* Understanding the locking behavior of msync
@ 2021-03-24 11:56 Maximilian Böther
2021-03-27 1:57 ` Mulyadi Santosa
0 siblings, 1 reply; 2+ messages in thread
From: Maximilian Böther @ 2021-03-24 11:56 UTC (permalink / raw)
To: kernelnewbies
Hello!
I am investigating an application that writes random data in fixed-size
chunks (e.g. 4k) to random locations in a large buffer file. I have
several processes (not threads) doing that, each process has its own
buffer file assigned.
If I use mmap+msync to write and persist data to disk, I see a
performance spike for 16 processes, and a performance drop for more
threads (32 processes). The CPU has 32 logical cores in total, and we
are not CPU bound.
If I use open+write+fsync, I do not see such a spike, instead a
performance plateau (and mmap is slower than open/write).
I've read multiple times [1,2] that both mmap and msync can take locks.
With vtune, I analyzed that we are indeed spinlocking, and spending the
most time in clear_page_erms and xas_load functions.
However, when reading the source code for msync [3], I cannot understand
whether these locks are global or per-file. The paper [2] states that
the locks are on radix-trees within the kernel that are per-file,
however, as I do observe some spinlocks in the kernel, I believe that
some locks may be global, as I have one file per process.
Do you have an explanation on why we have such a spike at 16 processes
for mmap and input on the locking behavior of msync?
Thank you!
Best,
Maximilian Böther
[1]
https://kb.pmem.io/development/100000025-Why-msync-is-less-optimal-for-persistent-memory/
- I know it's about PMem, but the lock argument is important
[2] Optimizing Memory-mapped I/O for Fast Storage Devices, Papagiannis
et al., ATC '20
[3] https://elixir.bootlin.com/linux/latest/source/mm/msync.c
_______________________________________________
Kernelnewbies mailing list
Kernelnewbies@kernelnewbies.org
https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: Understanding the locking behavior of msync
2021-03-24 11:56 Understanding the locking behavior of msync Maximilian Böther
@ 2021-03-27 1:57 ` Mulyadi Santosa
0 siblings, 0 replies; 2+ messages in thread
From: Mulyadi Santosa @ 2021-03-27 1:57 UTC (permalink / raw)
To: kernelnewbies
[-- Attachment #1.1: Type: text/plain, Size: 2070 bytes --]
On Fri, Mar 26, 2021, 22:05 Maximilian Böther <
maximilian.boether@student.hpi.de> wrote:
> Hello!
>
> I am investigating an application that writes random data in fixed-size
> chunks (e.g. 4k) to random locations in a large buffer file. I have
> several processes (not threads) doing that, each process has its own
> buffer file assigned.
>
> If I use mmap+msync to write and persist data to disk, I see a
> performance spike for 16 processes, and a performance drop for more
> threads (32 processes). The CPU has 32 logical cores in total, and we
> are not CPU bound.
>
> If I use open+write+fsync, I do not see such a spike, instead a
> performance plateau (and mmap is slower than open/write).
>
> I've read multiple times [1,2] that both mmap and msync can take locks.
> With vtune, I analyzed that we are indeed spinlocking, and spending the
> most time in clear_page_erms and xas_load functions.
>
> However, when reading the source code for msync [3], I cannot understand
> whether these locks are global or per-file. The paper [2] states that
> the locks are on radix-trees within the kernel that are per-file,
> however, as I do observe some spinlocks in the kernel, I believe that
> some locks may be global, as I have one file per process.
>
> Do you have an explanation on why we have such a spike at 16 processes
> for mmap and input on the locking behavior of msync?
>
> Thank you!
>
> Best,
> Maximilian Böther
>
> [1]
>
> https://kb.pmem.io/development/100000025-Why-msync-is-less-optimal-for-persistent-memory/
> - I know it's about PMem, but the lock argument is important
>
> [2] Optimizing Memory-mapped I/O for Fast Storage Devices, Papagiannis
> et al., ATC '20
>
> [3] https://elixir.bootlin.com/linux/latest/source/mm/msync.c
>
> _______________________________________________
> Kernelnewbies mailing list
> Kernelnewbies@kernelnewbies.org
> https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbi
> <https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies>es
>
Is it NUMA?
[-- Attachment #1.2: Type: text/html, Size: 3126 bytes --]
[-- Attachment #2: Type: text/plain, Size: 170 bytes --]
_______________________________________________
Kernelnewbies mailing list
Kernelnewbies@kernelnewbies.org
https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2021-03-27 1:59 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-03-24 11:56 Understanding the locking behavior of msync Maximilian Böther
2021-03-27 1:57 ` Mulyadi Santosa
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).