >>>>>>> <rongwei.wang@linux.alibaba.com> wrote:
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> On 4/27/22 7:19 AM, Barry Song wrote:
> >>>>>>>>> Hi SeongJae & Andrew,
> >>>>>>>>> (also Cc-ed main damon developers)
> >>>>>>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
> >>>>>>>>> that vaddr regions don't split well on large Android Apps though
> >>>>>>>>> everything works well on native Apps.
> >>>>>>>>>
> >>>>>>>>> I have tried the below two cases on an Android phone with 12GB memory
> >>>>>>>>> and snapdragon 888 CPU.
> >>>>>>>>> 1. a native program with small memory working set as below,
> >>>>>>>>> #define size (1024*1024*100)
> >>>>>>>>> main()
> >>>>>>>>> {
> >>>>>>>>> volatile int *p = malloc(size);
> >>>>>>>>> memset(p, 0x55, size);
> >>>>>>>>>
> >>>>>>>>> while(1) {
> >>>>>>>>> int i;
> >>>>>>>>> for (i = 0; i < size / 4; i++)
> >>>>>>>>> (void)*(p + i);
> >>>>>>>>> usleep(1000);
> >>>>>>>>>
> >>>>>>>>> for (i = 0; i < size / 16; i++)
> >>>>>>>>> (void)*(p + i);
> >>>>>>>>> usleep(1000);
> >>>>>>>>>
> >>>>>>>>> }
> >>>>>>>>> }
> >>>>>>>>> For this application, the Damon vaddr monitor works very well.
> >>>>>>>>> I have modified monitor.py in the damo userspace tool a little bit to
> >>>>>>>>> show the raw data getting from the kernel.
> >>>>>>>>> Regions can split decently on this kind of applications, a typical raw
> >>>>>>>>> data is as below,
> >>>>>>>>>
> >>>>>>>>> monitoring_start: 2.224 s
> >>>>>>>>> monitoring_end: 2.329 s
> >>>>>>>>> monitoring_duration: 104.336 ms
> >>>>>>>>> target_id: 0
> >>>>>>>>> nr_regions: 24
> >>>>>>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
> >>>>>>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
> >>>>>>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
> >>>>>>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
> >>>>>>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
> >>>>>>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
> >>>>>>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
> >>>>>>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
> >>>>>>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
> >>>>>>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
> >>>>>>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
> >>>>>>>>> 0076623e4000-007662837000( 4.324 MiB): 2
> >>>>>>>>> 007662837000-0076630f1000( 8.727 MiB): 3
> >>>>>>>>> 0076630f1000-007663494000( 3.637 MiB): 2
> >>>>>>>>> 007663494000-007663753000( 2.746 MiB): 1
> >>>>>>>>> 007663753000-007664251000( 10.992 MiB): 3
> >>>>>>>>> 007664251000-0076666fd000( 36.672 MiB): 2
> >>>>>>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
> >>>>>>>>> 007666e73000-007667c89000( 14.086 MiB): 2
> >>>>>>>>> 007667c89000-007667f97000( 3.055 MiB): 0
> >>>>>>>>> 007667f97000-007668112000( 1.480 MiB): 1
> >>>>>>>>> 007668112000-00766820f000(1012.000 KiB): 0
> >>>>>>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
> >>>>>>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
> >>>>>>>>>
> >>>>>>>>> 2. a large Android app like Asphalt 9
> >>>>>>>>> For this case, basically regions can't split very well, but monitor
> >>>>>>>>> works on small vma:
> >>>>>>>>>
> >>>>>>>>> monitoring_start: 2.220 s
> >>>>>>>>> monitoring_end: 2.318 s
> >>>>>>>>> monitoring_duration: 98.576 ms
> >>>>>>>>> target_id: 0
> >>>>>>>>> nr_regions: 15
> >>>>>>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
> >>>>>>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
> >>>>>>>>> 000371b6c000-000400000000( 2.223 GiB): 0
> >>>>>>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
> >>>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> >>>>>>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
> >>>>>>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
> >>>>>>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
> >>>>>>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
> >>>>>>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
> >>>>>>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
> >>>>>>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
> >>>>>>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
> >>>>>>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
> >>>>>>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
> >>>>>>>>>
> >>>>>>>>> As you can see, we have some regions which are very very big and they
> >>>>>>>>> are losing the chance to be splitted. But
> >>>>>>>>> Damon can still monitor memory access for those small VMA areas very well like:
> >>>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
> >>>>>>>> Hi, Barry
> >>>>>>>>
> >>>>>>>> Actually, we also had found the same problem in redis by ourselves
> >>>>>>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
> >>>>>>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
> >>>>>>>> hot areas to been monitored or found by DAMON, likes one or more address
> >>>>>>>> choose by DAMON not been accessed during sample period.
> >>>>>>>
> >>>>>>> Hi Rongwei,
> >>>>>>> Thanks for your comments and thanks for sharing your tools.
> >>>>>>>
> >>>>>>> I guess the cause might be:
> >>>>>>> in case a region is very big like 10GiB, we have only 1MiB hot pages
> >>>>>>> in this large region.
> >>>>>>> damon will randomly pick one page to sample, but the page has only
> >>>>>>> 1MiB/10GiB, thus
> >>>>>>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
> >>>>>>> 10000 sample periods
> >>>>>>> to hit the hot 1MiB in order to split this large region?
> >>>>>>>
> >>>>>>> @SeongJae, please correct me if I am wrong.
> >>>>>>>
> >>>>>>>>
> >>>>>>>> I'm not sure whether sets init_regions can deal with the above problem,
> >>>>>>>> or dynamic choose one or limited number VMA to monitor.
> >>>>>>>>
> >>>>>>>
> >>>>>>> I won't set a limited number of VMA as this will make the damon too hard to use
> >>>>>>> as nobody wants to make such complex operations, especially an Android
> >>>>>>> app might have more than 8000 VMAs.
> >>>>>>>
> >>>>>>> I agree init_regions might be the right place to enhance the situation.
> >>>>>>>
> >>>>>>>> I'm not sure, just share my idea.
> >>>>>>>>
> >>>>>>>> [1] https://github.com/aliyun/data-profile-tools.git
> >>>>>>>
> >>>>>>> I suppose this tool is based on damon? How do you finally resolve the problem
> >>>>>>> that large anon VMAs can't be splitted?
> >>>>>>> Anyway, I will give your tool a try.
> >>>>>>
> >>>>>> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
> >>>>>> though autogen.sh
> >>>>>> runs successfully.
