diff options
Diffstat (limited to 'sys')
123 files changed, 1429 insertions, 622 deletions
diff --git a/sys/amd64/include/vmm.h b/sys/amd64/include/vmm.h index 66d8991d36e8..ad67510fecf3 100644 --- a/sys/amd64/include/vmm.h +++ b/sys/amd64/include/vmm.h @@ -237,7 +237,7 @@ extern u_int vm_maxcpu; /* maximum virtual cpus */ int vm_create(const char *name, struct vm **retvm); struct vcpu *vm_alloc_vcpu(struct vm *vm, int vcpuid); void vm_disable_vcpu_creation(struct vm *vm); -void vm_slock_vcpus(struct vm *vm); +void vm_lock_vcpus(struct vm *vm); void vm_unlock_vcpus(struct vm *vm); void vm_destroy(struct vm *vm); int vm_reinit(struct vm *vm); @@ -362,6 +362,7 @@ enum vcpu_state { }; int vcpu_set_state(struct vcpu *vcpu, enum vcpu_state state, bool from_idle); +int vcpu_set_state_all(struct vm *vm, enum vcpu_state state); enum vcpu_state vcpu_get_state(struct vcpu *vcpu, int *hostcpu); static int __inline diff --git a/sys/amd64/pt/pt.c b/sys/amd64/pt/pt.c index c7b75767680a..6b2296de049c 100644 --- a/sys/amd64/pt/pt.c +++ b/sys/amd64/pt/pt.c @@ -42,15 +42,15 @@ */ #include <sys/systm.h> +#include <sys/bus.h> #include <sys/hwt.h> +#include <sys/interrupt.h> #include <sys/kernel.h> #include <sys/lock.h> #include <sys/malloc.h> #include <sys/module.h> #include <sys/mutex.h> -#include <sys/sdt.h> #include <sys/smp.h> -#include <sys/taskqueue.h> #include <vm/vm.h> #include <vm/vm_page.h> @@ -94,12 +94,7 @@ MALLOC_DEFINE(M_PT, "pt", "Intel Processor Trace"); -SDT_PROVIDER_DEFINE(pt); -SDT_PROBE_DEFINE(pt, , , topa__intr); - -TASKQUEUE_FAST_DEFINE_THREAD(pt); - -static void pt_send_buffer_record(void *arg, int pending __unused); +static void pt_send_buffer_record(void *arg); static int pt_topa_intr(struct trapframe *tf); /* @@ -122,29 +117,24 @@ struct pt_buffer { size_t size; struct mtx lock; /* Lock for fields below. */ vm_offset_t offset; - uint64_t wrap_count; - int curpage; }; struct pt_ctx { int id; struct pt_buffer buf; /* ToPA buffer metadata */ - struct task task; /* ToPA buffer notification task */ struct hwt_context *hwt_ctx; uint8_t *save_area; /* PT XSAVE area */ }; /* PT tracing contexts used for CPU mode. */ static struct pt_ctx *pt_pcpu_ctx; -enum pt_cpu_state { - PT_DISABLED = 0, - PT_STOPPED, - PT_ACTIVE -}; +enum pt_cpu_state { PT_INACTIVE = 0, PT_ACTIVE }; static struct pt_cpu { struct pt_ctx *ctx; /* active PT tracing context */ enum pt_cpu_state state; /* used as part of trace stop protocol */ + void *swi_cookie; /* Software interrupt handler context */ + int in_pcint_handler; } *pt_pcpu; /* @@ -199,31 +189,28 @@ static __inline void pt_update_buffer(struct pt_buffer *buf) { uint64_t reg; - int curpage; + uint64_t offset; /* Update buffer offset. */ reg = rdmsr(MSR_IA32_RTIT_OUTPUT_MASK_PTRS); - curpage = (reg & PT_TOPA_PAGE_MASK) >> PT_TOPA_PAGE_SHIFT; - mtx_lock_spin(&buf->lock); - /* Check if the output wrapped. */ - if (buf->curpage > curpage) - buf->wrap_count++; - buf->curpage = curpage; - buf->offset = reg >> 32; - mtx_unlock_spin(&buf->lock); - - dprintf("%s: wrap_cnt: %lu, curpage: %d, offset: %zu\n", __func__, - buf->wrap_count, buf->curpage, buf->offset); + offset = ((reg & PT_TOPA_PAGE_MASK) >> PT_TOPA_PAGE_SHIFT) * PAGE_SIZE; + offset += (reg >> 32); + + atomic_store_rel_64(&buf->offset, offset); } static __inline void pt_fill_buffer_record(int id, struct pt_buffer *buf, struct hwt_record_entry *rec) { + vm_offset_t offset; + + offset = atomic_load_acq_64(&buf->offset); + rec->record_type = HWT_RECORD_BUFFER; rec->buf_id = id; - rec->curpage = buf->curpage; - rec->offset = buf->offset + (buf->wrap_count * buf->size); + rec->curpage = offset / PAGE_SIZE; + rec->offset = offset & PAGE_MASK; } /* @@ -273,9 +260,9 @@ pt_cpu_start(void *dummy) MPASS(cpu->ctx != NULL); dprintf("%s: curcpu %d\n", __func__, curcpu); + pt_cpu_set_state(curcpu, PT_ACTIVE); load_cr4(rcr4() | CR4_XSAVE); wrmsr(MSR_IA32_RTIT_STATUS, 0); - pt_cpu_set_state(curcpu, PT_ACTIVE); pt_cpu_toggle_local(cpu->ctx->save_area, true); } @@ -291,16 +278,16 @@ pt_cpu_stop(void *dummy) struct pt_cpu *cpu; struct pt_ctx *ctx; - /* Shutdown may occur before PT gets properly configured. */ - if (pt_cpu_get_state(curcpu) == PT_DISABLED) - return; - cpu = &pt_pcpu[curcpu]; ctx = cpu->ctx; - MPASS(ctx != NULL); - dprintf("%s: curcpu %d\n", __func__, curcpu); - pt_cpu_set_state(curcpu, PT_STOPPED); + dprintf("%s: curcpu %d\n", __func__, curcpu); + /* Shutdown may occur before PT gets properly configured. */ + if (ctx == NULL) { + dprintf("%s: missing context on cpu %d; bailing\n", __func__, + curcpu); + return; + } pt_cpu_toggle_local(cpu->ctx->save_area, false); pt_update_buffer(&ctx->buf); } @@ -406,13 +393,11 @@ pt_init_ctx(struct pt_ctx *pt_ctx, struct hwt_vm *vm, int ctx_id) return (ENOMEM); dprintf("%s: preparing ToPA buffer\n", __func__); if (pt_topa_prepare(pt_ctx, vm) != 0) { - dprintf("%s: failed to prepare ToPA buffer\n", __func__); free(pt_ctx->save_area, M_PT); return (ENOMEM); } pt_ctx->id = ctx_id; - TASK_INIT(&pt_ctx->task, 0, pt_send_buffer_record, pt_ctx); return (0); } @@ -426,7 +411,6 @@ pt_deinit_ctx(struct pt_ctx *pt_ctx) if (pt_ctx->save_area != NULL) free(pt_ctx->save_area, M_PT); memset(pt_ctx, 0, sizeof(*pt_ctx)); - pt_ctx->buf.topa_hw = NULL; } /* @@ -519,7 +503,6 @@ pt_backend_configure(struct hwt_context *ctx, int cpu_id, int thread_id) XSTATE_XCOMP_BV_COMPACT; pt_ext->rtit_ctl |= RTIT_CTL_TRACEEN; pt_pcpu[cpu_id].ctx = pt_ctx; - pt_cpu_set_state(cpu_id, PT_STOPPED); return (0); } @@ -549,12 +532,19 @@ pt_backend_disable(struct hwt_context *ctx, int cpu_id) if (ctx->mode == HWT_MODE_CPU) return; - KASSERT(curcpu == cpu_id, ("%s: attempting to disable PT on another cpu", __func__)); + + cpu = &pt_pcpu[cpu_id]; + + dprintf("%s: waiting for cpu %d to exit interrupt handler\n", __func__, + cpu_id); + pt_cpu_set_state(cpu_id, PT_INACTIVE); + while (atomic_cmpset_int(&cpu->in_pcint_handler, 1, 0)) + ; + pt_cpu_stop(NULL); CPU_CLR(cpu_id, &ctx->cpu_map); - cpu = &pt_pcpu[cpu_id]; cpu->ctx = NULL; } @@ -564,14 +554,14 @@ pt_backend_disable(struct hwt_context *ctx, int cpu_id) static int pt_backend_enable_smp(struct hwt_context *ctx) { - dprintf("%s\n", __func__); + + KASSERT(ctx->mode == HWT_MODE_CPU, + ("%s: should only be used for CPU mode", __func__)); if (ctx->mode == HWT_MODE_CPU && atomic_swap_32(&cpu_mode_ctr, 1) != 0) return (-1); - KASSERT(ctx->mode == HWT_MODE_CPU, - ("%s: should only be used for CPU mode", __func__)); smp_rendezvous_cpus(ctx->cpu_map, NULL, pt_cpu_start, NULL, NULL); return (0); @@ -583,6 +573,7 @@ pt_backend_enable_smp(struct hwt_context *ctx) static int pt_backend_disable_smp(struct hwt_context *ctx) { + struct pt_cpu *cpu; dprintf("%s\n", __func__); if (ctx->mode == HWT_MODE_CPU && @@ -593,6 +584,14 @@ pt_backend_disable_smp(struct hwt_context *ctx) dprintf("%s: empty cpu map\n", __func__); return (-1); } + CPU_FOREACH_ISSET(cpu_id, &ctx->cpu_map) { + cpu = &pt_pcpu[cpu_id]; + dprintf("%s: waiting for cpu %d to exit interrupt handler\n", + __func__, cpu_id); + pt_cpu_set_state(cpu_id, PT_INACTIVE); + while (atomic_cmpset_int(&cpu->in_pcint_handler, 1, 0)) + ; + } smp_rendezvous_cpus(ctx->cpu_map, NULL, pt_cpu_stop, NULL, NULL); return (0); @@ -611,13 +610,13 @@ pt_backend_init(struct hwt_context *ctx) int error; dprintf("%s\n", __func__); - if (ctx->mode == HWT_MODE_CPU) { - TAILQ_FOREACH(hwt_cpu, &ctx->cpus, next) { - error = pt_init_ctx(&pt_pcpu_ctx[hwt_cpu->cpu_id], - hwt_cpu->vm, hwt_cpu->cpu_id); - if (error) - return (error); - } + if (ctx->mode != HWT_MODE_CPU) + return (0); + TAILQ_FOREACH(hwt_cpu, &ctx->cpus, next) { + error = pt_init_ctx(&pt_pcpu_ctx[hwt_cpu->cpu_id], hwt_cpu->vm, + hwt_cpu->cpu_id); + if (error) + return (error); } return (0); @@ -647,20 +646,16 @@ pt_backend_deinit(struct hwt_context *ctx) pt_deinit_ctx(pt_ctx); } } else { - CPU_FOREACH(cpu_id) { - if (!CPU_ISSET(cpu_id, &ctx->cpu_map)) + CPU_FOREACH_ISSET(cpu_id, &ctx->cpu_map) { + if (pt_pcpu[cpu_id].ctx == NULL) continue; - if (pt_pcpu[cpu_id].ctx != NULL) { - KASSERT(pt_pcpu[cpu_id].ctx == - &pt_pcpu_ctx[cpu_id], - ("%s: CPU mode tracing with non-cpu mode PT" - "context active", - __func__)); - pt_pcpu[cpu_id].ctx = NULL; - } - pt_ctx = &pt_pcpu_ctx[cpu_id]; - pt_deinit_ctx(pt_ctx); - memset(&pt_pcpu[cpu_id], 0, sizeof(struct pt_cpu)); + KASSERT(pt_pcpu[cpu_id].ctx == &pt_pcpu_ctx[cpu_id], + ("%s: CPU mode tracing with non-cpu mode PT" + "context active", + __func__)); + pt_deinit_ctx(pt_pcpu[cpu_id].ctx); + pt_pcpu[cpu_id].ctx = NULL; + atomic_set_int(&pt_pcpu[cpu_id].in_pcint_handler, 0); } } @@ -675,15 +670,15 @@ pt_backend_read(struct hwt_vm *vm, int *curpage, vm_offset_t *curpage_offset, uint64_t *data) { struct pt_buffer *buf; + uint64_t offset; if (vm->ctx->mode == HWT_MODE_THREAD) buf = &((struct pt_ctx *)vm->thr->private)->buf; else buf = &pt_pcpu[vm->cpu->cpu_id].ctx->buf; - mtx_lock_spin(&buf->lock); - *curpage = buf->curpage; - *curpage_offset = buf->offset + (buf->wrap_count * vm->ctx->bufsize); - mtx_unlock_spin(&buf->lock); + offset = atomic_load_acq_64(&buf->offset); + *curpage = offset / PAGE_SIZE; + *curpage_offset = offset & PAGE_MASK; return (0); } @@ -762,15 +757,13 @@ static struct hwt_backend backend = { * Used as a taskqueue routine from the ToPA interrupt handler. */ static void -pt_send_buffer_record(void *arg, int pending __unused) +pt_send_buffer_record(void *arg) { + struct pt_cpu *cpu = (struct pt_cpu *)arg; struct hwt_record_entry record; - struct pt_ctx *ctx = (struct pt_ctx *)arg; - /* Prepare buffer record. */ - mtx_lock_spin(&ctx->buf.lock); + struct pt_ctx *ctx = cpu->ctx; pt_fill_buffer_record(ctx->id, &ctx->buf, &record); - mtx_unlock_spin(&ctx->buf.lock); hwt_record_ctx(ctx->hwt_ctx, &record, M_ZERO | M_NOWAIT); } static void @@ -795,36 +788,40 @@ static int pt_topa_intr(struct trapframe *tf) { struct pt_buffer *buf; + struct pt_cpu *cpu; struct pt_ctx *ctx; uint64_t reg; - SDT_PROBE0(pt, , , topa__intr); - - if (pt_cpu_get_state(curcpu) != PT_ACTIVE) { - return (0); - } + cpu = &pt_pcpu[curcpu]; reg = rdmsr(MSR_IA_GLOBAL_STATUS); if ((reg & GLOBAL_STATUS_FLAG_TRACETOPAPMI) == 0) { - /* ACK spurious or leftover interrupt. */ pt_topa_status_clear(); + return (0); + } + + if (pt_cpu_get_state(curcpu) != PT_ACTIVE) { return (1); } + atomic_set_int(&cpu->in_pcint_handler, 1); - ctx = pt_pcpu[curcpu].ctx; + ctx = cpu->ctx; + KASSERT(ctx != NULL, + ("%s: cpu %d: ToPA PMI interrupt without an active context", + __func__, curcpu)); buf = &ctx->buf; KASSERT(buf->topa_hw != NULL, - ("%s: ToPA PMI interrupt with invalid buffer", __func__)); - + ("%s: cpu %d: ToPA PMI interrupt with invalid buffer", __func__, + curcpu)); pt_cpu_toggle_local(ctx->save_area, false); pt_update_buffer(buf); pt_topa_status_clear(); - taskqueue_enqueue_flags(taskqueue_pt, &ctx->task, - TASKQUEUE_FAIL_IF_PENDING); if (pt_cpu_get_state(curcpu) == PT_ACTIVE) { + swi_sched(cpu->swi_cookie, SWI_FROMNMI); pt_cpu_toggle_local(ctx->save_area, true); lapic_reenable_pcint(); } + atomic_set_int(&cpu->in_pcint_handler, 0); return (1); } @@ -839,7 +836,7 @@ static int pt_init(void) { u_int cp[4]; - int error; + int error, i; dprintf("pt: Enumerating part 1\n"); cpuid_count(CPUID_PT_LEAF, 0, cp); @@ -869,20 +866,38 @@ pt_init(void) pt_pcpu_ctx = mallocarray(mp_ncpus, sizeof(struct pt_ctx), M_PT, M_ZERO | M_WAITOK); + for (i = 0; i < mp_ncpus; i++) { + error = swi_add(&clk_intr_event, "pt", pt_send_buffer_record, + &pt_pcpu[i], SWI_CLOCK, INTR_MPSAFE, + &pt_pcpu[i].swi_cookie); + if (error != 0) { + dprintf( + "%s: failed to add interrupt handler for cpu: %d\n", + __func__, error); + goto err; + } + } + nmi_register_handler(pt_topa_intr); - if (!lapic_enable_pcint()) { - nmi_remove_handler(pt_topa_intr); - hwt_backend_unregister(&backend); - free(pt_pcpu, M_PT); - free(pt_pcpu_ctx, M_PT); - pt_pcpu = NULL; - pt_pcpu_ctx = NULL; + if (lapic_enable_pcint()) { + initialized = true; + return (0); + } else printf("pt: failed to setup interrupt line\n"); - return (error); +err: + nmi_remove_handler(pt_topa_intr); + hwt_backend_unregister(&backend); + + for (i = 0; i < mp_ncpus; i++) { + if (pt_pcpu[i].swi_cookie != 0) + swi_remove(pt_pcpu[i].swi_cookie); } - initialized = true; + free(pt_pcpu, M_PT); + free(pt_pcpu_ctx, M_PT); + pt_pcpu = NULL; + pt_pcpu_ctx = NULL; - return (0); + return (error); } /* @@ -941,14 +956,24 @@ pt_supported(void) static void pt_deinit(void) { + int i; + struct pt_cpu *cpu; + if (!initialized) return; nmi_remove_handler(pt_topa_intr); lapic_disable_pcint(); hwt_backend_unregister(&backend); + + for (i = 0; i < mp_ncpus; i++) { + cpu = &pt_pcpu[i]; + swi_remove(cpu->swi_cookie); + } + free(pt_pcpu, M_PT); free(pt_pcpu_ctx, M_PT); pt_pcpu = NULL; + pt_pcpu_ctx = NULL; initialized = false; } diff --git a/sys/amd64/vmm/vmm.c b/sys/amd64/vmm/vmm.c index 2ac076551165..f7c59847140b 100644 --- a/sys/amd64/vmm/vmm.c +++ b/sys/amd64/vmm/vmm.c @@ -562,9 +562,9 @@ vm_alloc_vcpu(struct vm *vm, int vcpuid) } void -vm_slock_vcpus(struct vm *vm) +vm_lock_vcpus(struct vm *vm) { - sx_slock(&vm->vcpus_init_lock); + sx_xlock(&vm->vcpus_init_lock); } void @@ -990,6 +990,54 @@ save_guest_fpustate(struct vcpu *vcpu) static VMM_STAT(VCPU_IDLE_TICKS, "number of ticks vcpu was idle"); +/* + * Invoke the rendezvous function on the specified vcpu if applicable. Return + * true if the rendezvous is finished, false otherwise. + */ +static bool +vm_rendezvous(struct vcpu *vcpu) +{ + struct vm *vm = vcpu->vm; + int vcpuid; + + mtx_assert(&vcpu->vm->rendezvous_mtx, MA_OWNED); + KASSERT(vcpu->vm->rendezvous_func != NULL, + ("vm_rendezvous: no rendezvous pending")); + + /* 'rendezvous_req_cpus' must be a subset of 'active_cpus' */ + CPU_AND(&vm->rendezvous_req_cpus, &vm->rendezvous_req_cpus, + &vm->active_cpus); + + vcpuid = vcpu->vcpuid; + if (CPU_ISSET(vcpuid, &vm->rendezvous_req_cpus) && + !CPU_ISSET(vcpuid, &vm->rendezvous_done_cpus)) { + VMM_CTR0(vcpu, "Calling rendezvous func"); + (*vm->rendezvous_func)(vcpu, vm->rendezvous_arg); + CPU_SET(vcpuid, &vm->rendezvous_done_cpus); + } + if (CPU_CMP(&vm->rendezvous_req_cpus, + &vm->rendezvous_done_cpus) == 0) { + VMM_CTR0(vcpu, "Rendezvous completed"); + CPU_ZERO(&vm->rendezvous_req_cpus); + vm->rendezvous_func = NULL; + wakeup(&vm->rendezvous_func); + return (true); + } + return (false); +} + +static void +vcpu_wait_idle(struct vcpu *vcpu) +{ + KASSERT(vcpu->state != VCPU_IDLE, ("vcpu already idle")); + + vcpu->reqidle = 1; + vcpu_notify_event_locked(vcpu, false); + VMM_CTR1(vcpu, "vcpu state change from %s to " + "idle requested", vcpu_state2str(vcpu->state)); + msleep_spin(&vcpu->state, &vcpu->mtx, "vmstat", hz); +} + static int vcpu_set_state_locked(struct vcpu *vcpu, enum vcpu_state newstate, bool from_idle) @@ -1004,13 +1052,8 @@ vcpu_set_state_locked(struct vcpu *vcpu, enum vcpu_state newstate, * ioctl() operating on a vcpu at any point. */ if (from_idle) { - while (vcpu->state != VCPU_IDLE) { - vcpu->reqidle = 1; - vcpu_notify_event_locked(vcpu, false); - VMM_CTR1(vcpu, "vcpu state change from %s to " - "idle requested", vcpu_state2str(vcpu->state)); - msleep_spin(&vcpu->state, &vcpu->mtx, "vmstat", hz); - } + while (vcpu->state != VCPU_IDLE) + vcpu_wait_idle(vcpu); } else { KASSERT(vcpu->state != VCPU_IDLE, ("invalid transition from " "vcpu idle state")); @@ -1062,6 +1105,95 @@ vcpu_set_state_locked(struct vcpu *vcpu, enum vcpu_state newstate, return (0); } +/* + * Try to lock all of the vCPUs in the VM while taking care to avoid deadlocks + * with vm_smp_rendezvous(). + * + * The complexity here suggests that the rendezvous mechanism needs a rethink. + */ +int +vcpu_set_state_all(struct vm *vm, enum vcpu_state newstate) +{ + cpuset_t locked; + struct vcpu *vcpu; + int error, i; + uint16_t maxcpus; + + KASSERT(newstate != VCPU_IDLE, + ("vcpu_set_state_all: invalid target state %d", newstate)); + + error = 0; + CPU_ZERO(&locked); + maxcpus = vm->maxcpus; + + mtx_lock(&vm->rendezvous_mtx); +restart: + if (vm->rendezvous_func != NULL) { + /* + * If we have a pending rendezvous, then the initiator may be + * blocked waiting for other vCPUs to execute the callback. The + * current thread may be a vCPU thread so we must not block + * waiting for the initiator, otherwise we get a deadlock. + * Thus, execute the callback on behalf of any idle vCPUs. + */ + for (i = 0; i < maxcpus; i++) { + vcpu = vm_vcpu(vm, i); + if (vcpu == NULL) + continue; + vcpu_lock(vcpu); + if (vcpu->state == VCPU_IDLE) { + (void)vcpu_set_state_locked(vcpu, VCPU_FROZEN, + true); + CPU_SET(i, &locked); + } + if (CPU_ISSET(i, &locked)) { + /* + * We can safely execute the callback on this + * vCPU's behalf. + */ + vcpu_unlock(vcpu); + (void)vm_rendezvous(vcpu); + vcpu_lock(vcpu); + } + vcpu_unlock(vcpu); + } + } + + /* + * Now wait for remaining vCPUs to become idle. This may include the + * initiator of a rendezvous that is currently blocked on the rendezvous + * mutex. + */ + CPU_FOREACH_ISCLR(i, &locked) { + if (i >= maxcpus) + break; + vcpu = vm_vcpu(vm, i); + if (vcpu == NULL) + continue; + vcpu_lock(vcpu); + while (vcpu->state != VCPU_IDLE) { + mtx_unlock(&vm->rendezvous_mtx); + vcpu_wait_idle(vcpu); + vcpu_unlock(vcpu); + mtx_lock(&vm->rendezvous_mtx); + if (vm->rendezvous_func != NULL) + goto restart; + vcpu_lock(vcpu); + } + error = vcpu_set_state_locked(vcpu, newstate, true); + vcpu_unlock(vcpu); + if (error != 0) { + /* Roll back state changes. */ + CPU_FOREACH_ISSET(i, &locked) + (void)vcpu_set_state(vcpu, VCPU_IDLE, false); + break; + } + CPU_SET(i, &locked); + } + mtx_unlock(&vm->rendezvous_mtx); + return (error); +} + static void vcpu_require_state(struct vcpu *vcpu, enum vcpu_state newstate) { @@ -1083,36 +1215,23 @@ vcpu_require_state_locked(struct vcpu *vcpu, enum vcpu_state newstate) static int vm_handle_rendezvous(struct vcpu *vcpu) { - struct vm *vm = vcpu->vm; + struct vm *vm; struct thread *td; - int error, vcpuid; - error = 0; - vcpuid = vcpu->vcpuid; td = curthread; + vm = vcpu->vm; + mtx_lock(&vm->rendezvous_mtx); while (vm->rendezvous_func != NULL) { - /* 'rendezvous_req_cpus' must be a subset of 'active_cpus' */ - CPU_AND(&vm->rendezvous_req_cpus, &vm->rendezvous_req_cpus, &vm->active_cpus); - - if (CPU_ISSET(vcpuid, &vm->rendezvous_req_cpus) && - !CPU_ISSET(vcpuid, &vm->rendezvous_done_cpus)) { - VMM_CTR0(vcpu, "Calling rendezvous func"); - (*vm->rendezvous_func)(vcpu, vm->rendezvous_arg); - CPU_SET(vcpuid, &vm->rendezvous_done_cpus); - } - if (CPU_CMP(&vm->rendezvous_req_cpus, - &vm->rendezvous_done_cpus) == 0) { - VMM_CTR0(vcpu, "Rendezvous completed"); - CPU_ZERO(&vm->rendezvous_req_cpus); - vm->rendezvous_func = NULL; - wakeup(&vm->rendezvous_func); + if (vm_rendezvous(vcpu)) break; - } + VMM_CTR0(vcpu, "Wait for rendezvous completion"); mtx_sleep(&vm->rendezvous_func, &vm->rendezvous_mtx, 0, "vmrndv", hz); if (td_ast_pending(td, TDA_SUSPEND)) { + int error; + mtx_unlock(&vm->rendezvous_mtx); error = thread_check_susp(td, true); if (error != 0) diff --git a/sys/arm/allwinner/aw_sid.c b/sys/arm/allwinner/aw_sid.c index ba5faca33c5e..932c2f189e51 100644 --- a/sys/arm/allwinner/aw_sid.c +++ b/sys/arm/allwinner/aw_sid.c @@ -297,7 +297,7 @@ aw_sid_attach(device_t dev) /* Register ourself so device can resolve who we are */ OF_device_register_xref(OF_xref_from_node(node), dev); - for (i = 0; i < sc->sid_conf->nfuses ;i++) {\ + for (i = 0; i < sc->sid_conf->nfuses; i++) { SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), SYSCTL_CHILDREN(device_get_sysctl_tree(dev)), OID_AUTO, sc->sid_conf->efuses[i].name, diff --git a/sys/arm/arm/elf_machdep.c b/sys/arm/arm/elf_machdep.c index ea6437f320ce..881c4fcff475 100644 --- a/sys/arm/arm/elf_machdep.c +++ b/sys/arm/arm/elf_machdep.c @@ -106,7 +106,7 @@ struct sysentvec elf32_freebsd_sysvec = { }; INIT_SYSENTVEC(elf32_sysvec, &elf32_freebsd_sysvec); -static Elf32_Brandinfo freebsd_brand_info = { +static const Elf32_Brandinfo freebsd_brand_info = { .brand = ELFOSABI_FREEBSD, .machine = EM_ARM, .compat_3_brand = "FreeBSD", @@ -118,7 +118,7 @@ static Elf32_Brandinfo freebsd_brand_info = { .header_supported= elf32_arm_abi_supported, }; -SYSINIT(elf32, SI_SUB_EXEC, SI_ORDER_FIRST, +C_SYSINIT(elf32, SI_SUB_EXEC, SI_ORDER_FIRST, (sysinit_cfunc_t) elf32_insert_brand_entry, &freebsd_brand_info); diff --git a/sys/arm/ti/ti_pruss.c b/sys/arm/ti/ti_pruss.c index 4e9f2022240c..bae1de9f2ddf 100644 --- a/sys/arm/ti/ti_pruss.c +++ b/sys/arm/ti/ti_pruss.c @@ -793,6 +793,7 @@ static const struct filterops ti_pruss_kq_read = { .f_isfd = 1, .f_detach = ti_pruss_irq_kqread_detach, .f_event = ti_pruss_irq_kqevent, + .f_copy = knote_triv_copy, }; static void diff --git a/sys/arm64/arm64/cpu_errata.c b/sys/arm64/arm64/cpu_errata.c index 989924bc0567..b876703a2a15 100644 --- a/sys/arm64/arm64/cpu_errata.c +++ b/sys/arm64/arm64/cpu_errata.c @@ -52,56 +52,11 @@ struct cpu_quirks { u_int flags; }; -static enum { - SSBD_FORCE_ON, - SSBD_FORCE_OFF, - SSBD_KERNEL, -} ssbd_method = SSBD_KERNEL; - -static cpu_quirk_install install_psci_bp_hardening; -static cpu_quirk_install install_ssbd_workaround; static cpu_quirk_install install_thunderx_bcast_tlbi_workaround; static struct cpu_quirks cpu_quirks[] = { { .