From 5b7c4cabbb65f5c469464da6c5f614cbd7f730f2 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Tue, 21 Feb 2023 18:24:12 -0800 Subject: Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next Pull networking updates from Jakub Kicinski: "Core: - Add dedicated kmem_cache for typical/small skb->head, avoid having to access struct page at kfree time, and improve memory use. - Introduce sysctl to set default RPS configuration for new netdevs. - Define Netlink protocol specification format which can be used to describe messages used by each family and auto-generate parsers. Add tools for generating kernel data structures and uAPI headers. - Expose all net/core sysctls inside netns. - Remove 4s sleep in netpoll if carrier is instantly detected on boot. - Add configurable limit of MDB entries per port, and port-vlan. - Continue populating drop reasons throughout the stack. - Retire a handful of legacy Qdiscs and classifiers. Protocols: - Support IPv4 big TCP (TSO frames larger than 64kB). - Add IP_LOCAL_PORT_RANGE socket option, to control local port range on socket by socket basis. - Track and report in procfs number of MPTCP sockets used. - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path manager. - IPv6: don't check net.ipv6.route.max_size and rely on garbage collection to free memory (similarly to IPv4). - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986). - ICMP: add per-rate limit counters. - Add support for user scanning requests in ieee802154. - Remove static WEP support. - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate reporting. - WiFi 7 EHT channel puncturing support (client & AP). BPF: - Add a rbtree data structure following the "next-gen data structure" precedent set by recently added linked list, that is, by using kfunc + kptr instead of adding a new BPF map type. - Expose XDP hints via kfuncs with initial support for RX hash and timestamp metadata. - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to better support decap on GRE tunnel devices not operating in collect metadata. - Improve x86 JIT's codegen for PROBE_MEM runtime error checks. - Remove the need for trace_printk_lock for bpf_trace_printk and bpf_trace_vprintk helpers. - Extend libbpf's bpf_tracing.h support for tracing arguments of kprobes/uprobes and syscall as a special case. - Significantly reduce the search time for module symbols by livepatch and BPF. - Enable cpumasks to be used as kptrs, which is useful for tracing programs tracking which tasks end up running on which CPUs in different time intervals. - Add support for BPF trampoline on s390x and riscv64. - Add capability to export the XDP features supported by the NIC. - Add __bpf_kfunc tag for marking kernel functions as kfuncs. - Add cgroup.memory=nobpf kernel parameter option to disable BPF memory accounting for container environments. Netfilter: - Remove the CLUSTERIP target. It has been marked as obsolete for years, and we still have WARN splats wrt races of the out-of-band /proc interface installed by this target. - Add 'destroy' commands to nf_tables. They are identical to the existing 'delete' commands, but do not return an error if the referenced object (set, chain, rule...) did not exist. Driver API: - Improve cpumask_local_spread() locality to help NICs set the right IRQ affinity on AMD platforms. - Separate C22 and C45 MDIO bus transactions more clearly. - Introduce new DCB table to control DSCP rewrite on egress. - Support configuration of Physical Layer Collision Avoidance (PLCA) Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of shared medium Ethernet. - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing preemption of low priority frames by high priority frames. - Add support for controlling MACSec offload using netlink SET. - Rework devlink instance refcounts to allow registration and de-registration under the instance lock. Split the code into multiple files, drop some of the unnecessarily granular locks and factor out common parts of netlink operation handling. - Add TX frame aggregation parameters (for USB drivers). - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning messages with