> >>>>>>
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
> >>>>>> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> >>>>>> reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
> >>>>>> reference to `wgetch'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
> >>>>>> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> >>>>>> reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
> >>>>>> reference to `subwin'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
> >>>>>> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
> >>>>>> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
> >>>>>> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
> >>>>>> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
> >>>>>> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
> >>>>>> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
> >>>>>> reference to `wattr_off'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
> >>>>>> reference to `wattr_on'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
> >>>>>> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> >>>>>> reference to `wattr_off'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
> >>>>>> reference to `mvwprintw'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
> >>>>>> reference to `wattr_off'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
> >>>>>> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> >>>>>> reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
> >>>>>> reference to `wclear'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
> >>>>>> reference to `wrefresh'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
> >>>>>> reference to `endwin'
> >>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
> >>>>>> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
> >>>>>> reference to `initscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> >>>>>> reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
> >>>>>> reference to `wrefresh'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
> >>>>>> reference to `use_default_colors'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
> >>>>>> reference to `start_color'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
> >>>>>> reference to `keypad'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
> >>>>>> reference to `nonl'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
> >>>>>> reference to `cbreak'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
> >>>>>> reference to `noecho'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
> >>>>>> reference to `curs_set'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> >>>>>> reference to `stdscr'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
> >>>>>> reference to `mvwprintw'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
> >>>>>> reference to `mvwprintw'
> >>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
> >>>>>> reference to `wrefresh'
> >>>>>> collect2: error: ld returned 1 exit status
> >>>>>> make[1]: *** [Makefile:592: datop] Error 1
> >>>>>> make[1]: Leaving directory '/root/data-profile-tools'
> >>>>>> make: *** [Makefile:438: all] Error 2
> >>>>> Hi, Barry
> >>>>>
> >>>>> Now, the question made me realize that the compatibility of this tool is
> >>>>> very poor. I built a ubuntu environment at yesterday, and fixed above
> >>>>> errors by:
> >>>>>
> >>>>> diff --git a/configure.ac b/configure.ac
> >>>>> index 7922f27..1ed823c 100644
> >>>>> --- a/configure.ac
> >>>>> +++ b/configure.ac
> >>>>> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
> >>>>> AC_CHECK_LIB([numa], [numa_free])
> >>>>> AC_CHECK_LIB([pthread], [pthread_create])
> >>>>>
> >>>>> -PKG_CHECK_MODULES([CHECK], [check])
> >>>>> -
> >>>>> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
> >>>>> $ncurses_LIBS"], [
> >>>>> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
> >>>>> - AC_MSG_ERROR([ncurses is required but was not found])
> >>>>> - ], [])
> >>>>> -])
> >>>>> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
> >>>>> + AC_MSG_ERROR([required library libncurses or ncurses not found])
> >>>>> + ])
> >>>>>
> >>>>
> >>>> I can confirm the patch fixed the issue I reported yesterday, thanks!
> >>>>
> >>>>> It works. But I found an another thing will hinder you using this tool.
> >>>>> We had developed other patches about DAMON base on upstream. This tool
> >>>>> only works well in ourselves kernel(anolis kernel, already open source).
> >>>>> Of course, I think it's unnecessary for you to change kernel, just let
> >>>>> you know this tool still has this problem.
> >>>>>
> >>>>
> >>>> Although I can't use this tool directly as I am not a NUMA right now,
> >>>> ~/data-profile-tools # ./datop --help
> >>>> Not support NUMA fault stat (DAMON)!
> >>>>
> >>>
> >>> I wonder if you can extend it to non-numa by setting "remote" to 0%
> >>> and local to "100%" always for non-numa machines rather than death.
> >> Hi Barry
> >>
> >> That's a great suggestion. Actually, I have removed 'numa_stat' check in
> >> datop. Maybe you can found. It does not enable numa stat when
> >> 'numa_stat' sysfs not found in the current system.
> >
> > yep. i am able to run it on a non-numa machine, but datop immediately crashes
> > due to some memory corruption issues:
> >
> > Monitoring 270 processes (interval: 5.5s)
> Barry, it's known bug. I remember the maximum number of processes that
> is 32 in datop. The reason that setting like this is that I feel
> impossible to monitor so many processes at the beginning.
>
> And it seems that the error message should been printed here, instead of
> crash. Thank you for reminding me.
> >
> > PID PROC TYPE START END
> > SIZE(KiB) ACCESS AGE
> > 1693 Binder:1693 ---- 0 0
> > 0 0 0
> > 428 ueventd ---- 0 0
> > 0 0 0
> > 28654 adbd ---- 0 0
> > 0 0 0
> > 971 usb@1.2-ser ---- 0 0
> > 0 0 0
> > 619 logd ---- 0 0
> > 0 0 0
> > 4311 a...
> >
> > <- Hotkey for sorting: 1(PID), 2(START), 3(SIZE), 4(ACCESS), 5(RMA) ->
> > CPU% = system CPU utilization
> >
> > Q: Quit; H: Home; B: Back; R: Refresh; D: DAMON
> > double free or corruption (!prev)
> >
> > Aborted
> >
> > if i move to monitor only one process, datop doesn't crash but it
> > doesn't show any
> > data either:
> >
> > # pgrep youtube
> > 4311
> > # ./datop -p 4311
> >
> > Monitoring 1 processes (interval: 5.0s)
> Oh, it's ever happen to me. Does It always show like this when
> monitoring one process in your environment?
right . I have never succeeded in using your datop.
~/damo # pgrep youtube
21287
~/damo # ./damo monitor --report_type=wss --count=5 21287
# <percentile> <wss>
# target_id 0
# avr: 28.285 MiB
0 0 B |
|
25 0 B |
|
50 0 B |
|
75 18.547 MiB |******
|
100 174.414 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 49.857 MiB
0 0 B |
|
25 0 B |
|
50 0 B |
|
75 124.180 MiB |****************************
|
100 256.375 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 35.999 MiB
0 0 B |
|
25 0 B |
|
50 0 B |
|
75 59.605 MiB |****************
|
100 218.191 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 35.638 MiB
0 0 B |
|
25 0 B |
|
50 0 B |
|
75 27.668 MiB |*******
|
100 230.922 MiB
|***********************************************************|
# <percentile> <wss>
# target_id 0
# avr: 60.585 MiB
0 0 B |
|
25 0 B |
|
50 21.297 MiB |****
|
75 122.703 MiB |**************************
|
100 272.973 MiB
|***********************************************************|
datop:
~/data-profile-tools # ./datop -p 21287
Monitoring 1 processes (interval: 5.1s)
PID PROC TYPE START END
SIZE(KiB) *ACCESS AGE
21287 android.you ---- 0 0
0 0 0
nothing is shown here.
Is it because your tool doesn't turn on related tracers automatically?
i have dumped
the values while running datop:
~/data-profile-tools # ./datop -p 21287
Start monitoring 21287 ...
DamonTOP is starting ...