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, - .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A57,0,0), - .quirk_install = install_psci_bp_hardening, - .flags = CPU_QUIRK_POST_DEVICE, - }, - { - .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, - .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A72,0,0), - .quirk_install = install_psci_bp_hardening, - .flags = CPU_QUIRK_POST_DEVICE, - }, - { - .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, - .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A73,0,0), - .quirk_install = install_psci_bp_hardening, - .flags = CPU_QUIRK_POST_DEVICE, - }, - { - .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, - .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A75,0,0), - .quirk_install = install_psci_bp_hardening, - .flags = CPU_QUIRK_POST_DEVICE, - }, - { - .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, - .midr_value = - CPU_ID_RAW(CPU_IMPL_CAVIUM, CPU_PART_THUNDERX2, 0,0), - .quirk_install = install_psci_bp_hardening, - .flags = CPU_QUIRK_POST_DEVICE, - }, - { - .midr_mask = 0, - .midr_value = 0, - .quirk_install = install_ssbd_workaround, - .flags = CPU_QUIRK_POST_DEVICE, - }, - { - .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, .midr_value = CPU_ID_RAW(CPU_IMPL_CAVIUM, CPU_PART_THUNDERX, 0, 0), .quirk_install = install_thunderx_bcast_tlbi_workaround, @@ -114,57 +69,6 @@ static struct cpu_quirks cpu_quirks[] = { }, }; -static void -install_psci_bp_hardening(void) -{ - /* SMCCC depends on PSCI. If PSCI is missing so is SMCCC */ - if (!psci_present) - return; - - if (smccc_arch_features(SMCCC_ARCH_WORKAROUND_1) != SMCCC_RET_SUCCESS) - return; - - PCPU_SET(bp_harden, smccc_arch_workaround_1); -} - -static void -install_ssbd_workaround(void) -{ - char *env; - - if (PCPU_GET(cpuid) == 0) { - env = kern_getenv("kern.cfg.ssbd"); - if (env != NULL) { - if (strcmp(env, "force-on") == 0) { - ssbd_method = SSBD_FORCE_ON; - } else if (strcmp(env, "force-off") == 0) { - ssbd_method = SSBD_FORCE_OFF; - } - } - } - - /* SMCCC depends on PSCI. If PSCI is missing so is SMCCC */ - if (!psci_present) - return; - - /* Enable the workaround on this CPU if it's enabled in the firmware */ - if (smccc_arch_features(SMCCC_ARCH_WORKAROUND_2) != SMCCC_RET_SUCCESS) - return; - - switch(ssbd_method) { - case SSBD_FORCE_ON: - smccc_arch_workaround_2(1); - break; - case SSBD_FORCE_OFF: - smccc_arch_workaround_2(0); - break; - case SSBD_KERNEL: - default: - PCPU_SET(ssbd, smccc_arch_workaround_2); - break; - } -} - /* * Workaround Cavium erratum 27456. * diff --git a/sys/arm64/arm64/spec_workaround.c b/sys/arm64/arm64/spec_workaround.c new file mode 100644 index 000000000000..7f4f86cdb48c --- /dev/null +++ b/sys/arm64/arm64/spec_workaround.c @@ -0,0 +1,166 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause + * + * Copyright (c) 2025 Arm Ltd + * Copyright (c) 2018 Andrew Turner + * + * This software was developed by SRI International and the University of + * Cambridge Computer Laboratory under DARPA/AFRL contract FA8750-10-C-0237 + * ("CTSRD"), as part of the DARPA CRASH research programme. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include <sys/param.h> +#include <sys/kernel.h> +#include <sys/pcpu.h> +#include <sys/systm.h> + +#include <machine/cpu.h> +#include <machine/cpu_feat.h> + +#include <dev/psci/psci.h> +#include <dev/psci/smccc.h> + +static enum { + SSBD_FORCE_ON, + SSBD_FORCE_OFF, + SSBD_KERNEL, +} ssbd_method = SSBD_KERNEL; + +struct psci_bp_hardening_impl { + u_int midr_mask; + u_int midr_value; +}; + +static struct psci_bp_hardening_impl psci_bp_hardening_impl[] = { + { + .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, + .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A57,0,0), + }, + { + .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, + .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A72,0,0), + }, + { + .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, + .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A73,0,0), + }, + { + .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, + .midr_value = CPU_ID_RAW(CPU_IMPL_ARM, CPU_PART_CORTEX_A75,0,0), + }, + { + .midr_mask = CPU_IMPL_MASK | CPU_PART_MASK, + .midr_value = + CPU_ID_RAW(CPU_IMPL_CAVIUM, CPU_PART_THUNDERX2, 0,0), + } +}; + +static cpu_feat_en +psci_bp_hardening_check(const struct cpu_feat *feat __unused, u_int midr) +{ + size_t i; + + for (i = 0; i < nitems(psci_bp_hardening_impl); i++) { + if ((midr & psci_bp_hardening_impl[i].midr_mask) == + psci_bp_hardening_impl[i].midr_value) { + /* SMCCC depends on PSCI. If PSCI is missing so is SMCCC */ + if (!psci_present) + return (FEAT_ALWAYS_DISABLE); + + if (smccc_arch_features(SMCCC_ARCH_WORKAROUND_1) != + SMCCC_RET_SUCCESS) + return (FEAT_ALWAYS_DISABLE); + + return (FEAT_DEFAULT_ENABLE); + } + } + + return (FEAT_ALWAYS_DISABLE); +} + +static bool +psci_bp_hardening_enable(const struct cpu_feat *feat __unused, + cpu_feat_errata errata_status __unused, u_int *errata_list __unused, + u_int errata_count __unused) +{ + PCPU_SET(bp_harden, smccc_arch_workaround_1); + + return (true); +} + +CPU_FEAT(feat_csv2_missing, "Branch Predictor Hardening", + psci_bp_hardening_check, NULL, psci_bp_hardening_enable, NULL, + CPU_FEAT_AFTER_DEV | CPU_FEAT_PER_CPU); + +static cpu_feat_en +ssbd_workaround_check(const struct cpu_feat *feat __unused, u_int midr __unused) +{ + char *env; + + if (PCPU_GET(cpuid) == 0) { + env = kern_getenv("kern.cfg.ssbd"); + if (env != NULL) { + if (strcmp(env, "force-on") == 0) { + ssbd_method = SSBD_FORCE_ON; + } else if (strcmp(env, "force-off") == 0) { + ssbd_method = SSBD_FORCE_OFF; + } + } + } + + /* SMCCC depends on PSCI. If PSCI is missing so is SMCCC */ + if (!psci_present) + return (FEAT_ALWAYS_DISABLE); + + /* Enable the workaround on this CPU if it's enabled in the firmware */ + if (smccc_arch_features(SMCCC_ARCH_WORKAROUND_2) != SMCCC_RET_SUCCESS) + return (FEAT_ALWAYS_DISABLE); + + return (FEAT_DEFAULT_ENABLE); +} + +static bool +ssbd_workaround_enable(const struct cpu_feat *feat __unused, + cpu_feat_errata errata_status __unused, u_int *errata_list __unused, + u_int errata_count __unused) +{ + switch(ssbd_method) { + case SSBD_FORCE_ON: + smccc_arch_workaround_2(1); + break; + case SSBD_FORCE_OFF: + smccc_arch_workaround_2(0); + break; + case SSBD_KERNEL: + default: + PCPU_SET(ssbd, smccc_arch_workaround_2); + break; + } + + return (true); +} + +CPU_FEAT(feat_ssbs_missing, "Speculator Store Bypass Disable Workaround", + ssbd_workaround_check, NULL, ssbd_workaround_enable, NULL, + CPU_FEAT_AFTER_DEV | CPU_FEAT_PER_CPU); diff --git a/sys/arm64/include/vmm.h b/sys/arm64/include/vmm.h index 84b286a60b38..696a69669a2a 100644 --- a/sys/arm64/include/vmm.h +++ b/sys/arm64/include/vmm.h @@ -177,7 +177,7 @@ DECLARE_VMMOPS_FUNC(int, restore_tsc, (void *vcpui, uint64_t now)); int vm_create(const char *name, struct vm **retvm); struct vcpu *vm_alloc_vcpu(struct vm *vm, int vcpuid); void vm_disable_vcpu_creation(struct vm *vm); -void vm_slock_vcpus(struct vm *vm); +void vm_lock_vcpus(struct vm *vm); void vm_unlock_vcpus(struct vm *vm); void vm_destroy(struct vm *vm); int vm_reinit(struct vm *vm); diff --git a/sys/arm64/vmm/vmm.c b/sys/arm64/vmm/vmm.c index aeda689f3b1a..bf52dc0fe916 100644 --- a/sys/arm64/vmm/vmm.c +++ b/sys/arm64/vmm/vmm.c @@ -469,9 +469,9 @@ vm_alloc_vcpu(struct vm *vm, int vcpuid) } void -vm_slock_vcpus(struct vm *vm) +vm_lock_vcpus(struct vm *vm) { - sx_slock(&vm->vcpus_init_lock); + sx_xlock(&vm->vcpus_init_lock); } void diff --git a/sys/cam/scsi/scsi_all.c b/sys/cam/scsi/scsi_all.c index 9c097f52d136..fd128e69f1f1 100644 --- a/sys/cam/scsi/scsi_all.c +++ b/sys/cam/scsi/scsi_all.c @@ -686,7 +686,7 @@ scsi_op_desc(uint16_t opcode, struct scsi_inquiry_data *inq_data) opmask = 1 << pd_type; for (j = 0; j < num_tables; j++) { - for (i = 0;i < num_ops[j] && table[j][i].opcode <= opcode; i++){ + for (i = 0; i < num_ops[j] && table[j][i].opcode <= opcode; i++) { if ((table[j][i].opcode == opcode) && ((table[j][i].opmask & opmask) != 0)) return(table[j][i].desc); diff --git a/sys/cam/scsi/scsi_enc_ses.c b/sys/cam/scsi/scsi_enc_ses.c index 435874a9874a..3a362eaf11a4 100644 --- a/sys/cam/scsi/scsi_enc_ses.c +++ b/sys/cam/scsi/scsi_enc_ses.c @@ -2302,7 +2302,7 @@ ses_print_addl_data_sas_type0(char *sesname, struct sbuf *sbp, sbuf_putc(sbp, '\n'); if (addl->proto_data.sasdev_phys == NULL) return; - for (i = 0;i < addl->proto_hdr.sas->base_hdr.num_phys;i++) { + for (i = 0; i < addl->proto_hdr.sas->base_hdr.num_phys; i++) { phy = &addl->proto_data.sasdev_phys[i]; sbuf_printf(sbp, "%s: phy %d:", sesname, i); if (ses_elm_sas_dev_phy_sata_dev(phy)) @@ -2349,7 +2349,7 @@ ses_print_addl_data_sas_type1(char *sesname, struct sbuf *sbp, sbuf_printf(sbp, "Expander: %d phys", num_phys); if (addl->proto_data.sasexp_phys == NULL) return; - for (i = 0;i < num_phys;i++) { + for (i = 0; i < num_phys; i++) { exp_phy = &addl->proto_data.sasexp_phys[i]; sbuf_printf(sbp, "%s: phy %d: connector %d other %d\n", sesname, i, exp_phy->connector_index, @@ -2360,7 +2360,7 @@ ses_print_addl_data_sas_type1(char *sesname, struct sbuf *sbp, sbuf_printf(sbp, "Port: %d phys", num_phys); if (addl->proto_data.sasport_phys == NULL) return; - for (i = 0;i < num_phys;i++) { + for (i = 0; i < num_phys; i++) { port_phy = &addl->proto_data.sasport_phys[i]; sbuf_printf(sbp, "%s: phy %d: id %d connector %d other %d\n", diff --git a/sys/cam/scsi/scsi_pass.c b/sys/cam/scsi/scsi_pass.c index c3587421c176..b44ab866dfe7 100644 --- a/sys/cam/scsi/scsi_pass.c +++ b/sys/cam/scsi/scsi_pass.c @@ -206,7 +206,8 @@ static struct cdevsw pass_cdevsw = { static const struct filterops passread_filtops = { .f_isfd = 1, .f_detach = passreadfiltdetach, - .f_event = passreadfilt + .f_event = passreadfilt, + .f_copy = knote_triv_copy, }; static MALLOC_DEFINE(M_SCSIPASS, "scsi_pass", "scsi passthrough buffers"); diff --git a/sys/cam/scsi/scsi_target.c b/sys/cam/scsi/scsi_target.c index 21c78e35dadc..39ce2bcea8f4 100644 --- a/sys/cam/scsi/scsi_target.c +++ b/sys/cam/scsi/scsi_target.c @@ -108,6 +108,7 @@ static const struct filterops targread_filtops = { .f_isfd = 1, .f_detach = targreadfiltdetach, .f_event = targreadfilt, + .f_copy = knote_triv_copy, }; static struct cdevsw targ_cdevsw = { diff --git a/sys/compat/linprocfs/linprocfs.c b/sys/compat/linprocfs/linprocfs.c index 95b212be1306..7ac48786c77b 100644 --- a/sys/compat/linprocfs/linprocfs.c +++ b/sys/compat/linprocfs/linprocfs.c @@ -2216,6 +2216,67 @@ linprocfs_dosysvipc_shm(PFS_FILL_ARGS) return (0); } +static int +linprocfs_doinotify(const char *sysctl, PFS_FILL_ARGS) +{ + size_t size; + int error, val; + + if (uio->uio_rw == UIO_READ) { + size = sizeof(val); + error = kernel_sysctlbyname(curthread, + __DECONST(void *, sysctl), &val, &size, NULL, 0, 0, 0); + if (error == 0) + sbuf_printf(sb, "%d\n", val); + } else { + char *endp, *newval; + long vall; + + sbuf_trim(sb); + sbuf_finish(sb); + newval = sbuf_data(sb); + vall = strtol(newval, &endp, 10); + if (vall < 0 || vall > INT_MAX || endp == newval || + *endp != '\0') + return (EINVAL); + val = (int)vall; + error = kernel_sysctlbyname(curthread, + __DECONST(void *, sysctl), NULL, NULL, + &val, sizeof(val), 0, 0); + } + return (error); +} + +/* + * Filler function for proc/sys/fs/inotify/max_queued_events + */ +static int +linprocfs_doinotify_max_queued_events(PFS_FILL_ARGS) +{ + return (linprocfs_doinotify("vfs.inotify.max_queued_events", + PFS_FILL_ARGNAMES)); +} + +/* + * Filler function for proc/sys/fs/inotify/max_user_instances + */ +static int +linprocfs_doinotify_max_user_instances(PFS_FILL_ARGS) +{ + return (linprocfs_doinotify("vfs.inotify.max_user_instances", + PFS_FILL_ARGNAMES)); +} + +/* + * Filler function for proc/sys/fs/inotify/max_user_watches + */ +static int +linprocfs_doinotify_max_user_watches(PFS_FILL_ARGS) +{ + return (linprocfs_doinotify("vfs.inotify.max_user_watches", + PFS_FILL_ARGNAMES)); +} + /* * Filler function for proc/sys/fs/mqueue/msg_default */ @@ -2313,9 +2374,7 @@ linprocfs_domqueue_queues_max(PFS_FILL_ARGS) static int linprocfs_init(PFS_INIT_ARGS) { - struct pfs_node *root; - struct pfs_node *dir; - struct pfs_node *sys; + struct pfs_node *dir, *fs, *root, *sys; root = pi->pi_root; @@ -2466,10 +2525,18 @@ linprocfs_init(PFS_INIT_ARGS) NULL, PFS_RD); /* /proc/sys/fs/... */ - pfs_create_dir(sys, &dir, "fs", NULL, NULL, NULL, 0); + pfs_create_dir(sys, &fs, "fs", NULL, NULL, NULL, 0); + + pfs_create_dir(fs, &dir, "inotify", NULL, NULL, NULL, 0); + pfs_create_file(dir, NULL, "max_queued_events", + &linprocfs_doinotify_max_queued_events, NULL, NULL, NULL, PFS_RDWR); + pfs_create_file(dir, NULL, "max_user_instances", + &linprocfs_doinotify_max_user_instances, NULL, NULL, NULL, PFS_RDWR); + pfs_create_file(dir, NULL, "max_user_watches", + &linprocfs_doinotify_max_user_watches, NULL, NULL, NULL, PFS_RDWR); /* /proc/sys/fs/mqueue/... */ - pfs_create_dir(dir, &dir, "mqueue", NULL, NULL, NULL, 0); + pfs_create_dir(fs, &dir, "mqueue", NULL, NULL, NULL, 0); pfs_create_file(dir, NULL, "msg_default", &linprocfs_domqueue_msg_default, NULL, NULL, NULL, PFS_RD); pfs_create_file(dir, NULL, "msgsize_default", diff --git a/sys/compat/linux/linux.c b/sys/compat/linux/linux.c index 61b207070963..a40f110634f7 100644 --- a/sys/compat/linux/linux.c +++ b/sys/compat/linux/linux.c @@ -578,8 +578,13 @@ bsd_to_linux_sockaddr(const struct sockaddr *sa, struct l_sockaddr **lsa, return (0); } +/* + * If sap is NULL, then osa points at already copied in linux sockaddr that + * should be edited in place. Otherwise memory is allocated, sockaddr + * copied in and returned in *sap. + */ int -linux_to_bsd_sockaddr(const struct l_sockaddr *osa, struct sockaddr **sap, +linux_to_bsd_sockaddr(struct l_sockaddr *osa, struct sockaddr **sap, socklen_t *len) { struct sockaddr *sa; @@ -609,10 +614,12 @@ linux_to_bsd_sockaddr(const struct l_sockaddr *osa, struct sockaddr **sap, } #endif - kosa = malloc(salen, M_SONAME, M_WAITOK); - - if ((error = copyin(osa, kosa, *len))) - goto out; + if (sap != NULL) { + kosa = malloc(salen, M_SONAME, M_WAITOK); + if ((error = copyin(osa, kosa, *len))) + goto out; + } else + kosa = osa; bdom = linux_to_bsd_domain(kosa->sa_family); if (bdom == AF_UNKNOWN) { @@ -686,12 +693,15 @@ linux_to_bsd_sockaddr(const struct l_sockaddr *osa, struct sockaddr **sap, sa->sa_family = bdom; sa->sa_len = salen; - *sap = sa; - *len = salen; + if (sap != NULL) { + *sap = sa; + *len = salen; + } return (0); out: - free(kosa, M_SONAME); + if (sap != NULL) + free(kosa, M_SONAME); return (error); } diff --git a/sys/compat/linux/linux_common.h b/sys/compat/linux/linux_common.h index 97f5a259f300..814c183b338a 100644 --- a/sys/compat/linux/linux_common.h +++ b/sys/compat/linux/linux_common.h @@ -43,7 +43,7 @@ sa_family_t bsd_to_linux_domain(sa_family_t domain); #define AF_UNKNOWN UINT8_MAX int bsd_to_linux_sockaddr(const struct sockaddr *sa, struct l_sockaddr **lsa, socklen_t len); -int linux_to_bsd_sockaddr(const struct l_sockaddr *lsa, +int linux_to_bsd_sockaddr(struct l_sockaddr *lsa, struct sockaddr **sap, socklen_t *len); void linux_to_bsd_poll_events(struct thread *td, int fd, short lev, short *bev); diff --git a/sys/compat/linux/linux_event.c b/sys/compat/linux/linux_event.c index e88791659f1f..fc3ef7c3e90a 100644 --- a/sys/compat/linux/linux_event.c +++ b/sys/compat/linux/linux_event.c @@ -104,7 +104,7 @@ static int epoll_create_common(struct thread *td, int flags) { - return (kern_kqueue(td, flags, NULL)); + return (kern_kqueue(td, flags, false, NULL)); } #ifdef LINUX_LEGACY_SYSCALLS diff --git a/sys/compat/linux/linux_socket.c b/sys/compat/linux/linux_socket.c index 0e07b0a60ced..b1a483ce611c 100644 --- a/sys/compat/linux/linux_socket.c +++ b/sys/compat/linux/linux_socket.c @@ -2146,7 +2146,8 @@ linux_setsockopt(struct thread *td, struct linux_setsockopt_args *args) return (ENOPROTOOPT); } - if (name == IPV6_NEXTHOP) { + switch (name) { + case