Hi, Miklos,
I noticed a performance regression of reading/writing files in mergeddir
caused by commit a6518f73e60e5044656d1ba587e7463479a9381a (vfs: don't
open real), using unixbench fstime.
Reproduce Steps:
1. cd /mnt/lower/ && git clone
https://github.com/kdlucas/byte-unixbench.git
2. mount -t overlay overlay
-olowerdir=/mnt/lower,upperdir=/mnt/upper,workdir=/mnt/work /mnt/merge
3. cd /mnt/merge/byte-unixbench/UnixBench && ./Run -c 1 -i 1 fstime
The score is 2870 before applying the patch, while it is 1780 after
applying the patch, causing a 40% performance regression.
The testcase repeatedly reads 1024 bytes from one file and writes the
readed data into another file, while both these two files
are created under /mnt/merge/tmp. I have testsed the latest kernel
5.4.0-rc4+, same results.
The perf shows that there's extra one call of file_remove_privs(),
override_creds() and revert_creds() every write() syscall,
among which file_remove_privs() is pretty expensive.
- perf data before applying the patch.
```
- 53.00% 0.93% fstime [kernel.kallsyms] [k] __vfs_write
- 52.08% __vfs_write
- 51.94% ext4_file_write_iter
+ 48.89% __generic_file_write_iter
0.83% down_write_trylock
0.79% up_write
```
- perf data after applying the patch.
```
+ 94.88% 0.00% fstime [kernel.kallsyms] [k]
entry_SYSCALL_64_after_hwframe
+ 94.88% 4.67% fstime [kernel.kallsyms] [k] do_syscall_64
+ 66.08% 1.60% fstime libc-2.17.so [.] __GI___libc_write
+ 62.37% 0.23% fstime [kernel.kallsyms] [k] ksys_write
+ 61.74% 0.62% fstime [kernel.kallsyms] [k] vfs_write
- 60.10% 0.49% fstime [kernel.kallsyms] [k] __vfs_write
- 59.61% __vfs_write
- 59.56% ovl_write_iter
- 33.81% do_iter_write
- 32.50% do_iter_readv_writev
+ ext4_file_write_iter
+ 19.15% file_remove_privs
2.15% revert_creds
2.02% override_creds
0.64% down_write
0.63% up_write
```
Regards.
Jeffle