[1]+ Stopped ./datop -p 21287
~/data-profile-tools # cat /sys/kernel/debug/damon/target_ids
21287
~/data-profile-tools # cat /sys/kernel/debug/damon/monitor_on
on
~/data-profile-tools # cat
/sys/kernel/debug/tracing/events/damon/damon_aggregated/enable
0
~/data-profile-tools # cat /sys/kernel/debug/tracing/tracing_on
0
so i enable them manually:
~/data-profile-tools # echo 1 >
/sys/kernel/debug/tracing/events/damon/damon_aggregated/enable
~/data-profile-tools # echo 1 > /sys/kernel/debug/tracing/tracing_on
but datop still shows nothing while kernel begins to report data correctly:
~/data-profile-tools # cat /sys/kernel/debug/tracing/trace | more
# tracer: nop
#
# WARNING: FUNCTION TRACING IS CORRUPTED
# MAY BE MISSING FUNCTION EVENTS
# entries-in-buffer/entries-written: 11599/11599 #P:8
#
# _-----=> irqs-off
# / _----=> need-resched
# | / _---=> hardirq/softirq
# || / _--=> preempt-depth
# ||| / delay
# TASK-PID CPU# |||| TIMESTAMP FUNCTION
# | | | |||| | |
kdamond.0-26895 [002] .... 160409.911650: damon_aggregated:
target_id=0 nr_regions=12 314572800-3954012160: 0 185
kdamond.0-26895 [002] .... 160409.911670: damon_aggregated:
target_id=0 nr_regions=12 394992549888-394992590848: 0 12
kdamond.0-26895 [002] .... 160409.911675: damon_aggregated:
target_id=0 nr_regions=12 488092233728-494587068416: 0 311
kdamond.0-26895 [002] .... 160409.911679: damon_aggregated:
target_id=0 nr_regions=12 494587068416-501072220160: 0 375
kdamond.0-26895 [002] .... 160409.911683: damon_aggregated:
target_id=0 nr_regions=12 501072220160-507561463808: 0 415
kdamond.0-26895 [002] .... 160409.911687: damon_aggregated:
target_id=0 nr_regions=12 507561463808-514046730240: 0 2091
kdamond.0-26895 [002] .... 160409.911691: damon_aggregated:
target_id=0 nr_regions=12 514046730240-520535441408: 0 2263
kdamond.0-26895 [002] .... 160409.911696: damon_aggregated:
target_id=0 nr_regions=12 520535441408-527022301184: 0 2349
kdamond.0-26895 [002] .... 160409.911700: damon_aggregated:
target_id=0 nr_regions=12 527022301184-533491294208: 0 2373
kdamond.0-26895 [002] .... 160409.911704: damon_aggregated:
target_id=0 nr_regions=12 533491294208-539965886464: 0 2378
kdamond.0-26895 [002] .... 160409.911708: damon_aggregated:
target_id=0 nr_regions=12 539965886464-546468855808: 0 2380
kdamond.0-26895 [002] .... 160409.911712: damon_aggregated:
target_id=0 nr_regions=12 546468855808-549620240384: 0 2381
kdamond.0-26895 [000] .... 160410.014673: damon_aggregated:
target_id=0 nr_regions=13 314572800-3954012160: 0 186
kdamond.0-26895 [000] .... 160410.014694: damon_aggregated:
target_id=0 nr_regions=13 394992549888-394992578560: 1 0
kdamond.0-26895 [000] .... 160410.014699: damon_aggregated:
target_id=0 nr_regions=13 394992578560-394992590848: 0 13
kdamond.0-26895 [000] .... 160410.014704: damon_aggregated:
target_id=0 nr_regions=13 488092233728-494587068416: 0 312
kdamond.0-26895 [000] .... 160410.014709: damon_aggregated:
target_id=0 nr_regions=13 494587068416-501072220160: 0 376
kdamond.0-26895 [000] .... 160410.014714: damon_aggregated:
target_id=0 nr_regions=13 501072220160-507561463808: 0 416
kdamond.0-26895 [000] .... 160410.014718: damon_aggregated:
target_id=0 nr_regions=13 507561463808-514046730240: 0 2092
kdamond.0-26895 [000] .... 160410.014723: damon_aggregated:
target_id=0 nr_regions=13 514046730240-520535441408: 0 2264
kdamond.0-26895 [000] .... 160410.014727: damon_aggregated:
target_id=0 nr_regions=13 520535441408-527022301184: 0 2350
kdamond.0-26895 [000] .... 160410.014732: damon_aggregated:
target_id=0 nr_regions=13 527022301184-533491294208: 0 2374
kdamond.0-26895 [000] .... 160410.014736: damon_aggregated:
target_id=0 nr_regions=13 533491294208-539965886464: 0 2379
kdamond.0-26895 [000] .... 160410.014740: damon_aggregated:
target_id=0 nr_regions=13 539965886464-546468855808: 0 2381
kdamond.0-26895 [000] .... 160410.014745: damon_aggregated:
target_id=0 nr_regions=13 546468855808-549620240384: 0 2382
kdamond.0-26895 [001] .... 160410.112316: damon_aggregated:
target_id=0 nr_regions=12 314572800-3954012160: 0 187
kdamond.0-26895 [001] .... 160410.112338: damon_aggregated:
target_id=0 nr_regions=12 394992549888-394992590848: 0 4
kdamond.0-26895 [001] .... 160410.112343: damon_aggregated:
target_id=0 nr_regions=12 488092233728-494587068416: 0 313
kdamond.0-26895 [001] .... 160410.112348: damon_aggregated:
target_id=0 nr_regions=12 494587068416-501072220160: 0 377
kdamond.0-26895 [001] .... 160410.112353: damon_aggregated:
target_id=0 nr_regions=12 501072220160-507561463808: 0 417
> >
> > PID PROC TYPE START END
> > SIZE(KiB) *ACCESS AGE
> > 4311 youtube ---- 0 0 0
> > 0 0
> >
> >
> >>
> >> What's more, a new hot key 'f' will be introduced which can enable some
> >> features dynamically, such as numa stat. Others features can be used
> >> only in our internal version, likes 'f' in top, and will be open source
> >> when stable.
> >>
> >>> as your tools can map regions to .so, which seems to be quite useful.
> >> enen, I'm agree with you. But you know, one region maybe covers one or
> >> more VMAs, hard to map access count of regions to the related .so or
> >> anon. A lazy way used by me now. I still think it's valuable in the future.
> >>
> >
> > it seems really an interesting topic worth more investigation. I wonder if
> > damon vaddr monitor should actually take vmas, or at least the types of
> > vmas into consideration while splitting.
> >
> > Different vma types should be inherently different in hotness. for example,
> > if 1mb text and 1mb data are put in the same region, the monitored data
> > to reflect the hotness for the whole 2mb seems to be pointless at all.
> >
> > Hi SeongJae,
> > what do you think about it?
> >
> >> Anyway, any idea are welcome.
> >>
> >> Thanks,
> >> -wrw
> >>
> >>>
> >>>> I am still quite interested in your design and the purpose of this project.
> >>>> Unfortunately the project seems to be lacking some design doc.
> >>>>
> >>>> And would you like to send patches to lkml regarding what you
> >>>> have changed atop DAMON?
> >>>>
> >>>>> Anyway, the question that you reported was valuable, made me realize
> >>>>> what we need to improve next.
> >>>>>
> >>>>> Thanks,
> >>>>> Rongwei Wang
> >>>>>>
> >>>>>>>
> >>>>>>>>>
> >>>>>>>>> Typical characteristics of a large Android app is that it has
> >>>>>>>>> thousands of vma and very large virtual address spaces:
> >>>>>>>>> ~/damo # pmap 2550 | wc -l
> >>>>>>>>> 8522
> >>>>>>>>>
> >>>>>>>>> ~/damo # pmap 2550
> >>>>>>>>> ...
> >>>>>>>>> 0000007992bbe000 4K r---- [ anon ]
> >>>>>>>>> 0000007992bbf000 24K rw--- [ anon ]
> >>>>>>>>> 0000007fe8753000 4K ----- [ anon ]
> >>>>>>>>> 0000007fe8754000 8188K rw--- [ stack ]
> >>>>>>>>> total 36742112K
> >>>>>>>>>
> >>>>>>>>> Because the whole vma list is too long, I have put the list here for
> >>>>>>>>> you to download:
> >>>>>>>>> wget http://www.linuxep.com/patches/android-app-vmas
> >>>>>>>>>
> >>>>>>>>> I can reproduce this problem on other Apps like youtube as well.
> >>>>>>>>> I suppose we need to boost the algorithm of splitting regions for this
> >>>>>>>>> kind of application.
> >>>>>>>>> Any thoughts?
