Em Tue, Jul 28, 2020 at 08:51:52AM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Fri, Jul 24, 2020 at 12:19:59AM -0700, Ian Rogers escreveu:
> > for_each_set_bit, or similar functions like for_each_cpu, may be hot
> > within the kernel. If many bits were set then one could imagine on
> > Intel a "bt" instruction with every bit may be faster than the function
> > call and word length find_next_bit logic. Add a benchmark to measure
> > this.
>
> Thanks, applied.
>
> - Arnaldo
>
> > This benchmark on AMD rome and Intel skylakex shows "bt" is not a good
> > option except for very small bitmaps.
> > +++ b/tools/perf/bench/find-bit-bench.c
> > +#if defined(__i386__) || defined(__x86_64__)
> > +static bool asm_test_bit(long nr, const unsigned long *addr)
> > +{
> > + bool oldbit;
> > +
> > + asm volatile("bt %2,%1"
> > + : "=@ccc" (oldbit)
> > + : "m" (*(unsigned long *)addr), "Ir" (nr) : "memory");
> > +
> > + return oldbit;
Some old clang versions are not liking this:
clang version 3.8.0 (tags/RELEASE_380/final)
Target: x86_64-alpine-linux-musl
Thread model: posix
InstalledDir: /usr/bin
Found candidate GCC installation:
/usr/bin/../lib/gcc/x86_64-alpine-linux-musl/5.3.0
Found candidate GCC installation: /usr/lib/gcc/x86_64-alpine-linux-musl/5.3.0
Selected GCC installation: /usr/bin/../lib/gcc/x86_64-alpine-linux-musl/5.3.0
Candidate multilib: .;@m64
Selected multilib: .;@m64
+ make ARCH= CROSS_COMPILE= EXTRA_CFLAGS= -C /git/linux/tools/perf
O=/tmp/build/perf CC=clang
CC /tmp/build/perf/trace/beauty/pkey_alloc.o
CC /tmp/build/perf/tests/openat-syscall-tp-fields.o
bench/find-bit-bench.c:46:10: error: invalid output constraint '=@ccc' in asm
: "=@ccc" (oldbit)
^
1 error generated.
mv: can't rename '/tmp/build/perf/bench/.find-bit-bench.o.tmp': No such file or
directory
/git/linux/tools/build/Makefile.build:96: recipe for target
'/tmp/build/perf/bench/find-bit-bench.o' failed
make[4]: *** [/tmp/build/perf/bench/find-bit-bench.o] Error 1
/git/linux/tools/build/Makefile.build:139: recipe for target 'bench' failed
make[3]: *** [bench] Error 2
make[3]: *** Waiting for unfinished jobs....
MKDIR /tmp/build/perf/arch/x86/tests/
Also:
clang version 3.8.1 (tags/RELEASE_381/final)
clang version 4.0.0 (tags/RELEASE_400/final)
- Arnaldo
