diff options
Diffstat (limited to 'drivers/hv/vmbus_drv.c')
| -rw-r--r-- | drivers/hv/vmbus_drv.c | 610 |
1 files changed, 450 insertions, 160 deletions
diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c index 67f95a29aeca..a53af6fe81a6 100644 --- a/drivers/hv/vmbus_drv.c +++ b/drivers/hv/vmbus_drv.c @@ -36,6 +36,7 @@ #include <linux/syscore_ops.h> #include <linux/dma-map-ops.h> #include <linux/pci.h> +#include <linux/export.h> #include <clocksource/hyperv_timer.h> #include <asm/mshyperv.h> #include "hyperv_vmbus.h" @@ -45,7 +46,8 @@ struct vmbus_dynid { struct hv_vmbus_device_id id; }; -static struct device *hv_dev; +/* VMBus Root Device */ +static struct device *vmbus_root_device; static int hyperv_cpuhp_online; @@ -56,6 +58,18 @@ int vmbus_irq; int vmbus_interrupt; /* + * If the Confidential VMBus is used, the data on the "wire" is not + * visible to either the host or the hypervisor. + */ +static bool is_confidential; + +bool vmbus_is_confidential(void) +{ + return is_confidential; +} +EXPORT_SYMBOL_GPL(vmbus_is_confidential); + +/* * The panic notifier below is responsible solely for unloading the * vmbus connection, which is necessary in a panic event. * @@ -80,9 +94,15 @@ static struct resource *fb_mmio; static struct resource *hyperv_mmio; static DEFINE_MUTEX(hyperv_mmio_lock); +struct device *hv_get_vmbus_root_device(void) +{ + return vmbus_root_device; +} +EXPORT_SYMBOL_GPL(hv_get_vmbus_root_device); + static int vmbus_exists(void) { - if (hv_dev == NULL) + if (vmbus_root_device == NULL) return -ENODEV; return 0; @@ -131,7 +151,7 @@ static ssize_t id_show(struct device *dev, struct device_attribute *dev_attr, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", hv_dev->channel->offermsg.child_relid); + return sysfs_emit(buf, "%d\n", hv_dev->channel->offermsg.child_relid); } static DEVICE_ATTR_RO(id); @@ -142,7 +162,7 @@ static ssize_t state_show(struct device *dev, struct device_attribute *dev_attr, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", hv_dev->channel->state); + return sysfs_emit(buf, "%d\n", hv_dev->channel->state); } static DEVICE_ATTR_RO(state); @@ -153,7 +173,7 @@ static ssize_t monitor_id_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", hv_dev->channel->offermsg.monitorid); + return sysfs_emit(buf, "%d\n", hv_dev->channel->offermsg.monitorid); } static DEVICE_ATTR_RO(monitor_id); @@ -164,8 +184,8 @@ static ssize_t class_id_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "{%pUl}\n", - &hv_dev->channel->offermsg.offer.if_type); + return sysfs_emit(buf, "{%pUl}\n", + &hv_dev->channel->offermsg.offer.if_type); } static DEVICE_ATTR_RO(class_id); @@ -176,8 +196,8 @@ static ssize_t device_id_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "{%pUl}\n", - &hv_dev->channel->offermsg.offer.if_instance); + return sysfs_emit(buf, "{%pUl}\n", + &hv_dev->channel->offermsg.offer.if_instance); } static DEVICE_ATTR_RO(device_id); @@ -186,7 +206,7 @@ static ssize_t modalias_show(struct device *dev, { struct hv_device *hv_dev = device_to_hv_device(dev); - return sprintf(buf, "vmbus:%*phN\n", UUID_SIZE, &hv_dev->dev_type); + return sysfs_emit(buf, "vmbus:%*phN\n", UUID_SIZE, &hv_dev->dev_type); } static DEVICE_ATTR_RO(modalias); @@ -199,7 +219,7 @@ static ssize_t numa_node_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", cpu_to_node(hv_dev->channel->target_cpu)); + return sysfs_emit(buf, "%d\n", cpu_to_node(hv_dev->channel->target_cpu)); } static DEVICE_ATTR_RO(numa_node); #endif @@ -212,9 +232,8 @@ static ssize_t server_monitor_pending_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", - channel_pending(hv_dev->channel, - vmbus_connection.monitor_pages[0])); + return sysfs_emit(buf, "%d\n", channel_pending(hv_dev->channel, + vmbus_connection.monitor_pages[0])); } static DEVICE_ATTR_RO(server_monitor_pending); @@ -226,9 +245,8 @@ static ssize_t client_monitor_pending_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", - channel_pending(hv_dev->channel, - vmbus_connection.monitor_pages[1])); + return sysfs_emit(buf, "%d\n", channel_pending(hv_dev->channel, + vmbus_connection.monitor_pages[1])); } static DEVICE_ATTR_RO(client_monitor_pending); @@ -240,9 +258,8 @@ static ssize_t server_monitor_latency_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", - channel_latency(hv_dev->channel, - vmbus_connection.monitor_pages[0])); + return sysfs_emit(buf, "%d\n", channel_latency(hv_dev->channel, + vmbus_connection.monitor_pages[0])); } static DEVICE_ATTR_RO(server_monitor_latency); @@ -254,9 +271,8 @@ static ssize_t client_monitor_latency_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", - channel_latency(hv_dev->channel, - vmbus_connection.monitor_pages[1])); + return