> >>>>>>>>>
> >>>>>>>
> >>>
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-05-18 9:51 ` Barry Song
@ 2022-05-19 3:09 ` Rongwei Wang
0 siblings, 0 replies; 23+ messages in thread
From: Rongwei Wang @ 2022-05-19 3:09 UTC (permalink / raw)
To: Barry Song
Cc: sj, Andrew Morton, Linux-MM, LKML, Matthew Wilcox, shuah,
Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao, vxc5208
On 5/18/22 5:51 PM, Barry Song wrote:
> On Wed, May 18, 2022 at 3:03 PM Rongwei Wang
> <rongwei.wang@linux.alibaba.com> wrote:
>>
>>
>>
>> On 5/17/22 7:14 PM, Barry Song wrote:
>>> On Tue, May 17, 2022 at 3:00 AM Rongwei Wang
>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>
>>>>
>>>>
>>>> On 5/16/22 3:03 PM, Barry Song wrote:
>>>>> On Thu, Apr 28, 2022 at 7:37 PM Barry Song <21cnbao@gmail.com> wrote:
>>>>>>
>>>>>> On Thu, Apr 28, 2022 at 2:05 PM Rongwei Wang
>>>>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On 4/27/22 5:22 PM, Barry Song wrote:
>>>>>>>> On Wed, Apr 27, 2022 at 7:44 PM Barry Song <21cnbao@gmail.com> wrote:
>>>>>>>>>
>>>>>>>>> On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
>>>>>>>>> <rongwei.wang@linux.alibaba.com> wrote:
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On 4/27/22 7:19 AM, Barry Song wrote:
>>>>>>>>>>> Hi SeongJae & Andrew,
>>>>>>>>>>> (also Cc-ed main damon developers)
>>>>>>>>>>> On an Android phone, I tried to use the DAMON vaddr monitor and found
>>>>>>>>>>> that vaddr regions don't split well on large Android Apps though
>>>>>>>>>>> everything works well on native Apps.
>>>>>>>>>>>
>>>>>>>>>>> I have tried the below two cases on an Android phone with 12GB memory
>>>>>>>>>>> and snapdragon 888 CPU.
>>>>>>>>>>> 1. a native program with small memory working set as below,
>>>>>>>>>>> #define size (1024*1024*100)
>>>>>>>>>>> main()
>>>>>>>>>>> {
>>>>>>>>>>> volatile int *p = malloc(size);
>>>>>>>>>>> memset(p, 0x55, size);
>>>>>>>>>>>
>>>>>>>>>>> while(1) {
>>>>>>>>>>> int i;
>>>>>>>>>>> for (i = 0; i < size / 4; i++)
>>>>>>>>>>> (void)*(p + i);
>>>>>>>>>>> usleep(1000);
>>>>>>>>>>>
>>>>>>>>>>> for (i = 0; i < size / 16; i++)
>>>>>>>>>>> (void)*(p + i);
>>>>>>>>>>> usleep(1000);
>>>>>>>>>>>
>>>>>>>>>>> }
>>>>>>>>>>> }
>>>>>>>>>>> For this application, the Damon vaddr monitor works very well.
>>>>>>>>>>> I have modified monitor.py in the damo userspace tool a little bit to
>>>>>>>>>>> show the raw data getting from the kernel.
>>>>>>>>>>> Regions can split decently on this kind of applications, a typical raw
>>>>>>>>>>> data is as below,
>>>>>>>>>>>
>>>>>>>>>>> monitoring_start: 2.224 s
>>>>>>>>>>> monitoring_end: 2.329 s
>>>>>>>>>>> monitoring_duration: 104.336 ms
>>>>>>>>>>> target_id: 0
>>>>>>>>>>> nr_regions: 24
>>>>>>>>>>> 005fb37b2000-005fb734a000( 59.594 MiB): 0
>>>>>>>>>>> 005fb734a000-005fbaf95000( 60.293 MiB): 0
>>>>>>>>>>> 005fbaf95000-005fbec0b000( 60.461 MiB): 0
>>>>>>>>>>> 005fbec0b000-005fc2910000( 61.020 MiB): 0
>>>>>>>>>>> 005fc2910000-005fc6769000( 62.348 MiB): 0
>>>>>>>>>>> 005fc6769000-005fca33f000( 59.836 MiB): 0
>>>>>>>>>>> 005fca33f000-005fcdc8b000( 57.297 MiB): 0
>>>>>>>>>>> 005fcdc8b000-005fd115a000( 52.809 MiB): 0
>>>>>>>>>>> 005fd115a000-005fd45bd000( 52.387 MiB): 0
>>>>>>>>>>> 007661c59000-007661ee4000( 2.543 MiB): 2
>>>>>>>>>>> 007661ee4000-0076623e4000( 5.000 MiB): 3
>>>>>>>>>>> 0076623e4000-007662837000( 4.324 MiB): 2
>>>>>>>>>>> 007662837000-0076630f1000( 8.727 MiB): 3
>>>>>>>>>>> 0076630f1000-007663494000( 3.637 MiB): 2
>>>>>>>>>>> 007663494000-007663753000( 2.746 MiB): 1
>>>>>>>>>>> 007663753000-007664251000( 10.992 MiB): 3
>>>>>>>>>>> 007664251000-0076666fd000( 36.672 MiB): 2
>>>>>>>>>>> 0076666fd000-007666e73000( 7.461 MiB): 1
>>>>>>>>>>> 007666e73000-007667c89000( 14.086 MiB): 2
>>>>>>>>>>> 007667c89000-007667f97000( 3.055 MiB): 0
>>>>>>>>>>> 007667f97000-007668112000( 1.480 MiB): 1
>>>>>>>>>>> 007668112000-00766820f000(1012.000 KiB): 0
>>>>>>>>>>> 007ff27b7000-007ff27d6000( 124.000 KiB): 0
>>>>>>>>>>> 007ff27d6000-007ff27d8000( 8.000 KiB): 8
>>>>>>>>>>>
>>>>>>>>>>> 2. a large Android app like Asphalt 9
>>>>>>>>>>> For this case, basically regions can't split very well, but monitor
>>>>>>>>>>> works on small vma:
>>>>>>>>>>>
>>>>>>>>>>> monitoring_start: 2.220 s
>>>>>>>>>>> monitoring_end: 2.318 s
>>>>>>>>>>> monitoring_duration: 98.576 ms
>>>>>>>>>>> target_id: 0
>>>>>>>>>>> nr_regions: 15
>>>>>>>>>>> 000012c00000-0001c301e000( 6.754 GiB): 0
>>>>>>>>>>> 0001c301e000-000371b6c000( 6.730 GiB): 0
>>>>>>>>>>> 000371b6c000-000400000000( 2.223 GiB): 0
>>>>>>>>>>> 005c6759d000-005c675a2000( 20.000 KiB): 0
>>>>>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>>>>>>> 005c675a3000-005c675a7000( 16.000 KiB): 0
>>>>>>>>>>> 0072f1e14000-0074928d4000( 6.510 GiB): 0
>>>>>>>>>>> 0074928d4000-00763c71f000( 6.655 GiB): 0
>>>>>>>>>>> 00763c71f000-0077e863e000( 6.687 GiB): 0
>>>>>>>>>>> 0077e863e000-00798e214000( 6.590 GiB): 0
>>>>>>>>>>> 00798e214000-007b0e48a000( 6.002 GiB): 0
>>>>>>>>>>> 007b0e48a000-007c62f00000( 5.323 GiB): 0
>>>>>>>>>>> 007c62f00000-007defb19000( 6.199 GiB): 0
>>>>>>>>>>> 007defb19000-007f794ef000( 6.150 GiB): 0
>>>>>>>>>>> 007f794ef000-007fe8f53000( 1.745 GiB): 0
>>>>>>>>>>>
>>>>>>>>>>> As you can see, we have some regions which are very very big and they
>>>>>>>>>>> are losing the chance to be splitted. But
>>>>>>>>>>> Damon can still monitor memory access for those small VMA areas very well like:
>>>>>>>>>>> 005c675a2000-005c675a3000( 4.000 KiB): 3
>>>>>>>>>> Hi, Barry
>>>>>>>>>>
>>>>>>>>>> Actually, we also had found the same problem in redis by ourselves
>>>>>>>>>> tool[1]. The DAMON can not split the large anon VMA well, and the anon
>>>>>>>>>> VMA has 10G~20G memory. I guess the whole region doesn't have sufficient
>>>>>>>>>> hot areas to been monitored or found by DAMON, likes one or more address
>>>>>>>>>> choose by DAMON not been accessed during sample period.