IPV6_NEXTHOP: { len = args->optlen; error = linux_to_bsd_sockaddr(PTRIN(args->optval), &sa, &len); if (error != 0) @@ -2155,7 +2156,34 @@ linux_setsockopt(struct thread *td, struct linux_setsockopt_args *args) error = kern_setsockopt(td, args->s, level, name, sa, UIO_SYSSPACE, len); free(sa, M_SONAME); - } else { + break; + } + case MCAST_JOIN_GROUP: + case MCAST_LEAVE_GROUP: + case MCAST_JOIN_SOURCE_GROUP: + case MCAST_LEAVE_SOURCE_GROUP: { + struct group_source_req req; + size_t size; + + size = (name == MCAST_JOIN_SOURCE_GROUP || + name == MCAST_LEAVE_SOURCE_GROUP) ? + sizeof(struct group_source_req) : sizeof(struct group_req); + + if ((error = copyin(PTRIN(args->optval), &req, size))) + return (error); + len = sizeof(struct sockaddr_storage); + if ((error = linux_to_bsd_sockaddr( + (struct l_sockaddr *)&req.gsr_group, NULL, &len))) + return (error); + if (size == sizeof(struct group_source_req) && + (error = linux_to_bsd_sockaddr( + (struct l_sockaddr *)&req.gsr_source, NULL, &len))) + return (error); + error = kern_setsockopt(td, args->s, level, name, &req, + UIO_SYSSPACE, size); + break; + } + default: error = kern_setsockopt(td, args->s, level, name, PTRIN(args->optval), UIO_USERSPACE, args->optlen); } diff --git a/sys/compat/linuxkpi/common/src/linux_compat.c b/sys/compat/linuxkpi/common/src/linux_compat.c index 458744a9fec6..ff0f477ea8cc 100644 --- a/sys/compat/linuxkpi/common/src/linux_compat.c +++ b/sys/compat/linuxkpi/common/src/linux_compat.c @@ -1171,12 +1171,14 @@ static const struct filterops linux_dev_kqfiltops_read = { .f_isfd = 1, .f_detach = linux_file_kqfilter_detach, .f_event = linux_file_kqfilter_read_event, + .f_copy = knote_triv_copy, }; static const struct filterops linux_dev_kqfiltops_write = { .f_isfd = 1, .f_detach = linux_file_kqfilter_detach, .f_event = linux_file_kqfilter_write_event, + .f_copy = knote_triv_copy, }; static void diff --git a/sys/conf/NOTES b/sys/conf/NOTES index a25ee8f6e1af..9944375c3615 100644 --- a/sys/conf/NOTES +++ b/sys/conf/NOTES @@ -2801,7 +2801,7 @@ options MAXFILES=999 # Random number generator # Alternative algorithm. -#options RANDOM_FENESTRASX +options RANDOM_FENESTRASX # Allow the CSPRNG algorithm to be loaded as a module. #options RANDOM_LOADABLE # Select this to allow high-rate but potentially expensive diff --git a/sys/conf/files.arm64 b/sys/conf/files.arm64 index 856ea3af1372..2f412fa3cb1b 100644 --- a/sys/conf/files.arm64 +++ b/sys/conf/files.arm64 @@ -73,6 +73,7 @@ arm64/arm64/pmap.c standard arm64/arm64/ptrace_machdep.c standard arm64/arm64/sdt_machdep.c optional kdtrace_hooks arm64/arm64/sigtramp.S standard +arm64/arm64/spec_workaround.c standard arm64/arm64/stack_machdep.c optional ddb | stack arm64/arm64/strcmp.S standard arm64/arm64/strncmp.S standard diff --git a/sys/contrib/libnv/bsd_nvpair.c b/sys/contrib/libnv/bsd_nvpair.c index c73bc2189121..b884dd260b84 100644 --- a/sys/contrib/libnv/bsd_nvpair.c +++ b/sys/contrib/libnv/bsd_nvpair.c @@ -985,13 +985,13 @@ nvpair_unpack_string_array(bool isbe __unused, nvpair_t *nvp, size = nvp->nvp_datasize; tmp = (const char *)ptr; for (ii = 0; ii < nvp->nvp_nitems; ii++) { - len = strnlen(tmp, size - 1) + 1; - size -= len; - if (tmp[len - 1] != '\0') { + if (size <= 0) { ERRNO_SET(EINVAL); return (NULL); } - if (size < 0) { + len = strnlen(tmp, size - 1) + 1; + size -= len; + if (tmp[len - 1] != '\0') { ERRNO_SET(EINVAL); return (NULL); } diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c index ace2360c032d..ebc2c0eeb6d2 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/sysctl_os.c @@ -163,6 +163,13 @@ param_set_arc_int(SYSCTL_HANDLER_ARGS) return (0); } +static void +warn_deprecated_sysctl(const char *old, const char *new) +{ + printf("WARNING: sysctl vfs.zfs.%s is deprecated. Use vfs.zfs.%s instead.\n", + old, new); +} + int param_set_arc_max(SYSCTL_HANDLER_ARGS) { @@ -185,12 +192,15 @@ param_set_arc_max(SYSCTL_HANDLER_ARGS) if (val != 0) zfs_arc_max = arc_c_max; + if (arg2 != 0) + warn_deprecated_sysctl("arc_max", "arc.max"); + return (0); } SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_max, CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, - NULL, 0, param_set_arc_max, "LU", + NULL, 1, param_set_arc_max, "LU", "Maximum ARC size in bytes (LEGACY)"); int @@ -214,12 +224,15 @@ param_set_arc_min(SYSCTL_HANDLER_ARGS) if (val != 0) zfs_arc_min = arc_c_min; + if (arg2 != 0) + warn_deprecated_sysctl("arc_min", "arc.min"); + return (0); } SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_min, CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, - NULL, 0, param_set_arc_min, "LU", + NULL, 1, param_set_arc_min, "LU", "Minimum ARC size in bytes (LEGACY)"); extern uint_t zfs_arc_free_target; @@ -242,6 +255,9 @@ param_set_arc_free_target(SYSCTL_HANDLER_ARGS) zfs_arc_free_target = val; + if (arg2 != 0) + warn_deprecated_sysctl("arc_free_target", "arc.free_target"); + return (0); } @@ -251,7 +267,7 @@ param_set_arc_free_target(SYSCTL_HANDLER_ARGS) */ SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_free_target, CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE, - NULL, 0, param_set_arc_free_target, "IU", + NULL, 1, param_set_arc_free_target, "IU", "Desired number of free pages below which ARC triggers reclaim" " (LEGACY)"); @@ -270,12 +286,15 @@ param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS) arc_no_grow_shift = val; + if (arg2 != 0) + warn_deprecated_sysctl("arc_no_grow_shift", "arc.no_grow_shift"); + return (0); } SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_no_grow_shift, CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, - NULL, 0, param_set_arc_no_grow_shift, "I", + NULL, 1, param_set_arc_no_grow_shift, "I", "log2(fraction of ARC which must be free to allow growing) (LEGACY)"); extern uint64_t l2arc_write_max; @@ -746,12 +765,15 @@ param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS) zfs_vdev_min_auto_ashift = val; + if (arg2 != 0) + warn_deprecated_sysctl("min_auto_ashift", + "vdev.min_auto_ashift"); + return (0); } SYSCTL_PROC(_vfs_zfs, OID_AUTO, min_auto_ashift, - CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, - &zfs_vdev_min_auto_ashift, sizeof (zfs_vdev_min_auto_ashift), + CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, NULL, 1, param_set_min_auto_ashift, "IU", "Min ashift used when creating new top-level vdev. (LEGACY)"); @@ -771,12 +793,15 @@ param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS) zfs_vdev_max_auto_ashift = val; + if (arg2 != 0) + warn_deprecated_sysctl("max_auto_ashift", + "vdev.max_auto_ashift"); + return (0); } SYSCTL_PROC(_vfs_zfs, OID_AUTO, max_auto_ashift, - CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, - &zfs_vdev_max_auto_ashift, sizeof (zfs_vdev_max_auto_ashift), + CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, NULL, 1, param_set_max_auto_ashift, "IU", "Max ashift used when optimizing for logical -> physical sector size on" " new top-level vdevs. (LEGACY)"); diff --git a/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c b/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c index 0dd2ecd7fd8d..3ddbfcb97184 100644 --- a/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c +++ b/sys/contrib/openzfs/module/os/freebsd/zfs/zvol_os.c @@ -183,6 +183,7 @@ static struct filterops zvol_filterops_vnode = { .f_isfd = 1, .f_detach = zvol_filter_detach, .f_event = zvol_filter_vnode, + .f_copy = knote_triv_copy, }; extern uint_t zfs_geom_probe_vdev_key; diff --git a/sys/crypto/chacha20/chacha.c b/sys/crypto/chacha20/chacha.c index 52f7e18c651c..cb06003b0ecf 100644 --- a/sys/crypto/chacha20/chacha.c +++ b/sys/crypto/chacha20/chacha.c @@ -138,7 +138,7 @@ chacha_encrypt_bytes(chacha_ctx *x,const u8 *m,u8 *c,u32 bytes) for (;;) { if (bytes < 64) { #ifndef KEYSTREAM_ONLY - for (i = 0;i < bytes;++i) tmp[i] = m[i]; + for (i = 0; i < bytes; ++i) tmp[i] = m[i]; m = tmp; #endif ctarget = c; @@ -160,7 +160,7 @@ chacha_encrypt_bytes(chacha_ctx *x,const u8 *m,u8 *c,u32 bytes) x13 = j13; x14 = j14; x15 = j15; - for (i = 20;i > 0;i -= 2) { + for (i = 20; i > 0; i -= 2) { QUARTERROUND( x0, x4, x8,x12) QUARTERROUND( x1, x5, x9,x13) QUARTERROUND( x2, x6,x10,x14) @@ -240,7 +240,7 @@ chacha_encrypt_bytes(chacha_ctx *x,const u8 *m,u8 *c,u32 bytes) if (bytes <= 64) { if (bytes < 64) { - for (i = 0;i < bytes;++i) ctarget[i] = c[i]; + for (i = 0; i < bytes; ++i) ctarget[i] = c[i]; } x->input[12] = j12; x->input[13] = j13; diff --git a/sys/crypto/openssl/ossl_sha256.c b/sys/crypto/openssl/ossl_sha256.c index 4613a9409b44..50cb9739d114 100644 --- a/sys/crypto/openssl/ossl_sha256.c +++ b/sys/crypto/openssl/ossl_sha256.c @@ -74,11 +74,11 @@ ossl_sha256_init(void *c_) unsigned int nn; \ switch ((c)->md_len) \ { case SHA224_DIGEST_LENGTH: \ - for (nn=0;nn<SHA224_DIGEST_LENGTH/4;nn++) \ + for (nn=0; nn < SHA224_DIGEST_LENGTH / 4; nn++) \ { ll=(c)->h[nn]; (void)HOST_l2c(ll,(s)); } \ break; \ case SHA256_DIGEST_LENGTH: \ - for (nn=0;nn<SHA256_DIGEST_LENGTH/4;nn++) \ + for (nn=0; nn < SHA256_DIGEST_LENGTH / 4; nn++) \ { ll=(c)->h[nn]; (void)HOST_l2c(ll,(s)); } \ break; \ default: \ diff --git a/sys/dev/ahci/ahciem.c b/sys/dev/ahci/ahciem.c index c9e6c35f4233..012796aa5d6f 100644 --- a/sys/dev/ahci/ahciem.c +++ b/sys/dev/ahci/ahciem.c @@ -479,7 +479,7 @@ ahci_em_emulate_ses_on_led(device_t dev, union ccb *ccb) else ads->common.bytes[0] |= SES_OBJSTAT_NOTINSTALLED; if (ch->disablephy) - ads->common.bytes[3] |= SESCTL_DEVOFF; + ads->bytes[2] |= SESCTL_DEVOFF; ahci_putch(ch); } ccb->ccb_h.status = CAM_REQ_CMP; diff --git a/sys/dev/aic7xxx/aic79xx.c b/sys/dev/aic7xxx/aic79xx.c index 2b5015b20e41..cee45fa5cc8a 100644 --- a/sys/dev/aic7xxx/aic79xx.c +++ b/sys/dev/aic7xxx/aic79xx.c @@ -7788,8 +7788,8 @@ ahd_abort_scbs(struct ahd_softc *ahd, int target, char channel, } if (role != ROLE_TARGET) { - for (;i < maxtarget; i++) { - for (j = minlun;j < maxlun; j++) { + for (; i < maxtarget; i++) { + for (j = minlun; j < maxlun; j++) { u_int scbid; u_int tcl; diff --git a/sys/dev/aic7xxx/aic7xxx.c b/sys/dev/aic7xxx/aic7xxx.c index c09876e9f589..18f68b806948 100644 --- a/sys/dev/aic7xxx/aic7xxx.c +++ b/sys/dev/aic7xxx/aic7xxx.c @@ -5903,8 +5903,8 @@ ahc_abort_scbs(struct ahc_softc *ahc, int target, char channel, } if (role != ROLE_TARGET) { - for (;i < maxtarget; i++) { - for (j = minlun;j < maxlun; j++) { + for (; i < maxtarget; i++) { + for (j = minlun; j < maxlun; j++) { u_int scbid; u_int tcl; diff --git a/sys/dev/atkbdc/psm.c b/sys/dev/atkbdc/psm.c index 8563b5f93aa2..137758b104d3 100644 --- a/sys/dev/atkbdc/psm.c +++ b/sys/dev/atkbdc/psm.c @@ -5287,6 +5287,7 @@ static const struct filterops psmfiltops = { .f_isfd = 1, .f_detach = psmfilter_detach, .f_event = psmfilter, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/dev/cyapa/cyapa.c b/sys/dev/cyapa/cyapa.c index ed755f992949..464b03c0ab64 100644 --- a/sys/dev/cyapa/cyapa.c +++ b/sys/dev/cyapa/cyapa.c @@ -1121,7 +1121,8 @@ static int cyapafilt(struct knote *, long); static const struct filterops cyapa_filtops = { .f_isfd = 1, .f_detach = cyapafiltdetach, - .f_event = cyapafilt + .f_event = cyapafilt, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/dev/dc/if_dc.c b/sys/dev/dc/if_dc.c index bed74c3b6181..5c1d7ff30976 100644 --- a/sys/dev/dc/if_dc.c +++ b/sys/dev/dc/if_dc.c @@ -999,7 +999,7 @@ dc_setfilt_21143(struct dc_softc *sc) else DC_CLRBIT(sc, DC_NETCFG, DC_NETCFG_RX_ALLMULTI); - if_foreach_llmaddr(ifp, dc_hash_maddr_21143, sp); + if_foreach_llmaddr(ifp, dc_hash_maddr_21143, sc); if (if_getflags(ifp) & IFF_BROADCAST) { h = dc_mchash_le(sc, if_getbroadcastaddr(ifp)); diff --git a/sys/dev/enetc/if_enetc.c b/sys/dev/enetc/if_enetc.c index 808397b229a7..53002f9d73ce 100644 --- a/sys/dev/enetc/if_enetc.c +++ b/sys/dev/enetc/if_enetc.c @@ -848,7 +848,7 @@ enetc_hash_vid(uint16_t vid) bool bit; int i; - for (i = 0;i < 6;i++) { + for (i = 0; i < 6; i++) { bit = vid & BIT(i); bit ^= !!(vid & BIT(i + 6)); hash |= bit << i; @@ -1020,7 +1020,7 @@ enetc_msix_intr_assign(if_ctx_t ctx, int msix) ENETC_RBICR0_ICEN | ENETC_RBICR0_SET_ICPT(ENETC_RX_INTR_PKT_THR)); } vector = 0; - for (i = 0;i < sc->tx_num_queues; i++, vector++) { + for (i = 0; i < sc->tx_num_queues; i++, vector++) { tx_queue = &sc->tx_queues[i]; snprintf(irq_name, sizeof(irq_name), "txq%d", i); iflib_softirq_alloc_generic(ctx, &tx_queue->irq, @@ -1130,7 +1130,7 @@ enetc_isc_txd_encap(void *data, if_pkt_info_t ipi) } /* Now add remaining descriptors. */ - for (;i < ipi->ipi_nsegs; i++) { + for (; i < ipi->ipi_nsegs; i++) { desc = &queue->ring[pidx]; bzero(desc, sizeof(*desc)); desc->addr = segs[i].ds_addr; diff --git a/sys/dev/evdev/cdev.c b/sys/dev/evdev/cdev.c index 9fe1299a0937..dd4115cdfc71 100644 --- a/sys/dev/evdev/cdev.c +++ b/sys/dev/evdev/cdev.c @@ -96,6 +96,7 @@ static const struct filterops evdev_cdev_filterops = { .f_attach = NULL, .f_detach = evdev_kqdetach, .f_event = evdev_kqread, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/dev/evdev/uinput.c b/sys/dev/evdev/uinput.c index 9ac9fee8a157..76a530479c02 100644 --- a/sys/dev/evdev/uinput.c +++ b/sys/dev/evdev/uinput.c @@ -104,6 +104,7 @@ static const struct filterops uinput_filterops = { .f_attach = NULL, .f_detach = uinput_kqdetach, .f_event = uinput_kqread, + .f_copy = knote_triv_copy, }; struct uinput_cdev_state diff --git a/sys/dev/gpio/gpioc.c b/sys/dev/gpio/gpioc.c index 6c6f79227166..517f7752daad 100644 --- a/sys/dev/gpio/gpioc.c +++ b/sys/dev/gpio/gpioc.c @@ -158,7 +158,8 @@ static const struct filterops gpioc_read_filterops = { .f_attach = NULL, .f_detach = gpioc_kqdetach, .f_event = gpioc_kqread, - .f_touch = NULL + .f_touch = NULL, + .f_copy = knote_triv_copy, }; static struct gpioc_pin_event * diff --git a/sys/dev/hid/hidraw.c b/sys/dev/hid/hidraw.c index 4855843cd265..5b5e9b58f8bd 100644 --- a/sys/dev/hid/hidraw.c +++ b/sys/dev/hid/hidraw.c @@ -182,6 +182,7 @@ static const struct filterops hidraw_filterops_read = { .f_isfd = 1, .f_detach = hidraw_kqdetach, .f_event = hidraw_kqread, + .f_copy = knote_triv_copy, }; static void diff --git a/sys/dev/hid/u2f.c b/sys/dev/hid/u2f.c index 08f1a5ceedba..e1f696d72f01 100644 --- a/sys/dev/hid/u2f.c +++ b/sys/dev/hid/u2f.c @@ -132,6 +132,7 @@ static struct filterops u2f_filterops_read = { .f_isfd = 1, .f_detach = u2f_kqdetach, .f_event = u2f_kqread, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/dev/hptmv/entry.c b/sys/dev/hptmv/entry.c index 5c4718bf582f..f3d58f285b39 100644 --- a/sys/dev/hptmv/entry.c +++ b/sys/dev/hptmv/entry.c @@ -430,7 +430,7 @@ static void device_change(IAL_ADAPTER_T *pAdapter , MV_U8 channelIndex, int plug if(pVDev->pParent) { int iMember; - for(iMember = 0; iMember < pVDev->pParent->u.array.bArnMember; iMember++) + for (iMember = 0; iMember < pVDev->pParent->u.array.bArnMember; iMember++) if((PVDevice)pVDev->pParent->u.array.pMember[iMember] == pVDev) pVDev->pParent->u.array.pMember[iMember] = NULL; pVDev->pParent = NULL; @@ -984,7 +984,7 @@ fRegisterVdevice(IAL_ADAPTER_T *pAdapter) PVBus pVBus; int i,j; - for(i=0;i<MV_SATA_CHANNELS_NUM;i++) { + for (i = 0; i < MV_SATA_CHANNELS_NUM; i++) { pPhysical = &(pAdapter->VDevices[i]); pLogical = pPhysical; while (pLogical->pParent) pLogical = pLogical->pParent; @@ -1027,8 +1027,7 @@ GetSpareDisk(_VBUS_ARG PVDevice pArray) PVDevice pVDevice, pFind = NULL; int i; - for(i=0;i<MV_SATA_CHANNELS_NUM;i++) - { + for (i=0; i < MV_SATA_CHANNELS_NUM; i++) { pVDevice = &pAdapter->VDevices[i]; if(!pVDevice) continue; @@ -1356,7 +1355,7 @@ unregister: goto unregister; } - for (i=0; i<MAX_COMMAND_BLOCKS_FOR_EACH_VBUS; i++) { + for (i = 0; i < MAX_COMMAND_BLOCKS_FOR_EACH_VBUS; i++) { FreeCommand(_VBUS_P &(pAdapter->pCommandBlocks[i])); } @@ -1370,7 +1369,7 @@ unregister: memset((void *)pAdapter->pbus_dmamap, 0, sizeof(struct _BUS_DMAMAP) * MAX_QUEUE_COMM); pAdapter->pbus_dmamap_list = 0; - for (i=0; i < MAX_QUEUE_COMM; i++) { + for (i = 0; i < MAX_QUEUE_COMM; i++) { PBUS_DMAMAP pmap = &(pAdapter->pbus_dmamap[i]); pmap->pAdapter = pAdapter; dmamap_put(pmap); @@ -1398,7 +1397,7 @@ unregister: pAdapter->prdTableAlignedAddr = (PUCHAR)(((ULONG_PTR)pAdapter->prdTableAddr + 0x1f) & ~(ULONG_PTR)0x1fL); { PUCHAR PRDTable = pAdapter->prdTableAlignedAddr; - for (i=0; i<PRD_TABLES_FOR_VBUS; i++) + for (i = 0; i < PRD_TABLES_FOR_VBUS; i++) { /* KdPrint(("i=%d,pAdapter->pFreePRDLink=%p\n",i,pAdapter->pFreePRDLink)); */ FreePRDTable(pAdapter, PRDTable); @@ -1447,7 +1446,7 @@ unregister: } #ifdef SUPPORT_ARRAY - for(i = MAX_ARRAY_DEVICE - 1; i >= 0; i--) { + for (i = MAX_ARRAY_DEVICE - 1; i >= 0; i--) { pVDev = ArrayTables(i); mArFreeArrayTable(pVDev); } @@ -1467,7 +1466,7 @@ unregister: _vbus_p->nInstances = 1; fRegisterVdevice(pAdapter); - for (channel=0;channel<MV_SATA_CHANNELS_NUM;channel++) { + for (channel = 0; channel < MV_SATA_CHANNELS_NUM; channel++) { pVDev = _vbus_p->pVDevice[channel]; if (pVDev && pVDev->vf_online) fCheckBootable(pVDev); @@ -1567,7 +1566,7 @@ fResetActiveCommands(PVBus _vbus_p) { MV_SATA_ADAPTER *pMvSataAdapter = &((IAL_ADAPTER_T *)_vbus_p->OsExt)->mvSataAdapter; MV_U8 channel; - for (channel=0;channel< MV_SATA_CHANNELS_NUM;channel++) { + for (channel = 0; channel < MV_SATA_CHANNELS_NUM; channel++) { if (pMvSataAdapter->sataChannel[channel] && pMvSataAdapter->sataChannel[channel]->outstandingCommands) MvSataResetChannel(pMvSataAdapter,channel); } @@ -1590,7 +1589,7 @@ check_cmds: dataxfer_poll(); xor_poll(); #endif - for (channel=0;channel< MV_SATA_CHANNELS_NUM;channel++) { + for (channel = 0; channel < MV_SATA_CHANNELS_NUM; channel++) { pMvSataChannel = pMvSataAdapter->sataChannel[channel]; if (pMvSataChannel && pMvSataChannel->outstandingCommands) { @@ -1716,7 +1715,7 @@ fDeviceSendCommand(_VBUS_ARG PCommand pCmd) MV_BOOLEAN is48bit; MV_U8 channel; - int i=0; + int i = 0; DECLARE_BUFFER(FPSCAT_GATH, tmpSg); @@ -2141,7 +2140,7 @@ FlushAdapter(IAL_ADAPTER_T *pAdapter) hpt_printk(("flush all devices\n")); /* flush all devices */ - for (i=0; i<MAX_VDEVICE_PER_VBUS; i++) { + for (i = 0; i < MAX_VDEVICE_PER_VBUS; i++) { PVDevice pVDev = pAdapter->VBus.pVDevice[i]; if(pVDev) fFlushVDev(pVDev); } @@ -2174,7 +2173,7 @@ Check_Idle_Call(IAL_ADAPTER_T *pAdapter) { int i; PVDevice pArray; - for(i = 0; i < MAX_ARRAY_PER_VBUS; i++){ + for (i = 0; i < MAX_ARRAY_PER_VBUS; i++) { if ((pArray=ArrayTables(i))->u.array.dArStamp==0) continue; else if (pArray->u.array.rf_auto_rebuild) { @@ -2378,7 +2377,7 @@ hpt_free_ccb(union ccb **ccb_Q, union ccb *ccb) static void hpt_worker_thread(void) { - for(;;) { + for (;;) { mtx_lock(&DpcQueue_Lock); while (DpcQueue_First!