sysfs_emit(buf, "%d\n", channel_latency(hv_dev->channel, + vmbus_connection.monitor_pages[1])); } static DEVICE_ATTR_RO(client_monitor_latency); @@ -268,9 +284,8 @@ static ssize_t server_monitor_conn_id_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", - channel_conn_id(hv_dev->channel, - vmbus_connection.monitor_pages[0])); + return sysfs_emit(buf, "%d\n", channel_conn_id(hv_dev->channel, + vmbus_connection.monitor_pages[0])); } static DEVICE_ATTR_RO(server_monitor_conn_id); @@ -282,9 +297,8 @@ static ssize_t client_monitor_conn_id_show(struct device *dev, if (!hv_dev->channel) return -ENODEV; - return sprintf(buf, "%d\n", - channel_conn_id(hv_dev->channel, - vmbus_connection.monitor_pages[1])); + return sysfs_emit(buf, "%d\n", channel_conn_id(hv_dev->channel, + vmbus_connection.monitor_pages[1])); } static DEVICE_ATTR_RO(client_monitor_conn_id); @@ -303,7 +317,7 @@ static ssize_t out_intr_mask_show(struct device *dev, if (ret < 0) return ret; - return sprintf(buf, "%d\n", outbound.current_interrupt_mask); + return sysfs_emit(buf, "%d\n", outbound.current_interrupt_mask); } static DEVICE_ATTR_RO(out_intr_mask); @@ -321,7 +335,7 @@ static ssize_t out_read_index_show(struct device *dev, &outbound); if (ret < 0) return ret; - return sprintf(buf, "%d\n", outbound.current_read_index); + return sysfs_emit(buf, "%u\n", outbound.current_read_index); } static DEVICE_ATTR_RO(out_read_index); @@ -340,7 +354,7 @@ static ssize_t out_write_index_show(struct device *dev, &outbound); if (ret < 0) return ret; - return sprintf(buf, "%d\n", outbound.current_write_index); + return sysfs_emit(buf, "%u\n", outbound.current_write_index); } static DEVICE_ATTR_RO(out_write_index); @@ -359,7 +373,7 @@ static ssize_t out_read_bytes_avail_show(struct device *dev, &outbound); if (ret < 0) return ret; - return sprintf(buf, "%d\n", outbound.bytes_avail_toread); + return sysfs_emit(buf, "%d\n", outbound.bytes_avail_toread); } static DEVICE_ATTR_RO(out_read_bytes_avail); @@ -378,7 +392,7 @@ static ssize_t out_write_bytes_avail_show(struct device *dev, &outbound); if (ret < 0) return ret; - return sprintf(buf, "%d\n", outbound.bytes_avail_towrite); + return sysfs_emit(buf, "%d\n", outbound.bytes_avail_towrite); } static DEVICE_ATTR_RO(out_write_bytes_avail); @@ -396,7 +410,7 @@ static ssize_t in_intr_mask_show(struct device *dev, if (ret < 0) return ret; - return sprintf(buf, "%d\n", inbound.current_interrupt_mask); + return sysfs_emit(buf, "%d\n", inbound.current_interrupt_mask); } static DEVICE_ATTR_RO(in_intr_mask); @@ -414,7 +428,7 @@ static ssize_t in_read_index_show(struct device *dev, if (ret < 0) return ret; - return sprintf(buf, "%d\n", inbound.current_read_index); + return sysfs_emit(buf, "%d\n", inbound.current_read_index); } static DEVICE_ATTR_RO(in_read_index); @@ -432,7 +446,7 @@ static ssize_t in_write_index_show(struct device *dev, if (ret < 0) return ret; - return sprintf(buf, "%d\n", inbound.current_write_index); + return sysfs_emit(buf, "%d\n", inbound.current_write_index); } static DEVICE_ATTR_RO(in_write_index); @@ -451,7 +465,7 @@ static ssize_t in_read_bytes_avail_show(struct device *dev, if (ret < 0) return ret; - return sprintf(buf, "%d\n", inbound.bytes_avail_toread); + return sysfs_emit(buf, "%d\n", inbound.bytes_avail_toread); } static DEVICE_ATTR_RO(in_read_bytes_avail); @@ -470,7 +484,7 @@ static ssize_t in_write_bytes_avail_show(struct device *dev, if (ret < 0) return ret; - return sprintf(buf, "%d\n", inbound.bytes_avail_towrite); + return sysfs_emit(buf, "%d\n", inbound.bytes_avail_towrite); } static DEVICE_ATTR_RO(in_write_bytes_avail); @@ -480,7 +494,7 @@ static ssize_t channel_vp_mapping_show(struct device *dev, { struct hv_device *hv_dev = device_to_hv_device(dev); struct vmbus_channel *channel = hv_dev->channel, *cur_sc; - int buf_size = PAGE_SIZE, n_written, tot_written; + int n_written; struct list_head *cur; if (!channel) @@ -488,25 +502,21 @@ static ssize_t channel_vp_mapping_show(struct device *dev, mutex_lock(&vmbus_connection.channel_mutex); - tot_written = snprintf(buf, buf_size, "%u:%u\n", - channel->offermsg.child_relid, channel->target_cpu); + n_written = sysfs_emit(buf, "%u:%u\n", + channel->offermsg.child_relid, + channel->target_cpu); list_for_each(cur, &channel->sc_list) { - if (tot_written >= buf_size - 1) - break; cur_sc = list_entry(cur, struct vmbus_channel, sc_list); - n_written = scnprintf(buf + tot_written, - buf_size - tot_written, - "%u:%u\n", - cur_sc->offermsg.child_relid, - cur_sc->target_cpu); - tot_written += n_written; + n_written += sysfs_emit_at(buf, n_written, "%u:%u\n", + cur_sc->offermsg.child_relid, + cur_sc->target_cpu); } mutex_unlock(&vmbus_connection.channel_mutex); - return