>>>>>>>>>
>>>>>>>>> Hi Rongwei,
>>>>>>>>> Thanks for your comments and thanks for sharing your tools.
>>>>>>>>>
>>>>>>>>> I guess the cause might be:
>>>>>>>>> in case a region is very big like 10GiB, we have only 1MiB hot pages
>>>>>>>>> in this large region.
>>>>>>>>> damon will randomly pick one page to sample, but the page has only
>>>>>>>>> 1MiB/10GiB, thus
>>>>>>>>> less than 1/10000 chance to hit the hot 1MiB. so probably we need
>>>>>>>>> 10000 sample periods
>>>>>>>>> to hit the hot 1MiB in order to split this large region?
>>>>>>>>>
>>>>>>>>> @SeongJae, please correct me if I am wrong.
>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> I'm not sure whether sets init_regions can deal with the above problem,
>>>>>>>>>> or dynamic choose one or limited number VMA to monitor.
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> I won't set a limited number of VMA as this will make the damon too hard to use
>>>>>>>>> as nobody wants to make such complex operations, especially an Android
>>>>>>>>> app might have more than 8000 VMAs.
>>>>>>>>>
>>>>>>>>> I agree init_regions might be the right place to enhance the situation.
>>>>>>>>>
>>>>>>>>>> I'm not sure, just share my idea.
>>>>>>>>>>
>>>>>>>>>> [1] https://github.com/aliyun/data-profile-tools.git
>>>>>>>>>
>>>>>>>>> I suppose this tool is based on damon? How do you finally resolve the problem
>>>>>>>>> that large anon VMAs can't be splitted?
>>>>>>>>> Anyway, I will give your tool a try.
>>>>>>>>
>>>>>>>> Unfortunately, data-profile-tools.git doesn't build on aarch64 ubuntu
>>>>>>>> though autogen.sh
>>>>>>>> runs successfully.
>>>>>>>>
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(disp.o): in function `cons_handler':
>>>>>>>> /root/data-profile-tools/src/disp.c:625: undefined reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>>>>>>> reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/disp.c:625: undefined
>>>>>>>> reference to `wgetch'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_create':
>>>>>>>> /root/data-profile-tools/src/reg.c:108: undefined reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>>>>>>> reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:108: undefined
>>>>>>>> reference to `subwin'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_erase':
>>>>>>>> /root/data-profile-tools/src/reg.c:161: undefined reference to `werase'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh':
>>>>>>>> /root/data-profile-tools/src/reg.c:171: undefined reference to `wrefresh'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_refresh_nout':
>>>>>>>> /root/data-profile-tools/src/reg.c:182: undefined reference to `wnoutrefresh'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_update_all':
>>>>>>>> /root/data-profile-tools/src/reg.c:191: undefined reference to `doupdate'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_win_destroy':
>>>>>>>> /root/data-profile-tools/src/reg.c:200: undefined reference to `delwin'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_line_write':
>>>>>>>> /root/data-profile-tools/src/reg.c:226: undefined reference to `mvwprintw'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:230: undefined
>>>>>>>> reference to `wattr_off'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:217: undefined
>>>>>>>> reference to `wattr_on'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_highlight_write':
>>>>>>>> /root/data-profile-tools/src/reg.c:245: undefined reference to `wattr_on'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>>>>>>> reference to `wattr_off'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:252: undefined
>>>>>>>> reference to `mvwprintw'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:255: undefined
>>>>>>>> reference to `wattr_off'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_fini':
>>>>>>>> /root/data-profile-tools/src/reg.c:367: undefined reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>>>>>>> reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:367: undefined
>>>>>>>> reference to `wclear'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:368: undefined
>>>>>>>> reference to `wrefresh'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:369: undefined
>>>>>>>> reference to `endwin'
>>>>>>>> /usr/bin/ld: ./.libs/libdatop.a(reg.o): in function `reg_curses_init':
>>>>>>>> /root/data-profile-tools/src/reg.c:382: undefined reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:381: undefined
>>>>>>>> reference to `initscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>>>>>>> reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:382: undefined
>>>>>>>> reference to `wrefresh'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:383: undefined
>>>>>>>> reference to `use_default_colors'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:384: undefined
>>>>>>>> reference to `start_color'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:385: undefined
>>>>>>>> reference to `keypad'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:386: undefined
>>>>>>>> reference to `nonl'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:387: undefined
>>>>>>>> reference to `cbreak'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:388: undefined
>>>>>>>> reference to `noecho'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:389: undefined
>>>>>>>> reference to `curs_set'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>>>>>>> reference to `stdscr'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:401: undefined
>>>>>>>> reference to `mvwprintw'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:403: undefined
>>>>>>>> reference to `mvwprintw'
>>>>>>>> /usr/bin/ld: /root/data-profile-tools/src/reg.c:405: undefined
>>>>>>>> reference to `wrefresh'
>>>>>>>> collect2: error: ld returned 1 exit status
>>>>>>>> make[1]: *** [Makefile:592: datop] Error 1
>>>>>>>> make[1]: Leaving directory '/root/data-profile-tools'
>>>>>>>> make: *** [Makefile:438: all] Error 2
>>>>>>> Hi, Barry
>>>>>>>
>>>>>>> Now, the question made me realize that the compatibility of this tool is
>>>>>>> very poor. I built a ubuntu environment at yesterday, and fixed above
>>>>>>> errors by:
>>>>>>>
>>>>>>> diff --git a/configure.ac b/configure.ac
>>>>>>> index 7922f27..1ed823c 100644
>>>>>>> --- a/configure.ac
>>>>>>> +++ b/configure.ac
>>>>>>> @@ -21,13 +21,9 @@ AC_PROG_INSTALL
>>>>>>> AC_CHECK_LIB([numa], [numa_free])
>>>>>>> AC_CHECK_LIB([pthread], [pthread_create])
>>>>>>>
>>>>>>> -PKG_CHECK_MODULES([CHECK], [check])
>>>>>>> -
>>>>>>> -PKG_CHECK_MODULES([NCURSES], [ncursesw ncurses], [LIBS="$LIBS
>>>>>>> $ncurses_LIBS"], [
>>>>>>> - AC_SEARCH_LIBS([delwin], [ncursesw ncurses], [], [
>>>>>>> - AC_MSG_ERROR([ncurses is required but was not found])
>>>>>>> - ], [])
>>>>>>> -])
>>>>>>> +AC_SEARCH_LIBS([stdscr], [ncurses ncursesw], [], [
>>>>>>> + AC_MSG_ERROR([required library libncurses or ncurses not found])
>>>>>>> + ])
>>>>>>>
>>>>>>
>>>>>> I can confirm the patch fixed the issue I reported yesterday, thanks!
