[PATCH v2 3/3] event/cnxk: use WFE in Tx fc wait
Stephen Hemminger
stephen at networkplumber.org
Thu Jun 15 17:28:57 CEST 2023
On Tue, 13 Jun 2023 14:55:48 +0530
<pbhagavatula at marvell.com> wrote:
> static __rte_always_inline void
> cn10k_sso_txq_fc_wait(const struct cn10k_eth_txq *txq)
> {
> +#ifdef RTE_ARCH_ARM64
> + uint64_t space;
> +
> + asm volatile(PLT_CPU_FEATURE_PREAMBLE
> + " ldxr %[space], [%[addr]] \n"
> + " cmp %[adj], %[space] \n"
> + " b.hi .Ldne%= \n"
> + " sevl \n"
> + ".Lrty%=: wfe \n"
> + " ldxr %[space], [%[addr]] \n"
> + " cmp %[adj], %[space] \n"
> + " b.ls .Lrty%= \n"
> + ".Ldne%=: \n"
> + : [space] "=&r"(space)
> + : [adj] "r"(txq->nb_sqb_bufs_adj), [addr] "r"(txq->fc_mem)
> + : "memory");
> +#else
> while ((uint64_t)txq->nb_sqb_bufs_adj <=
> __atomic_load_n(txq->fc_mem, __ATOMIC_RELAXED))
> ;
> +#endif
> }
Rather than introduce assembly here, please extend existing rte_pause
functions and then other drivers could benefit and it would fit
existing WFE usage.
Something like:
static __rte_always_inline void
rte_wait_until_le_32(volatile uint32_t *addr, uint32_t expected,
int memorder);
Direct assembly in drivers is hard to maintain and should be forbidden.
More information about the dev
mailing list