tot_written; + return n_written; } static DEVICE_ATTR_RO(channel_vp_mapping); @@ -516,7 +526,7 @@ static ssize_t vendor_show(struct device *dev, { struct hv_device *hv_dev = device_to_hv_device(dev); - return sprintf(buf, "0x%x\n", hv_dev->vendor_id); + return sysfs_emit(buf, "0x%x\n", hv_dev->vendor_id); } static DEVICE_ATTR_RO(vendor); @@ -526,7 +536,7 @@ static ssize_t device_show(struct device *dev, { struct hv_device *hv_dev = device_to_hv_device(dev); - return sprintf(buf, "0x%x\n", hv_dev->device_id); + return sysfs_emit(buf, "0x%x\n", hv_dev->device_id); } static DEVICE_ATTR_RO(device); @@ -551,7 +561,7 @@ static ssize_t driver_override_show(struct device *dev, ssize_t len; device_lock(dev); - len = snprintf(buf, PAGE_SIZE, "%s\n", hv_dev->driver_override); + len = sysfs_emit(buf, "%s\n", hv_dev->driver_override); device_unlock(dev); return len; @@ -695,7 +705,7 @@ static const struct hv_vmbus_device_id vmbus_device_null; * Return a matching hv_vmbus_device_id pointer. * If there is no match, return NULL. */ -static const struct hv_vmbus_device_id *hv_vmbus_get_id(struct hv_driver *drv, +static const struct hv_vmbus_device_id *hv_vmbus_get_id(const struct hv_driver *drv, struct hv_device *dev) { const guid_t *guid = &dev->dev_type; @@ -706,7 +716,7 @@ static const struct hv_vmbus_device_id *hv_vmbus_get_id(struct hv_driver *drv, return NULL; /* Look at the dynamic ids first, before the static ones */ - id = hv_vmbus_dynid_match(drv, guid); + id = hv_vmbus_dynid_match((struct hv_driver *)drv, guid); if (!id) id = hv_vmbus_dev_match(drv->id_table, guid); @@ -717,7 +727,30 @@ static const struct hv_vmbus_device_id *hv_vmbus_get_id(struct hv_driver *drv, return id; } -/* vmbus_add_dynid - add a new device ID to this driver and re-probe devices */ +/* vmbus_add_dynid - add a new device ID to this driver and re-probe devices + * + * This function can race with vmbus_device_register(). This function is + * typically running on a user thread in response to writing to the "new_id" + * sysfs entry for a driver. vmbus_device_register() is running on a + * workqueue thread in response to the Hyper-V host offering a device to the + * guest. This function calls driver_attach(), which looks for an existing + * device matching the new id, and attaches the driver to which the new id + * has been assigned. vmbus_device_register() calls device_register(), which + * looks for a driver that matches the device being registered. If both + * operations are running simultaneously, the device driver probe function runs + * on whichever thread establishes the linkage between the driver and device. + * + * In most cases, it doesn't matter which thread runs the driver probe + * function. But if vmbus_device_register() does not find a matching driver, + * it proceeds to create the "channels" subdirectory and numbered per-channel + * subdirectory in sysfs. While that multi-step creation is in progress, this + * function could run the driver probe function. If the probe function checks + * for, or operates on, entries in the "channels" subdirectory, including by + * calling hv_create_ring_sysfs(), the operation may or may not succeed + * depending on the race. The race can't create a kernel failure in VMBus + * or device subsystem code, but probe functions in VMBus drivers doing such + * operations must be prepared for the failure case. + */ static int vmbus_add_dynid(struct hv_driver *drv, guid_t *guid) { struct vmbus_dynid *dynid; @@ -819,9 +852,9 @@ ATTRIBUTE_GROUPS(vmbus_drv); /* * vmbus_match - Attempt to match the specified device to the specified driver */ -static int vmbus_match(struct device *device, struct device_driver *driver) +static int vmbus_match(struct device *device, const struct device_driver *driver) { - struct hv_driver *drv = drv_to_hv_drv(driver); + const struct hv_driver *drv = drv_to_hv_drv(driver); struct hv_device *hv_dev = device_to_hv_device(device); /* The hv_sock driver handles all hv_sock offers. */ @@ -871,7 +904,7 @@ static int vmbus_dma_configure(struct device *child_device) * On x86/x64 coherence is assumed and these calls have no effect. */ hv_setup_dma_ops(child_device, - device_get_dma_attr(hv_dev) == DEV_DMA_COHERENT); + device_get_dma_attr(vmbus_root_device) == DEV_DMA_COHERENT); return 0; } @@ -988,7 +1021,7 @@ static const struct dev_pm_ops vmbus_pm = { }; /* The one and only one */ -static struct bus_type hv_bus = { +static const struct bus_type hv_bus = { .name = "vmbus", .match = vmbus_match, .shutdown = vmbus_shutdown, @@ -1025,12 +1058,9 @@ static void vmbus_onmessage_work(struct work_struct *work) kfree(ctx); } -void vmbus_on_msg_dpc(unsigned long data) +static void __vmbus_on_msg_dpc(void *message_page_addr) { - struct hv_per_cpu_context *hv_cpu = (void *)data; - void *page_addr = hv_cpu->synic_message_page; - struct hv_message msg_copy, *msg = (struct hv_message *)page_addr + - VMBUS_MESSAGE_SINT; + struct hv_message msg_copy, *msg; struct vmbus_channel_message_header *hdr; enum vmbus_channel_message_type msgtype; const struct vmbus_channel_message_table_entry *entry; @@ -1038,6 +1068,10 @@ void vmbus_on_msg_dpc(unsigned long data) __u8 payload_size; u32 message_type; + if (!message_page_addr) + return; + msg = (struct hv_message *)message_page_addr + VMBUS_MESSAGE_SINT; + /* * 'enum vmbus_channel_message_type' is supposed to always be 'u32' as * it is being used in 'struct vmbus_channel_message_header' definition @@ -1163,6 +1197,14 @@ msg_handled: vmbus_signal_eom(msg, message_type); } +void vmbus_on_msg_dpc(unsigned long data) +{ + struct hv_per_cpu_context *hv_cpu = (void *)data; + + __vmbus_on_msg_dpc(hv_cpu->hyp_synic_message_page); + __vmbus_on_msg_dpc(hv_cpu->para_synic_message_page); +} + #ifdef CONFIG_PM_SLEEP /* * Fake RESCIND_CHANNEL messages to clean up hv_sock channels by force for @@ -1201,21 +1243,19 @@ static void vmbus_force_channel_rescinded(struct vmbus_channel *channel) #endif /* CONFIG_PM_SLEEP */ /* - * Schedule all channels with events pending + * Schedule all channels with events pending. + * The event page can be directly checked to get the id of + * the channel that has the interrupt pending. */ -static void vmbus_chan_sched(struct hv_per_cpu_context *hv_cpu) +static void vmbus_chan_sched(void *event_page_addr) { unsigned long *recv_int_page; u32 maxbits, relid; + union hv_synic_event_flags *event; - /* - * The event page can be directly checked to get the id of - * the channel that has the interrupt pending. - */ - void *page_addr = hv_cpu->synic_event_page; - union hv_synic_event_flags *event - = (union hv_synic_event_flags *)page_addr + - VMBUS_MESSAGE_SINT; + if (!event_page_addr) + return; + event = (union hv_synic_event_flags *)event_page_addr + VMBUS_MESSAGE_SINT; maxbits = HV_EVENT_FLAGS_COUNT; recv_int_page = event->flags; @@ -1223,6 +1263,11 @@ static void vmbus_chan_sched(struct hv_per_cpu_context *hv_cpu) if (unlikely(!recv_int_page)) return; + /* + * Suggested-by: Michael Kelley <mhklinux@outlook.com> + * One possible optimization would be to keep track of the largest relID that's in use, + * and only scan up to that relID. + */ for_each_set_bit(relid, recv_int_page, maxbits) { void (*callback_fn)(void *context); struct vmbus_channel *channel; @@ -1286,29 +1331,39 @@ sched_unlock_rcu: } } -static void vmbus_isr(void) +static void vmbus_message_sched(struct hv_per_cpu_context *hv_cpu, void *message_page_addr) { - struct hv_per_cpu_context *hv_cpu - = this_cpu_ptr(hv_context.cpu_context); - void *page_addr; struct hv_message *msg; - vmbus_chan_sched(hv_cpu); - - page_addr = hv_cpu->synic_message_page; - msg = (struct hv_message *)page_addr + VMBUS_MESSAGE_SINT; + if (!message_page_addr) + return; + msg = (struct hv_message *)message_page_addr + VMBUS_MESSAGE_SINT; /* Check if there are actual msgs to be processed */ if (msg->header.message_type != HVMSG_NONE) { if (msg->header.message_type == HVMSG_TIMER_EXPIRED) { hv_stimer0_isr(); vmbus_signal_eom(msg, HVMSG_TIMER_EXPIRED); - } else + } else { tasklet_schedule(&hv_cpu->msg_dpc); + } } +} + +void vmbus_isr(void) +{ + struct hv_per_cpu_context *hv_cpu + = this_cpu_ptr(hv_context.cpu_context); + + vmbus_chan_sched(hv_cpu->hyp_synic_event_page); + vmbus_chan_sched(hv_cpu->para_synic_event_page); + + vmbus_message_sched(hv_cpu, hv_cpu->hyp_synic_message_page); + vmbus_message_sched(hv_cpu, hv_cpu->para_synic_message_page); add_interrupt_randomness(vmbus_interrupt); } +EXPORT_SYMBOL_FOR_MODULES(vmbus_isr, "mshv_vtl"); static irqreturn_t vmbus_percpu_isr(int irq, void *dev_id) { @@ -1316,6 +1371,66 @@ static irqreturn_t vmbus_percpu_isr(int irq, void *dev_id) return IRQ_HANDLED; } +static void vmbus_percpu_work(struct work_struct *work) +{ + unsigned int cpu = smp_processor_id(); + + hv_synic_init(cpu); +} + +static int vmbus_alloc_synic_and_connect(void) +{ + int ret, cpu; + struct work_struct __percpu *works; + int hyperv_cpuhp_online; + + ret = hv_synic_alloc(); + if (ret < 0) + goto err_alloc; + + works = alloc_percpu(struct work_struct); + if (!works) { + ret = -ENOMEM; + goto err_alloc; + } + + /* + * Initialize the per-cpu interrupt state and stimer state. + * Then connect to the host. + */ + cpus_read_lock(); + for_each_online_cpu(cpu) { + struct work_struct *work = per_cpu_ptr(works, cpu); + + INIT_WORK(work, vmbus_percpu_work); + schedule_work_on(cpu, work); + } + + for_each_online_cpu(cpu) + flush_work(per_cpu_ptr(works, cpu)); + + /* Register