| Message ID | 1542542538-11938-1-git-send-email-laoar.shao@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | mm/filemap.c: minor optimization in write_iter file operation | expand |
On Sun, Nov 18, 2018 at 08:02:18PM +0800, Yafang Shao wrote: > This little adjustment on bitwise operation could make the code a little > faster. > As write_iter is used in lots of critical path, so this code change is > useful for performance. Did you check the before/after code generation with this patch applied? $ diff -u before.S after.S --- before.S 2018-11-18 07:11:48.031096768 -0500 +++ after.S 2018-11-18 07:11:36.883069103 -0500 @@ -1,5 +1,5 @@ -before.o: file format elf32-i386 +after.o: file format elf32-i386 Disassembly of section .text: with gcc 8.2.0, I see no difference, indicating that the compiler already makes this optimisation.
On Sun, Nov 18, 2018 at 8:13 PM Matthew Wilcox <willy@infradead.org> wrote: > > On Sun, Nov 18, 2018 at 08:02:18PM +0800, Yafang Shao wrote: > > This little adjustment on bitwise operation could make the code a little > > faster. > > As write_iter is used in lots of critical path, so this code change is > > useful for performance. > > Did you check the before/after code generation with this patch applied? > Yes, I did. My oompiler is gcc-4.8.5, a litte old, and with CONFIG_CC_OPTIMIZE_FOR_SIZE on. The output file is differrent. > $ diff -u before.S after.S > --- before.S 2018-11-18 07:11:48.031096768 -0500 > +++ after.S 2018-11-18 07:11:36.883069103 -0500 > @@ -1,5 +1,5 @@ > > -before.o: file format elf32-i386 > +after.o: file format elf32-i386 > > > Disassembly of section .text: > > with gcc 8.2.0, I see no difference, indicating that the compiler already > makes this optimisation. Could pls. try set CONFIG_CC_OPTIMIZE_FOR_SIZE on and then compare them again ? Thanks Yafang
On Sun, Nov 18, 2018 at 11:02:19PM +0800, Yafang Shao wrote: > On Sun, Nov 18, 2018 at 8:13 PM Matthew Wilcox <willy@infradead.org> wrote: > > Did you check the before/after code generation with this patch applied? > > Yes, I did. > My oompiler is gcc-4.8.5, a litte old, and with CONFIG_CC_OPTIMIZE_FOR_SIZE on. > > with gcc 8.2.0, I see no difference, indicating that the compiler already > > makes this optimisation. > > Could pls. try set CONFIG_CC_OPTIMIZE_FOR_SIZE on and then compare them again ? Actually it was already on: # CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE is not set CONFIG_CC_OPTIMIZE_FOR_SIZE=y I happened to build it in my build-tiny output tree.
diff --git a/mm/filemap.c b/mm/filemap.c index 81adec8..a65056ea 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -2881,7 +2881,8 @@ inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from) if (iocb->ki_flags & IOCB_APPEND) iocb->ki_pos = i_size_read(inode); - if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT)) + if ((iocb->ki_flags & (IOCB_NOWAIT | IOCB_DIRECT)) == + IOCB_NOWAIT) return -EINVAL; count = iov_iter_count(from);
This little adjustment on bitwise operation could make the code a little faster. As write_iter is used in lots of critical path, so this code change is useful for performance. Signed-off-by: Yafang Shao <laoar.shao@gmail.com> --- mm/filemap.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)