notifications for debug. - Allow offloading of UDP NEW connections via act_ct. - Add support for per action HW stats in TC. - Support hardware miss to TC action (continue processing in SW from a specific point in the action chain). - Warn if old Wireless Extension user space interface is used with modern cfg80211/mac80211 drivers. Do not support Wireless Extensions for Wi-Fi 7 devices at all. Everyone should switch to using nl80211 interface instead. - Improve the CAN bit timing configuration. Use extack to return error messages directly to user space, update the SJW handling, including the definition of a new default value that will benefit CAN-FD controllers, by increasing their oscillator tolerance. New hardware / drivers: - Ethernet: - nVidia BlueField-3 support (control traffic driver) - Ethernet support for imx93 SoCs - Motorcomm yt8531 gigabit Ethernet PHY - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA) - Microchip LAN8841 PHY (incl. cable diagnostics and PTP) - Amlogic gxl MDIO mux - WiFi: - RealTek RTL8188EU (rtl8xxxu) - Qualcomm Wi-Fi 7 devices (ath12k) - CAN: - Renesas R-Car V4H Drivers: - Bluetooth: - Set Per Platform Antenna Gain (PPAG) for Intel controllers. - Ethernet NICs: - Intel (1G, igc): - support TSN / Qbv / packet scheduling features of i226 model - Intel (100G, ice): - use GNSS subsystem instead of TTY - multi-buffer XDP support - extend support for GPIO pins to E823 devices - nVidia/Mellanox: - update the shared buffer configuration on PFC commands - implement PTP adjphase function for HW offset control - TC support for Geneve and GRE with VF tunnel offload - more efficient crypto key management method - multi-port eswitch support - Netronome/Corigine: - add DCB IEEE support - support IPsec offloading for NFP3800 - Freescale/NXP (enetc): - support XDP_REDIRECT for XDP non-linear buffers - improve reconfig, avoid link flap and waiting for idle - support MAC Merge layer - Other NICs: - sfc/ef100: add basic devlink support for ef100 - ionic: rx_push mode operation (writing descriptors via MMIO) - bnxt: use the auxiliary bus abstraction for RDMA - r8169: disable ASPM and reset bus in case of tx timeout - cpsw: support QSGMII mode for J721e CPSW9G - cpts: support pulse-per-second output - ngbe: add an mdio bus driver - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing - r8152: handle devices with FW with NCM support - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation - virtio-net: support multi buffer XDP - virtio/vsock: replace virtio_vsock_pkt with sk_buff - tsnep: XDP support - Ethernet high-speed switches: - nVidia/Mellanox (mlxsw): - add support for latency TLV (in FW control messages) - Microchip (sparx5): - separate explicit and implicit traffic forwarding rules, make the implicit rules always active - add support for egress DSCP rewrite - IS0 VCAP support (Ingress Classification) - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS etc.) - ES2 VCAP support (Egress Access Control) - support for Per-Stream Filtering and Policing (802.1Q, 8.6.5.1) - Ethernet embedded switches: - Marvell (mv88e6xxx): - add MAB (port auth) offload support - enable PTP receive for mv88e6390 - NXP (ocelot): - support MAC Merge layer - support for the the vsc7512 internal copper phys - Microchip: - lan9303: convert to PHYLINK - lan966x: support TC flower filter statistics - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x - lan937x: support Credit Based Shaper configuration - ksz9477: support Energy Efficient Ethernet - other: - qca8k: convert to regmap read/write API, use bulk operations - rswitch: Improve TX timestamp accuracy - Intel WiFi (iwlwifi): - EHT (Wi-Fi 7) rate reporting - STEP equalizer support: transfer some STEP (connection to radio on platforms with integrated wifi) related parameters from the BIOS to the firmware. - Qualcomm 802.11ax WiFi (ath11k): - IPQ5018 support - Fine Timing Measurement (FTM) responder role support - channel 177 support - MediaTek WiFi (mt76): - per-PHY LED support - mt7996: EHT (Wi-Fi 7) support - Wireless Ethernet Dispatch (WED) reset support - switch to using page pool allocator - RealTek WiFi (rtw89): - support