=DpcQueue_Last) { ST_HPT_DPC p; @@ -2418,7 +2417,7 @@ static void hpt_worker_thread(void) mtx_lock(&pAdapter->lock); _vbus_p = &pAdapter->VBus; - for (i=0;i<MAX_ARRAY_PER_VBUS;i++) + for (i = 0; i < MAX_ARRAY_PER_VBUS; i++) { if ((pArray=ArrayTables(i))->u.array.dArStamp==0) continue; @@ -2472,7 +2471,7 @@ launch_worker_thread(void) int i; PVDevice pVDev; - for(i = 0; i < MAX_ARRAY_PER_VBUS; i++) + for (i = 0; i < MAX_ARRAY_PER_VBUS; i++) if ((pVDev=ArrayTables(i))->u.array.dArStamp==0) continue; else{ diff --git a/sys/dev/hptmv/gui_lib.c b/sys/dev/hptmv/gui_lib.c index d78fdcca69d2..f11044db733a 100644 --- a/sys/dev/hptmv/gui_lib.c +++ b/sys/dev/hptmv/gui_lib.c @@ -86,8 +86,7 @@ check_VDevice_valid(PVDevice p) while(pAdapter != NULL) { _vbus_p = &pAdapter->VBus; - for (i=0;i<MAX_ARRAY_PER_VBUS;i++) - { + for (i = 0; i<MAX_ARRAY_PER_VBUS; i++) { pVDevice=ArrayTables(i); if ((pVDevice->u.array.dArStamp != 0) && (pVDevice == p)) return 0; @@ -244,9 +243,9 @@ static void get_array_info(PVDevice pVDevice, PHPT_ARRAY_INFO pArrayInfo) if(pVDevice->u.array.pMember[i] != NULL) pArrayInfo->Members[pArrayInfo->nDisk++] = VDEV_TO_ID(pVDevice->u.array.pMember[i]); - for(i=pArrayInfo->nDisk; i<MAX_ARRAY_MEMBERS; i++) + for (i = pArrayInfo->nDisk; i < MAX_ARRAY_MEMBERS; i++) pArrayInfo->Members[i] = INVALID_DEVICEID; - } +} static void get_array_info_v2(PVDevice pVDevice, PHPT_ARRAY_INFO_V2 pArrayInfo) { @@ -266,7 +265,7 @@ static void get_array_info_v2(PVDevice pVDevice, PHPT_ARRAY_INFO_V2 pArrayInfo) if(pVDevice->u.array.pMember[i] != NULL) pArrayInfo->Members[pArrayInfo->nDisk++] = VDEV_TO_ID(pVDevice->u.array.pMember[i]); - for(i=pArrayInfo->nDisk; i<MAX_ARRAY_MEMBERS_V2; i++) + for (i = pArrayInfo->nDisk; i < MAX_ARRAY_MEMBERS_V2; i++) pArrayInfo->Members[i] = INVALID_DEVICEID; } #endif @@ -461,8 +460,7 @@ found: pInfo->IoPort = 0; pInfo->ControlPort = 0; - for (i=0; i<2 ;i++) - { + for (i = 0; i < 2; i++) { pInfo->Devices[i] = (DEVICEID)INVALID_DEVICEID; } diff --git a/sys/dev/hptmv/hptproc.c b/sys/dev/hptmv/hptproc.c index 38fe61ee7e04..328750d9034c 100644 --- a/sys/dev/hptmv/hptproc.c +++ b/sys/dev/hptmv/hptproc.c @@ -107,7 +107,7 @@ hpt_set_asc_info(IAL_ADAPTER_T *pAdapter, char *buffer,int length) return -EINVAL; } - for (i=0;i<MV_SATA_CHANNELS_NUM;i++) + for (i = 0; i < MV_SATA_CHANNELS_NUM; i++) if(i == ichan) goto rebuild; diff --git a/sys/dev/ixl/if_ixl.c b/sys/dev/ixl/if_ixl.c index 261f76055901..bfaf6cd69e58 100644 --- a/sys/dev/ixl/if_ixl.c +++ b/sys/dev/ixl/if_ixl.c @@ -1480,17 +1480,33 @@ ixl_if_multi_set(if_ctx_t ctx) struct ixl_pf *pf = iflib_get_softc(ctx); struct ixl_vsi *vsi = &pf->vsi; struct i40e_hw *hw = vsi->hw; + enum i40e_status_code status; int mcnt; + if_t ifp = iflib_get_ifp(ctx); IOCTL_DEBUGOUT("ixl_if_multi_set: begin"); /* Delete filters for removed multicast addresses */ ixl_del_multi(vsi, false); - mcnt = min(if_llmaddr_count(iflib_get_ifp(ctx)), MAX_MULTICAST_ADDR); + mcnt = min(if_llmaddr_count(ifp), MAX_MULTICAST_ADDR); if (__predict_false(mcnt == MAX_MULTICAST_ADDR)) { - i40e_aq_set_vsi_multicast_promiscuous(hw, + /* Check if promisc mode is already enabled, if yes return */ + if (vsi->flags & IXL_FLAGS_MC_PROMISC) + return; + + status = i40e_aq_set_vsi_multicast_promiscuous(hw, vsi->seid, TRUE, NULL); + if (status != I40E_SUCCESS) + if_printf(ifp, "Failed to enable multicast promiscuous " + "mode, status: %s\n", i40e_stat_str(hw, status)); + else { + if_printf(ifp, "Enabled multicast promiscuous mode\n"); + + /* Set the flag to track promiscuous mode */ + vsi->flags |= IXL_FLAGS_MC_PROMISC; + } + /* Delete all existing MC filters */ ixl_del_multi(vsi, true); return; } @@ -1693,6 +1709,13 @@ ixl_if_promisc_set(if_ctx_t ctx, int flags) return (err); err = i40e_aq_set_vsi_multicast_promiscuous(hw, vsi->seid, multi, NULL); + + /* Update the multicast promiscuous flag based on the new state */ + if (multi) + vsi->flags |= IXL_FLAGS_MC_PROMISC; + else + vsi->flags &= ~IXL_FLAGS_MC_PROMISC; + return (err); } diff --git a/sys/dev/ixl/ixl.h b/sys/dev/ixl/ixl.h index 95379448b570..ab0f38307d90 100644 --- a/sys/dev/ixl/ixl.h +++ b/sys/dev/ixl/ixl.h @@ -202,6 +202,7 @@ #define IXL_FLAGS_KEEP_TSO6 (1 << 1) #define IXL_FLAGS_USES_MSIX (1 << 2) #define IXL_FLAGS_IS_VF (1 << 3) +#define IXL_FLAGS_MC_PROMISC (1 << 4) #define IXL_VSI_IS_PF(v) ((v->flags & IXL_FLAGS_IS_VF) == 0) #define IXL_VSI_IS_VF(v) ((v->flags & IXL_FLAGS_IS_VF) != 0) diff --git a/sys/dev/ixl/ixl_pf_main.c b/sys/dev/ixl/ixl_pf_main.c index 1752efc02fff..b62619ced5cb 100644 --- a/sys/dev/ixl/ixl_pf_main.c +++ b/sys/dev/ixl/ixl_pf_main.c @@ -593,24 +593,29 @@ ixl_add_maddr(void *arg, struct sockaddr_dl *sdl, u_int cnt) * Routines for multicast and vlan filter management. * *********************************************************************/ + +/** + * ixl_add_multi - Add multicast filters to the hardware + * @vsi: The VSI structure + * + * In case number of multicast filters in the IFP exceeds 127 entries, + * multicast promiscuous mode will be enabled and the filters will be removed + * from the hardware + */ void ixl_add_multi(struct ixl_vsi *vsi) { if_t ifp = vsi->ifp; - struct i40e_hw *hw = vsi->hw; int mcnt = 0; struct ixl_add_maddr_arg cb_arg; IOCTL_DEBUGOUT("ixl_add_multi: begin"); - mcnt = if_llmaddr_count(ifp); - if (__predict_false(mcnt >= MAX_MULTICAST_ADDR)) { - i40e_aq_set_vsi_multicast_promiscuous(hw, - vsi->seid, TRUE, NULL); - /* delete all existing MC filters */ - ixl_del_multi(vsi, true); - return; - } + /* + * There is no need to check if the number of multicast addresses + * exceeds the MAX_MULTICAST_ADDR threshold and set promiscuous mode + * here, as all callers already handle this case. + */ cb_arg.vsi = vsi; LIST_INIT(&cb_arg.to_add); @@ -633,30 +638,103 @@ ixl_match_maddr(void *arg, struct sockaddr_dl *sdl, u_int cnt) return (0); } +/** + * ixl_dis_multi_promisc - Disable multicast promiscuous mode + * @vsi: The VSI structure + * @vsi_mcnt: Number of multicast filters in the VSI + * + * Disable multicast promiscuous mode based on number of entries in the IFP + * and the VSI, then re-add multicast filters. + * + */ +static void +ixl_dis_multi_promisc(struct ixl_vsi *vsi, int vsi_mcnt) +{ + struct ifnet *ifp = vsi->ifp; + struct i40e_hw *hw = vsi->hw; + int ifp_mcnt = 0; + enum i40e_status_code status; + + /* + * Check if multicast promiscuous mode was actually enabled. + * If promiscuous mode was not enabled, don't attempt to disable it. + * Also, don't disable if IFF_PROMISC or IFF_ALLMULTI is set. + */ + if (!(vsi->flags & IXL_FLAGS_MC_PROMISC) || + (if_getflags(ifp) & (IFF_PROMISC | IFF_ALLMULTI))) + return; + + ifp_mcnt = if_llmaddr_count(ifp); + /* + * Equal lists or empty ifp list mean the list has not been changed + * and in such case avoid disabling multicast promiscuous mode as it + * was not previously enabled. Case where multicast promiscuous mode has + * been enabled is when vsi_mcnt == 0 && ifp_mcnt > 0. + */ + if (ifp_mcnt == vsi_mcnt || ifp_mcnt == 0 || + ifp_mcnt >= MAX_MULTICAST_ADDR) + return; + + status = i40e_aq_set_vsi_multicast_promiscuous(hw, vsi->seid, + FALSE, NULL); + if (status != I40E_SUCCESS) { + if_printf(ifp, "Failed to disable multicast promiscuous " + "mode, status: %s\n", i40e_stat_str(hw, status)); + + return; + } + + /* Clear the flag since promiscuous mode is now disabled */ + vsi->flags &= ~IXL_FLAGS_MC_PROMISC; + if_printf(ifp, "Disabled multicast promiscuous mode\n"); + + ixl_add_multi(vsi); +} + +/** + * ixl_del_multi - Delete multicast filters from the hardware + * @vsi: The VSI structure + * @all: Bool to determine if all the multicast filters should be removed + * + * In case number of multicast filters in the IFP drops to 127 entries, + * multicast promiscuous mode will be disabled and the filters will be reapplied + * to the hardware. + */ void ixl_del_multi(struct ixl_vsi *vsi, bool all) { - struct ixl_ftl_head to_del; + int to_del_cnt = 0, vsi_mcnt = 0; if_t ifp = vsi->ifp; struct ixl_mac_filter *f, *fn; - int mcnt = 0; + struct ixl_ftl_head to_del; IOCTL_DEBUGOUT("ixl_del_multi: begin"); LIST_INIT(&to_del); /* Search for removed multicast addresses */ LIST_FOREACH_SAFE(f, &vsi->ftl, ftle, fn) { - if ((f->flags & IXL_FILTER_MC) == 0 || - (!all && (if_foreach_llmaddr(ifp, ixl_match_maddr, f) == 0))) + if ((f->flags & IXL_FILTER_MC) == 0) + continue; + + /* Count all the multicast filters in the VSI for comparison */ + vsi_mcnt++; + + if (!all && if_foreach_llmaddr(ifp, ixl_match_maddr, f) != 0) continue; LIST_REMOVE(f, ftle); LIST_INSERT_HEAD(&to_del, f, ftle); - mcnt++; + to_del_cnt++; } - if (mcnt > 0) - ixl_del_hw_filters(vsi, &to_del, mcnt); + if (to_del_cnt > 0) { + ixl_del_hw_filters(vsi, &to_del, to_del_cnt); + return; + } + + ixl_dis_multi_promisc(vsi, vsi_mcnt); + + IOCTL_DEBUGOUT("ixl_del_multi: end"); } void diff --git a/sys/dev/mps/mps_sas.c b/sys/dev/mps/mps_sas.c index d69c8ea5fded..fa0f817ed67b 100644 --- a/sys/dev/mps/mps_sas.c +++ b/sys/dev/mps/mps_sas.c @@ -858,7 +858,7 @@ mps_detach_sas(struct mps_softc *sc) if (sassc->devq != NULL) cam_simq_free(sassc->devq); - for(i=0; i< sassc->maxtargets ;i++) { + for (i = 0; i < sassc->maxtargets; i++) { targ = &sassc->targets[i]; SLIST_FOREACH_SAFE(lun, &targ->luns, lun_link, lun_tmp) { free(lun, M_MPT2); @@ -3396,7 +3396,7 @@ mpssas_realloc_targets(struct mps_softc *sc, int maxtargets) * the allocated LUNs for each target and then the target buffer * itself. */ - for (i=0; i< maxtargets; i++) { + for (i = 0; i < maxtargets; i++) { targ = &sassc->targets[i]; SLIST_FOREACH_SAFE(lun, &targ->luns, lun_link, lun_tmp) { free(lun, M_MPT2); diff --git a/sys/dev/mpt/mpt_raid.c b/sys/dev/mpt/mpt_raid.c index 5ff08ffcf2b3..2b868f6ef070 100644 --- a/sys/dev/mpt/mpt_raid.c +++ b/sys/dev/mpt/mpt_raid.c @@ -830,7 +830,7 @@ mpt_is_raid_volume(struct mpt_softc *mpt, target_id_t tgt) } ioc_vol = mpt->ioc_page2->RaidVolume; ioc_last_vol = ioc_vol + mpt->ioc_page2->NumActiveVolumes; - for (;ioc_vol != ioc_last_vol; ioc_vol++) { + for (; ioc_vol != ioc_last_vol; ioc_vol++) { if (ioc_vol->VolumeID == tgt) { return (1); } @@ -1406,7 +1406,7 @@ mpt_refresh_raid_data(struct mpt_softc *mpt) ioc_vol = mpt->ioc_page2->RaidVolume; ioc_last_vol = ioc_vol + mpt->ioc_page2->NumActiveVolumes; - for (;ioc_vol != ioc_last_vol; ioc_vol++) { + for (; ioc_vol != ioc_last_vol; ioc_vol++) { struct mpt_raid_volume *mpt_vol; mpt_vol = mpt->raid_volumes + ioc_vol->VolumePageNumber; diff --git a/sys/dev/netmap/netmap_freebsd.c b/sys/dev/netmap/netmap_freebsd.c index 8cc543d54c2e..9fb4370129f3 100644 --- a/sys/dev/netmap/netmap_freebsd.c +++ b/sys/dev/netmap/netmap_freebsd.c @@ -738,6 +738,7 @@ nm_os_extmem_create(unsigned long p, struct nmreq_pools_info *pi, int *perror) out_rem: vm_map_remove(kernel_map, e->kva, e->kva + e->size); + e->obj = NULL; /* reference consumed by vm_map_remove() */ out_rel: vm_object_deallocate(e->obj); e->obj = NULL; @@ -1406,19 +1407,34 @@ netmap_knwrite(struct knote *kn, long hint) return netmap_knrw(kn, hint, POLLOUT); } +static int +netmap_kncopy(struct knote *kn, struct proc *p1) +{ + struct netmap_priv_d *priv; + struct nm_selinfo *si; + + priv = kn->kn_hook; + si = priv->np_si[kn->kn_filter == EVFILT_WRITE ? NR_TX : NR_RX]; + NMG_LOCK(); + si->kqueue_users++; + NMG_UNLOCK(); + return (0); +} + static const struct filterops netmap_rfiltops = { .f_isfd = 1, .f_detach = netmap_knrdetach, .f_event = netmap_knread, + .f_copy = netmap_kncopy, }; static const struct filterops netmap_wfiltops = { .f_isfd = 1, .f_detach = netmap_knwdetach, .f_event = netmap_knwrite, + .f_copy = netmap_kncopy, }; - /* * This is called when a thread invokes kevent() to record * a change in the configuration of the kqueue(). diff --git a/sys/dev/nfe/if_nfe.c b/sys/dev/nfe/if_nfe.c index 4625c2616562..265181ef7ad0 100644 --- a/sys/dev/nfe/if_nfe.c +++ b/sys/dev/nfe/if_nfe.c @@ -2078,7 +2078,7 @@ nfe_rxeof(struct nfe_softc *sc, int count, int *rx_npktsp) bus_dmamap_sync(sc->rxq.rx_desc_tag, sc->rxq.rx_desc_map, BUS_DMASYNC_POSTREAD); - for (prog = 0;;NFE_INC(sc->rxq.cur, NFE_RX_RING_COUNT), vtag = 0) { + for (prog = 0; ; NFE_INC(sc->rxq.cur, NFE_RX_RING_COUNT), vtag = 0) { if (count <= 0) break; count--; @@ -2192,7 +2192,7 @@ nfe_jrxeof(struct nfe_softc *sc, int count, int *rx_npktsp) bus_dmamap_sync(sc->jrxq.jrx_desc_tag, sc->jrxq.jrx_desc_map, BUS_DMASYNC_POSTREAD); - for (prog = 0;;NFE_INC(sc->jrxq.jcur, NFE_JUMBO_RX_RING_COUNT), + for (prog = 0; ; NFE_INC(sc->jrxq.jcur, NFE_JUMBO_RX_RING_COUNT), vtag = 0) { if (count <= 0) break; diff --git a/sys/dev/null/null.c b/sys/dev/null/null.c index 8525eb9543c3..b5725de30bef 100644 --- a/sys/dev/null/null.c +++ b/sys/dev/null/null.c @@ -61,12 +61,14 @@ static int zero_ev(struct knote *kn, long hint); static const struct filterops one_fop = { .f_isfd = 1, - .f_event = one_ev + .f_event = one_ev, + .f_copy = knote_triv_copy, }; static const struct filterops zero_fop = { .f_isfd = 1, - .f_event = zero_ev + .f_event = zero_ev, + .f_copy = knote_triv_copy, }; static struct cdevsw full_cdevsw = { diff --git a/sys/dev/ocs_fc/ocs_mgmt.c b/sys/dev/ocs_fc/ocs_mgmt.c index 726b499f28ba..5b7f6557c017 100644 --- a/sys/dev/ocs_fc/ocs_mgmt.c +++ b/sys/dev/ocs_fc/ocs_mgmt.c @@ -226,7 +226,7 @@ ocs_mgmt_get_list(ocs_t *ocs, ocs_textbuf_t *textbuf) ocs_mgmt_start_unnumbered_section(textbuf, "ocs"); - for (i=0;i<ARRAY_SIZE(mgmt_table);i++) { + for (i = 0; i < ARRAY_SIZE(mgmt_table); i++) { access = 0; if (mgmt_table[i].get_handler) { access |= MGMT_MODE_RD; @@ -305,7 +305,7 @@ ocs_mgmt_get(ocs_t *ocs, char *name, ocs_textbuf_t *textbuf) if (ocs_strncmp(name, qualifier, strlen(qualifier)) == 0) { char *unqualified_name = name + strlen(qualifier) + 1; - for (i=0;i<ARRAY_SIZE(mgmt_table);i++) { + for (i = 0; i < ARRAY_SIZE(mgmt_table); i++) { if (ocs_strcmp(unqualified_name, mgmt_table[i].name) == 0) { if (mgmt_table[i].get_handler) { mgmt_table[i].get_handler(ocs, name, textbuf); @@ -387,7 +387,7 @@ ocs_mgmt_set(ocs_t *ocs, char *name, char *value) char *unqualified_name = name + strlen(qualifier) +1; /* See if it's a value I can set */ - for (i=0;i<ARRAY_SIZE(mgmt_table);i++) { + for (i = 0; i < ARRAY_SIZE(mgmt_table); i++) { if (ocs_strcmp(unqualified_name, mgmt_table[i].name) == 0) { if (mgmt_table[i].set_handler) { return mgmt_table[i].set_handler(ocs, name, value); @@ -469,7 +469,7 @@ ocs_mgmt_exec(ocs_t *ocs, char *action, void *arg_in, char *unqualified_name = action + strlen(qualifier) +1; /* See if it's an action I can perform */ - for (i=0;i<ARRAY_SIZE(mgmt_table); i++) { + for (i = 0; i < ARRAY_SIZE(mgmt_table); i++) { if (ocs_strcmp(unqualified_name, mgmt_table[i].name) == 0) { if (mgmt_table[i].action_handler) { return mgmt_table[i].action_handler(ocs, action, arg_in, arg_in_length, @@ -527,7 +527,7 @@ ocs_mgmt_get_all(ocs_t *ocs, ocs_textbuf_t *textbuf) ocs_mgmt_start_unnumbered_section(textbuf, "ocs"); - for (i=0;i<ARRAY_SIZE(mgmt_table);i++) { + for (i = 0; i < ARRAY_SIZE(mgmt_table); i++) { if (mgmt_table[i].get_handler) { mgmt_table[i].get_handler(ocs, mgmt_table[i].name, textbuf); } else if (mgmt_table[i].action_handler) { @@ -1212,7 +1212,7 @@ get_sfp_a2(ocs_t *ocs, char *name, ocs_textbuf_t *textbuf) int buffer_remaining = (SFP_PAGE_SIZE * 3) + 1; int bytes_added; - for (i=0; i < bytes_read; i++) { + for (i = 0; i < bytes_read; i++) { bytes_added = ocs_snprintf(d, buffer_remaining, "%02x ", *s); ++s; d += bytes_added; @@ -2040,7 +2040,7 @@ get_profile_list(ocs_t *ocs, char *name, ocs_textbuf_t *textbuf) result_buf = ocs_malloc(ocs, BUFFER_SIZE, OCS_M_ZERO); bytes_left = BUFFER_SIZE; - for (i=0; i<result.list->num_descriptors; i++) { + for (i = 0; i < result.list->num_descriptors; i++) { sprintf(result_line, "0x%02x:%s\n", result.list->descriptors[i].profile_id, result.list->descriptors[i].profile_description); if (strlen(result_line) < bytes_left) { diff --git a/sys/dev/pci/controller/pci_n1sdp.c b/sys/dev/pci/controller/pci_n1sdp.c index 487041bc78e4..22f0ea27d45b 100644 --- a/sys/dev/pci/controller/pci_n1sdp.c +++ b/sys/dev/pci/controller/pci_n1sdp.c @@ -345,6 +345,17 @@ n1sdp_pcie_write_config(device_t dev, u_int bus, u_int slot, bus_space_write_4(t, h, offset & ~3, data); } +static int +n1sdp_pcie_acpi_request_feature(device_t pcib __unused, device_t dev __unused, + enum pci_feature feature __unused) +{ + /* + * HotPlug isn't supported on the N1SDP as it causes an interrupt storm + */ + return (EINVAL); +} + + static device_method_t n1sdp_pcie_acpi_methods[] = { DEVMETHOD(device_probe, n1sdp_pcie_acpi_probe), DEVMETHOD(device_attach, n1sdp_pcie_acpi_attach), @@ -352,6 +363,7 @@ static device_method_t n1sdp_pcie_acpi_methods[] = { /* pcib interface */ DEVMETHOD(pcib_read_config, n1sdp_pcie_read_config), DEVMETHOD(pcib_write_config, n1sdp_pcie_write_config), + DEVMETHOD(pcib_request_feature, n1sdp_pcie_acpi_request_feature), DEVMETHOD_END }; diff --git a/sys/dev/ppc/ppc.c b/sys/dev/ppc/ppc.c index 9870379e2eba..de75f4747709 100644 --- a/sys/dev/ppc/ppc.c +++ b/sys/dev/ppc/ppc.c @@ -1389,7 +1389,7 @@ ppc_exec_microseq(device_t dev, struct ppb_microseq **p_msq) /* let's suppose the next instr. is the same */ prefetch: - for (;mi->opcode == MS_OP_RASSERT; INCR_PC) + for (; mi->opcode == MS_OP_RASSERT; INCR_PC) w_reg(mi->arg[0].i, ppc, (char)mi->arg[1].i); if (mi->opcode == MS_OP_DELAY) { diff --git a/sys/dev/qat/qat_common/adf_freebsd_dev_processes.c b/sys/dev/qat/qat_common/adf_freebsd_dev_processes.c index 67e1d4ad2cab..c5b745bb78fb 100644 --- a/sys/dev/qat/qat_common/adf_freebsd_dev_processes.c +++ b/sys/dev/qat/qat_common/adf_freebsd_dev_processes.c @@ -89,6 +89,7 @@ static struct filterops adf_state_read_filterops = { .f_attach = NULL, .f_detach = adf_state_kqread_detach, .f_event = adf_state_kqread_event, + .f_copy = knote_triv_copy, }; static struct cdev *adf_processes_dev; diff --git a/sys/dev/smartpqi/smartpqi_event.c b/sys/dev/smartpqi/smartpqi_event.c index f000d9ce9db3..88dcf45dd08a 100644 --- a/sys/dev/smartpqi/smartpqi_event.c +++ b/sys/dev/smartpqi/smartpqi_event.c @@ -115,7 +115,7 @@ pqisrc_ack_all_events(void *arg1) pending_event = &softs->pending_events[0]; - for (i=0; i < PQI_NUM_SUPPORTED_EVENTS; i++) { + for (i = 0; i < PQI_NUM_SUPPORTED_EVENTS; i++) { if (pending_event->pending == true) { pending_event->pending = false; pqisrc_acknowledge_event(softs, pending_event); @@ -417,7 +417,7 @@ pqisrc_report_event_config(pqisrc_softstate_t *softs) softs->event_config.num_event_descriptors = MIN(event_config_p->num_event_descriptors, PQI_MAX_EVENT_DESCRIPTORS) ; - for (i=0; i < softs->event_config.num_event_descriptors ;i++){ + for (i = 0; i < softs->event_config.num_event_descriptors; i++) { softs->event_config.descriptors[i].event_type = event_config_p->descriptors[i].event_type; } @@ -477,7 +477,7 @@ pqisrc_set_event_config(pqisrc_softstate_t *softs) event_config_p->num_event_descriptors = softs->event_config.num_event_descriptors; - for (i=0; i < softs->event_config.num_event_descriptors ; i++){ + for (i = 0; i < softs->event_config.num_event_descriptors; i++) { event_config_p->descriptors[i].event_type = softs->event_config.descriptors[i].event_type; if( pqisrc_event_type_to_event_index(event_config_p->descriptors[i].event_type) != -1) diff --git a/sys/dev/smartpqi/smartpqi_queue.c b/sys/dev/smartpqi/smartpqi_queue.c index 2e80b01b5436..f05c951cd4f9 100644 --- a/sys/dev/smartpqi/smartpqi_queue.c +++ b/sys/dev/smartpqi/smartpqi_queue.c @@ -700,7 +700,7 @@ pqisrc_create_op_obq(pqisrc_softstate_t *softs, } else { int i = 0; DBG_WARN("Error Status Descriptors\n"); - for(i = 0; i < 4;i++) + for (i = 0; i < 4; i++) DBG_WARN(" %x ",admin_resp.resp_type.create_op_oq.status_desc[i]); } @@ -743,7 +743,7 @@ pqisrc_create_op_ibq(pqisrc_softstate_t *softs, } else { int i = 0; DBG_WARN("Error Status Decsriptors\n"); - for(i = 0; i < 4;i++) + for (i = 0; i < 4; i++) DBG_WARN(" %x ",admin_resp.resp_type.create_op_iq.status_desc[i]); } diff --git a/sys/dev/sym/sym_hipd.c b/sys/dev/sym/sym_hipd.c index fa65d544e17d..b4e5c1075fb4 100644 --- a/sys/dev/sym/sym_hipd.c +++ b/sys/dev/sym/sym_hipd.c @@ -3266,7 +3266,7 @@ static void sym_init (hcb_p np, int reason) * Reinitialize usrwide. * Prepare sync negotiation according to actual SCSI bus mode. */ - for (i=0;i<SYM_CONF_MAX_TARGET;i++) { + for (i = 0; i < SYM_CONF_MAX_TARGET; i++) { tcb_p tp = &np->target[i]; tp->to_reset = 0; @@ -3715,7 +3715,7 @@ static void sym_log_hard_error(hcb_p np, u_short sist, u_char dstat) } printf ("%s: regdump:", sym_name(np)); - for (i=0; i<24;i++) + for (i = 0; i < 24; i++) printf (" %02x", (unsigned)INB_OFF(i)); printf (".