the callbacks for possible CPU online/offline'ing */ + ret = cpuhp_setup_state_nocalls_cpuslocked(CPUHP_AP_ONLINE_DYN, "hyperv/vmbus:online", + hv_synic_init, hv_synic_cleanup); + cpus_read_unlock(); + free_percpu(works); + if (ret < 0) + goto err_alloc; + hyperv_cpuhp_online = ret; + + ret = vmbus_connect(); + if (ret) + goto err_connect; + return 0; + +err_connect: + cpuhp_remove_state(hyperv_cpuhp_online); + return -ENODEV; +err_alloc: + hv_synic_free(); + return -ENOMEM; +} + /* * vmbus_bus_init -Main vmbus driver initialization routine. * @@ -1361,21 +1476,15 @@ static int vmbus_bus_init(void) } } - ret = hv_synic_alloc(); - if (ret) - goto err_alloc; - /* - * Initialize the per-cpu interrupt state and stimer state. - * Then connect to the host. + * Cache the value as getting it involves a VM exit on x86(_64), and + * doing that on each VP while initializing SynIC's wastes time. */ - ret = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "hyperv/vmbus:online", - hv_synic_init, hv_synic_cleanup); - if (ret < 0) - goto err_alloc; - hyperv_cpuhp_online = ret; - - ret = vmbus_connect(); + is_confidential = ms_hyperv.confidential_vmbus_available; + if (is_confidential) + pr_info("Establishing connection to the confidential VMBus\n"); + hv_para_set_sint_proxy(!is_confidential); + ret = vmbus_alloc_synic_and_connect(); if (ret) goto err_connect; @@ -1391,9 +1500,6 @@ static int vmbus_bus_init(void) return 0; err_connect: - cpuhp_remove_state(hyperv_cpuhp_online); -err_alloc: - hv_synic_free(); if (vmbus_irq == -1) { hv_remove_vmbus_handler(); } else { @@ -1593,16 +1699,16 @@ static ssize_t target_cpu_show(struct vmbus_channel *channel, char *buf) { return sprintf(buf, "%u\n", channel->target_cpu); } -static ssize_t target_cpu_store(struct vmbus_channel *channel, - const char *buf, size_t count) + +int vmbus_channel_set_cpu(struct vmbus_channel *channel, u32 target_cpu) { - u32 target_cpu, origin_cpu; - ssize_t ret = count; + u32 origin_cpu; + int ret = 0; - if (vmbus_proto_version < VERSION_WIN10_V4_1) - return -EIO; + lockdep_assert_cpus_held(); + lockdep_assert_held(&vmbus_connection.channel_mutex); - if (sscanf(buf, "%uu", &target_cpu) != 1) + if (vmbus_proto_version < VERSION_WIN10_V4_1) return -EIO; /* Validate target_cpu for the cpumask_test_cpu() operation below. */ @@ -1612,22 +1718,17 @@ static ssize_t target_cpu_store(struct vmbus_channel *channel, if (!cpumask_test_cpu(target_cpu, housekeeping_cpumask(HK_TYPE_MANAGED_IRQ))) return -EINVAL; - /* No CPUs should come up or down during this. */ - cpus_read_lock(); - - if (!cpu_online(target_cpu)) { - cpus_read_unlock(); + if (!cpu_online(target_cpu)) return -EINVAL; - } /* - * Synchronizes target_cpu_store() and channel closure: + * Synchronizes vmbus_channel_set_cpu() and channel closure: * * { Initially: state = CHANNEL_OPENED } * * CPU1 CPU2 * - * [target_cpu_store()] [vmbus_disconnect_ring()] + * [vmbus_channel_set_cpu()] [vmbus_disconnect_ring()] * * LOCK channel_mutex LOCK channel_mutex * LOAD r1 = state LOAD r2 = state @@ -1642,7 +1743,6 @@ static ssize_t target_cpu_store(struct vmbus_channel *channel, * Note. The host processes the channel messages "sequentially", in * the order in which they are received on a per-partition basis. */ - mutex_lock(&vmbus_connection.channel_mutex); /* * Hyper-V will ignore MODIFYCHANNEL messages for "non-open" channels; @@ -1650,17 +1750,17 @@ static ssize_t target_cpu_store(struct vmbus_channel *channel, */ if (channel->state != CHANNEL_OPENED_STATE) { ret = -EIO; - goto cpu_store_unlock; + goto end; } origin_cpu = channel->target_cpu; if (target_cpu == origin_cpu) - goto cpu_store_unlock; + goto end; if (vmbus_send_modifychannel(channel, hv_cpu_number_to_vp_number(target_cpu))) { ret = -EIO; - goto cpu_store_unlock; + goto end; } /* @@ -1690,10 +1790,26 @@ static ssize_t target_cpu_store(struct vmbus_channel *channel, origin_cpu, target_cpu); } -cpu_store_unlock: +end: + return ret; +} + +static ssize_t target_cpu_store(struct vmbus_channel *channel, + const char *buf, size_t count) +{ + u32 target_cpu; + ssize_t ret; + + if (sscanf(buf, "%u", &target_cpu) != 1) + return -EIO; + + cpus_read_lock(); + mutex_lock(&vmbus_connection.channel_mutex); + ret = vmbus_channel_set_cpu(channel, target_cpu); mutex_unlock(&vmbus_connection.channel_mutex); cpus_read_unlock(); - return ret; + + return ret ?: count; } static VMBUS_CHAN_ATTR(cpu, 0644, target_cpu_show, target_cpu_store); @@ -1774,6 +1890,26 @@ static ssize_t subchannel_id_show(struct vmbus_channel *channel, } static VMBUS_CHAN_ATTR_RO(subchannel_id); +static int hv_mmap_ring_buffer_wrapper(struct file *filp, struct kobject *kobj, + const struct