Re: [linus:master] [page cache] 9425c591e0: vm-scalability.throughput -20.0% regression

[Mike Kravetz]

On 06/21/23 15:19, kernel test robot wrote:
> 
> 
> Hello,
> 
> kernel test robot noticed a -20.0% regression of vm-scalability.throughput on:
> 
> 
> commit: 9425c591e06a9ab27a145ba655fb50532cf0bcc9 ("page cache: fix page_cache_next/prev_miss off by one")
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
> 
> testcase: vm-scalability
> test machine: 96 threads 2 sockets Intel(R) Xeon(R) Platinum 8260L CPU @ 2.40GHz (Cascade Lake) with 128G memory
> parameters:
> 
> 	runtime: 300s
> 	test: lru-file-readonce
> 	cpufreq_governor: performance
> 
> test-description: The motivation behind this suite is to exercise functions and regions of the mm/ of the Linux kernel which are of interest to us.
> test-url: https://git.kernel.org/cgit/linux/kernel/git/wfg/vm-scalability.git/
> 
> In addition to that, the commit also has significant impact on the following tests:
> 
> +------------------+----------------------------------------------------------------------------------------------------+
> | testcase: change | vm-scalability: vm-scalability.throughput -18.9% regression                                        |
> | test machine     | 96 threads 2 sockets Intel(R) Xeon(R) Platinum 8260L CPU @ 2.40GHz (Cascade Lake) with 128G memory |
> | test parameters  | cpufreq_governor=performance                                                                       |
> |                  | debug-setup=no-monitor                                                                             |
> |                  | runtime=300s                                                                                       |
> |                  | test=lru-file-readonce                                                                             |
> +------------------+----------------------------------------------------------------------------------------------------+
> | testcase: change | vm-scalability: vm-scalability.throughput -52.8% regression                                        |
> | test machine     | 224 threads 2 sockets Intel(R) Xeon(R) Platinum 8480CTDX (Sapphire Rapids) with 256G memory        |
> | test parameters  | cpufreq_governor=performance                                                                       |
> |                  | runtime=300s                                                                                       |
> |                  | test=lru-file-readonce                                                                             |
> +------------------+----------------------------------------------------------------------------------------------------+
> | testcase: change | vm-scalability: vm-scalability.throughput -54.0% regression                                        |
> | test machine     | 224 threads 2 sockets Intel(R) Xeon(R) Platinum 8480CTDX (Sapphire Rapids) with 256G memory        |
> | test parameters  | cpufreq_governor=performance                                                                       |
> |                  | debug-setup=no-monitor                                                                             |
> |                  | runtime=300s                                                                                       |
> |                  | test=lru-file-readonce                                                                             |
> +------------------+----------------------------------------------------------------------------------------------------+
> 

Ouch!

I suspected this change could impact page_cache_next/prev_miss users, but had
no idea how much.

Unless someone sees something wrong in 9425c591e06a, the best approach
might be to revert and then add a simple interface to check for 'folio at
a given index in the cache' as suggested by Ackerley Tng.
https://lore.kernel.org/linux-mm/98624c2f481966492b4eb8272aef747790229b73.1683069252.git.ackerleytng@xxxxxxxxxx/
-- 
Mike Kravetz