new version of Bluetooth co-existance - Mobile: - rmnet: support TX aggregation" * tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits) page_pool: add a comment explaining the fragment counter usage net: ethtool: fix __ethtool_dev_mm_supported() implementation ethtool: pse-pd: Fix double word in comments xsk: add linux/vmalloc.h to xsk.c sefltests: netdevsim: wait for devlink instance after netns removal selftest: fib_tests: Always cleanup before exit net/mlx5e: Align IPsec ASO result memory to be as required by hardware net/mlx5e: TC, Set CT miss to the specific ct action instance net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG net/mlx5: Refactor tc miss handling to a single function net/mlx5: Kconfig: Make tc offload depend on tc skb extension net/sched: flower: Support hardware miss to tc action net/sched: flower: Move filter handle initialization earlier net/sched: cls_api: Support hardware miss to tc action net/sched: Rename user cookie and act cookie sfc: fix builds without CONFIG_RTC_LIB sfc: clean up some inconsistent indentings net/mlx4_en: Introduce flexible array to silence overflow warning net: lan966x: Fix possible deadlock inside PTP net/ulp: Remove redundant ->clone() test in inet_clone_ulp(). ... --- drivers/virt/fsl_hypervisor.c | 932 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 932 insertions(+) create mode 100644 drivers/virt/fsl_hypervisor.c (limited to 'drivers/virt/fsl_hypervisor.c') diff --git a/drivers/virt/fsl_hypervisor.c b/drivers/virt/fsl_hypervisor.c new file mode 100644 index 000000000..f8b4389d6 --- /dev/null +++ b/drivers/virt/fsl_hypervisor.c @@ -0,0 +1,932 @@ +/* + * Freescale Hypervisor Management Driver + + * Copyright (C) 2008-2011 Freescale Semiconductor, Inc. + * Author: Timur Tabi + * + * This file is licensed under the terms of the GNU General Public License + * version 2. This program is licensed "as is" without any warranty of any + * kind, whether express or implied. + * + * The Freescale hypervisor management driver provides several services to + * drivers and applications related to the Freescale hypervisor: + * + * 1. An ioctl interface for querying and managing partitions. + * + * 2. A file interface to reading incoming doorbells. + * + * 3. An interrupt handler for shutting down the partition upon receiving the + * shutdown doorbell from a manager partition. + * + * 4. A kernel interface for receiving callbacks when a managed partition + * shuts down. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include + +static BLOCKING_NOTIFIER_HEAD(failover_subscribers); + +/* + * Ioctl interface for FSL_HV_IOCTL_PARTITION_RESTART + * + * Restart a running partition + */ +static long ioctl_restart(struct fsl_hv_ioctl_restart __user *p) +{ + struct fsl_hv_ioctl_restart param; + + /* Get the parameters from the user */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_restart))) + return -EFAULT; + + param.ret = fh_partition_restart(param.partition); + + if (copy_to_user(&p->ret, ¶m.ret, sizeof(__u32))) + return -EFAULT; + + return 0; +} + +/* + * Ioctl interface for FSL_HV_IOCTL_PARTITION_STATUS + * + * Query the status of a partition + */ +static long ioctl_status(struct fsl_hv_ioctl_status __user *p) +{ + struct fsl_hv_ioctl_status param; + u32 status; + + /* Get the parameters from the user */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_status))) + return -EFAULT; + + param.ret = fh_partition_get_status(param.partition, &status); + if (!param.ret) + param.status = status; + + if (copy_to_user(p, ¶m, sizeof(struct fsl_hv_ioctl_status))) + return -EFAULT; + + return 0; +} + +/* + * Ioctl interface for FSL_HV_IOCTL_PARTITION_START + * + * Start a stopped partition. + */ +static long ioctl_start(struct fsl_hv_ioctl_start __user *p) +{ + struct fsl_hv_ioctl_start param; + + /* Get the parameters from the user */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_start))) + return -EFAULT; + + param.ret = fh_partition_start(param.partition, param.entry_point, + param.load); + + if (copy_to_user(&p->ret, ¶m.ret, sizeof(__u32))) + return -EFAULT; + + return 0; +} + +/* + * Ioctl interface for