[RFC PATCH v3 11/12] lib: vdso: split clock verification out of __arch_get_hw_counter()
From: Christophe Leroy <hidden>
Date: 2020-01-13 17:09:05
Also in:
linux-arm-kernel, linuxppc-dev, lkml
Subsystem:
generic vdso library, library code, the rest · Maintainers:
Andy Lutomirski, Thomas Gleixner, Vincenzo Frascino, Andrew Morton, Linus Torvalds
__arch_get_hw_counter() returns the current value of the counter if the counter is valid or a negative number if the counter is not valid. This is suboptimal because the validity is checked twice: once before reading the counter and once after reading the counter. Optionaly split the verification out of __arch_get_hw_counter() by providing an optional __arch_is_hw_counter_valid() function. Signed-off-by: Christophe Leroy <redacted> --- lib/vdso/gettimeofday.c | 7 +++++++ 1 file changed, 7 insertions(+)
diff --git a/lib/vdso/gettimeofday.c b/lib/vdso/gettimeofday.c
index ea1a55507af5..001f6329e846 100644
--- a/lib/vdso/gettimeofday.c
+++ b/lib/vdso/gettimeofday.c@@ -67,11 +67,18 @@ static __always_inline int do_hres(const struct vdso_data *vd, clockid_t clk, do { seq = vdso_read_begin(vd); +#ifdef __arch_is_hw_counter_valid + if (unlikely(!__arch_is_hw_counter_valid(vd->clock_mode))) + return -1; +#endif + cycles = __arch_get_hw_counter(vd->clock_mode); ns = vdso_ts->nsec; last = vd->cycle_last; +#ifndef __arch_is_hw_counter_valid if (unlikely((s64)cycles < 0)) return -1; +#endif ns += vdso_calc_delta(cycles, last, vd->mask, vd->mult); nsec = ns >> vd->shift;
--
2.13.3