diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2021-06-29 11:21:35 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2021-06-29 11:21:35 -0700 |
commit | b694011a4aec3e8df98bc59fdb78e018b09de79d (patch) | |
tree | bbbe4c92f12cfb7a50d85dbf1509e51acb19f497 /drivers/hv | |
parent | c54b245d011855ea91c5beff07f1db74143ce614 (diff) | |
parent | 7d815f4afa87f2032b650ae1bba7534b550a6b8b (diff) | |
download | lwn-b694011a4aec3e8df98bc59fdb78e018b09de79d.tar.gz lwn-b694011a4aec3e8df98bc59fdb78e018b09de79d.zip |
Merge tag 'hyperv-next-signed-20210629' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux
Pull hyperv updates from Wei Liu:
"Just a few minor enhancement patches and bug fixes"
* tag 'hyperv-next-signed-20210629' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux:
PCI: hv: Add check for hyperv_initialized in init_hv_pci_drv()
Drivers: hv: Move Hyper-V extended capability check to arch neutral code
drivers: hv: Fix missing error code in vmbus_connect()
x86/hyperv: fix logical processor creation
hv_utils: Fix passing zero to 'PTR_ERR' warning
scsi: storvsc: Use blk_mq_unique_tag() to generate requestIDs
Drivers: hv: vmbus: Copy packets sent by Hyper-V out of the ring buffer
hv_balloon: Remove redundant assignment to region_start
Diffstat (limited to 'drivers/hv')
-rw-r--r-- | drivers/hv/Makefile | 3 | ||||
-rw-r--r-- | drivers/hv/channel.c | 23 | ||||
-rw-r--r-- | drivers/hv/connection.c | 4 | ||||
-rw-r--r-- | drivers/hv/hv_balloon.c | 1 | ||||
-rw-r--r-- | drivers/hv/hv_common.c | 66 | ||||
-rw-r--r-- | drivers/hv/hv_fcopy.c | 1 | ||||
-rw-r--r-- | drivers/hv/hv_kvp.c | 1 | ||||
-rw-r--r-- | drivers/hv/hv_util.c | 4 | ||||
-rw-r--r-- | drivers/hv/hyperv_vmbus.h | 2 | ||||
-rw-r--r-- | drivers/hv/ring_buffer.c | 95 |
10 files changed, 168 insertions, 32 deletions
diff --git a/drivers/hv/Makefile b/drivers/hv/Makefile index 94daf8240c95..d76df5c8c2a9 100644 --- a/drivers/hv/Makefile +++ b/drivers/hv/Makefile @@ -11,3 +11,6 @@ hv_vmbus-y := vmbus_drv.o \ channel_mgmt.o ring_buffer.o hv_trace.o hv_vmbus-$(CONFIG_HYPERV_TESTING) += hv_debugfs.o hv_utils-y := hv_util.o hv_kvp.o hv_snapshot.o hv_fcopy.o hv_utils_transport.o + +# Code that must be built-in +obj-$(subst m,y,$(CONFIG_HYPERV)) += hv_common.o diff --git a/drivers/hv/channel.c b/drivers/hv/channel.c index c2635e913a92..f3761c73b074 100644 --- a/drivers/hv/channel.c +++ b/drivers/hv/channel.c @@ -662,12 +662,15 @@ static int __vmbus_open(struct vmbus_channel *newchannel, newchannel->onchannel_callback = onchannelcallback; newchannel->channel_callback_context = context; - err = hv_ringbuffer_init(&newchannel->outbound, page, send_pages); + if (!newchannel->max_pkt_size) + newchannel->max_pkt_size = VMBUS_DEFAULT_MAX_PKT_SIZE; + + err = hv_ringbuffer_init(&newchannel->outbound, page, send_pages, 0); if (err) goto error_clean_ring; - err = hv_ringbuffer_init(&newchannel->inbound, - &page[send_pages], recv_pages); + err = hv_ringbuffer_init(&newchannel->inbound, &page[send_pages], + recv_pages, newchannel->max_pkt_size); if (err) goto error_clean_ring; @@ -1186,15 +1189,14 @@ EXPORT_SYMBOL_GPL(vmbus_recvpacket_raw); * vmbus_next_request_id - Returns a new request id. It is also * the index at which the guest memory address is stored. * Uses a spin lock to avoid race conditions. - * @rqstor: Pointer to the requestor struct + * @channel: Pointer to the VMbus channel struct * @rqst_add: Guest memory address to be stored in the array */ -u64 vmbus_next_request_id(struct vmbus_requestor *rqstor, u64 rqst_addr) +u64 vmbus_next_request_id(struct vmbus_channel *channel, u64 rqst_addr) { + struct vmbus_requestor *rqstor = &channel->requestor; unsigned long flags; u64 current_id; - const struct vmbus_channel *channel = - container_of(rqstor, const struct vmbus_channel, requestor); /* Check rqstor has been initialized */ if (!channel->rqstor_size) @@ -1228,16 +1230,15 @@ EXPORT_SYMBOL_GPL(vmbus_next_request_id); /* * vmbus_request_addr - Returns the memory address stored at @trans_id * in @rqstor. Uses a spin lock to avoid race conditions. - * @rqstor: Pointer to the requestor struct + * @channel: Pointer to the VMbus channel struct * @trans_id: Request id sent back from Hyper-V. Becomes the requestor's * next request id. */ -u64 vmbus_request_addr(struct vmbus_requestor *rqstor, u64 trans_id) +u64 vmbus_request_addr(struct vmbus_channel *channel, u64 trans_id) { + struct vmbus_requestor *rqstor = &channel->requestor; unsigned long flags; u64 req_addr; - const struct vmbus_channel *channel = - container_of(rqstor, const struct vmbus_channel, requestor); /* Check rqstor has been initialized */ if (!channel->rqstor_size) diff --git a/drivers/hv/connection.c b/drivers/hv/connection.c index 311cd005b3be..5e479d54918c 100644 --- a/drivers/hv/connection.c +++ b/drivers/hv/connection.c @@ -232,8 +232,10 @@ int vmbus_connect(void) */ for (i = 0; ; i++) { - if (i == ARRAY_SIZE(vmbus_versions)) + if (i == ARRAY_SIZE(vmbus_versions)) { + ret = -EDOM; goto cleanup; + } version = vmbus_versions[i]; if (version > max_version) diff --git a/drivers/hv/hv_balloon.c b/drivers/hv/hv_balloon.c index 58af84e30144..7f11ea07d698 100644 --- a/drivers/hv/hv_balloon.c +++ b/drivers/hv/hv_balloon.c @@ -1010,7 +1010,6 @@ static void hot_add_req(struct work_struct *dummy) * that need to be hot-added while ensuring the alignment * and size requirements of Linux as it relates to hot-add. */ - region_start = pg_start; region_size = (pfn_cnt / HA_CHUNK) * HA_CHUNK; if (pfn_cnt % HA_CHUNK) region_size += HA_CHUNK; diff --git a/drivers/hv/hv_common.c b/drivers/hv/hv_common.c new file mode 100644 index 000000000000..7f42da98d377 --- /dev/null +++ b/drivers/hv/hv_common.c @@ -0,0 +1,66 @@ +// SPDX-License-Identifier: GPL-2.0 + +/* + * Architecture neutral utility routines for interacting with + * Hyper-V. This file is specifically for code that must be + * built-in to the kernel image when CONFIG_HYPERV is set + * (vs. being in a module) because it is called from architecture + * specific code under arch/. + * + * Copyright (C) 2021, Microsoft, Inc. + * + * Author : Michael Kelley <mikelley@microsoft.com> + */ + +#include <linux/types.h> +#include <linux/export.h> +#include <linux/bitfield.h> +#include <asm/hyperv-tlfs.h> +#include <asm/mshyperv.h> + + +/* Bit mask of the extended capability to query: see HV_EXT_CAPABILITY_xxx */ +bool hv_query_ext_cap(u64 cap_query) +{ + /* + * The address of the 'hv_extended_cap' variable will be used as an + * output parameter to the hypercall below and so it should be + * compatible with 'virt_to_phys'. Which means, it's address should be + * directly mapped. Use 'static' to keep it compatible; stack variables + * can be virtually mapped, making them incompatible with + * 'virt_to_phys'. + * Hypercall input/output addresses should also be 8-byte aligned. + */ + static u64 hv_extended_cap __aligned(8); + static bool hv_extended_cap_queried; + u64 status; + + /* + * Querying extended capabilities is an extended hypercall. Check if the + * partition supports extended hypercall, first. + */ + if (!(ms_hyperv.priv_high & HV_ENABLE_EXTENDED_HYPERCALLS)) + return false; + + /* Extended capabilities do not change at runtime. */ + if (hv_extended_cap_queried) + return hv_extended_cap & cap_query; + + status = hv_do_hypercall(HV_EXT_CALL_QUERY_CAPABILITIES, NULL, + &hv_extended_cap); + + /* + * The query extended capabilities hypercall should not fail under + * any normal circumstances. Avoid repeatedly making the hypercall, on + * error. + */ + hv_extended_cap_queried = true; + if (!hv_result_success(status)) { + pr_err("Hyper-V: Extended query capabilities hypercall failed 0x%llx\n", + status); + return false; + } + + return hv_extended_cap & cap_query; +} +EXPORT_SYMBOL_GPL(hv_query_ext_cap); diff --git a/drivers/hv/hv_fcopy.c b/drivers/hv/hv_fcopy.c index 59ce85e00a02..660036da7449 100644 --- a/drivers/hv/hv_fcopy.c +++ b/drivers/hv/hv_fcopy.c @@ -349,6 +349,7 @@ int hv_fcopy_init(struct hv_util_service *srv) { recv_buffer = srv->recv_buffer; fcopy_transaction.recv_channel = srv->channel; + fcopy_transaction.recv_channel->max_pkt_size = HV_HYP_PAGE_SIZE * 2; /* * When this driver loads, the user level daemon that diff --git a/drivers/hv/hv_kvp.c b/drivers/hv/hv_kvp.c index b49962d312ce..c698592b83e4 100644 --- a/drivers/hv/hv_kvp.c +++ b/drivers/hv/hv_kvp.c @@ -757,6 +757,7 @@ hv_kvp_init(struct hv_util_service *srv) { recv_buffer = srv->recv_buffer; kvp_transaction.recv_channel = srv->channel; + kvp_transaction.recv_channel->max_pkt_size = HV_HYP_PAGE_SIZE * 4; /* * When this driver loads, the user level daemon that diff --git a/drivers/hv/hv_util.c b/drivers/hv/hv_util.c index e4aefeb330da..136576cba26f 100644 --- a/drivers/hv/hv_util.c +++ b/drivers/hv/hv_util.c @@ -750,8 +750,8 @@ static int hv_timesync_init(struct hv_util_service *srv) */ hv_ptp_clock = ptp_clock_register(&ptp_hyperv_info, NULL); if (IS_ERR_OR_NULL(hv_ptp_clock)) { - pr_err("cannot register PTP clock: %ld\n", - PTR_ERR(hv_ptp_clock)); + pr_err("cannot register PTP clock: %d\n", + PTR_ERR_OR_ZERO(hv_ptp_clock)); hv_ptp_clock = NULL; } diff --git a/drivers/hv/hyperv_vmbus.h b/drivers/hv/hyperv_vmbus.h index 9416e09ebd58..42f3d9d123a1 