how is clock_gettime implemented on linux? - linux

Question

When I strace this:
#include <stdio.h>
#include <time.h>
int main()
{
struct timespec ts;
fprintf(stderr, "start!\n");
clock_gettime(CLOCK_REALTIME, &ts);
fprintf(stderr, "realtime %lu %lu\n", ts.tv_sec, ts.tv_nsec);
clock_gettime(CLOCK_MONOTONIC, &ts);
fprintf(stderr, "monotonic %lu %lu\n", ts.tv_sec, ts.tv_nsec);
return 0;
}
I get this:
execve("/tmp/x", ["/tmp/x"], [/* 72 vars */]) = 0
brk(NULL) = 0x13d0000
access("/etc/ld.so.nohwcap", F_OK) = -1 ENOENT (No such file or directory)
mmap(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fed9a0d1000
access("/etc/ld.so.preload", R_OK) = -1 ENOENT (No such file or directory)
open("/etc/ld.so.cache", O_RDONLY|O_CLOEXEC) = 3
fstat(3, {st_mode=S_IFREG|0644, st_size=126501, ...}) = 0
mmap(NULL, 126501, PROT_READ, MAP_PRIVATE, 3, 0) = 0x7fed9a0b2000
close(3) = 0
access("/etc/ld.so.nohwcap", F_OK) = -1 ENOENT (No such file or directory)
open("/lib/x86_64-linux-gnu/libc.so.6", O_RDONLY|O_CLOEXEC) = 3
read(3, "\177ELF\2\1\1\3\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0P\t\2\0\0\0\0\0"..., 832) = 832
fstat(3, {st_mode=S_IFREG|0755, st_size=1864888, ...}) = 0
mmap(NULL, 3967488, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0x7fed99ae5000
mprotect(0x7fed99ca5000, 2093056, PROT_NONE) = 0
mmap(0x7fed99ea4000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x1bf000) = 0x7fed99ea4000
mmap(0x7fed99eaa000, 14848, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x7fed99eaa000
close(3) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fed9a0b1000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fed9a0b0000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7fed9a0af000
arch_prctl(ARCH_SET_FS, 0x7fed9a0b0700) = 0
mprotect(0x7fed99ea4000, 16384, PROT_READ) = 0
mprotect(0x600000, 4096, PROT_READ) = 0
mprotect(0x7fed9a0d3000, 4096, PROT_READ) = 0
munmap(0x7fed9a0b2000, 126501) = 0
write(2, "start!\n", 7start!
) = 7
write(2, "realtime 1488919932 97097045\n", 29realtime 1488919932 97097045
) = 29
write(2, "monotonic 258985 170149836\n", 27monotonic 258985 170149836
) = 27
exit_group(0) = ?
+++ exited with 0 +++
There doesn't seem to be a system call involved in producing either CLOCK_REALTIME or CLOCK_MONOTONIC values. How is this implemented? I tried to step through it in assembler but I must've not noticed the critical part because I can't figure out how it's done.

Answer 1

This is a virtual system call. That is, calling it does not require switching to the kernel mode -- it is executed in user mode to improve performance. If you have ever wondered what does linux-vdso.so.1 mean in the ldd output, it is where these virtual system calls are implemented. You can learn more here.