Date: Wed, 20 Dec 2017 01:03:34 +0000 (UTC) From: Stephen Hurd <shurd@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r327013 - head/sys/net Message-ID: <201712200103.vBK13YgS025896@repo.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: shurd Date: Wed Dec 20 01:03:34 2017 New Revision: 327013 URL: https://svnweb.freebsd.org/changeset/base/327013 Log: Support attaching tx queues to cpus This will attempt to use a different thread/core on the same L2 cache when possible, or use the same cpu as the rx thread when not. If SMP isn't enabled, don't go looking for cores to use. This is mostly useful when using shared TX/RX queues. Reviewed by: sbruno Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D12446 Modified: head/sys/net/iflib.c Modified: head/sys/net/iflib.c ============================================================================== --- head/sys/net/iflib.c Wed Dec 20 00:49:08 2017 (r327012) +++ head/sys/net/iflib.c Wed Dec 20 01:03:34 2017 (r327013) @@ -31,6 +31,7 @@ __FBSDID("$FreeBSD$"); #include "opt_inet.h" #include "opt_inet6.h" #include "opt_acpi.h" +#include "opt_sched.h" #include <sys/param.h> #include <sys/types.h> @@ -5044,25 +5045,136 @@ iflib_irq_alloc(if_ctx_t ctx, if_irq_t irq, int rid, return (_iflib_irq_alloc(ctx, irq, rid, filter, handler, arg, name)); } +#ifdef SMP static int -find_nth(if_ctx_t ctx, cpuset_t *cpus, int qid) +find_nth(if_ctx_t ctx, int qid) { + cpuset_t cpus; int i, cpuid, eqid, count; - CPU_COPY(&ctx->ifc_cpus, cpus); - count = CPU_COUNT(cpus); + CPU_COPY(&ctx->ifc_cpus, &cpus); + count = CPU_COUNT(&cpus); eqid = qid % count; /* clear up to the qid'th bit */ for (i = 0; i < eqid; i++) { - cpuid = CPU_FFS(cpus); + cpuid = CPU_FFS(&cpus); MPASS(cpuid != 0); - CPU_CLR(cpuid-1, cpus); + CPU_CLR(cpuid-1, &cpus); } - cpuid = CPU_FFS(cpus); + cpuid = CPU_FFS(&cpus); MPASS(cpuid != 0); return (cpuid-1); } +#ifdef SCHED_ULE +extern struct cpu_group *cpu_top; /* CPU topology */ + +static int +find_child_with_core(int cpu, struct cpu_group *grp) +{ + int i; + + if (grp->cg_children == 0) + return -1; + + MPASS(grp->cg_child); + for (i = 0; i < grp->cg_children; i++) { + if (CPU_ISSET(cpu, &grp->cg_child[i].cg_mask)) + return i; + } + + return -1; +} + +/* + * Find the nth thread on the specified core + */ +static int +find_thread(int cpu, int thread_num) +{ + struct cpu_group *grp; + int i; + cpuset_t cs; + + grp = cpu_top; + if (grp == NULL) + return cpu; + i = 0; + while ((i = find_child_with_core(cpu, grp)) != -1) { + /* If the child only has one cpu, don't descend */ + if (grp->cg_child[i].cg_count <= 1) + break; + grp = &grp->cg_child[i]; + } + + /* If they don't share at least an L2 cache, use the same CPU */ + if (grp->cg_level > CG_SHARE_L2 || grp->cg_level == CG_SHARE_NONE) + return cpu; + + /* Now pick one */ + CPU_COPY(&grp->cg_mask, &cs); + for (i = thread_num % grp->cg_count; i > 0; i--) { + MPASS(CPU_FFS(&cs)); + CPU_CLR(CPU_FFS(&cs) - 1, &cs); + } + MPASS(CPU_FFS(&cs)); + return