100644 --- a/drivers/hv/hyperv_vmbus.h +++ b/drivers/hv/hyperv_vmbus.h @@ -174,7 +174,7 @@ extern int hv_synic_cleanup(unsigned int cpu); void hv_ringbuffer_pre_init(struct vmbus_channel *channel); int hv_ringbuffer_init(struct hv_ring_buffer_info *ring_info, - struct page *pages, u32 pagecnt); + struct page *pages, u32 pagecnt, u32 max_pkt_size); void hv_ringbuffer_cleanup(struct hv_ring_buffer_info *ring_info); diff --git a/drivers/hv/ring_buffer.c b/drivers/hv/ring_buffer.c index 374f8afbf8a5..2aee356840a2 100644 --- a/drivers/hv/ring_buffer.c +++ b/drivers/hv/ring_buffer.c @@ -181,7 +181,7 @@ void hv_ringbuffer_pre_init(struct vmbus_channel *channel) /* Initialize the ring buffer. */ int hv_ringbuffer_init(struct hv_ring_buffer_info *ring_info, - struct page *pages, u32 page_cnt) + struct page *pages, u32 page_cnt, u32 max_pkt_size) { int i; struct page **pages_wraparound; @@ -223,6 +223,14 @@ int hv_ringbuffer_init(struct hv_ring_buffer_info *ring_info, sizeof(struct hv_ring_buffer); ring_info->priv_read_index = 0; + /* Initialize buffer that holds copies of incoming packets */ + if (max_pkt_size) { + ring_info->pkt_buffer = kzalloc(max_pkt_size, GFP_KERNEL); + if (!ring_info->pkt_buffer) + return -ENOMEM; + ring_info->pkt_buffer_size = max_pkt_size; + } + spin_lock_init(&ring_info->ring_lock); return 0; @@ -235,6 +243,9 @@ void hv_ringbuffer_cleanup(struct hv_ring_buffer_info *ring_info) vunmap(ring_info->ring_buffer); ring_info->ring_buffer = NULL; mutex_unlock(&ring_info->ring_buffer_mutex); + + kfree(ring_info->pkt_buffer); + ring_info->pkt_buffer_size = 0; } /* Write to the ring buffer. */ @@ -301,10 +312,12 @@ int hv_ringbuffer_write(struct vmbus_channel *channel, */ if (desc->flags == VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED) { - rqst_id = vmbus_next_request_id(&channel->requestor, requestid); - if (rqst_id == VMBUS_RQST_ERROR) { - spin_unlock_irqrestore(&outring_info->ring_lock, flags); - return -EAGAIN; + if (channel->next_request_id_callback != NULL) { + rqst_id = channel->next_request_id_callback(channel, requestid); + if (rqst_id == VMBUS_RQST_ERROR) { + spin_unlock_irqrestore(&outring_info->ring_lock, flags); + return -EAGAIN; + } } } desc = hv_get_ring_buffer(outring_info) + old_write; @@ -332,7 +345,8 @@ int hv_ringbuffer_write(struct vmbus_channel *channel, if (channel->rescind) { if (rqst_id != VMBUS_NO_RQSTOR) { /* Reclaim request ID to avoid leak of IDs */ - vmbus_request_addr(&channel->requestor, rqst_id); + if (channel->request_addr_callback != NULL) + channel->request_addr_callback(channel, rqst_id); } return -ENODEV; } @@ -375,7 +389,7 @@ int hv_ringbuffer_read(struct vmbus_channel *channel, memcpy(buffer, (const char *)desc + offset, packetlen); /* Advance ring index to next packet descriptor */ - __hv_pkt_iter_next(channel, desc); + __hv_pkt_iter_next(channel, desc, true); /* Notify host of update */ hv_pkt_iter_close(channel); @@ -402,6 +416,22 @@ static u32 hv_pkt_iter_avail(const struct hv_ring_buffer_info *rbi) } /* + * Get first vmbus packet without copying it out of the