\n"); @@ -5527,8 +5527,8 @@ static int sym_show_msg (u_char * msg) u_char i; printf ("%x",*msg); if (*msg==M_EXTENDED) { - for (i=1;i<8;i++) { - if (i-1>msg[1]) break; + for (i = 1; i < 8; i++) { + if (i - 1 > msg[1]) break; printf ("-%x",msg[i]); } return (i+1); @@ -6744,10 +6744,10 @@ restart_test: /* * Wait 'til done (with timeout) */ - for (i=0; i<SYM_SNOOP_TIMEOUT; i++) + for (i = 0; i < SYM_SNOOP_TIMEOUT; i++) if (INB(nc_istat) & (INTF|SIP|DIP)) break; - if (i>=SYM_SNOOP_TIMEOUT) { + if (i >= SYM_SNOOP_TIMEOUT) { printf ("CACHE TEST FAILED: timeout.\n"); return (0x20); } diff --git a/sys/dev/tws/tws.c b/sys/dev/tws/tws.c index af151c8c4f06..fccd6689a6aa 100644 --- a/sys/dev/tws/tws.c +++ b/sys/dev/tws/tws.c @@ -311,7 +311,7 @@ attach_fail_4: if (sc->cmd_tag) bus_dma_tag_destroy(sc->cmd_tag); attach_fail_3: - for(i=0;i<sc->irqs;i++) { + for (i = 0; i < sc->irqs; i++) { if ( sc->irq_res[i] ){ if (bus_release_resource(sc->tws_dev, SYS_RES_IRQ, sc->irq_res_id[i], sc->irq_res[i])) @@ -369,7 +369,7 @@ tws_detach(device_t dev) tws_teardown_intr(sc); /* Release irq resource */ - for(i=0;i<sc->irqs;i++) { + for (i = 0; i < sc->irqs; i++) { if ( sc->irq_res[i] ){ if (bus_release_resource(sc->tws_dev, SYS_RES_IRQ, sc->irq_res_id[i], sc->irq_res[i])) @@ -402,7 +402,7 @@ tws_detach(device_t dev) TWS_TRACE(sc, "bus release mem resource", 0, sc->reg_res_id); } - for ( i=0; i< tws_queue_depth; i++) { + for (i = 0; i < tws_queue_depth; i++) { if (sc->reqs[i].dma_map) bus_dmamap_destroy(sc->data_tag, sc->reqs[i].dma_map); callout_drain(&sc->reqs[i].timeout); @@ -432,7 +432,7 @@ tws_setup_intr(struct tws_softc *sc, int irqs) { int i, error; - for(i=0;i<irqs;i++) { + for (i = 0; i < irqs; i++) { if (!(sc->intr_handle[i])) { if ((error = bus_setup_intr(sc->tws_dev, sc->irq_res[i], INTR_TYPE_CAM | INTR_MPSAFE, @@ -452,7 +452,7 @@ tws_teardown_intr(struct tws_softc *sc) { int i; - for(i=0;i<sc->irqs;i++) { + for (i = 0; i < sc->irqs; i++) { if (sc->intr_handle[i]) { bus_teardown_intr(sc->tws_dev, sc->irq_res[i], sc->intr_handle[i]); @@ -669,8 +669,7 @@ tws_init_reqs(struct tws_softc *sc, u_int32_t dma_mem_size) bzero(cmd_buf, dma_mem_size); TWS_TRACE_DEBUG(sc, "phy cmd", sc->dma_mem_phys, 0); mtx_lock(&sc->q_lock); - for ( i=0; i< tws_queue_depth; i++) - { + for (i = 0; i < tws_queue_depth; i++) { if (bus_dmamap_create(sc->data_tag, 0, &sc->reqs[i].dma_map)) { /* log a ENOMEM failure msg here */ mtx_unlock(&sc->q_lock); diff --git a/sys/dev/tws/tws_services.c b/sys/dev/tws/tws_services.c index da8bbacc39f7..e5c3d45c533f 100644 --- a/sys/dev/tws/tws_services.c +++ b/sys/dev/tws/tws_services.c @@ -200,7 +200,7 @@ tws_init_qs(struct tws_softc *sc) { mtx_lock(&sc->q_lock); - for(int i=0;i<TWS_MAX_QS;i++) { + for (int i = 0; i < TWS_MAX_QS; i++) { sc->q_head[i] = NULL; sc->q_tail[i] = NULL; } diff --git a/sys/dev/usb/usb_dev.c b/sys/dev/usb/usb_dev.c index 293b0c72587f..e58d6a674ec0 100644 --- a/sys/dev/usb/usb_dev.c +++ b/sys/dev/usb/usb_dev.c @@ -1231,12 +1231,14 @@ static const struct filterops usb_filtops_write = { .f_isfd = 1, .f_detach = usb_filter_detach, .f_event = usb_filter_write, + .f_copy = knote_triv_copy, }; static const struct filterops usb_filtops_read = { .f_isfd = 1, .f_detach = usb_filter_detach, .f_event = usb_filter_read, + .f_copy = knote_triv_copy, }; /* ARGSUSED */ diff --git a/sys/dev/vmm/vmm_dev.c b/sys/dev/vmm/vmm_dev.c index 460a508a60dc..4961b21180e1 100644 --- a/sys/dev/vmm/vmm_dev.c +++ b/sys/dev/vmm/vmm_dev.c @@ -120,18 +120,18 @@ vcpu_unlock_one(struct vcpu *vcpu) vcpu_set_state(vcpu, VCPU_IDLE, false); } +#ifndef __amd64__ static int -vcpu_lock_all(struct vmmdev_softc *sc) +vcpu_set_state_all(struct vm *vm, enum vcpu_state newstate) { struct vcpu *vcpu; int error; uint16_t i, j, maxcpus; error = 0; - vm_slock_vcpus(sc->vm); - maxcpus = vm_get_maxcpus(sc->vm); + maxcpus = vm_get_maxcpus(vm); for (i = 0; i < maxcpus; i++) { - vcpu = vm_vcpu(sc->vm, i); + vcpu = vm_vcpu(vm, i); if (vcpu == NULL) continue; error = vcpu_lock_one(vcpu); @@ -141,16 +141,32 @@ vcpu_lock_all(struct vmmdev_softc *sc) if (error) { for (j = 0; j < i; j++) { - vcpu = vm_vcpu(sc->vm, j); + vcpu = vm_vcpu(vm, j); if (vcpu == NULL) continue; vcpu_unlock_one(vcpu); } - vm_unlock_vcpus(sc->vm); } return (error); } +#endif + +static int +vcpu_lock_all(struct vmmdev_softc *sc) +{ + int error; + + /* + * Serialize vcpu_lock_all() callers. Individual vCPUs are not locked + * in a consistent order so we need to serialize to avoid deadlocks. + */ + vm_lock_vcpus(sc->vm); + error = vcpu_set_state_all(sc->vm, VCPU_FROZEN); + if (error != 0) + vm_unlock_vcpus(sc->vm); + return (error); +} static void vcpu_unlock_all(struct vmmdev_softc *sc) diff --git a/sys/fs/cuse/cuse.c b/sys/fs/cuse/cuse.c index d63a7d4691cf..b2524324584a 100644 --- a/sys/fs/cuse/cuse.c +++ b/sys/fs/cuse/cuse.c @@ -195,12 +195,14 @@ static const struct filterops cuse_client_kqfilter_read_ops = { .f_isfd = 1, .f_detach = cuse_client_kqfilter_read_detach, .f_event = cuse_client_kqfilter_read_event, + .f_copy = knote_triv_copy, }; static const struct filterops cuse_client_kqfilter_write_ops = { .f_isfd = 1, .f_detach = cuse_client_kqfilter_write_detach, .f_event = cuse_client_kqfilter_write_event, + .f_copy = knote_triv_copy, }; static d_open_t cuse_client_open; diff --git a/sys/fs/devfs/devfs_dir.c b/sys/fs/devfs/devfs_dir.c index 3dc87538017d..aad87606e738 100644 --- a/sys/fs/devfs/devfs_dir.c +++ b/sys/fs/devfs/devfs_dir.c @@ -162,7 +162,7 @@ int devfs_pathpath(const char *p1, const char *p2) { - for (;;p1++, p2++) { + for (;; p1++, p2++) { if (*p1 != *p2) { if (*p1 == '/' && *p2 == '\0') return (1); diff --git a/sys/fs/fuse/fuse_device.c b/sys/fs/fuse/fuse_device.c index 57b3559731f7..75bc0357571f 100644 --- a/sys/fs/fuse/fuse_device.c +++ b/sys/fs/fuse/fuse_device.c @@ -126,11 +126,13 @@ static const struct filterops fuse_device_rfiltops = { .f_isfd = 1, .f_detach = fuse_device_filt_detach, .f_event = fuse_device_filt_read, + .f_copy = knote_triv_copy, }; static const struct filterops fuse_device_wfiltops = { .f_isfd = 1, .f_event = fuse_device_filt_write, + .f_copy = knote_triv_copy, }; /**************************** diff --git a/sys/fs/udf/osta.c b/sys/fs/udf/osta.c index f79b86993367..1a083d8c26b1 100644 --- a/sys/fs/udf/osta.c +++ b/sys/fs/udf/osta.c @@ -383,7 +383,7 @@ int UDFTransName( int maxFilenameLen; /* Translate extension, and store it in ext. */ for(index = 0; index<EXT_SIZE && - extIndex + index +1 < udfLen; index++ ) { + extIndex + index +1 < udfLen; index++) { current = udfName[extIndex + index + 1]; if (IsIllegal(current) || !UnicodeIsPrint(current)) { @@ -432,7 +432,7 @@ int UDFTransName( /* Place a translated extension at end, if found. */ if (hasExt) { newName[newIndex++] = PERIOD; - for (index = 0;index < localExtIndex ;index++ ) { + for (index = 0; index < localExtIndex; index++) { newName[newIndex++] = ext[index]; } } diff --git a/sys/fs/unionfs/union_vnops.c b/sys/fs/unionfs/union_vnops.c index 26fa14603c85..66fee97a07d5 100644 --- a/sys/fs/unionfs/union_vnops.c +++ b/sys/fs/unionfs/union_vnops.c @@ -2208,7 +2208,6 @@ unionfs_lock_restart: vholdnz(tvp); VI_UNLOCK(vp); error = VOP_LOCK(tvp, flags); - vdrop(tvp); if (error == 0 && (lvp_locked || VTOUNIONFS(vp) == NULL)) { /* * After dropping the interlock above, there exists a window @@ -2234,6 +2233,7 @@ unionfs_lock_restart: unp = VTOUNIONFS(vp); if (unp == NULL || unp->un_uppervp != NULL) { VOP_UNLOCK(tvp); + vdrop(tvp); /* * If we previously held the lock, the upgrade may * have temporarily dropped the lock, in which case @@ -2249,6 +2249,7 @@ unionfs_lock_restart: goto unionfs_lock_restart; } } + vdrop(tvp); return (error); } @@ -2259,7 +2260,6 @@ unionfs_unlock(struct vop_unlock_args *ap) struct vnode *vp; struct vnode *tvp; struct unionfs_node *unp; - int error; KASSERT_UNIONFS_VNODE(ap->a_vp); @@ -2271,11 +2271,7 @@ unionfs_unlock(struct vop_unlock_args *ap) tvp = (unp->un_uppervp != NULL ? unp->un_uppervp : unp->un_lowervp); - vholdnz(tvp); - error = VOP_UNLOCK(tvp); - vdrop(tvp); - - return (error); + return (VOP_UNLOCK(tvp)); } static int diff --git a/sys/geom/geom_dev.c b/sys/geom/geom_dev.c index 27c65f15d5e3..db0bc77a752f 100644 --- a/sys/geom/geom_dev.c +++ b/sys/geom/geom_dev.c @@ -82,6 +82,7 @@ static const struct filterops gdev_filterops_vnode = { .f_isfd = 1, .f_detach = gdev_filter_detach, .f_event = gdev_filter_vnode, + .f_copy = knote_triv_copy, }; static struct cdevsw g_dev_cdevsw = { diff --git a/sys/i386/i386/elf_machdep.c b/sys/i386/i386/elf_machdep.c index 13769af0fbca..14c942942d08 100644 --- a/sys/i386/i386/elf_machdep.c +++ b/sys/i386/i386/elf_machdep.c @@ -92,7 +92,7 @@ struct sysentvec elf32_freebsd_sysvec = { }; INIT_SYSENTVEC(elf32_sysvec, &elf32_freebsd_sysvec); -static Elf32_Brandinfo freebsd_brand_info = { +static const Elf32_Brandinfo freebsd_brand_info = { .brand = ELFOSABI_FREEBSD, .machine = EM_386, .compat_3_brand = "FreeBSD", @@ -103,11 +103,11 @@ static Elf32_Brandinfo freebsd_brand_info = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -SYSINIT(elf32, SI_SUB_EXEC, SI_ORDER_FIRST, +C_SYSINIT(elf32, SI_SUB_EXEC, SI_ORDER_FIRST, (sysinit_cfunc_t) elf32_insert_brand_entry, &freebsd_brand_info); -static Elf32_Brandinfo freebsd_brand_oinfo = { +static const Elf32_Brandinfo freebsd_brand_oinfo = { .brand = ELFOSABI_FREEBSD, .machine = EM_386, .compat_3_brand = "FreeBSD", @@ -118,11 +118,11 @@ static Elf32_Brandinfo freebsd_brand_oinfo = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -SYSINIT(oelf32, SI_SUB_EXEC, SI_ORDER_ANY, +C_SYSINIT(oelf32, SI_SUB_EXEC, SI_ORDER_ANY, (sysinit_cfunc_t) elf32_insert_brand_entry, &freebsd_brand_oinfo); -static Elf32_Brandinfo kfreebsd_brand_info = { +static const Elf32_Brandinfo kfreebsd_brand_info = { .brand = ELFOSABI_FREEBSD, .machine = EM_386, .compat_3_brand = "FreeBSD", @@ -133,7 +133,7 @@ static Elf32_Brandinfo kfreebsd_brand_info = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE_MANDATORY }; -SYSINIT(kelf32, SI_SUB_EXEC, SI_ORDER_ANY, +C_SYSINIT(kelf32, SI_SUB_EXEC, SI_ORDER_ANY, (sysinit_cfunc_t) elf32_insert_brand_entry, &kfreebsd_brand_info); diff --git a/sys/i386/i386/in_cksum_machdep.c b/sys/i386/i386/in_cksum_machdep.c index 27ab09d82da0..b658d85bc892 100644 --- a/sys/i386/i386/in_cksum_machdep.c +++ b/sys/i386/i386/in_cksum_machdep.c @@ -84,7 +84,7 @@ in_cksum_skip(struct mbuf *m, int len, int skip) } } - for (;m && len; m = m->m_next) { + for (; m && len; m = m->m_next) { if (m->m_len == 0) continue; w = mtod(m, u_short *); diff --git a/sys/i386/linux/linux_sysvec.c b/sys/i386/linux/linux_sysvec.c index 85877bf40997..14b5f64388d2 100644 --- a/sys/i386/linux/linux_sysvec.c +++ b/sys/i386/linux/linux_sysvec.c @@ -796,7 +796,7 @@ linux_vdso_reloc(char *mapping, Elf_Addr offset) } } -static Elf_Brandnote linux_brandnote = { +static const Elf_Brandnote linux_brandnote = { .hdr.n_namesz = sizeof(GNU_ABI_VENDOR), .hdr.n_descsz = 16, /* XXX at least 16 */ .hdr.n_type = 1, @@ -805,7 +805,7 @@ static Elf_Brandnote linux_brandnote = { .trans_osrel = linux_trans_osrel }; -static Elf32_Brandinfo linux_brand = { +static const Elf32_Brandinfo linux_brand = { .brand = ELFOSABI_LINUX, .machine = EM_386, .compat_3_brand = "Linux", @@ -816,7 +816,7 @@ static Elf32_Brandinfo linux_brand = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -static Elf32_Brandinfo linux_glibc2brand = { +static const Elf32_Brandinfo linux_glibc2brand = { .brand = ELFOSABI_LINUX, .machine = EM_386, .compat_3_brand = "Linux", @@ -827,7 +827,7 @@ static Elf32_Brandinfo linux_glibc2brand = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -static Elf32_Brandinfo linux_muslbrand = { +static const Elf32_Brandinfo linux_muslbrand = { .brand = ELFOSABI_LINUX, .machine = EM_386, .compat_3_brand = "Linux", @@ -839,7 +839,7 @@ static Elf32_Brandinfo linux_muslbrand = { LINUX_BI_FUTEX_REQUEUE }; -Elf32_Brandinfo *linux_brandlist[] = { +const Elf32_Brandinfo *linux_brandlist[] = { &linux_brand, &linux_glibc2brand, &linux_muslbrand, @@ -849,7 +849,7 @@ Elf32_Brandinfo *linux_brandlist[] = { static int linux_elf_modevent(module_t mod, int type, void *data) { - Elf32_Brandinfo **brandinfo; + const Elf32_Brandinfo **brandinfo; int error; struct linux_ioctl_handler **lihp; diff --git a/sys/kern/imgact_elf.c b/sys/kern/imgact_elf.c index c53707a1286c..779158b41221 100644 --- a/sys/kern/imgact_elf.c +++ b/sys/kern/imgact_elf.c @@ -231,7 +231,7 @@ static const Elf_Brandinfo *elf_brand_list[MAX_BRANDS]; #define aligned(a, t) (rounddown2((u_long)(a), sizeof(t)) == (u_long)(a)) -Elf_Brandnote __elfN(freebsd_brandnote) = { +const Elf_Brandnote __elfN(freebsd_brandnote) = { .hdr.n_namesz = sizeof(FREEBSD_ABI_VENDOR), .hdr.n_descsz = sizeof(int32_t), .hdr.n_type = NT_FREEBSD_ABI_TAG, @@ -254,7 +254,7 @@ __elfN(freebsd_trans_osrel)(const Elf_Note *note, int32_t *osrel) static int GNU_KFREEBSD_ABI_DESC = 3; -Elf_Brandnote __elfN(kfreebsd_brandnote) = { +const Elf_Brandnote __elfN(kfreebsd_brandnote) = { .hdr.n_namesz = sizeof(GNU_ABI_VENDOR), .hdr.n_descsz = 16, /* XXX at least 16 */ .hdr.n_type = 1, @@ -2831,7 +2831,7 @@ __elfN(parse_notes)(const struct image_params *imgp, const Elf_Note *checknote, } if ((const char *)note_end - (const char *)note < sizeof(Elf_Note)) { - uprintf("ELF note to short\n"); + uprintf("ELF note too short\n"); goto retf; } if (note->n_namesz != checknote->n_namesz || @@ -2839,9 +2839,9 @@ __elfN(parse_notes)(const struct image_params *imgp, const Elf_Note *checknote, note->n_type != checknote->n_type) goto nextnote; note_name = (const char *)(note + 1); - if (note_name + checknote->n_namesz >= - (const char *)note_end || strncmp(note_vendor, - note_name, checknote->n_namesz) != 0) + if (note_name + roundup2(note->n_namesz, ELF_NOTE_ROUNDSIZE) + + note->n_descsz >= (const char *)note_end || + strncmp(note_vendor, note_name, checknote->n_namesz) != 0) goto nextnote; if (cb(note, cb_arg, &res)) diff --git a/sys/kern/kern_descrip.c b/sys/kern/kern_descrip.c index 19118eb7f275..a71a601733e5 100644 --- a/sys/kern/kern_descrip.c +++ b/sys/kern/kern_descrip.c @@ -2486,7 +2486,7 @@ fdunshare(struct thread *td) if (refcount_load(&p->p_fd->fd_refcnt) == 1) return; - tmp = fdcopy(p->p_fd); + tmp = fdcopy(p->p_fd, p); fdescfree(td); p->p_fd = tmp; } @@ -2515,14 +2515,17 @@ pdunshare(struct thread *td) * this is to ease callers, not catch errors. */ struct filedesc * -fdcopy(struct filedesc *fdp) +fdcopy(struct filedesc *fdp, struct proc *p1) { struct filedesc *newfdp; struct filedescent *nfde, *ofde; + struct file *fp; int i, lastfile; + bool fork_pass; MPASS(fdp != NULL); + fork_pass = false; newfdp = fdinit(); FILEDESC_SLOCK(fdp); for (;;) { @@ -2533,10 +2536,35 @@ fdcopy(struct filedesc *fdp) fdgrowtable(newfdp, lastfile + 1); FILEDESC_SLOCK(fdp); } - /* copy all passable descriptors (i.e. not kqueue) */ + + /* + * Copy all passable descriptors (i.e. not kqueue), and + * prepare to handle copyable but not passable descriptors + * (kqueues). + * + * The pass to handle copying is performed after all passable + * files are installed into the new file descriptor's table, + * since kqueues need all referenced file descriptors already + * valid, including other kqueues. For the same reason the + * copying is done in two passes by itself, first installing + * not fully initialized ('empty') copyable files into the new + * fd table, and then giving the subsystems a second chance to + * really fill the copied file backing structure with the + * content. + */ newfdp->fd_freefile = fdp->fd_freefile; FILEDESC_FOREACH_FDE(fdp, i, ofde) { - if ((ofde->fde_file->f_ops->fo_flags & DFLAG_PASSABLE) == 0 || + const struct fileops *ops; + + ops = ofde->fde_file->f_ops; + fp = NULL; + if ((ops->fo_flags & DFLAG_FORK) != 0 && + (ofde->fde_flags & UF_FOCLOSE) == 0) { + if (ops->fo_fork(newfdp, ofde->fde_file, &fp, p1, + curthread) != 0) + continue; + fork_pass = true; + } else if ((ops->fo_flags & DFLAG_PASSABLE) == 0 || (ofde->fde_flags & UF_FOCLOSE) != 0 || !fhold(ofde->fde_file)) { if (newfdp->fd_freefile == fdp->fd_freefile) @@ -2545,11 +2573,30 @@ fdcopy(struct filedesc *fdp) } nfde = &newfdp->fd_ofiles[i]; *nfde = *ofde; + if (fp != NULL) + nfde->fde_file = fp; filecaps_copy(&ofde->fde_caps, &nfde->fde_caps, true); fdused_init(newfdp, i); } MPASS(newfdp->fd_freefile != -1); FILEDESC_SUNLOCK(fdp); + + /* + * Now handle copying kqueues, since all fds, including + * kqueues, are in place. + */ + if (__predict_false(fork_pass)) { + FILEDESC_FOREACH_FDE(newfdp, i, nfde) { + const struct