>>>>>>
>>>>>>> It works. But I found an another thing will hinder you using this tool.
>>>>>>> We had developed other patches about DAMON base on upstream. This tool
>>>>>>> only works well in ourselves kernel(anolis kernel, already open source).
>>>>>>> Of course, I think it's unnecessary for you to change kernel, just let
>>>>>>> you know this tool still has this problem.
>>>>>>>
>>>>>>
>>>>>> Although I can't use this tool directly as I am not a NUMA right now,
>>>>>> ~/data-profile-tools # ./datop --help
>>>>>> Not support NUMA fault stat (DAMON)!
>>>>>>
>>>>>
>>>>> I wonder if you can extend it to non-numa by setting "remote" to 0%
>>>>> and local to "100%" always for non-numa machines rather than death.
>>>> Hi Barry
>>>>
>>>> That's a great suggestion. Actually, I have removed 'numa_stat' check in
>>>> datop. Maybe you can found. It does not enable numa stat when
>>>> 'numa_stat' sysfs not found in the current system.
>>>
>>> yep. i am able to run it on a non-numa machine, but datop immediately crashes
>>> due to some memory corruption issues:
>>>
>>> Monitoring 270 processes (interval: 5.5s)
>> Barry, it's known bug. I remember the maximum number of processes that
>> is 32 in datop. The reason that setting like this is that I feel
>> impossible to monitor so many processes at the beginning.
>>
>> And it seems that the error message should been printed here, instead of
>> crash. Thank you for reminding me.
>>>
>>> PID PROC TYPE START END
>>> SIZE(KiB) ACCESS AGE
>>> 1693 Binder:1693 ---- 0 0
>>> 0 0 0
>>> 428 ueventd ---- 0 0
>>> 0 0 0
>>> 28654 adbd ---- 0 0
>>> 0 0 0
>>> 971 usb@1.2-ser ---- 0 0
>>> 0 0 0
>>> 619 logd ---- 0 0
>>> 0 0 0
>>> 4311 a...
>>>
>>> <- Hotkey for sorting: 1(PID), 2(START), 3(SIZE), 4(ACCESS), 5(RMA) ->
>>> CPU% = system CPU utilization
>>>
>>> Q: Quit; H: Home; B: Back; R: Refresh; D: DAMON
>>> double free or corruption (!prev)
>>>
>>> Aborted
>>>
>>> if i move to monitor only one process, datop doesn't crash but it
>>> doesn't show any
>>> data either:
>>>
>>> # pgrep youtube
>>> 4311
>>> # ./datop -p 4311
>>>
>>> Monitoring 1 processes (interval: 5.0s)
>> Oh, it's ever happen to me. Does It always show like this when
>> monitoring one process in your environment?
>
>
> right . I have never succeeded in using your datop.
OK, can you help to create a issue here:
https://github.com/aliyun/data-profile-tools/issues. And you can use
'datop -p <pid> -l 2 -f error.log' to record some exception messages. If
available, that's very helpful to me.
Thanks.
>
> ~/damo # pgrep youtube
> 21287
>
> ~/damo # ./damo monitor --report_type=wss --count=5 21287
> # <percentile> <wss>
> # target_id 0
> # avr: 28.285 MiB
> 0 0 B |
> |
> 25 0 B |
> |
> 50 0 B |
> |
> 75 18.547 MiB |******
> |
> 100 174.414 MiB
> |***********************************************************|
>
> # <percentile> <wss>
> # target_id 0
> # avr: 49.857 MiB
> 0 0 B |
> |
> 25 0 B |
> |
> 50 0 B |
> |
> 75 124.180 MiB |****************************
> |
> 100 256.375 MiB
> |***********************************************************|
>
> # <percentile> <wss>
> # target_id 0
> # avr: 35.999 MiB
> 0 0 B |
> |
> 25 0 B |
> |
> 50 0 B |
> |
> 75 59.605 MiB |****************
> |
> 100 218.191 MiB
> |***********************************************************|
>
>
> # <percentile> <wss>
> # target_id 0
> # avr: 35.638 MiB
> 0 0 B |
> |
> 25 0 B |
> |
> 50 0 B |
> |
> 75 27.668 MiB |*******
> |
> 100 230.922 MiB
> |***********************************************************|
>
> # <percentile> <wss>
> # target_id 0
> # avr: 60.585 MiB
> 0 0 B |
> |
> 25 0 B |
> |
> 50 21.297 MiB |****
> |
> 75 122.703 MiB |**************************
> |
> 100 272.973 MiB
> |***********************************************************|
>
>
> datop:
>
> ~/data-profile-tools # ./datop -p 21287
>
> Monitoring 1 processes (interval: 5.1s)
>
> PID PROC TYPE START END
> SIZE(KiB) *ACCESS AGE
> 21287 android.you ---- 0 0
> 0 0 0
>
>
> nothing is shown here.
>
> Is it because your tool doesn't turn on related tracers automatically?
enen, I'm sure that datop or DAMON not needs to turn on tracers. But,
datop will to require event ID from
'/sys/kernel/debug/tracing/events/damon/damon_aggregated/format' to
start perf.
-wrw
> i have dumped
> the values while running datop:
>
> ~/data-profile-tools # ./datop -p 21287
> Start monitoring 21287 ...
> DamonTOP is starting ...