bin_attribute *attr, + struct vm_area_struct *vma) +{ + struct vmbus_channel *channel = container_of(kobj, struct vmbus_channel, kobj); + + /* + * hv_(create|remove)_ring_sysfs implementation ensures that mmap_ring_buffer + * is not NULL. + */ + return channel->mmap_ring_buffer(channel, vma); +} + +static struct bin_attribute chan_attr_ring_buffer = { + .attr = { + .name = "ring", + .mode = 0600, + }, + .mmap = hv_mmap_ring_buffer_wrapper, +}; static struct attribute *vmbus_chan_attrs[] = { &chan_attr_out_mask.attr, &chan_attr_in_mask.attr, @@ -1793,6 +1929,11 @@ static struct attribute *vmbus_chan_attrs[] = { NULL }; +static const struct bin_attribute *vmbus_chan_bin_attrs[] = { + &chan_attr_ring_buffer, + NULL +}; + /* * Channel-level attribute_group callback function. Returns the permission for * each attribute, and returns 0 if an attribute is not visible. @@ -1813,16 +1954,99 @@ static umode_t vmbus_chan_attr_is_visible(struct kobject *kobj, return attr->mode; } -static struct attribute_group vmbus_chan_group = { +static umode_t vmbus_chan_bin_attr_is_visible(struct kobject *kobj, + const struct bin_attribute *attr, int idx) +{ + const struct vmbus_channel *channel = + container_of(kobj, struct vmbus_channel, kobj); + + /* Hide ring attribute if channel's ring_sysfs_visible is set to false */ + if (attr == &chan_attr_ring_buffer && !channel->ring_sysfs_visible) + return 0; + + return attr->attr.mode; +} + +static size_t vmbus_chan_bin_size(struct kobject *kobj, + const struct bin_attribute *bin_attr, int a) +{ + const struct vmbus_channel *channel = + container_of(kobj, struct vmbus_channel, kobj); + + return channel->ringbuffer_pagecount << PAGE_SHIFT; +} + +static const struct attribute_group vmbus_chan_group = { .attrs = vmbus_chan_attrs, - .is_visible = vmbus_chan_attr_is_visible + .bin_attrs = vmbus_chan_bin_attrs, + .is_visible = vmbus_chan_attr_is_visible, + .is_bin_visible = vmbus_chan_bin_attr_is_visible, + .bin_size = vmbus_chan_bin_size, }; -static struct kobj_type vmbus_chan_ktype = { +static const struct kobj_type vmbus_chan_ktype = { .sysfs_ops = &vmbus_chan_sysfs_ops, .release = vmbus_chan_release, }; +/** + * hv_create_ring_sysfs() - create "ring" sysfs entry corresponding to ring buffers for a channel. + * @channel: Pointer to vmbus_channel structure + * @hv_mmap_ring_buffer: function pointer for initializing the function to be called on mmap of + * channel's "ring" sysfs node, which is for the ring buffer of that channel. + * Function pointer is of below type: + * int (*hv_mmap_ring_buffer)(struct vmbus_channel *channel, + * struct vm_area_struct *vma)) + * This has a pointer to the channel and a pointer to vm_area_struct, + * used for mmap, as arguments. + * + * Sysfs node for ring buffer of a channel is created along with other fields, however its + * visibility is disabled by default. Sysfs creation needs to be controlled when the use-case + * is running. + * For example, HV_NIC device is used either by uio_hv_generic or hv_netvsc at any given point of + * time, and "ring" sysfs is needed only when uio_hv_generic is bound to that device. To avoid + * exposing the ring buffer by default, this function is responsible to enable visibility of + * ring for userspace to use. + * Note: Race conditions can happen with userspace and it is not encouraged to create new + * use-cases for this. This was added to maintain backward compatibility, while solving + * one of the race conditions in uio_hv_generic while creating sysfs. See comments with + * vmbus_add_dynid() and vmbus_device_register(). + * + * Returns 0 on success or error code on failure. + */ +int hv_create_ring_sysfs(struct vmbus_channel *channel, + int (*hv_mmap_ring_buffer)(struct vmbus_channel *channel, + struct vm_area_struct *vma)) +{ + struct kobject *kobj = &channel->kobj; + + channel->mmap_ring_buffer = hv_mmap_ring_buffer; + channel->ring_sysfs_visible = true; + + return sysfs_update_group(kobj, &vmbus_chan_group); +} +EXPORT_SYMBOL_GPL(hv_create_ring_sysfs); + +/** + * hv_remove_ring_sysfs() - remove ring sysfs entry corresponding to ring buffers for a channel. + * @channel: Pointer to vmbus_channel structure + * + * Hide "ring" sysfs for a channel by changing its is_visible attribute and updating sysfs group. + * + * Returns 0 on success or error code on failure. + */ +int hv_remove_ring_sysfs(struct vmbus_channel *channel) +{ + struct kobject *kobj = &channel->kobj; + int ret; + + channel->ring_sysfs_visible = false; + ret = sysfs_update_group(kobj, &vmbus_chan_group); + channel->mmap_ring_buffer = NULL; + return ret; +} +EXPORT_SYMBOL_GPL(hv_remove_ring_sysfs); + /* * vmbus_add_channel_kobj - setup a sub-directory under device/channels */ @@ -1902,7 +2126,7 @@ int vmbus_device_register(struct hv_device *child_device_obj) &child_device_obj->channel->offermsg.offer.if_instance); child_device_obj->device.bus = &hv_bus; - child_device_obj->device.parent = hv_dev; + child_device_obj->device.parent = vmbus_root_device; child_device_obj->device.release = vmbus_device_release; child_device_obj->device.dma_parms = &child_device_obj->dma_parms; @@ -1920,6 +2144,20 @@ int vmbus_device_register(struct hv_device *child_device_obj) return ret; } + /* + * If device_register() found a driver to assign to the device, the + * driver's probe function has already run at this point. If that + * probe function accesses or operates on the "channels" subdirectory + * in sysfs, those operations will have failed because the "channels" + * subdirectory doesn't exist until the code below runs. Or if the + * probe function creates a /dev entry, a user space program could + * find and open the /dev entry, and then create a race by accessing + * the "channels" subdirectory while the creation steps are in progress + * here. The race can't result in a kernel failure, but the user space + * program may get an error in accessing "channels" or its + * subdirectories. See also comments with vmbus_add_dynid() about a + * related race condition. + */ child_device_obj->channels_kset = kset_create_and_add("channels", NULL, kobj); if (!child_device_obj->channels_kset) { @@ -1930,7 +2168,7 @@ int vmbus_device_register(struct hv_device *child_device_obj) ret = vmbus_add_channel_kobj(child_device_obj, child_device_obj->channel); if (ret) { - pr_err("Unable to register primary channeln"); + pr_err("Unable to register primary channel\n"); goto err_kset_unregister; } hv_debug_add_dev_dir(child_device_obj); @@ -1962,6 +2200,7 @@ void vmbus_device_unregister(struct hv_device *device_obj) */ device_unregister(&device_obj->device); } +EXPORT_SYMBOL_GPL(vmbus_device_unregister); #ifdef CONFIG_ACPI /* @@ -2100,8 +2339,10 @@ static void __maybe_unused vmbus_reserve_fb(void) if (efi_enabled(EFI_BOOT)) { /* Gen2 VM: get FB base from EFI framebuffer */ - start = screen_info.lfb_base; - size = max_t(__u32, screen_info.lfb_size, 0x800000); + if (IS_ENABLED(CONFIG_SYSFB)) { + start = screen_info.lfb_base; + size = max_t(__u32, screen_info.lfb_size, 0x800000); + } } else { /* Gen1 VM: get FB base from PCI */ pdev = pci_get_device(PCI_VENDOR_ID_MICROSOFT, @@ -2241,12 +2482,25 @@ void vmbus_free_mmio(resource_size_t start, resource_size_t size) struct resource *iter; mutex_lock(&hyperv_mmio_lock); + + /* + * If all bytes of the MMIO range to be released are within the + * special case fb_mmio shadow region, skip releasing the shadow + * region since no corresponding __request_region() was done + * in vmbus_allocate_mmio(). + */ + if (fb_mmio && start >= fb_mmio->start && + (start + size - 1 <= fb_mmio->end)) + goto skip_shadow_release; + for (iter = hyperv_mmio; iter; iter = iter->sibling) { if ((iter->start >= start + size) || (iter->end <= start)) continue; __release_region(iter, start, size); } + +skip_shadow_release: release_mem_region(start, size); mutex_unlock(&hyperv_mmio_lock); @@ -2261,7 +2515,7 @@ static int vmbus_acpi_add(struct platform_device *pdev) struct acpi_device *ancestor; struct acpi_device *device = ACPI_COMPANION(&pdev->dev); - hv_dev = &device->dev; + vmbus_root_device = &device->dev; /* * Older versions of Hyper-V for ARM64 fail to include the _CCA @@ -2287,7 +2541,8 @@ static int vmbus_acpi_add(struct platform_device *pdev) * Some ancestor of the vmbus acpi device (Gen1 or Gen2 * firmware) is the VMOD that has the mmio ranges. Get that. */ - for (ancestor = acpi_dev_parent(device); ancestor; + for (ancestor = acpi_dev_parent(device); + ancestor && ancestor->handle != ACPI_ROOT_OBJECT; ancestor = acpi_dev_parent(ancestor)) { result = acpi_walk_resources(ancestor->handle, METHOD_NAME__CRS, vmbus_walk_resources, NULL); @@ -2312,6 +2567,32 @@ static int vmbus_acpi_add(struct platform_device *pdev) return 0; } #endif +#ifndef HYPERVISOR_CALLBACK_VECTOR +static int vmbus_set_irq(struct platform_device *pdev) +{ + struct irq_data *data; + int irq; + irq_hw_number_t hwirq; + + irq = platform_get_irq(pdev, 0); + /* platform_get_irq() may not return 0. */ + if (irq < 0) + return irq; + + data = irq_get_irq_data(irq); + if (!data) { + pr_err("No interrupt data for VMBus virq %d\n", irq); + return -ENODEV; + } + hwirq = irqd_to_hwirq(data); + + vmbus_irq = irq; + vmbus_interrupt = hwirq; + pr_debug("VMBus virq %d, hwirq %d\n", vmbus_irq, vmbus_interrupt); + + return 0; +} +#endif static int vmbus_device_add(struct platform_device *pdev) { @@ -2321,12 +2602,17 @@ static int vmbus_device_add(struct