On 11/06/2018 10:23 AM, Aubrey Li wrote:
> +static inline void update_avx_state(struct avx_state *avx)
> +{
> + /*
> + * Check if XGETBV with ECX = 1 supported. XGETBV with ECX = 1
> + * returns the logical-AND of XCR0 and XINUSE. XINUSE is a bitmap
> + * by which the processor tracks the status of various components.
> + */
> + if (!use_xgetbv1()) {
> + avx->state = 0;
> + return;
> + }
> + /*
> + * XINUSE is dynamic to track component state because VZEROUPPER
> + * happens on every function end and reset the bitmap to the
> + * initial configuration.
> + *
> + * State decay is introduced to solve the race condition between
> + * context switch and a function end. State is aggressively set
> + * once it's detected but need to be cleared by decay 3 context
> + * switches
> + */
> + if (xgetbv(XINUSE_STATE_BITMAP_INDEX) & XFEATURE_MASK_Hi16_ZMM) {
> + avx->state = 1;
> + avx->decay_count = AVX_STATE_DECAY_COUNT;
> + } else {
> + if (!avx->decay_count)
Seems like the check should be
if (avx->decay_count)
as we decrement the decay_count if it is non-zero.
> + avx->decay_count--;
> + else
> + avx->state = 0;
> + }
> +}
Tim
