Re: [linus:master] [btrfs] bddf57a707: stress-ng.sync-file.ops_per_sec 44.2% regression

[David Sterba <dsterba@suse.cz>]

On Wed, Sep 03, 2025 at 06:18:01PM +0930, Qu Wenruo wrote:
> 在 2025/9/3 18:14, kernel test robot 写道:
> > 
> > Hello,
> > 
> > kernel test robot noticed a 44.2% regression of stress-ng.sync-file.ops_per_sec on:
> > 
> > 
> > commit: bddf57a70781ef8821d415200bdbcb71f443993a ("btrfs: delay btrfs_open_devices() until super block is created")
> > https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
> > 
> > [still regression on      linus/master fb679c832b6497f19fffb8274c419783909c0912]
> > [still regression on linux-next/master 3cace99d63192a7250461b058279a42d91075d0c]
> > 
> > testcase: stress-ng
> > config: x86_64-rhel-9.4
> > compiler: gcc-12
> > test machine: 64 threads 2 sockets Intel(R) Xeon(R) Gold 6346 CPU @ 3.10GHz (Ice Lake) with 256G memory
> > parameters:
> > 
> > 	nr_threads: 100%
> > 	disk: 1HDD
> > 	testtime: 60s
> > 	fs: btrfs
> > 	test: sync-file
> > 	cpufreq_governor: performance
> > 
> > 
> > 
> > 
> > If you fix the issue in a separate patch/commit (i.e. not just a new version of
> > the same patch/commit), kindly add following tags
> > | Reported-by: kernel test robot <oliver.sang@intel.com>
> > | Closes: https://lore.kernel.org/oe-lkp/202509031643.303d114c-lkp@intel.com
> > 
> > 
> > Details are as below:
> > -------------------------------------------------------------------------------------------------->
> > 
> > 
> > The kernel config and materials to reproduce are available at:
> > https://download.01.org/0day-ci/archive/20250903/202509031643.303d114c-lkp@intel.com
> > 
> > =========================================================================================
> > compiler/cpufreq_governor/disk/fs/kconfig/nr_threads/rootfs/tbox_group/test/testcase/testtime:
> >    gcc-12/performance/1HDD/btrfs/x86_64-rhel-9.4/100%/debian-12-x86_64-20240206.cgz/lkp-icl-2sp8/sync-file/stress-ng/60s
> > 
> > commit:
> >    de339cbfb4 ("btrfs: call bdev_fput() to reclaim the blk_holder immediately")
> >    bddf57a707 ("btrfs: delay btrfs_open_devices() until super block is created")
> 
> This doesn't sound sane to me.
> 
> The two commits are only affecting btrfs mounting/unmounting, I can not 
> make any sense on why they would affect performance.
> 
> Or does stress-ng doing a lot of mounting/unmounting?

Yeah, unless there's some indirect way how mount affects the tests the
numbers do not match the identified patches. The difference is roughly
consistent in all the stats to be about 40% less so it's like it's doing
half of the work. Delayed device opening does not explain that.