fileops *ops; + + ops = nfde->fde_file->f_ops; + if ((ops->fo_flags & DFLAG_FORK) == 0 || + nfde->fde_file == NULL) + continue; + ops->fo_fork(newfdp, NULL, &nfde->fde_file, p1, + curthread); + } + } return (newfdp); } diff --git a/sys/kern/kern_devctl.c b/sys/kern/kern_devctl.c index a1696225df32..a37cb23efed8 100644 --- a/sys/kern/kern_devctl.c +++ b/sys/kern/kern_devctl.c @@ -130,6 +130,7 @@ static const struct filterops devctl_rfiltops = { .f_isfd = 1, .f_detach = filt_devctl_detach, .f_event = filt_devctl_read, + .f_copy = knote_triv_copy, }; static struct cdev *devctl_dev; diff --git a/sys/kern/kern_event.c b/sys/kern/kern_event.c index a6333d8011b1..1baa24d278bf 100644 --- a/sys/kern/kern_event.c +++ b/sys/kern/kern_event.c @@ -134,6 +134,7 @@ static fo_kqfilter_t kqueue_kqfilter; static fo_stat_t kqueue_stat; static fo_close_t kqueue_close; static fo_fill_kinfo_t kqueue_fill_kinfo; +static fo_fork_t kqueue_fork; static const struct fileops kqueueops = { .fo_read = invfo_rdwr, @@ -148,7 +149,9 @@ static const struct fileops kqueueops = { .fo_chown = invfo_chown, .fo_sendfile = invfo_sendfile, .fo_cmp = file_kcmp_generic, + .fo_fork = kqueue_fork, .fo_fill_kinfo = kqueue_fill_kinfo, + .fo_flags = DFLAG_FORK, }; static int knote_attach(struct knote *kn, struct kqueue *kq); @@ -176,6 +179,7 @@ static void filt_timerdetach(struct knote *kn); static void filt_timerstart(struct knote *kn, sbintime_t to); static void filt_timertouch(struct knote *kn, struct kevent *kev, u_long type); +static int filt_timercopy(struct knote *kn, struct proc *p1); static int filt_timervalidate(struct knote *kn, sbintime_t *to); static int filt_timer(struct knote *kn, long hint); static int filt_userattach(struct knote *kn); @@ -187,11 +191,13 @@ static void filt_usertouch(struct knote *kn, struct kevent *kev, static const struct filterops file_filtops = { .f_isfd = 1, .f_attach = filt_fileattach, + .f_copy = knote_triv_copy, }; static const struct filterops kqread_filtops = { .f_isfd = 1, .f_detach = filt_kqdetach, .f_event = filt_kqueue, + .f_copy = knote_triv_copy, }; /* XXX - move to kern_proc.c? */ static const struct filterops proc_filtops = { @@ -199,12 +205,14 @@ static const struct filterops proc_filtops = { .f_attach = filt_procattach, .f_detach = filt_procdetach, .f_event = filt_proc, + .f_copy = knote_triv_copy, }; static const struct filterops jail_filtops = { .f_isfd = 0, .f_attach = filt_jailattach, .f_detach = filt_jaildetach, .f_event = filt_jail, + .f_copy = knote_triv_copy, }; static const struct filterops timer_filtops = { .f_isfd = 0, @@ -212,12 +220,14 @@ static const struct filterops timer_filtops = { .f_detach = filt_timerdetach, .f_event = filt_timer, .f_touch = filt_timertouch, + .f_copy = filt_timercopy, }; static const struct filterops user_filtops = { .f_attach = filt_userattach, .f_detach = filt_userdetach, .f_event = filt_user, .f_touch = filt_usertouch, + .f_copy = knote_triv_copy, }; static uma_zone_t knote_zone; @@ -347,6 +357,7 @@ filt_nullattach(struct knote *kn) static const struct filterops null_filtops = { .f_isfd = 0, .f_attach = filt_nullattach, + .f_copy = knote_triv_copy, }; /* XXX - make SYSINIT to add these, and move into respective modules. */ @@ -940,6 +951,30 @@ filt_timerattach(struct knote *kn) return (0); } +static int +filt_timercopy(struct knote *kn, struct proc *p) +{ + struct kq_timer_cb_data *kc_src, *kc; + + if (atomic_fetchadd_int(&kq_ncallouts, 1) + 1 > kq_calloutmax) { + atomic_subtract_int(&kq_ncallouts, 1); + return (ENOMEM); + } + + kn->kn_status &= ~KN_DETACHED; + kc_src = kn->kn_ptr.p_v; + kn->kn_ptr.p_v = kc = malloc(sizeof(*kc), M_KQUEUE, M_WAITOK); + kc->kn = kn; + kc->p = p; + kc->flags = kc_src->flags & ~KQ_TIMER_CB_ENQUEUED; + kc->next = kc_src->next; + kc->to = kc_src->to; + kc->cpuid = PCPU_GET(cpuid); + callout_init(&kc->c, 1); + kqtimer_sched_callout(kc); + return (0); +} + static void filt_timerstart(struct knote *kn, sbintime_t to) { @@ -1151,7 +1186,7 @@ int sys_kqueue(struct thread *td, struct kqueue_args *uap) { - return (kern_kqueue(td, 0, NULL)); + return (kern_kqueue(td, 0, false, NULL)); } int @@ -1159,55 +1194,76 @@ sys_kqueuex(struct thread *td, struct kqueuex_args *uap) { int flags; - if ((uap->flags & ~(KQUEUE_CLOEXEC)) != 0) + if ((uap->flags & ~(KQUEUE_CLOEXEC | KQUEUE_CPONFORK)) != 0) return (EINVAL); flags = 0; if ((uap->flags & KQUEUE_CLOEXEC) != 0) flags |= O_CLOEXEC; - return (kern_kqueue(td, flags, NULL)); + return (kern_kqueue(td, flags, (uap->flags & KQUEUE_CPONFORK) != 0, + NULL)); } static void -kqueue_init(struct kqueue *kq) +kqueue_init(struct kqueue *kq, bool cponfork) { mtx_init(&kq->kq_lock, "kqueue", NULL, MTX_DEF | MTX_DUPOK); TAILQ_INIT(&kq->kq_head); knlist_init_mtx(&kq->kq_sel.si_note, &kq->kq_lock); TASK_INIT(&kq->kq_task, 0, kqueue_task, kq); + if (cponfork) + kq->kq_state |= KQ_CPONFORK; } -int -kern_kqueue(struct thread *td, int flags, struct filecaps *fcaps) +static int +kern_kqueue_alloc(struct thread *td, struct filedesc *fdp, int *fdip, + struct file **fpp, int flags, struct filecaps *fcaps, bool cponfork, + struct kqueue **kqp) { - struct filedesc *fdp; - struct kqueue *kq; - struct file *fp; struct ucred *cred; - int fd, error; + struct kqueue *kq; + int error; - fdp = td->td_proc->p_fd; cred = td->td_ucred; if (!chgkqcnt(cred->cr_ruidinfo, 1, lim_cur(td, RLIMIT_KQUEUES))) return (ENOMEM); - error = falloc_caps(td, &fp, &fd, flags, fcaps); + error = fdip != NULL ? falloc_caps(td, fpp, fdip, flags, fcaps) : + _falloc_noinstall(td, fpp, 1); if (error != 0) { chgkqcnt(cred->cr_ruidinfo, -1, 0); return (error); } /* An extra reference on `fp' has been held for us by falloc(). */ - kq = malloc(sizeof *kq, M_KQUEUE, M_WAITOK | M_ZERO); - kqueue_init(kq); + kq = malloc(sizeof(*kq), M_KQUEUE, M_WAITOK | M_ZERO); + kqueue_init(kq, cponfork); kq->kq_fdp = fdp; kq->kq_cred = crhold(cred); - FILEDESC_XLOCK(fdp); + if (fdip != NULL) + FILEDESC_XLOCK(fdp); TAILQ_INSERT_HEAD(&fdp->fd_kqlist, kq, kq_list); - FILEDESC_XUNLOCK(fdp); + if (fdip != NULL) + FILEDESC_XUNLOCK(fdp); + + finit(*fpp, FREAD | FWRITE, DTYPE_KQUEUE, kq, &kqueueops); + *kqp = kq; + return (0); +} + +int +kern_kqueue(struct thread *td, int flags, bool cponfork, struct filecaps *fcaps) +{ + struct kqueue *kq; + struct file *fp; + int fd, error; + + error = kern_kqueue_alloc(td, td->td_proc->p_fd, &fd, &fp, flags, + fcaps, cponfork, &kq); + if (error != 0) + return (error); - finit(fp, FREAD | FWRITE, DTYPE_KQUEUE, kq, &kqueueops); fdrop(fp, td); td->td_retval[0] = fd; @@ -1488,7 +1544,7 @@ kern_kevent_anonymous(struct thread *td, int nevents, struct kqueue kq = {}; int error; - kqueue_init(&kq); + kqueue_init(&kq, false); kq.kq_refcnt = 1; error = kqueue_kevent(&kq, td, nevents, nevents, k_ops, NULL); kqueue_drain(&kq, td); @@ -1576,7 +1632,7 @@ kqueue_fo_release(int filt) mtx_lock(&filterops_lock); KASSERT(sysfilt_ops[~filt].for_refcnt > 0, - ("filter object refcount not valid on release")); + ("filter object %d refcount not valid on release", filt)); sysfilt_ops[~filt].for_refcnt--; mtx_unlock(&filterops_lock); } @@ -1855,17 +1911,8 @@ done: } static int -kqueue_acquire(struct file *fp, struct kqueue **kqp) +kqueue_acquire_ref(struct kqueue *kq) { - int error; - struct kqueue *kq; - - error = 0; - - kq = fp->f_data; - if (fp->f_type != DTYPE_KQUEUE || kq == NULL) - return (EINVAL); - *kqp = kq; KQ_LOCK(kq); if ((kq->kq_state & KQ_CLOSING) == KQ_CLOSING) { KQ_UNLOCK(kq); @@ -1873,8 +1920,22 @@ kqueue_acquire(struct file *fp, struct kqueue **kqp) } kq->kq_refcnt++; KQ_UNLOCK(kq); + return (0); +} - return error; +static int +kqueue_acquire(struct file *fp, struct kqueue **kqp) +{ + struct kqueue *kq; + int error; + + kq = fp->f_data; + if (fp->f_type != DTYPE_KQUEUE || kq == NULL) + return (EINVAL); + error = kqueue_acquire_ref(kq); + if (error == 0) + *kqp = kq; + return (error); } static void @@ -2937,6 +2998,152 @@ noacquire: return (error); } +static int +kqueue_fork_alloc(struct filedesc *fdp, struct file *fp, struct file **fp1, + struct thread *td) +{ + struct kqueue *kq, *kq1; + int error; + + MPASS(fp->f_type == DTYPE_KQUEUE); + kq = fp->f_data; + if ((kq->kq_state & KQ_CPONFORK) == 0) + return (EOPNOTSUPP); + error = kqueue_acquire_ref(kq); + if (error != 0) + return (error); + error = kern_kqueue_alloc(td, fdp, NULL, fp1, 0, NULL, true, &kq1); + if (error == 0) { + kq1->kq_forksrc = kq; + (*fp1)->f_flag = fp->f_flag & (FREAD | FWRITE | FEXEC | + O_CLOEXEC | O_CLOFORK); + } else { + kqueue_release(kq, 0); + } + return (error); +} + +static void +kqueue_fork_copy_knote(struct kqueue *kq1, struct knote *kn, struct proc *p1, + struct filedesc *fdp) +{ + struct knote *kn1; + const struct filterops *fop; + int error; + + fop = kn->kn_fop; + if (fop->f_copy == NULL || (fop->f_isfd && + fdp->fd_files->fdt_ofiles[kn->kn_kevent.ident].fde_file == NULL)) + return; + error = kqueue_expand(kq1, fop, kn->kn_kevent.ident, M_WAITOK); + if (error != 0) + return; + + kn1 = knote_alloc(M_WAITOK); + *kn1 = *kn; + kn1->kn_status |= KN_DETACHED; + kn1->kn_status &= ~KN_QUEUED; + kn1->kn_kq = kq1; + error = fop->f_copy(kn1, p1); + if (error != 0) { + knote_free(kn1); + return; + } + (void)kqueue_fo_find(kn->kn_kevent.filter); + if (fop->f_isfd && !fhold(kn1->kn_fp)) { + fop->f_detach(kn1); + kqueue_fo_release(kn->kn_kevent.filter); + knote_free(kn1); + return; + } + if (kn->kn_knlist != NULL) + knlist_add(kn->kn_knlist, kn1, 0); + KQ_LOCK(kq1); + knote_attach(kn1, kq1); + kn1->kn_influx = 0; + if ((kn->kn_status & KN_QUEUED) != 0) + knote_enqueue(kn1); + KQ_UNLOCK(kq1); +} + +static void +kqueue_fork_copy_list(struct klist *knlist, struct knote *marker, + struct kqueue *kq, struct kqueue *kq1, struct proc *p1, + struct filedesc *fdp) +{ + struct knote *kn; + + KQ_OWNED(kq); + kn = SLIST_FIRST(knlist); + while (kn != NULL) { + if ((kn->kn_status & KN_DETACHED) != 0 || + (kn_in_flux(kn) && (kn->kn_status & KN_SCAN) == 0)) { + kn = SLIST_NEXT(kn, kn_link); + continue; + } + kn_enter_flux(kn); + SLIST_INSERT_AFTER(kn, marker, kn_link); + KQ_UNLOCK(kq); + kqueue_fork_copy_knote(kq1, kn, p1, fdp); + KQ_LOCK(kq); + kn_leave_flux(kn); + kn = SLIST_NEXT(marker, kn_link); + /* XXXKIB switch kn_link to LIST? */ + SLIST_REMOVE(knlist, marker, knote, kn_link); + } +} + +static int +kqueue_fork_copy(struct filedesc *fdp, struct file *fp, struct file *fp1, + struct proc *p1, struct thread *td) +{ + struct kqueue *kq, *kq1; + struct knote *marker; + int error, i; + + error = 0; + MPASS(fp == NULL); + MPASS(fp1->f_type == DTYPE_KQUEUE); + + kq1 = fp1->f_data; + kq = kq1->kq_forksrc; + marker = knote_alloc(M_WAITOK); + marker->kn_status = KN_MARKER; + + KQ_LOCK(kq); + for (i = 0; i < kq->kq_knlistsize; i++) { + kqueue_fork_copy_list(&kq->kq_knlist[i], marker, kq, kq1, + p1, fdp); + } + if (kq->kq_knhashmask != 0) { + for (i = 0; i <= kq->kq_knhashmask; i++) { + kqueue_fork_copy_list(&kq->kq_knhash[i], marker, kq, + kq1, p1, fdp); + } + } + kqueue_release(kq, 1); + kq1->kq_forksrc = NULL; + KQ_UNLOCK(kq); + + knote_free(marker); + return (error); +} + +static int +kqueue_fork(struct filedesc *fdp, struct file *fp, struct file **fp1, + struct proc *p1, struct thread *td) +{ + if (*fp1 == NULL) + return (kqueue_fork_alloc(fdp, fp, fp1, td)); + return (kqueue_fork_copy(fdp, fp, *fp1, p1, td)); +} + +int +knote_triv_copy(struct knote *kn __unused, struct proc *p1 __unused) +{ + return (0); +} + struct knote_status_export_bit { int kn_status_bit; int knt_status_bit; diff --git a/sys/kern/kern_exit.c b/sys/kern/kern_exit.c index ab8ed32ad189..c4b1c8201ff2 100644 --- a/sys/kern/kern_exit.c +++ b/sys/kern/kern_exit.c @@ -807,7 +807,7 @@ kern_abort2(struct thread *td, const char *why, int nargs, void **uargs) } if (nargs > 0) { sbuf_putc(sb, '('); - for (i = 0;i < nargs; i++) + for (i = 0; i < nargs; i++) sbuf_printf(sb, "%s%p", i == 0 ? "" : ", ", uargs[i]); sbuf_putc(sb, ')'); } diff --git a/sys/kern/kern_fork.c b/sys/kern/kern_fork.c index 7f6abae187b3..8b237b6dbd17 100644 --- a/sys/kern/kern_fork.c +++ b/sys/kern/kern_fork.c @@ -423,7 +423,7 @@ do_fork(struct thread *td, struct fork_req *fr, struct proc *p2, struct thread * pd = pdshare(p1->p_pd); else pd = pdcopy(p1->p_pd); - fd = fdcopy(p1->p_fd); + fd = fdcopy(p1->p_fd, p2); fdtol = NULL; } else { if (fr->fr_flags2 & FR2_SHARE_PATHS) diff --git a/sys/kern/kern_jaildesc.c b/sys/kern/kern_jaildesc.c index 3f322b271400..a564393d3366 100644 --- a/sys/kern/kern_jaildesc.c +++ b/sys/kern/kern_jaildesc.c @@ -344,6 +344,7 @@ static const struct filterops jaildesc_kqops = { .f_isfd = 1, .f_detach = jaildesc_kqops_detach, .f_event = jaildesc_kqops_event, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/kern/kern_malloc.c b/sys/kern/kern_malloc.c index e919b15543b2..fcbfbe64f854 100644 --- a/sys/kern/kern_malloc.c +++ b/sys/kern/kern_malloc.c @@ -1302,7 +1302,7 @@ mallocinit(void *dummy) #endif align, UMA_ZONE_MALLOC); } - for (;i <= size; i+= KMEM_ZBASE) + for (; i <= size; i+= KMEM_ZBASE) kmemsize[i >> KMEM_ZSHIFT] = indx; } } diff --git a/sys/kern/kern_sig.c b/sys/kern/kern_sig.c index 21f765b17f62..a55f3c761449 100644 --- a/sys/kern/kern_sig.c +++ b/sys/kern/kern_sig.c @@ -124,6 +124,7 @@ const struct filterops sig_filtops = { .f_attach = filt_sigattach, .f_detach = filt_sigdetach, .f_event = filt_signal, + .f_copy = knote_triv_copy, }; static int kern_forcesigexit = 1; diff --git a/sys/kern/subr_devstat.c b/sys/kern/subr_devstat.c index 07a9cc0f57be..c4d0223d484f 100644 --- a/sys/kern/subr_devstat.c +++ b/sys/kern/subr_devstat.c @@ -415,7 +415,7 @@ sysctl_devstat(SYSCTL_HANDLER_ARGS) if (error != 0) return (error); - for (;nds != NULL;) { + while (nds != NULL) { error = SYSCTL_OUT(req, nds, sizeof(struct devstat)); if (error != 0) return (error); diff --git a/sys/kern/subr_log.c b/sys/kern/subr_log.c index 5380902e602f..aac35a56130e 100644 --- a/sys/kern/subr_log.c +++ b/sys/kern/subr_log.c @@ -79,6 +79,7 @@ static const struct filterops log_read_filterops = { .f_attach = NULL, .f_detach = logkqdetach, .f_event = logkqread, + .f_copy = knote_triv_copy, }; static struct logsoftc { diff --git a/sys/kern/subr_prf.c b/sys/kern/subr_prf.c index db0ceb17b9f0..e2070ae3f865 100644 --- a/sys/kern/subr_prf.c +++ b/sys/kern/subr_prf.c @@ -766,7 +766,7 @@ reswitch: switch (ch = (u_char)*fmt++) { PCHAR(hex2ascii(*up & 0x0f)); up++; if (width) - for (q=p;*q;q++) + for (q = p; *q; q++) PCHAR(*q); } break; diff --git a/sys/kern/sys_eventfd.c b/sys/kern/sys_eventfd.c index c2a0f67cae85..04ed107c933d 100644 --- a/sys/kern/sys_eventfd.c +++ b/sys/kern/sys_eventfd.c @@ -85,13 +85,16 @@ static int filt_eventfdwrite(struct knote *kn, long hint); static const struct filterops eventfd_rfiltops = { .f_isfd = 1, .f_detach = filt_eventfddetach, - .f_event = filt_eventfdread + .f_event = filt_eventfdread, + .f_copy = knote_triv_copy, }; + static const struct filterops eventfd_wfiltops = { .f_isfd = 1, .f_detach = filt_eventfddetach, - .f_event = filt_eventfdwrite + .f_event = filt_eventfdwrite, + .f_copy = knote_triv_copy, }; struct eventfd { diff --git a/sys/kern/sys_pipe.c b/sys/kern/sys_pipe.c index 57ebe8dc85f0..6531cea31423 100644 --- a/sys/kern/sys_pipe.c +++ b/sys/kern/sys_pipe.c @@ -181,20 +181,23 @@ static int filt_pipedump(struct proc *p, struct knote *kn, static const struct filterops pipe_nfiltops = { .f_isfd = 1, .f_detach = filt_pipedetach_notsup, - .f_event = filt_pipenotsup + .f_event = filt_pipenotsup, /* no userdump */ + .f_copy = knote_triv_copy, }; static const struct filterops pipe_rfiltops = { .f_isfd = 1, .f_detach = filt_pipedetach, .f_event = filt_piperead, .f_userdump = filt_pipedump, + .f_copy = knote_triv_copy, }; static const struct filterops pipe_wfiltops = { .f_isfd = 1, .f_detach = filt_pipedetach, .f_event = filt_pipewrite, .f_userdump = filt_pipedump, + .f_copy = knote_triv_copy, }; /* diff --git a/sys/kern/sys_procdesc.c b/sys/kern/sys_procdesc.c index acaf1241cb2e..c5db21544b0f 100644 --- a/sys/kern/sys_procdesc.c +++ b/sys/kern/sys_procdesc.c @@ -486,6 +486,7 @@ static const struct filterops procdesc_kqops = { .f_isfd = 1, .f_detach = procdesc_kqops_detach, .f_event = procdesc_kqops_event, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/kern/tty.c b/sys/kern/tty.c index c8e2c561b7cf..067471eb949a 100644 --- a/sys/kern/tty.c +++ b/sys/kern/tty.c @@ -754,12 +754,14 @@ static const struct filterops tty_kqops_read = { .f_isfd = 1, .f_detach = tty_kqops_read_detach, .f_event = tty_kqops_read_event, + .f_copy = knote_triv_copy, }; static const struct filterops tty_kqops_write = { .f_isfd = 1, .f_detach = tty_kqops_write_detach, .f_event = tty_kqops_write_event, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/kern/tty_pts.c b/sys/kern/tty_pts.c index 1291770a9ccb..2672935c2d89 100644 --- a/sys/kern/tty_pts.c +++ b/sys/kern/tty_pts.c @@ -491,11 +491,13 @@ static const struct filterops pts_kqops_read = { .f_isfd = 1, .f_detach = pts_kqops_read_detach, .f_event = pts_kqops_read_event, + .f_copy = knote_triv_copy, }; static const struct filterops pts_kqops_write = { .f_isfd = 1, .f_detach = pts_kqops_write_detach, .f_event = pts_kqops_write_event, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/kern/uipc_mqueue.c b/sys/kern/uipc_mqueue.c index a8aec397b352..4c1bb1ff228e 100644 --- a/sys/kern/uipc_mqueue.c +++ b/sys/kern/uipc_mqueue.c @@ -281,11 +281,13 @@ static const struct filterops mq_rfiltops = { .f_isfd = 1, .f_detach = filt_mqdetach, .f_event = filt_mqread, + .f_copy = knote_triv_copy, }; static const struct filterops mq_wfiltops = { .f_isfd = 1, .f_detach = filt_mqdetach, .f_event = filt_mqwrite, + .f_copy = knote_triv_copy, }; /* diff --git a/sys/kern/uipc_socket.c b/sys/kern/uipc_socket.c index fe2d8d056062..eb9544628137 100644 --- a/sys/kern/uipc_socket.c +++ b/sys/kern/uipc_socket.c @@ -191,16 +191,19 @@ static const struct filterops soread_filtops = { .f_isfd = 1, .f_detach = filt_sordetach, .f_event = filt_soread, + .f_copy = knote_triv_copy, }; static const struct filterops sowrite_filtops = { .f_isfd = 1, .f_detach = filt_sowdetach, .f_event = filt_sowrite, + .f_copy = knote_triv_copy, }; static const struct filterops soempty_filtops = { .f_isfd = 1, .f_detach = filt_sowdetach, .f_event = filt_soempty, + .f_copy = knote_triv_copy, }; so_gen_t so_gencnt; /* generation count for sockets */ diff --git a/sys/kern/uipc_usrreq.c b/sys/kern/uipc_usrreq.c index c5fc1e84ce3f..807271488af2 100644 --- a/sys/kern/uipc_usrreq.c +++ b/sys/kern/uipc_usrreq.c @@ -1559,15 +1559,19 @@ restart: mc_init_m(&cmc, control); SOCK_RECVBUF_LOCK(so); - MPASS(!