> [1]+ Stopped ./datop -p 21287
>
> ~/data-profile-tools # cat /sys/kernel/debug/damon/target_ids
> 21287
> ~/data-profile-tools # cat /sys/kernel/debug/damon/monitor_on
> on
> ~/data-profile-tools # cat
> /sys/kernel/debug/tracing/events/damon/damon_aggregated/enable
> 0
> ~/data-profile-tools # cat /sys/kernel/debug/tracing/tracing_on
> 0
>
> so i enable them manually:
> ~/data-profile-tools # echo 1 >
> /sys/kernel/debug/tracing/events/damon/damon_aggregated/enable
> ~/data-profile-tools # echo 1 > /sys/kernel/debug/tracing/tracing_on
>
> but datop still shows nothing while kernel begins to report data correctly:
> ~/data-profile-tools # cat /sys/kernel/debug/tracing/trace | more
> # tracer: nop
> #
> # WARNING: FUNCTION TRACING IS CORRUPTED
> # MAY BE MISSING FUNCTION EVENTS
> # entries-in-buffer/entries-written: 11599/11599 #P:8
> #
> # _-----=> irqs-off
> # / _----=> need-resched
> # | / _---=> hardirq/softirq
> # || / _--=> preempt-depth
> # ||| / delay
> # TASK-PID CPU# |||| TIMESTAMP FUNCTION
> # | | | |||| | |
> kdamond.0-26895 [002] .... 160409.911650: damon_aggregated:
> target_id=0 nr_regions=12 314572800-3954012160: 0 185
> kdamond.0-26895 [002] .... 160409.911670: damon_aggregated:
> target_id=0 nr_regions=12 394992549888-394992590848: 0 12
> kdamond.0-26895 [002] .... 160409.911675: damon_aggregated:
> target_id=0 nr_regions=12 488092233728-494587068416: 0 311
> kdamond.0-26895 [002] .... 160409.911679: damon_aggregated:
> target_id=0 nr_regions=12 494587068416-501072220160: 0 375
> kdamond.0-26895 [002] .... 160409.911683: damon_aggregated:
> target_id=0 nr_regions=12 501072220160-507561463808: 0 415
> kdamond.0-26895 [002] .... 160409.911687: damon_aggregated:
> target_id=0 nr_regions=12 507561463808-514046730240: 0 2091
> kdamond.0-26895 [002] .... 160409.911691: damon_aggregated:
> target_id=0 nr_regions=12 514046730240-520535441408: 0 2263
> kdamond.0-26895 [002] .... 160409.911696: damon_aggregated:
> target_id=0 nr_regions=12 520535441408-527022301184: 0 2349
> kdamond.0-26895 [002] .... 160409.911700: damon_aggregated:
> target_id=0 nr_regions=12 527022301184-533491294208: 0 2373
> kdamond.0-26895 [002] .... 160409.911704: damon_aggregated:
> target_id=0 nr_regions=12 533491294208-539965886464: 0 2378
> kdamond.0-26895 [002] .... 160409.911708: damon_aggregated:
> target_id=0 nr_regions=12 539965886464-546468855808: 0 2380
> kdamond.0-26895 [002] .... 160409.911712: damon_aggregated:
> target_id=0 nr_regions=12 546468855808-549620240384: 0 2381
> kdamond.0-26895 [000] .... 160410.014673: damon_aggregated:
> target_id=0 nr_regions=13 314572800-3954012160: 0 186
> kdamond.0-26895 [000] .... 160410.014694: damon_aggregated:
> target_id=0 nr_regions=13 394992549888-394992578560: 1 0
> kdamond.0-26895 [000] .... 160410.014699: damon_aggregated:
> target_id=0 nr_regions=13 394992578560-394992590848: 0 13
> kdamond.0-26895 [000] .... 160410.014704: damon_aggregated:
> target_id=0 nr_regions=13 488092233728-494587068416: 0 312
> kdamond.0-26895 [000] .... 160410.014709: damon_aggregated:
> target_id=0 nr_regions=13 494587068416-501072220160: 0 376
> kdamond.0-26895 [000] .... 160410.014714: damon_aggregated:
> target_id=0 nr_regions=13 501072220160-507561463808: 0 416
> kdamond.0-26895 [000] .... 160410.014718: damon_aggregated:
> target_id=0 nr_regions=13 507561463808-514046730240: 0 2092
> kdamond.0-26895 [000] .... 160410.014723: damon_aggregated:
> target_id=0 nr_regions=13 514046730240-520535441408: 0 2264
> kdamond.0-26895 [000] .... 160410.014727: damon_aggregated:
> target_id=0 nr_regions=13 520535441408-527022301184: 0 2350
> kdamond.0-26895 [000] .... 160410.014732: damon_aggregated:
> target_id=0 nr_regions=13 527022301184-533491294208: 0 2374
> kdamond.0-26895 [000] .... 160410.014736: damon_aggregated:
> target_id=0 nr_regions=13 533491294208-539965886464: 0 2379
> kdamond.0-26895 [000] .... 160410.014740: damon_aggregated:
> target_id=0 nr_regions=13 539965886464-546468855808: 0 2381
> kdamond.0-26895 [000] .... 160410.014745: damon_aggregated:
> target_id=0 nr_regions=13 546468855808-549620240384: 0 2382
> kdamond.0-26895 [001] .... 160410.112316: damon_aggregated:
> target_id=0 nr_regions=12 314572800-3954012160: 0 187
> kdamond.0-26895 [001] .... 160410.112338: damon_aggregated:
> target_id=0 nr_regions=12 394992549888-394992590848: 0 4
> kdamond.0-26895 [001] .... 160410.112343: damon_aggregated:
> target_id=0 nr_regions=12 488092233728-494587068416: 0 313
> kdamond.0-26895 [001] .... 160410.112348: damon_aggregated:
> target_id=0 nr_regions=12 494587068416-501072220160: 0 377
> kdamond.0-26895 [001] .... 160410.112353: damon_aggregated:
> target_id=0 nr_regions=12 501072220160-507561463808: 0 417
>
>
>
>
>>>
>>> PID PROC TYPE START END
>>> SIZE(KiB) *ACCESS AGE
>>> 4311 youtube ---- 0 0 0
>>> 0 0
>>>
>>>
>>>>
>>>> What's more, a new hot key 'f' will be introduced which can enable some
>>>> features dynamically, such as numa stat. Others features can be used
>>>> only in our internal version, likes 'f' in top, and will be open source
>>>> when stable.
>>>>
>>>>> as your tools can map regions to .so, which seems to be quite useful.
>>>> enen, I'm agree with you. But you know, one region maybe covers one or
>>>> more VMAs, hard to map access count of regions to the related .so or
>>>> anon. A lazy way used by me now. I still think it's valuable in the future.
>>>>
>>>
>>> it seems really an interesting topic worth more investigation. I wonder if
>>> damon vaddr monitor should actually take vmas, or at least the types of
>>> vmas into consideration while splitting.
>>>
>>> Different vma types should be inherently different in hotness. for example,
>>> if 1mb text and 1mb data are put in the same region, the monitored data
>>> to reflect the hotness for the whole 2mb seems to be pointless at all.
>>>
>>> Hi SeongJae,
>>> what do you think about it?
>>>
>>>> Anyway, any idea are welcome.
>>>>
>>>> Thanks,
>>>> -wrw
>>>>
>>>>>
>>>>>> I am still quite interested in your design and the purpose of this project.
>>>>>> Unfortunately the project seems to be lacking some design doc.
>>>>>>
>>>>>> And would you like to send patches to lkml regarding what you
>>>>>> have changed atop DAMON?
>>>>>>
>>>>>>> Anyway, the question that you reported was valuable, made me realize
>>>>>>> what we need to improve next.
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Rongwei Wang
>>>>>>>>
>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Typical characteristics of a large Android app is that it has
>>>>>>>>>>> thousands of vma and very large virtual address spaces:
>>>>>>>>>>> ~/damo # pmap 2550 | wc -l
>>>>>>>>>>> 8522
>>>>>>>>>>>
>>>>>>>>>>> ~/damo # pmap 2550
>>>>>>>>>>> ...
>>>>>>>>>>> 0000007992bbe000 4K r---- [ anon ]
>>>>>>>>>>> 0000007992bbf000 24K rw--- [ anon ]
>>>>>>>>>>> 0000007fe8753000 4K ----- [ anon ]
>>>>>>>>>>> 0000007fe8754000 8188K rw--- [ stack ]
>>>>>>>>>>> total 36742112K
>>>>>>>>>>>
>>>>>>>>>>> Because the whole vma list is too long, I have put the list here for
>>>>>>>>>>> you to download:
>>>>>>>>>>> wget http://www.linuxep.com/patches/android-app-vmas
>>>>>>>>>>>
>>>>>>>>>>> I can reproduce this problem on other Apps like youtube as well.
>>>>>>>>>>> I suppose we need to boost the algorithm of splitting regions for this
>>>>>>>>>>> kind of application.
>>>>>>>>>>> Any thoughts?