ring buffer + */ +struct vmpacket_descriptor *hv_pkt_iter_first_raw(struct vmbus_channel *channel) +{ + struct hv_ring_buffer_info *rbi = &channel->inbound; + + hv_debug_delay_test(channel, MESSAGE_DELAY); + + if (hv_pkt_iter_avail(rbi) < sizeof(struct vmpacket_descriptor)) + return NULL; + + return (struct vmpacket_descriptor *)(hv_get_ring_buffer(rbi) + rbi->priv_read_index); +} +EXPORT_SYMBOL_GPL(hv_pkt_iter_first_raw); + +/* * Get first vmbus packet from ring buffer after read_index * * If ring buffer is empty, returns NULL and no other action needed. @@ -409,17 +439,49 @@ static u32 hv_pkt_iter_avail(const struct hv_ring_buffer_info *rbi) struct vmpacket_descriptor *hv_pkt_iter_first(struct vmbus_channel *channel) { struct hv_ring_buffer_info *rbi = &channel->inbound; - struct vmpacket_descriptor *desc; + struct vmpacket_descriptor *desc, *desc_copy; + u32 bytes_avail, pkt_len, pkt_offset; - hv_debug_delay_test(channel, MESSAGE_DELAY); - if (hv_pkt_iter_avail(rbi) < sizeof(struct vmpacket_descriptor)) + desc = hv_pkt_iter_first_raw(channel); + if (!desc) return NULL; - desc = hv_get_ring_buffer(rbi) + rbi->priv_read_index; - if (desc) - prefetch((char *)desc + (desc->len8 << 3)); + bytes_avail = min(rbi->pkt_buffer_size, hv_pkt_iter_avail(rbi)); + + /* + * Ensure the compiler does not use references to incoming Hyper-V values (which + * could change at any moment) when reading local variables later in the code + */ + pkt_len = READ_ONCE(desc->len8) << 3; + pkt_offset = READ_ONCE(desc->offset8) << 3; + + /* + * If pkt_len is invalid, set it to the smaller of hv_pkt_iter_avail() and + * rbi->pkt_buffer_size + */ + if (pkt_len < sizeof(struct vmpacket_descriptor) || pkt_len > bytes_avail) + pkt_len = bytes_avail; + + /* + * If pkt_offset is invalid, arbitrarily set it to + * the size of vmpacket_descriptor + */ + if (pkt_offset < sizeof(struct vmpacket_descriptor) || pkt_offset > pkt_len) + pkt_offset = sizeof(struct vmpacket_descriptor); + + /* Copy the Hyper-V packet out of the ring buffer */ + desc_copy = (struct vmpacket_descriptor *)rbi->pkt_buffer; + memcpy(desc_copy, desc, pkt_len); + + /* + * Hyper-V could still change len8 and offset8 after the earlier read. + * Ensure that desc_copy has legal values for len8 and offset8 that + * are consistent with the copy we just made + */ + desc_copy->len8 = pkt_len >> 3; + desc_copy->offset8 = pkt_offset >> 3; - return desc; + return desc_copy; } EXPORT_SYMBOL_GPL(hv_pkt_iter_first); @@ -431,7 +493,8 @@ EXPORT_SYMBOL_GPL(hv_pkt_iter_first); */ struct vmpacket_descriptor * __hv_pkt_iter_next(struct vmbus_channel *channel, - const struct vmpacket_descriptor *desc) + const struct vmpacket_descriptor *desc, + bool copy) { struct hv_ring_buffer_info *rbi = &channel->inbound; u32 packetlen = desc->len8 << 3; @@ -444,7 +507,7 @@ __hv_pkt_iter_next(struct vmbus_channel *channel, rbi->priv_read_index -= dsize; /* more data? */ - return hv_pkt_iter_first(channel); + return copy ? hv_pkt_iter_first(channel) : hv_pkt_iter_first_raw(channel); } EXPORT_SYMBOL_GPL(__hv_pkt_iter_next); |