platform_device *pdev) struct device_node *np = pdev->dev.of_node; int ret; - hv_dev = &pdev->dev; + vmbus_root_device = &pdev->dev; ret = of_range_parser_init(&parser, np); if (ret) return ret; +#ifndef HYPERVISOR_CALLBACK_VECTOR + ret = vmbus_set_irq(pdev); + if (ret) + return ret; +#endif for_each_of_range(&parser, &range) { struct resource *res; @@ -2356,10 +2642,9 @@ static int vmbus_platform_driver_probe(struct platform_device *pdev) return vmbus_acpi_add(pdev); } -static int vmbus_platform_driver_remove(struct platform_device *pdev) +static void vmbus_platform_driver_remove(struct platform_device *pdev) { vmbus_mmio_remove(); - return 0; } #ifdef CONFIG_PM_SLEEP @@ -2406,11 +2691,6 @@ static int vmbus_bus_suspend(struct device *dev) if (atomic_read(&vmbus_connection.nr_chan_close_on_suspend) > 0) wait_for_completion(&vmbus_connection.ready_for_suspend_event); - if (atomic_read(&vmbus_connection.nr_chan_fixup_on_resume) != 0) { - pr_err("Can not suspend due to a previous failed resuming\n"); - return -EBUSY; - } - mutex_lock(&vmbus_connection.channel_mutex); list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) { @@ -2435,22 +2715,18 @@ static int vmbus_bus_suspend(struct device *dev) pr_err("Sub-channel not deleted!\n"); WARN_ON_ONCE(1); } - - atomic_inc(&vmbus_connection.nr_chan_fixup_on_resume); } mutex_unlock(&vmbus_connection.channel_mutex); vmbus_initiate_unload(false); - /* Reset the event for the next resume. */ - reinit_completion(&vmbus_connection.ready_for_resume_event); - return 0; } static int vmbus_bus_resume(struct device *dev) { + struct vmbus_channel *channel; struct vmbus_channel_msginfo *msginfo; size_t msgsize; int ret; @@ -2481,13 +2757,23 @@ static int vmbus_bus_resume(struct device *dev) if (ret != 0) return ret; - WARN_ON(atomic_read(&vmbus_connection.nr_chan_fixup_on_resume) == 0); - vmbus_request_offers(); - if (wait_for_completion_timeout( - &vmbus_connection.ready_for_resume_event, 10 * HZ) == 0) - pr_err("Some vmbus device is missing after suspending?\n"); + mutex_lock(&vmbus_connection.channel_mutex); + list_for_each_entry(channel, &vmbus_connection.chn_list, listentry) { + if (channel->offermsg.child_relid != INVALID_RELID) + continue; + + /* hvsock channels are not expected to be present. */ + if (is_hvsock_channel(channel)) + continue; + + pr_err("channel %pUl/%pUl not present after resume.\n", + &channel->offermsg.offer.if_type, + &channel->offermsg.offer.if_instance); + /* ToDo: Cleanup these channels here */ + } + mutex_unlock(&vmbus_connection.channel_mutex); /* Reset the event for the next suspend. */ reinit_completion(&vmbus_connection.ready_for_suspend_event); @@ -2570,10 +2856,10 @@ static void hv_crash_handler(struct pt_regs *regs) */ cpu = smp_processor_id(); hv_stimer_cleanup(cpu); - hv_synic_disable_regs(cpu); + hv_hyp_synic_disable_regs(cpu); }; -static int hv_synic_suspend(void) +static int hv_synic_suspend(void *data) { /* * When we reach here, all the non-boot CPUs have been offlined. @@ -2595,14 +2881,14 @@ static int hv_synic_suspend(void) * interrupts-disabled context. */ - hv_synic_disable_regs(0); + hv_hyp_synic_disable_regs(0); return 0; } -static void hv_synic_resume(void) +static void hv_synic_resume(void *data) { - hv_synic_enable_regs(0); + hv_hyp_synic_enable_regs(0); /* * Note: we don't need to call hv_stimer_init(0), because the timer @@ -2612,11 +2898,15 @@ static void hv_synic_resume(void) } /* The callbacks run only on CPU0, with irqs_disabled. */ -static struct syscore_ops hv_synic_syscore_ops = { +static const struct syscore_ops hv_synic_syscore_ops = { .suspend = hv_synic_suspend, .resume = hv_synic_resume, }; +static struct syscore hv_synic_syscore = { + .ops = &hv_synic_syscore_ops, +}; + static int __init hv_acpi_init(void) { int ret; @@ -2624,7 +2914,7 @@ static int __init hv_acpi_init(void) if (!hv_is_hyperv_initialized()) return -ENODEV; - if (hv_root_partition && !hv_nested) + if (hv_root_partition() && !hv_nested) return 0; /* @@ -2634,7 +2924,7 @@ static int __init hv_acpi_init(void) if (ret) return ret; - if (!hv_dev) { + if (!vmbus_root_device) { ret = -ENODEV; goto cleanup; } @@ -2659,13 +2949,13 @@ static int __init hv_acpi_init(void) hv_setup_kexec_handler(hv_kexec_handler); hv_setup_crash_handler(hv_crash_handler); - register_syscore_ops(&hv_synic_syscore_ops); + register_syscore(&hv_synic_syscore); return 0; cleanup: platform_driver_unregister(&vmbus_platform_driver); - hv_dev = NULL; + vmbus_root_device = NULL; return ret; } @@ -2673,7 +2963,7 @@ static void __exit vmbus_exit(void) { int cpu; - unregister_syscore_ops(&hv_synic_syscore_ops); + unregister_syscore(&hv_synic_syscore); hv_remove_kexec_handler(); hv_remove_crash_handler(); |