(sb->sb_state & SBS_CANTRCVMORE)); - - if (__predict_false(cmc.mc_len + sb->sb_ccc + - sb->sb_ctl > sb->sb_hiwat)) { + if (__predict_false( + (sb->sb_state & SBS_CANTRCVMORE) || + cmc.mc_len + sb->sb_ccc + sb->sb_ctl > + sb->sb_hiwat)) { /* - * Too bad, while unp_externalize() was - * failing, the other side had filled - * the buffer and we can't prepend data - * back. Losing data! + * While the lock was dropped and we + * were failing in unp_externalize(), + * the peer could has a) disconnected, + * b) filled the buffer so that we + * can't prepend data back. + * These are two edge conditions that + * we just can't handle, so lose the + * data and return the error. */ SOCK_RECVBUF_UNLOCK(so); SOCK_IO_RECV_UNLOCK(so); @@ -1851,11 +1855,13 @@ static const struct filterops uipc_write_filtops = { .f_isfd = 1, .f_detach = uipc_filt_sowdetach, .f_event = uipc_filt_sowrite, + .f_copy = knote_triv_copy, }; static const struct filterops uipc_empty_filtops = { .f_isfd = 1, .f_detach = uipc_filt_sowdetach, .f_event = uipc_filt_soempty, + .f_copy = knote_triv_copy, }; static int diff --git a/sys/kern/vfs_aio.c b/sys/kern/vfs_aio.c index e63fa4c01434..60916a9fbd32 100644 --- a/sys/kern/vfs_aio.c +++ b/sys/kern/vfs_aio.c @@ -345,12 +345,14 @@ static const struct filterops aio_filtops = { .f_attach = filt_aioattach, .f_detach = filt_aiodetach, .f_event = filt_aio, + .f_copy = knote_triv_copy, }; static const struct filterops lio_filtops = { .f_isfd = 0, .f_attach = filt_lioattach, .f_detach = filt_liodetach, - .f_event = filt_lio + .f_event = filt_lio, + .f_copy = knote_triv_copy, }; static eventhandler_tag exit_tag, exec_tag; diff --git a/sys/kern/vfs_inotify.c b/sys/kern/vfs_inotify.c index b265a5ff3a62..e60d8426ee42 100644 --- a/sys/kern/vfs_inotify.c +++ b/sys/kern/vfs_inotify.c @@ -111,6 +111,7 @@ static const struct filterops inotify_rfiltops = { .f_isfd = 1, .f_detach = filt_inotifydetach, .f_event = filt_inotifyevent, + .f_copy = knote_triv_copy, }; static MALLOC_DEFINE(M_INOTIFY, "inotify", "inotify data structures"); diff --git a/sys/kern/vfs_subr.c b/sys/kern/vfs_subr.c index 73e110c05bc1..58975f7ac932 100644 --- a/sys/kern/vfs_subr.c +++ b/sys/kern/vfs_subr.c @@ -6545,6 +6545,7 @@ const struct filterops fs_filtops = { .f_attach = filt_fsattach, .f_detach = filt_fsdetach, .f_event = filt_fsevent, + .f_copy = knote_triv_copy, }; static int @@ -6624,24 +6625,28 @@ static int filt_vfsvnode(struct knote *kn, long hint); static void filt_vfsdetach(struct knote *kn); static int filt_vfsdump(struct proc *p, struct knote *kn, struct kinfo_knote *kin); +static int filt_vfscopy(struct knote *kn, struct proc *p1); static const struct filterops vfsread_filtops = { .f_isfd = 1, .f_detach = filt_vfsdetach, .f_event = filt_vfsread, .f_userdump = filt_vfsdump, + .f_copy = filt_vfscopy, }; static const struct filterops vfswrite_filtops = { .f_isfd = 1, .f_detach = filt_vfsdetach, .f_event = filt_vfswrite, .f_userdump = filt_vfsdump, + .f_copy = filt_vfscopy, }; static const struct filterops vfsvnode_filtops = { .f_isfd = 1, .f_detach = filt_vfsdetach, .f_event = filt_vfsvnode, .f_userdump = filt_vfsdump, + .f_copy = filt_vfscopy, }; static void @@ -6825,6 +6830,16 @@ filt_vfsdump(struct proc *p, struct knote *kn, struct kinfo_knote *kin) return (0); } +static int +filt_vfscopy(struct knote *kn, struct proc *p1) +{ + struct vnode *vp; + + vp = (struct vnode *)kn->kn_hook; + vhold(vp); + return (0); +} + int vfs_read_dirent(struct vop_readdir_args *ap, struct dirent *dp, off_t off) { diff --git a/sys/modules/Makefile b/sys/modules/Makefile index feb9778c23da..63a0b3260e6d 100644 --- a/sys/modules/Makefile +++ b/sys/modules/Makefile @@ -577,6 +577,7 @@ _mlx5ib= mlx5ib ${MACHINE_CPUARCH} == "i386" _ena= ena _gve= gve +_igc= igc # gcc13 and earlier lack __builtin_bitcountg used by linux emulation .if !(${COMPILER_TYPE} == "gcc" && ${COMPILER_VERSION} < 140000) _iwlwifi= iwlwifi @@ -747,7 +748,6 @@ _et= et _ftgpio= ftgpio _ftwd= ftwd _exca= exca -_igc= igc _io= io _itwd= itwd _ix= ix diff --git a/sys/net/bpf.c b/sys/net/bpf.c index a347dbe2eb73..f598733773d0 100644 --- a/sys/net/bpf.c +++ b/sys/net/bpf.c @@ -253,12 +253,14 @@ static const struct filterops bpfread_filtops = { .f_isfd = 1, .f_detach = filt_bpfdetach, .f_event = filt_bpfread, + .f_copy = knote_triv_copy, }; static const struct filterops bpfwrite_filtops = { .f_isfd = 1, .f_detach = filt_bpfdetach, .f_event = filt_bpfwrite, + .f_copy = knote_triv_copy, }; /* diff --git a/sys/net/if_tuntap.c b/sys/net/if_tuntap.c index c8dbb6aa8893..56bb90cce9bc 100644 --- a/sys/net/if_tuntap.c +++ b/sys/net/if_tuntap.c @@ -261,6 +261,7 @@ static const struct filterops tun_read_filterops = { .f_attach = NULL, .f_detach = tunkqdetach, .f_event = tunkqread, + .f_copy = knote_triv_copy, }; static const struct filterops tun_write_filterops = { @@ -268,6 +269,7 @@ static const struct filterops tun_write_filterops = { .f_attach = NULL, .f_detach = tunkqdetach, .f_event = tunkqwrite, + .f_copy = knote_triv_copy, }; static struct tuntap_driver { diff --git a/sys/netgraph/bluetooth/drivers/ubt/ng_ubt_rtl.c b/sys/netgraph/bluetooth/drivers/ubt/ng_ubt_rtl.c index 0181a67ac604..f35712cc8f69 100644 --- a/sys/netgraph/bluetooth/drivers/ubt/ng_ubt_rtl.c +++ b/sys/netgraph/bluetooth/drivers/ubt/ng_ubt_rtl.c @@ -81,9 +81,6 @@ const STRUCT_USB_HOST_ID ubt_rtl_devs[] = { USB_VPI(0x0bda, 0xb00c, 0) }, { USB_VPI(0x0bda, 0xc822, 0) }, - /* Realtek 8822CU Bluetooth devices */ - { USB_VPI(0x13d3, 0x3549, 0) }, - /* Realtek 8851BE Bluetooth devices */ { USB_VPI(0x13d3, 0x3600, 0) }, diff --git a/sys/netinet/in_mcast.c b/sys/netinet/in_mcast.c index f5b20c49ffd2..ba112afbf002 100644 --- a/sys/netinet/in_mcast.c +++ b/sys/netinet/in_mcast.c @@ -159,9 +159,6 @@ static struct ip_moptions * static int inp_get_source_filters(struct inpcb *, struct sockopt *); static int inp_join_group(struct inpcb *, struct sockopt *); static int inp_leave_group(struct inpcb *, struct sockopt *); -static struct ifnet * - inp_lookup_mcast_ifp(const struct inpcb *, - const struct sockaddr_in *, const struct in_addr); static int inp_block_unblock_source(struct inpcb *, struct sockopt *); static int inp_set_multicast_if(struct inpcb *, struct sockopt *); static int inp_set_source_filters(struct inpcb *, struct sockopt *); @@ -1832,69 +1829,55 @@ inp_getmoptions(struct inpcb *inp, struct sockopt *sopt) } /* - * Look up the ifnet to use for a multicast group membership, - * given the IPv4 address of an interface, and the IPv4 group address. - * - * This routine exists to support legacy multicast applications - * which do not understand that multicast memberships are scoped to - * specific physical links in the networking stack, or which need - * to join link-scope groups before IPv4 addresses are configured. - * - * Use this socket's current FIB number for any required FIB lookup. - * If ina is INADDR_ANY, look up the group address in the unicast FIB, - * and use its ifp; usually, this points to the default next-hop. - * - * If the FIB lookup fails, attempt to use the first non-loopback - * interface with multicast capability in the system as a - * last resort. The legacy IPv4 ASM API requires that we do - * this in order to allow groups to be joined when the routing - * table has not yet been populated during boot. - * - * Returns NULL if no ifp could be found, otherwise return referenced ifp. + * Look up the ifnet to join a multicast group membership via legacy + * IP_ADD_MEMBERSHIP or via more modern MCAST_JOIN_GROUP. * - * FUTURE: Implement IPv4 source-address selection. + * If the interface index was specified explicitly, just use it. If the + * address was specified (legacy), try to find matching interface. Else + * (index == 0 && no address) do a route lookup. If that fails for a modern + * MCAST_JOIN_GROUP return failure, for legacy IP_ADD_MEMBERSHIP find first + * multicast capable interface. */ static struct ifnet * -inp_lookup_mcast_ifp(const struct inpcb *inp, - const struct sockaddr_in *gsin, const struct in_addr ina) +inp_lookup_mcast_ifp(const struct inpcb *inp, const struct in_addr maddr, +const struct in_addr *ina, const u_int index) { struct ifnet *ifp; struct nhop_object *nh; NET_EPOCH_ASSERT(); - KASSERT(inp != NULL, ("%s: inp must not be NULL", __func__)); - KASSERT(gsin->sin_family == AF_INET, ("%s: not AF_INET", __func__)); - KASSERT(IN_MULTICAST(ntohl(gsin->sin_addr.s_addr)), - ("%s: not multicast", __func__)); - ifp = NULL; - if (!in_nullhost(ina)) { - INADDR_TO_IFP(ina, ifp); + if (index != 0) + return (ifnet_byindex_ref(index)); + + if (ina != NULL && !in_nullhost(*ina)) { + INADDR_TO_IFP(*ina, ifp); if (ifp != NULL) if_ref(ifp); - } else { - nh = fib4_lookup(inp->inp_inc.inc_fibnum, gsin->sin_addr, 0, NHR_NONE, 0); - if (nh != NULL) { - ifp = nh->nh_ifp; - if_ref(ifp); - } else { - struct in_ifaddr *ia; - struct ifnet *mifp; - - mifp = NULL; - CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { - mifp = ia->ia_ifp; - if (!(mifp->if_flags & IFF_LOOPBACK) && - (mifp->if_flags & IFF_MULTICAST)) { - ifp = mifp; - if_ref(ifp); - break; - } + return (ifp); + } + + nh = fib4_lookup(inp->inp_inc.inc_fibnum, maddr, 0, NHR_NONE, 0); + if (nh != NULL) { + ifp = nh->nh_ifp; + if_ref(ifp); + return (ifp); + } + + if (ina != NULL) { + struct in_ifaddr *ia; + + CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) { + if (!(ia->ia_ifp->if_flags & IFF_LOOPBACK) && + (ia->ia_ifp->if_flags & IFF_MULTICAST)) { + ifp = ia->ia_ifp; + if_ref(ifp); + return (ifp); } } } - return (ifp); + return (NULL); } /* @@ -1926,13 +1909,13 @@ inp_join_group(struct inpcb *inp, struct sockopt *sopt) switch (sopt->sopt_name) { case IP_ADD_MEMBERSHIP: { struct ip_mreqn mreqn; + bool mreq; - if (sopt->sopt_valsize == sizeof(struct ip_mreqn)) - error = sooptcopyin(sopt, &mreqn, - sizeof(struct ip_mreqn), sizeof(struct ip_mreqn)); - else - error = sooptcopyin(sopt, &mreqn, - sizeof(struct ip_mreq), sizeof(struct ip_mreq)); + mreq = (sopt->sopt_valsize != sizeof(struct ip_mreqn)); + + error = sooptcopyin(sopt, &mreqn, + mreq ? sizeof(struct ip_mreq) : sizeof(struct ip_mreqn), + mreq ? sizeof(struct ip_mreq) : sizeof(struct ip_mreqn)); if (error) return (error); @@ -1943,12 +1926,9 @@ inp_join_group(struct inpcb *inp, struct sockopt *sopt) return (EINVAL); NET_EPOCH_ENTER(et); - if (sopt->sopt_valsize == sizeof(struct ip_mreqn) && - mreqn.imr_ifindex != 0) - ifp = ifnet_byindex_ref(mreqn.imr_ifindex); - else - ifp = inp_lookup_mcast_ifp(inp, &gsa->sin, - mreqn.imr_address); + ifp = inp_lookup_mcast_ifp(inp, mreqn.imr_multiaddr, + mreq ? &mreqn.imr_address : NULL, + mreq ? 0 : mreqn.imr_ifindex); NET_EPOCH_EXIT(et); break; } @@ -1971,8 +1951,8 @@ inp_join_group(struct inpcb *inp, struct sockopt *sopt) ssa->sin.sin_addr = mreqs.imr_sourceaddr; NET_EPOCH_ENTER(et); - ifp = inp_lookup_mcast_ifp(inp, &gsa->sin, - mreqs.imr_interface); + ifp = inp_lookup_mcast_ifp(inp, mreqs.imr_multiaddr, + &mreqs.imr_interface, 0); NET_EPOCH_EXIT(et); CTR3(KTR_IGMPV3, "%s: imr_interface = 0x%08x, ifp = %p", __func__, ntohl(mreqs.imr_interface.s_addr), ifp); @@ -2013,7 +1993,8 @@ inp_join_group(struct inpcb *inp, struct sockopt *sopt) return (EINVAL); NET_EPOCH_ENTER(et); - ifp = ifnet_byindex_ref(gsr.gsr_interface); + ifp = inp_lookup_mcast_ifp(inp, gsa->sin.sin_addr, NULL, + gsr.gsr_interface); NET_EPOCH_EXIT(et); if (ifp == NULL) return (EADDRNOTAVAIL); diff --git a/sys/netinet/libalias/alias_db.c b/sys/netinet/libalias/alias_db.c index c143d74a2f45..41f0a328daec 100644 --- a/sys/netinet/libalias/alias_db.c +++ b/sys/netinet/libalias/alias_db.c @@ -2181,7 +2181,7 @@ LibAliasInit(struct libalias *la) #undef malloc /* XXX: ugly */ la = malloc(sizeof *la, M_ALIAS, M_WAITOK | M_ZERO); #else - la = calloc(sizeof *la, 1); + la = calloc(1, sizeof *la); if (la == NULL) return (la); #endif diff --git a/sys/netinet/siftr.c b/sys/netinet/siftr.c index 374b5595fcbc..5b89ca026e85 100644 --- a/sys/netinet/siftr.c +++ b/sys/netinet/siftr.c @@ -519,7 +519,7 @@ siftr_pkt_manager_thread(void *arg) if (log_buf != NULL) { alq_post_flags(siftr_alq, log_buf, 0); } - for (;cnt > 0; cnt--) { + for (; cnt > 0; cnt--) { pkt_node = STAILQ_FIRST(&tmp_pkt_queue); STAILQ_REMOVE_HEAD(&tmp_pkt_queue, nodes); free(pkt_node, M_SIFTR_PKTNODE); diff --git a/sys/netinet/tcp_hpts_test.c b/sys/netinet/tcp_hpts_test.c index bab5827e0572..c5dc9cb5b03b 100644 --- a/sys/netinet/tcp_hpts_test.c +++ b/sys/netinet/tcp_hpts_test.c @@ -27,6 +27,7 @@ #include <tests/ktest.h> #include <sys/cdefs.h> +#include "opt_inet.h" #include <sys/param.h> #include <sys/bus.h> #include <sys/interrupt.h> @@ -119,6 +120,8 @@ SYSCTL_INT(_net_inet_tcp_hpts_test, OID_AUTO, exit_on_failure, CTLFLAG_RW, } \ } while (0) +#ifdef TCP_HPTS_KTEST + static void dump_hpts_entry(struct ktest_test_context *ctx, struct tcp_hpts_entry *hpts) { @@ -1658,5 +1661,22 @@ static const struct ktest_test_info tests[] = { KTEST_INFO(generation_count_validation), }; +#else /* TCP_HPTS_KTEST */ + +/* + * Stub to indicate that the TCP HPTS ktest is not enabled. + */ +KTEST_FUNC(module_load_without_tests) +{ + KTEST_LOG(ctx, "Warning: TCP HPTS ktest is not enabled"); + return (0); +} + +static const struct ktest_test_info tests[] = { + KTEST_INFO(module_load_without_tests), +}; + +#endif + KTEST_MODULE_DECLARE(ktest_tcphpts, tests); KTEST_MODULE_DEPEND(ktest_tcphpts, tcphpts); diff --git a/sys/netinet/tcp_input.c b/sys/netinet/tcp_input.c index dd27ec77c1af..2146b0cac48f 100644 --- a/sys/netinet/tcp_input.c +++ b/sys/netinet/tcp_input.c @@ -219,7 +219,7 @@ SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_auto, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_do_autorcvbuf), 0, "Enable automatic receive buffer sizing"); -VNET_DEFINE(int, tcp_autorcvbuf_max) = 2*1024*1024; +VNET_DEFINE(int, tcp_autorcvbuf_max) = 8*1024*1024; SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbuf_max, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autorcvbuf_max), 0, "Max size of automatic receive buffer"); diff --git a/sys/netinet/tcp_output.c b/sys/netinet/tcp_output.c index 2dfb7faf56e3..208f72c4661c 100644 --- a/sys/netinet/tcp_output.c +++ b/sys/netinet/tcp_output.c @@ -123,7 +123,7 @@ SYSCTL_INT(_net_inet_tcp, OID_AUTO, sendbuf_inc, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autosndbuf_inc), 0, "Incrementor step size of automatic send buffer"); -VNET_DEFINE(int, tcp_autosndbuf_max) = 2*1024*1024; +VNET_DEFINE(int, tcp_autosndbuf_max) = 8*1024*1024; SYSCTL_INT(_net_inet_tcp, OID_AUTO, sendbuf_max, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(tcp_autosndbuf_max), 0, "Max size of automatic send buffer"); diff --git a/sys/netinet/tcp_stacks/rack.c b/sys/netinet/tcp_stacks/rack.c index c7962b57a69e..0320479e882e 100644 --- a/sys/netinet/tcp_stacks/rack.c +++ b/sys/netinet/tcp_stacks/rack.c @@ -6864,6 +6864,18 @@ rack_mark_lost(struct tcpcb *tp, } } +static inline void +rack_mark_nolonger_lost(struct tcp_rack *rack, struct rack_sendmap *rsm) +{ + KASSERT((rack->r_ctl.rc_considered_lost >= (rsm->r_end - rsm->r_start)), + ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); + rsm->r_flags &= ~RACK_WAS_LOST; + if (rack->r_ctl.rc_considered_lost >= (rsm->r_end - rsm->r_start)) + rack->r_ctl.rc_considered_lost -= rsm->r_end - rsm->r_start; + else + rack->r_ctl.rc_considered_lost = 0; +} + /* * RACK Timer, here we simply do logging and house keeping. * the normal rack_output() function will call the @@ -8130,13 +8142,7 @@ rack_update_rsm(struct tcpcb *tp, struct tcp_rack *rack, * remove the lost desgination and reduce the * bytes considered lost. */ - rsm->r_flags &= ~RACK_WAS_LOST; - KASSERT((rack->r_ctl.rc_considered_lost >= (rsm->r_end - rsm->r_start)), - ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); - if (rack->r_ctl.rc_considered_lost >= (rsm->r_end - rsm->r_start)) - rack->r_ctl.rc_considered_lost -= rsm->r_end - rsm->r_start; - else - rack->r_ctl.rc_considered_lost = 0; + rack_mark_nolonger_lost(rack, rsm); } idx = rsm->r_rtr_cnt - 1; rsm->r_tim_lastsent[idx] = ts; @@ -9492,6 +9498,11 @@ do_rest_ofb: if (rsm->r_flags & RACK_WAS_LOST) { int my_chg; + /* + * Note here we do not use our rack_mark_nolonger_lost() function + * since we are moving our data pointer around and the + * ack'ed side is already not considered lost. + */ my_chg = (nrsm->r_end - nrsm->r_start); KASSERT((rack->r_ctl.rc_considered_lost >= my_chg), ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); @@ -9659,16 +9670,11 @@ do_rest_ofb: changed += (rsm->r_end - rsm->r_start); /* You get a count for acking a whole segment or more */ if (rsm->r_flags & RACK_WAS_LOST) { - int my_chg; - - my_chg = (rsm->r_end - rsm->r_start); - rsm->r_flags &= ~RACK_WAS_LOST; - KASSERT((rack->r_ctl.rc_considered_lost >= my_chg), - ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); - if (my_chg <= rack->r_ctl.rc_considered_lost) - rack->r_ctl.rc_considered_lost -= my_chg; - else - rack->r_ctl.rc_considered_lost = 0; + /* + * Here we can use the inline function since + * the rsm is truly marked lost and now no longer lost. + */ + rack_mark_nolonger_lost(rack, rsm); } rack->r_ctl.rc_sacked += (rsm->r_end - rsm->r_start); if (rsm->r_in_tmap) /* should be true */ @@ -9851,6 +9857,10 @@ do_rest_ofb: if (rsm->r_flags & RACK_WAS_LOST) { int my_chg; + /* + * Note here we are using hookery again so we can't + * use our rack_mark_nolonger_lost() function. + */ my_chg = (nrsm->r_end - nrsm->r_start); KASSERT((rack->r_ctl.rc_considered_lost >= my_chg), ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); @@ -9952,16 +9962,10 @@ do_rest_ofb: rack_update_rtt(tp, rack, rsm, to, cts, SACKED, 0); changed += (rsm->r_end - rsm->r_start); if (rsm->r_flags & RACK_WAS_LOST) { - int my_chg; - - my_chg = (rsm->r_end - rsm->r_start); - rsm->r_flags &= ~RACK_WAS_LOST; - KASSERT((rack->r_ctl.rc_considered_lost >= my_chg), - ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); - if (my_chg <= rack->r_ctl.rc_considered_lost) - rack->r_ctl.rc_considered_lost -= my_chg; - else - rack->r_ctl.rc_considered_lost = 0; + /* + * Here it is safe to use our function. + */ + rack_mark_nolonger_lost(rack, rsm); } rack->r_ctl.rc_sacked += (rsm->r_end - rsm->r_start); @@ -10362,13 +10366,7 @@ more: * and yet before retransmitting we get an ack * which can happen due to reordering. */ - rsm->r_flags &= ~RACK_WAS_LOST; - KASSERT((rack->r_ctl.rc_considered_lost >= (rsm->r_end - rsm->r_start)), - ("rsm:%p rack:%p rc_considered_lost goes negative", rsm, rack)); - if (rack->r_ctl.rc_considered_lost >= (rsm->r_end - rsm->r_start)) - rack->r_ctl.rc_considered_lost -= rsm->r_end - rsm->r_start; - else - rack->r_ctl.rc_considered_lost = 0; + rack_mark_nolonger_lost(rack, rsm); } rack_log_map_chg(tp, rack, NULL, rsm, NULL, MAP_FREE, rsm->r_end, __LINE__); rack->r_ctl.rc_holes_rxt -= rsm->r_rtr_bytes; @@ -10476,12 +10474,7 @@ more: * which can happen due to reordering. In this * case its only a partial ack of the send. */ - KASSERT((rack->r_ctl.rc_considered_lost >= (th_ack - rsm->r_start)), - ("rsm:%p rack:%p rc_considered_lost goes negative th_ack:%u", rsm, rack, th_ack)); - if (rack->r_ctl.rc_considered_lost >= (th_ack - rsm->r_start)) - rack->r_ctl.rc_considered_lost -= th_ack - rsm->r_start; - else - rack->r_ctl.rc_considered_lost = 0; + rack_mark_nolonger_lost(rack, rsm); } /* * Clear the dup ack count for diff --git a/sys/netinet/tcp_syncache.c b/sys/netinet/tcp_syncache.c index f842a5678fa1..be20fb44a820 100644 --- a/sys/netinet/tcp_syncache.c +++ b/sys/netinet/tcp_syncache.c @@ -1046,6 +1046,8 @@ abort: * * On syncache_socket() success the newly created socket * has its underlying inp locked. + * + * *lsop is updated, if and only if 1 is returned. */ int syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, @@ -1094,12 +1096,14 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, */ SCH_UNLOCK(sch); TCPSTAT_INC(tcps_sc_spurcookie); - if ((s = tcp_log_addrs(inc, th, NULL, NULL))) + if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Spurious ACK, " "segment rejected " "(syncookies disabled)\n", s, __func__); - goto failed; + free(s, M_TCPLOG); + } + return (0); } if (sch->sch_last_overflow < time_uptime - SYNCOOKIE_LIFETIME) { @@ -1109,12 +1113,14 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, */ SCH_UNLOCK(sch); TCPSTAT_INC(tcps_sc_spurcookie); - if ((s = tcp_log_addrs(inc, th, NULL, NULL))) + if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Spurious ACK, " "segment rejected " "(no syncache entry)\n", s, __func__); - goto failed; + free(s, M_TCPLOG); + } + return (0); } SCH_UNLOCK(sch); } @@ -1128,11 +1134,13 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, TCPSTAT_INC(tcps_sc_recvcookie); } else { TCPSTAT_INC(tcps_sc_failcookie); - if ((s = tcp_log_addrs(inc, th, NULL, NULL))) + if ((s = tcp_log_addrs(inc, th, NULL, NULL))) { log(LOG_DEBUG, "%s; %s: Segment failed " "SYNCOOKIE authentication, segment rejected " "(probably spoofed)\n", s, __func__); - goto failed; + free(s, M_TCPLOG); + } + return (0); } #if defined(IPSEC_SUPPORT) || defined(TCP_SIGNATURE) /* If received ACK has MD5 signature, check it. */ @@ -1206,9 +1214,9 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, "%s; %s: SEG.TSval %u < TS.Recent %u, " "segment dropped\n", s, __func__, to->to_tsval, sc->sc_tsreflect); - free(s, M_TCPLOG); } SCH_UNLOCK(sch); + free(s, M_TCPLOG); return (-1); /* Do not send RST */ } @@ -1225,7 +1233,6 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, "expected, segment processed normally\n", s, __func__); free(s, M_TCPLOG); - s = NULL; } } @@ -1312,16 +1319,6 @@ syncache_expand(struct in_conninfo *inc, struct tcpopt *to, struct tcphdr *th, if (sc != &scs) syncache_free(sc); return (1); -failed: - if (sc != NULL) { - TCPSTATES_DEC(TCPS_SYN_RECEIVED); - if (sc != &scs) - syncache_free(sc); - } - if (s != NULL) - free(s, M_TCPLOG); - *lsop = NULL; - return (0); } static struct socket * diff --git a/sys/netpfil/ipfw/ip_dummynet.c b/sys/netpfil/ipfw/ip_dummynet.c index b3f52322425f..d522f9da0fbe 100644 --- a/sys/netpfil/ipfw/ip_dummynet.c +++ b/sys/netpfil/ipfw/ip_dummynet.c @@ -1150,7 +1150,7 @@ copy_data_helper(void *_o, void *_arg) return 0; /* not a pipe */ /* see if the object is within one of our ranges */ - for (;r < lim; r += 2) { + for (; r < lim; r += 2) { if (n < r[0] || n > r[1]) continue; /* Found a valid entry, copy and we are done */ @@ -1183,7 +1183,7 @@ copy_data_helper(void *_o, void *_arg) if (n >= DN_MAX_ID) return 0; /* see if the object is within one of our ranges */ - for (;r < lim; r += 2) { + for (; r < lim; r += 2) { if (n < r[0] || n > r[1]) continue; if (copy_flowset(a, fs, 0)) diff --git a/sys/powerpc/powerpc/elf32_machdep.c b/sys/powerpc/powerpc/elf32_machdep.c index e1118713bff0..6b904c02ea15 100644 --- a/sys/powerpc/powerpc/elf32_machdep.c +++ b/sys/powerpc/powerpc/elf32_machdep.c @@ -143,7 +143,7 @@ struct sysentvec elf32_freebsd_sysvec = { }; INIT_SYSENTVEC(elf32_sysvec, &elf32_freebsd_sysvec); -static Elf32_Brandinfo freebsd_brand_info = { +static const Elf32_Brandinfo freebsd_brand_info = { .brand = ELFOSABI_FREEBSD, .machine = EM_PPC, .compat_3_brand = "FreeBSD", @@ -158,11 +158,11 @@ static Elf32_Brandinfo freebsd_brand_info = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -SYSINIT(elf32, SI_SUB_EXEC, SI_ORDER_FIRST, +C_SYSINIT(elf32, SI_SUB_EXEC, SI_ORDER_FIRST, (sysinit_cfunc_t) elf32_insert_brand_entry, &freebsd_brand_info); -static Elf32_Brandinfo freebsd_brand_oinfo = { +static const Elf32_Brandinfo freebsd_brand_oinfo = { .brand = ELFOSABI_FREEBSD, .machine = EM_PPC, .compat_3_brand = "FreeBSD", @@ -173,7 +173,7 @@ static Elf32_Brandinfo freebsd_brand_oinfo = { .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -SYSINIT(oelf32, SI_SUB_EXEC, SI_ORDER_ANY, +C_SYSINIT(oelf32, SI_SUB_EXEC, SI_ORDER_ANY, (sysinit_cfunc_t) elf32_insert_brand_entry, &freebsd_brand_oinfo); diff --git a/sys/powerpc/powerpc/elf64_machdep.c b/sys/powerpc/powerpc/elf64_machdep.c index a999f742caeb..0be40bed69cb 100644 --- a/sys/powerpc/powerpc/elf64_machdep.c +++ b/sys/powerpc/powerpc/elf64_machdep.c @@ -154,7 +154,7 @@ static bool ppc64_elfv1_header_match(const struct image_params *params, static bool ppc64_elfv2_header_match(const struct image_params *params, const int32_t *, const uint32_t *); -static Elf64_Brandinfo freebsd_brand_info_elfv1 = { +static const Elf64_Brandinfo freebsd_brand_info_elfv1 = { .brand = ELFOSABI_FREEBSD, .machine = EM_PPC64, .compat_3_brand = "FreeBSD", @@ -166,11 +166,11 @@ static Elf64_Brandinfo freebsd_brand_info_elfv1 = { .header_supported = &ppc64_elfv1_header_match }; -SYSINIT(elf64v1, SI_SUB_EXEC, SI_ORDER_ANY, +C_SYSINIT(elf64v1, SI_SUB_EXEC, SI_ORDER_ANY, (sysinit_cfunc_t) elf64_insert_brand_entry, &freebsd_brand_info_elfv1); -static Elf64_Brandinfo freebsd_brand_info_elfv2 = { +static const Elf64_Brandinfo freebsd_brand_info_elfv2 = { .brand = ELFOSABI_FREEBSD, .machine = EM_PPC64, .compat_3_brand = "FreeBSD", @@ -182,11 +182,11 @@ static Elf64_Brandinfo freebsd_brand_info_elfv2 = { .header_supported = &ppc64_elfv2_header_match }; -SYSINIT(elf64v2, SI_SUB_EXEC, SI_ORDER_ANY, +C_SYSINIT(elf64v2, SI_SUB_EXEC, SI_ORDER_ANY, (sysinit_cfunc_t) elf64_insert_brand_entry, &freebsd_brand_info_elfv2); -static Elf64_Brandinfo freebsd_brand_oinfo = { +static const Elf64_Brandinfo freebsd_brand_oinfo = { .brand = ELFOSABI_FREEBSD, .machine = EM_PPC64, .compat_3_brand = "FreeBSD", @@ -198,7 +198,7 @@ static Elf64_Brandinfo freebsd_brand_oinfo = { .header_supported = &ppc64_elfv1_header_match }; -SYSINIT(oelf64, SI_SUB_EXEC, SI_ORDER_ANY, +C_SYSINIT(oelf64, SI_SUB_EXEC, SI_ORDER_ANY, (sysinit_cfunc_t) elf64_insert_brand_entry, &freebsd_brand_oinfo); diff --git a/sys/riscv/include/vmm.h b/sys/riscv/include/vmm.h index bc00474ed0fd..e227dd825966 100644 --- a/sys/riscv/include/vmm.h +++ b/sys/riscv/include/vmm.h @@ -149,7 +149,7 @@ DECLARE_VMMOPS_FUNC(void, vmspace_free, (struct vmspace *vmspace)); int vm_create(const char *name, struct vm **retvm); struct vcpu *vm_alloc_vcpu(struct vm *vm, int vcpuid); void vm_disable_vcpu_creation(struct vm *vm); -void vm_slock_vcpus(struct vm *vm); +void vm_lock_vcpus(struct vm *vm); void vm_unlock_vcpus(struct vm *vm); void vm_destroy(struct vm *vm); int vm_reinit(struct vm *vm); diff --git a/sys/riscv/riscv/elf_machdep.c b/sys/riscv/riscv/elf_machdep.c index 67b1fcc4c1a9..5bd4af4c15f8 100644 --- a/sys/riscv/riscv/elf_machdep.c +++ b/sys/riscv/riscv/elf_machdep.c @@ -100,7 +100,7 @@ static struct sysentvec elf64_freebsd_sysvec = { }; INIT_SYSENTVEC(elf64_sysvec, &elf64_freebsd_sysvec); -static Elf64_Brandinfo freebsd_brand_info = { +static const Elf64_Brandinfo freebsd_brand_info = { .brand = ELFOSABI_FREEBSD, .machine = EM_RISCV, .compat_3_brand = "FreeBSD", @@ -110,7 +110,7 @@ static Elf64_Brandinfo freebsd_brand_info = { .brand_note = &elf64_freebsd_brandnote, .flags = BI_CAN_EXEC_DYN | BI_BRAND_NOTE }; -SYSINIT(elf64, SI_SUB_EXEC, SI_ORDER_FIRST, +C_SYSINIT(elf64, SI_SUB_EXEC, SI_ORDER_FIRST, (sysinit_cfunc_t)elf64_insert_brand_entry, &freebsd_brand_info); static void diff --git a/sys/riscv/vmm/vmm.c b/sys/riscv/vmm/vmm.c index 790dcc576507..4c9b1fa53f7a 100644 --- a/sys/riscv/vmm/vmm.c +++ b/sys/riscv/vmm/vmm.c @@ -346,9 +346,9 @@ vm_alloc_vcpu(struct vm *vm, int vcpuid) } void -vm_slock_vcpus(struct vm *vm) +vm_lock_vcpus(struct vm *vm) { - sx_slock(&vm->vcpus_init_lock); + sx_xlock(&vm->vcpus_init_lock); } void diff --git a/sys/security/audit/audit_pipe.c b/sys/security/audit/audit_pipe.c index fb773fd04297..4d9815467e1a 100644 --- a/sys/security/audit/audit_pipe.c +++ b/sys/security/audit/audit_pipe.c @@ -243,6 +243,7 @@ static const struct filterops audit_pipe_read_filterops = { .f_attach = NULL, .f_detach = audit_pipe_kqdetach, .f_event = audit_pipe_kqread, + .f_copy = knote_triv_copy, }; /* diff --git a/sys/sys/event.h b/sys/sys/event.h index 084eaafcbdc0..ebbcdb703183 100644 --- a/sys/sys/event.h +++ b/sys/sys/event.h @@ -228,6 +228,7 @@ struct freebsd11_kevent32 { /* Flags for kqueuex(2) */ #define KQUEUE_CLOEXEC 0x00000001 /* close on exec */ +#define KQUEUE_CPONFORK 0x00000002 /* copy on fork */ struct knote; SLIST_HEAD(klist, knote); @@ -283,6 +284,7 @@ struct filterops { void (*f_touch)(struct knote *kn, struct kevent *kev, u_long type); int (*f_userdump)(struct proc *p, struct knote *kn, struct kinfo_knote *kin); + int (*f_copy)(struct knote *kn, struct proc *p1); }; /* @@ -346,6 +348,7 @@ struct rwlock; void knote(struct knlist *list, long hint, int lockflags); void knote_fork(struct knlist *list, int pid); +int knote_triv_copy(struct knote *kn, struct proc *p1); struct knlist *knlist_alloc(struct mtx *lock); void knlist_detach(struct knlist *knl); void knlist_add(struct knlist *knl, struct knote *kn, int islocked); diff --git a/sys/sys/eventvar.h b/sys/sys/eventvar.h index 7fec444447f9..7cb3269f1fdf 100644 --- a/sys/sys/eventvar.h +++ b/sys/sys/eventvar.h @@ -55,12 +55,14 @@ struct kqueue { #define KQ_CLOSING 0x10 #define KQ_TASKSCHED 0x20 /* task scheduled */ #define KQ_TASKDRAIN 0x40 /* waiting for task to drain */ +#define KQ_CPONFORK 0x80 int kq_knlistsize; /* size of knlist */ struct klist *kq_knlist; /* list of knotes */ u_long kq_knhashmask; /* size of knhash */ struct klist *kq_knhash; /* hash table for knotes */ struct task kq_task; struct ucred *kq_cred; + struct kqueue *kq_forksrc; }; #endif /* !_SYS_EVENTVAR_H_ */ diff --git a/sys/sys/file.h b/sys/sys/file.h index c44fd0f28929..e0195c7c6c2a 100644 --- a/sys/sys/file.h +++ b/sys/sys/file.h @@ -139,6 +139,8 @@ typedef int fo_fspacectl_t(struct file *fp, int cmd, off_t *offset, off_t *length, int flags, struct ucred *active_cred, struct thread *td); typedef int fo_cmp_t(struct file *fp, struct file *fp1, struct thread *td); +typedef int fo_fork_t(struct filedesc *fdp, struct file *fp, struct file **fp1, + struct proc *p1, struct thread *td); typedef int fo_spare_t(struct file *fp); typedef int fo_flags_t; @@ -163,12 +165,14 @@ struct fileops { fo_fallocate_t *fo_fallocate; fo_fspacectl_t *fo_fspacectl; fo_cmp_t *fo_cmp; + fo_fork_t *fo_fork; fo_spare_t *fo_spares[8]; /* Spare slots */ fo_flags_t fo_flags; /* DFLAG_* below */ }; #define DFLAG_PASSABLE 0x01 /* may be passed via unix sockets. */ #define DFLAG_SEEKABLE 0x02 /* seekable / nonsequential */ +#define DFLAG_FORK 0x04 /* copy on fork */ #endif /* _KERNEL */ #if defined(_KERNEL) || defined(_WANT_FILE) diff --git a/sys/sys/filedesc.h b/sys/sys/filedesc.h index 0a388c90de26..4817855443af 100644 --- a/sys/sys/filedesc.h +++ b/sys/sys/filedesc.h @@ -265,7 +265,7 @@ int fdcheckstd(struct thread *td); void fdclose(struct thread *td, struct file *fp, int idx); void fdcloseexec(struct thread *td); void fdsetugidsafety(struct thread *td); -struct filedesc *fdcopy(struct filedesc *fdp); +struct filedesc *fdcopy(struct filedesc *fdp, struct proc *p1); void fdunshare(struct thread *td); void fdescfree(struct thread *td); int fdlastfile(struct filedesc *fdp); diff --git a/sys/sys/imgact_elf.h b/sys/sys/imgact_elf.h index 9e2a233248b4..25ce7871ba7c 100644 --- a/sys/sys/imgact_elf.h +++ b/sys/sys/imgact_elf.h @@ -87,7 +87,7 @@ typedef struct { const char *interp_newpath; int flags; const Elf_Brandnote *brand_note; - bool (*header_supported)(const struct image_params *, + bool (*const header_supported)(const struct image_params *, const int32_t *, const uint32_t *); /* High 8 bits of flags is private to the ABI */ #define BI_CAN_EXEC_DYN 0x0001 @@ -132,8 +132,8 @@ bool __elfN(parse_notes)(const struct image_params *, const Elf_Note *, void __elfN(dump_thread)(struct thread *, void *, size_t *); extern int __elfN(fallback_brand); -extern Elf_Brandnote __elfN(freebsd_brandnote); -extern Elf_Brandnote __elfN(kfreebsd_brandnote); +extern const Elf_Brandnote __elfN(freebsd_brandnote); +extern const Elf_Brandnote __elfN(kfreebsd_brandnote); #endif /* _KERNEL */ #endif /* !_SYS_IMGACT_ELF_H_ */ diff --git a/sys/sys/sockbuf.h b/sys/sys/sockbuf.h index b4593f38f592..739723754b7d 100644 --- a/sys/sys/sockbuf.h +++ b/sys/sys/sockbuf.h @@ -62,7 +62,7 @@ #include <sys/_sx.h> #include <sys/_task.h> -#define SB_MAX (2*1024*1024) /* default for max chars in sockbuf */ +#define SB_MAX (8*1024*1024) /* default for max chars in sockbuf */ struct ktls_session; struct mbuf; diff --git a/sys/sys/sockopt.h b/sys/sys/sockopt.h index bfe12d8510d7..d2b0ff5ed2c8 100644 --- a/sys/sys/sockopt.h +++ b/sys/sys/sockopt.h @@ -57,8 +57,10 @@ struct sockopt { int sosetopt(struct socket *so, struct sockopt *sopt); int sogetopt(struct socket *so, struct sockopt *sopt); -int sooptcopyin(struct sockopt *sopt, void *buf, size_t len, size_t minlen); -int sooptcopyout(struct sockopt *sopt, const void *buf, size_t len); +int __result_use_check sooptcopyin(struct sockopt *sopt, void *buf, size_t len, + size_t minlen); +int __result_use_check sooptcopyout(struct sockopt *sopt, const void *buf, + size_t len); int soopt_getm(struct sockopt *sopt, struct mbuf **mp); int soopt_mcopyin(struct sockopt *sopt, struct mbuf *m); int soopt_mcopyout(struct sockopt *sopt, struct mbuf *m); diff --git a/sys/sys/syscallsubr.h b/sys/sys/syscallsubr.h index 8237165b84ce..d32690634059 100644 --- a/sys/sys/syscallsubr.h +++ b/sys/sys/syscallsubr.h @@ -211,7 +211,8 @@ int kern_kevent_fp(struct thread *td, struct file *fp, int nchanges, int nevents, struct kevent_copyops *k_ops, const struct timespec *timeout); int kern_kill(struct thread *td, pid_t pid, int signum); -int kern_kqueue(struct thread *td, int flags, struct filecaps *fcaps); +int kern_kqueue(struct thread *td, int flags, bool cponfork, + struct filecaps *fcaps); int kern_kldload(struct thread *td, const char *file, int *fileid); int kern_kldstat(struct thread *td, int fileid, struct kld_file_stat *stat); int kern_kldunload(struct thread *td, int fileid, int flags); diff --git a/sys/vm/vm_domainset.c b/sys/vm/vm_domainset.c index 9fa17da954f7..c25ed0cc2267 100644 --- a/sys/vm/vm_domainset.c +++ b/sys/vm/vm_domainset.c @@ -113,7 +113,6 @@ vm_domainset_iter_interleave(struct vm_domainset_iter *di, int *domain) int d; d = di->di_offset % di->di_domain->ds_cnt; - *di->di_iter = d; *domain = di->di_domain->ds_order[d]; } @@ -260,9 +259,14 @@ vm_domainset_iter_page_init(struct vm_domainset_iter *di, struct vm_object *obj, * are immutable and unsynchronized. Updates can race but pointer * loads are assumed to be atomic. */ - if (obj != NULL && obj->domain.dr_policy != NULL) + if (obj != NULL && obj->domain.dr_policy != NULL) { + /* + * This write lock protects non-atomic increments of the + * iterator index in vm_domainset_iter_rr(). + */ + VM_OBJECT_ASSERT_WLOCKED(obj); dr = &obj->domain; - else + } else dr = &curthread->td_domain; vm_domainset_iter_init(di, dr->dr_policy, &dr->dr_iter, obj, pindex); diff --git a/sys/vm/vm_glue.c b/sys/vm/vm_glue.c index e0f1807a1b32..18d789c59281 100644 --- a/sys/vm/vm_glue.c +++ b/sys/vm/vm_glue.c @@ -441,19 +441,16 @@ vm_thread_kstack_arena_release(void *arena, vmem_addr_t addr, vmem_size_t size) * Create the kernel stack for a new thread. */ static vm_offset_t -vm_thread_stack_create(struct domainset *ds, int pages) +vm_thread_stack_create(struct domainset *ds, int pages, int flags) { vm_page_t ma[KSTACK_MAX_PAGES]; struct vm_domainset_iter di; - int req = VM_ALLOC_NORMAL; - vm_object_t obj; + int req; vm_offset_t ks; int domain, i; - obj = vm_thread_kstack_size_to_obj(pages); - if (vm_ndomains > 1) - obj->domain.dr_policy = ds; - vm_domainset_iter_page_init(&di, obj, 0, &domain, &req); + vm_domainset_iter_policy_init(&di, ds, &domain, &flags); + req = malloc2vm_flags(flags); do { /* * Get a kernel virtual address for this thread's kstack. @@ -480,7 +477,7 @@ vm_thread_stack_create(struct domainset *ds, int pages) vm_page_valid(ma[i]); pmap_qenter(ks, ma, pages); return (ks); - } while (vm_domainset_iter_page(&di, obj, &domain, NULL) == 0); + } while (vm_domainset_iter_policy(&di, &domain) == 0); return (0); } @@ -532,15 +529,9 @@ vm_thread_new(struct thread *td, int pages) ks = 0; if (pages == kstack_pages && kstack_cache != NULL) ks = (vm_offset_t)uma_zalloc(kstack_cache, M_NOWAIT); - - /* - * Ensure that kstack objects can draw pages from any memory - * domain. Otherwise a local memory shortage can block a process - * swap-in. - */ if (ks == 0) ks = vm_thread_stack_create(DOMAINSET_PREF(PCPU_GET(domain)), - pages); + pages, M_NOWAIT); if (ks == 0) return (0); @@ -660,7 +651,8 @@ kstack_import(void *arg, void **store, int cnt, int domain, int flags) ds = DOMAINSET_PREF(domain); for (i = 0; i < cnt; i++) { - store[i] = (void *)vm_thread_stack_create(ds, kstack_pages); + store[i] = (void *)vm_thread_stack_create(ds, kstack_pages, + flags); if (store[i] == NULL) break; } diff --git a/sys/x86/acpica/acpi_apm.c b/sys/x86/acpica/acpi_apm.c index 8e5785cf0ed6..919f76949dd4 100644 --- a/sys/x86/acpica/acpi_apm.c +++ b/sys/x86/acpica/acpi_apm.c @@ -64,6 +64,7 @@ static const struct filterops apm_readfiltops = { .f_isfd = 1, .f_detach = apmreadfiltdetach, .f_event = apmreadfilt, + .f_copy = knote_triv_copy, }; static struct cdevsw apm_cdevsw = { |