>>>>>>>>>>>
>>>>>>>>>
>>>>>
>
> Thanks
> Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-04-27 17:50 ` sj
@ 2022-05-29 19:54 ` SeongJae Park
2022-05-30 5:04 ` Barry Song
0 siblings, 1 reply; 23+ messages in thread
From: SeongJae Park @ 2022-05-29 19:54 UTC (permalink / raw)
To: sj
Cc: Barry Song, Rongwei Wang, Andrew Morton, Linux-MM, LKML,
Matthew Wilcox, shuah, brendanhiggins, foersleo, sieberf,
Shakeel Butt, sjpark, tuhailong, Song Jiang,
张诗明(Simon Zhang),
李培锋(wink),
xhao, damon
On Wed, 27 Apr 2022 17:50:49 +0000 sj@kernel.org wrote:
> Hello Rongwei and Barry,
>
> On Wed, 27 Apr 2022 19:44:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
>
> > On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> > <rongwei.wang@linux.alibaba.com> wrote:
> > >
> > >
> > >
> > > On 4/27/22 7:19 AM, Barry Song wrote:
[...]
> >
> > I guess the cause might be:
> > in case a region is very big like 10GiB, we have only 1MiB hot pages
> > in this large region.
> > damon will randomly pick one page to sample, but the page has only
> > 1MiB/10GiB, thus
> > less than 1/10000 chance to hit the hot 1MiB. so probably we need
> > 10000 sample periods
> > to hit the hot 1MiB in order to split this large region?
> >
> > @SeongJae, please correct me if I am wrong.
>
> I think your theory makes sense. There was a similar concern, so we made DAMON
> to split regions into 3 sub-regions when we don't see advance[1]. My current
> rough idea for improving DAMON accuracy is making it more aggressive while
> keeping the monitoring overhead low.
>
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/sj/linux.git/tree/mm/damon/core.c?h=damon/next-2022-04-21-08-31-on-v5.18-rc3-mmots-2022-04-20-17-37#n1053
>
> >
> > >
> > > I'm not sure whether sets init_regions can deal with the above problem,
> > > or dynamic choose one or limited number VMA to monitor.
> > >
> >
> > I won't set a limited number of VMA as this will make the damon too hard to use
> > as nobody wants to make such complex operations, especially an Android
> > app might have more than 8000 VMAs.
> >
> > I agree init_regions might be the right place to enhance the situation.
>
> 'init_regions' has developed for the purpose, where user space knows some good
> information for starting point of the regions adjustment, and thus want to hint
> DAMON. Nevertheless, it might not work as expected, because DAMON
> automatically updates the target regions to cover all VMAs as much as it can.
> I have posted a patchset for the use case yesterday[1].
>
> [1] https://lore.kernel.org/linux-mm/20220426231750.48822-1-sj@kernel.org/
FWIW, the patchset for the fixed virtual address space ranges monitoring has
merged in the mainline[1].
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=98931dd95fd489fcbfa97da563505a6f071d7c77
Thanks,
SJ
[...]
^ permalink raw reply [flat|nested] 23+ messages in thread
* Re: DAMON VA regions don't split on an large Android APP
2022-05-29 19:54 ` SeongJae Park
@ 2022-05-30 5:04 ` Barry Song
0 siblings, 0 replies; 23+ messages in thread
From: Barry Song @ 2022-05-30 5:04 UTC (permalink / raw)
To: SeongJae Park
Cc: Rongwei Wang, Andrew Morton, Linux-MM, LKML, Matthew Wilcox,
shuah, Brendan Higgins, foersleo, sieberf, Shakeel Butt, sjpark,
tuhailong, Song Jiang, 张诗明(Simon Zhang),
李培锋(wink),
xhao, damon
On Mon, May 30, 2022 at 7:54 AM SeongJae Park <sj@kernel.org> wrote:
>
> On Wed, 27 Apr 2022 17:50:49 +0000 sj@kernel.org wrote:
>
> > Hello Rongwei and Barry,
> >
> > On Wed, 27 Apr 2022 19:44:23 +1200 Barry Song <21cnbao@gmail.com> wrote:
> >
> > > On Wed, Apr 27, 2022 at 6:56 PM Rongwei Wang
> > > <rongwei.wang@linux.alibaba.com> wrote:
> > > >
> > > >
> > > >
> > > > On 4/27/22 7:19 AM, Barry Song wrote:
> [...]
> > >
> > > I guess the cause might be:
> > > in case a region is very big like 10GiB, we have only 1MiB hot pages
> > > in this large region.
> > > damon will randomly pick one page to sample, but the page has only
> > > 1MiB/10GiB, thus
> > > less than 1/10000 chance to hit the hot 1MiB. so probably we need
> > > 10000 sample periods
> > > to hit the hot 1MiB in order to split this large region?
> > >
> > > @SeongJae, please correct me if I am wrong.
> >
> > I think your theory makes sense. There was a similar concern, so we made DAMON
> > to split regions into 3 sub-regions when we don't see advance[1]. My current
> > rough idea for improving DAMON accuracy is making it more aggressive while
> > keeping the monitoring overhead low.
> >
> > [1] https://git.kernel.org/pub/scm/linux/kernel/git/sj/linux.git/tree/mm/damon/core.c?h=damon/next-2022-04-21-08-31-on-v5.18-rc3-mmots-2022-04-20-17-37#n1053
> >
> > >
> > > >
> > > > I'm not sure whether sets init_regions can deal with the above problem,
> > > > or dynamic choose one or limited number VMA to monitor.
> > > >
> > >
> > > I won't set a limited number of VMA as this will make the damon too hard to use
> > > as nobody wants to make such complex operations, especially an Android
> > > app might have more than 8000 VMAs.
> > >
> > > I agree init_regions might be the right place to enhance the situation.
> >
> > 'init_regions' has developed for the purpose, where user space knows some good
> > information for starting point of the regions adjustment, and thus want to hint
> > DAMON. Nevertheless, it might not work as expected, because DAMON
> > automatically updates the target regions to cover all VMAs as much as it can.
> > I have posted a patchset for the use case yesterday[1].
> >
> > [1] https://lore.kernel.org/linux-mm/20220426231750.48822-1-sj@kernel.org/
>
> FWIW, the patchset for the fixed virtual address space ranges monitoring has
> merged in the mainline[1].
>
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=98931dd95fd489fcbfa97da563505a6f071d7c77
>
nice to know that. Thanks, though it doesn't fix my problem as I am
looking for a solution to
collect precise monitoring data automatically and economically.
>
> Thanks,
> SJ
Thanks
Barry
^ permalink raw reply [flat|nested] 23+ messages in thread
end of thread, other threads:[~2022-05-30 5:04 UTC | newest]
Thread overview: 23+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-04-26 23:19 DAMON VA regions don't split on an large Android APP Barry Song
2022-04-27 0:21 ` sj
2022-04-27 2:08 ` Barry Song
2022-04-27 17:39 ` sj
2022-04-28 1:27 ` Barry Song
2022-04-28 16:16 ` sj
2022-04-27 6:56 ` Rongwei Wang
2022-04-27 7:44 ` Barry Song
2022-04-27 9:22 ` Barry Song
2022-04-27 11:55 ` Rongwei Wang
2022-04-28 2:04 ` Rongwei Wang
2022-04-28 7:37 ` Barry Song
2022-04-28 11:19 ` Rongwei Wang
2022-05-16 7:03 ` Barry Song
2022-05-16 15:00 ` Rongwei Wang
2022-05-17 11:14 ` Barry Song
2022-05-18 3:03 ` Rongwei Wang
2022-05-18 9:51 ` Barry Song
2022-05-19 3:09 ` Rongwei Wang
2022-04-27 12:06 ` Rongwei Wang
2022-04-27 17:50 ` sj
2022-05-29 19:54 ` SeongJae Park
2022-05-30 5:04 ` Barry Song
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).