CPU_FFS(&cs) - 1; +} +#else +static int +find_thread(int cpu, int thread_num __unused) +{ + return cpu_id +} +#endif + +static int +get_thread_num(if_ctx_t ctx, iflib_intr_type_t type, int qid) +{ + switch (type) { + case IFLIB_INTR_TX: + /* TX queues get threads on the same core as the corresponding RX queue */ + /* XXX handle multiple RX threads per core and more than two threads per core */ + return qid / CPU_COUNT(&ctx->ifc_cpus) + 1; + case IFLIB_INTR_RX: + case IFLIB_INTR_RXTX: + /* RX queues get the first thread on their core */ + return qid / CPU_COUNT(&ctx->ifc_cpus); + default: + return -1; + } +} +#else +#define get_thread_num(ctx, type, qid) CPU_FIRST() +#define find_thread(cpuid, tid) CPU_FIRST() +#define find_nth(ctx, gid) CPU_FIRST() +#endif + +/* Just to avoid copy/paste */ +static inline int +iflib_irq_set_affinity(if_ctx_t ctx, int irq, iflib_intr_type_t type, int qid, + struct grouptask *gtask, struct taskqgroup *tqg, void *uniq, char *name) +{ + int cpuid; + int err, tid; + + cpuid = find_nth(ctx, qid); + tid = get_thread_num(ctx, type, qid); + MPASS(tid >= 0); + cpuid = find_thread(cpuid, tid); + err = taskqgroup_attach_cpu(tqg, gtask, uniq, cpuid, irq, name); + if (err) { + device_printf(ctx->ifc_dev, "taskqgroup_attach_cpu failed %d\n", err); + return (err); + } +#ifdef notyet + if (cpuid > ctx->ifc_cpuid_highest) + ctx->ifc_cpuid_highest = cpuid; +#endif + MPASS(gtask->gt_taskqueue != NULL); + return 0; +} + int iflib_irq_alloc_generic(if_ctx_t ctx, if_irq_t irq, int rid, iflib_intr_type_t type, driver_filter_t *filter, @@ -5071,9 +5183,8 @@ iflib_irq_alloc_generic(if_ctx_t ctx, if_irq_t irq, in struct grouptask *gtask; struct taskqgroup *tqg; iflib_filter_info_t info; - cpuset_t cpus; gtask_fn_t *fn; - int tqrid, err, cpuid; + int tqrid, err; driver_filter_t *intr_fast; void *q; @@ -5136,8 +5247,9 @@ iflib_irq_alloc_generic(if_ctx_t ctx, if_irq_t irq, in return (0); if (tqrid != -1) { - cpuid = find_nth(ctx, &cpus, qid); - taskqgroup_attach_cpu(tqg, gtask, q, cpuid, rman_get_start(irq->ii_res), name); + err = iflib_irq_set_affinity(ctx, rman_get_start(irq->ii_res), type, qid, gtask, tqg, q, name); + if (err) + return (err); } else { taskqgroup_attach(tqg, gtask, q, rman_get_start(irq->ii_res), name); } @@ -5153,6 +5265,7 @@ iflib_softirq_alloc_generic(if_ctx_t ctx, if_irq_t irq gtask_fn_t *fn; void *q; int irq_num = -1; + int err; switch (type) { case IFLIB_INTR_TX: @@ -5181,7 +5294,14 @@ iflib_softirq_alloc_generic(if_ctx_t ctx, if_irq_t irq panic("unknown net intr type"); } GROUPTASK_INIT(gtask, 0, fn, q); - taskqgroup_attach(tqg, gtask, q, irq_num, name); + if (irq_num != -1) { + err = iflib_irq_set_affinity(ctx, irq_num, type, qid, gtask, tqg, q, name); + if (err) + taskqgroup_attach(tqg, gtask, q, irq_num, name); + } + else { + taskqgroup_attach(tqg, gtask, q, irq_num, name); + } } void
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201712200103.vBK13YgS025896>