FSL_HV_IOCTL_PARTITION_STOP + * + * Stop a running partition + */ +static long ioctl_stop(struct fsl_hv_ioctl_stop __user *p) +{ + struct fsl_hv_ioctl_stop param; + + /* Get the parameters from the user */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_stop))) + return -EFAULT; + + param.ret = fh_partition_stop(param.partition); + + if (copy_to_user(&p->ret, ¶m.ret, sizeof(__u32))) + return -EFAULT; + + return 0; +} + +/* + * Ioctl interface for FSL_HV_IOCTL_MEMCPY + * + * The FH_MEMCPY hypercall takes an array of address/address/size structures + * to represent the data being copied. As a convenience to the user, this + * ioctl takes a user-create buffer and a pointer to a guest physically + * contiguous buffer in the remote partition, and creates the + * address/address/size array for the hypercall. + */ +static long ioctl_memcpy(struct fsl_hv_ioctl_memcpy __user *p) +{ + struct fsl_hv_ioctl_memcpy param; + + struct page **pages = NULL; + void *sg_list_unaligned = NULL; + struct fh_sg_list *sg_list = NULL; + + unsigned int num_pages; + unsigned long lb_offset; /* Offset within a page of the local buffer */ + + unsigned int i; + long ret = 0; + int num_pinned = 0; /* return value from get_user_pages_fast() */ + phys_addr_t remote_paddr; /* The next address in the remote buffer */ + uint32_t count; /* The number of bytes left to copy */ + + /* Get the parameters from the user */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_memcpy))) + return -EFAULT; + + /* + * One partition must be local, the other must be remote. In other + * words, if source and target are both -1, or are both not -1, then + * return an error. + */ + if ((param.source == -1) == (param.target == -1)) + return -EINVAL; + + /* + * The array of pages returned by get_user_pages_fast() covers only + * page-aligned memory. Since the user buffer is probably not + * page-aligned, we need to handle the discrepancy. + * + * We calculate the offset within a page of the S/G list, and make + * adjustments accordingly. This will result in a page list that looks + * like this: + * + * ---- <-- first page starts before the buffer + * | | + * |////|-> ---- + * |////| | | + * ---- | | + * | | + * ---- | | + * |////| | | + * |////| | | + * |////| | | + * ---- | | + * | | + * ---- | | + * |////| | | + * |////| | | + * |////| | | + * ---- | | + * | | + * ---- | | + * |////| | | + * |////|-> ---- + * | | <-- last page ends after the buffer + * ---- + * + * The distance between the start of the first page and the start of the + * buffer is lb_offset. The hashed (///) areas are the parts of the + * page list that contain the actual buffer. + * + * The advantage of this approach is that the number of pages is + * equal to the number of entries in the S/G list that we give to the + * hypervisor. + */ + lb_offset = param.local_vaddr & (PAGE_SIZE - 1); + if (param.count == 0 || + param.count > U64_MAX - lb_offset - PAGE_SIZE + 1) + return -EINVAL; + num_pages = (param.count + lb_offset + PAGE_SIZE - 1) >> PAGE_SHIFT; + + /* Allocate the buffers we need */ + + /* + * 'pages' is an array of struct page pointers that's initialized by + * get_user_pages_fast(). + */ + pages = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL); + if (!pages) { + pr_debug("fsl-hv: could not allocate page list\n"); + return -ENOMEM; + } + + /* + * sg_list is the list of fh_sg_list objects that we pass to the + * hypervisor. + */ + sg_list_unaligned = kmalloc(num_pages * sizeof(struct fh_sg_list) + + sizeof(struct fh_sg_list) - 1, GFP_KERNEL); + if (!sg_list_unaligned) { + pr_debug("fsl-hv: could not allocate S/G list\n"); + ret = -ENOMEM; + goto free_pages; + } + sg_list = PTR_ALIGN(sg_list_unaligned, sizeof(struct fh_sg_list)); + + /* Get the physical addresses of the source buffer */ + num_pinned = get_user_pages_fast(param.local_vaddr - lb_offset, + num_pages, param.source != -1 ? FOLL_WRITE : 0, pages); + + if (num_pinned != num_pages) { + pr_debug("fsl-hv: could not lock source buffer\n"); + ret = (num_pinned < 0) ? num_pinned : -EFAULT; + goto exit; + } + + /* + * Build the fh_sg_list[] array. The first page is special + * because it's misaligned. + */ + if (param.source == -1) { + sg_list[0].source = page_to_phys(pages[0]) + lb_offset; + sg_list[0].target = param.remote_paddr; + } else { + sg_list[0].source = param.remote_paddr; + sg_list[0].target = page_to_phys(pages[0]) + lb_offset; + } + sg_list[0].size = min_t(uint64_t, param.count, PAGE_SIZE - lb_offset); + + remote_paddr = param.remote_paddr + sg_list[0].size; + count = param.count - sg_list[0].size; + + for (i = 1; i < num_pages; i++) { + if (param.source == -1) { + /* local to remote */ + sg_list[i].source = page_to_phys(pages[i]); + sg_list[i].target = remote_paddr; + } else { + /* remote to local */ + sg_list[i].source = remote_paddr; + sg_list[i].target = page_to_phys(pages[i]); + } + sg_list[i].size = min_t(uint64_t, count, PAGE_SIZE); + + remote_paddr += sg_list[i].size; + count -= sg_list[i].size; + } + + param.ret = fh_partition_memcpy(param.source, param.target, + virt_to_phys(sg_list), num_pages); + +exit: + if (pages && (num_pinned > 0)) { + for (i = 0; i < num_pinned; i++) + put_page(pages[i]); + } + + kfree(sg_list_unaligned); +free_pages: + kfree(pages); + + if (!ret) + if (copy_to_user(&p->ret, ¶m.ret, sizeof(__u32))) + return -EFAULT; + + return ret; +} + +/* + * Ioctl interface for FSL_HV_IOCTL_DOORBELL + * + * Ring a doorbell + */ +static long ioctl_doorbell(struct fsl_hv_ioctl_doorbell __user *p) +{ + struct fsl_hv_ioctl_doorbell param; + + /* Get the parameters from the user. */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_doorbell))) + return -EFAULT; + + param.ret = ev_doorbell_send(param.doorbell); + + if (copy_to_user(&p->ret, ¶m.ret, sizeof(__u32))) + return -EFAULT; + + return 0; +} + +static long ioctl_dtprop(struct fsl_hv_ioctl_prop __user *p, int set) +{ + struct fsl_hv_ioctl_prop param; + char __user *upath, *upropname; + void __user *upropval; + char *path, *propname; + void *propval; + int ret = 0; + + /* Get the parameters from the user. */ + if (copy_from_user(¶m, p, sizeof(struct fsl_hv_ioctl_prop))) + return -EFAULT; + + upath = (char __user *)(uintptr_t)param.path; + upropname = (char __user *)(uintptr_t)param.propname; + upropval = (void __user *)(uintptr_t)param.propval; + + path = strndup_user(upath, FH_DTPROP_MAX_PATHLEN); + if (IS_ERR(path)) + return PTR_ERR(path); + + propname = strndup_user(upropname, FH_DTPROP_MAX_PATHLEN); + if (IS_ERR(propname)) { + ret = PTR_ERR(propname); + goto err_free_path; + } + + if (param.proplen > FH_DTPROP_MAX_PROPLEN) { + ret = -EINVAL; + goto err_free_propname; + } + + propval = kmalloc(param.proplen, GFP_KERNEL); + if (!propval) { + ret = -ENOMEM; + goto err_free_propname; + } + + if (set) { + if (copy_from_user(propval, upropval, param.proplen)) { + ret = -EFAULT; + goto err_free_propval; + } + + param.ret = fh_partition_set_dtprop(param.handle, + virt_to_phys(path), + virt_to_phys(propname), + virt_to_phys(propval), + param.proplen); + } else { + param.ret = fh_partition_get_dtprop(param.handle, + virt_to_phys(path), + virt_to_phys(propname), + virt_to_phys(propval), + ¶m.proplen); + + if (param.ret == 0) { + if (copy_to_user(upropval, propval, param.proplen) || + put_user(param.proplen, &p->proplen)) { + ret = -EFAULT; + goto err_free_propval; + } + } + } + + if (put_user(param.ret, &p->ret)) + ret = -EFAULT; + +err_free_propval: + kfree(propval); +err_free_propname: + kfree(propname); +err_free_path: + kfree(path); + + return ret; +} + +/* + * Ioctl main entry point + */ +static long fsl_hv_ioctl(struct file *file, unsigned int cmd, + unsigned long argaddr) +{ + void __user *arg = (void __user *)argaddr; + long ret; + + switch (cmd) { + case FSL_HV_IOCTL_PARTITION_RESTART: + ret = ioctl_restart(arg); + break; + case FSL_HV_IOCTL_PARTITION_GET_STATUS: + ret = ioctl_status(arg); + break; + case FSL_HV_IOCTL_PARTITION_START: + ret = ioctl_start(arg); + break; + case FSL_HV_IOCTL_PARTITION_STOP: + ret = ioctl_stop(arg); + break; + case FSL_HV_IOCTL_MEMCPY: + ret = ioctl_memcpy(arg); + break; + case FSL_HV_IOCTL_DOORBELL: + ret = ioctl_doorbell(arg); + break; + case FSL_HV_IOCTL_GETPROP: + ret = ioctl_dtprop(arg, 0); + break; + case FSL_HV_IOCTL_SETPROP: + ret = ioctl_dtprop(arg, 1); + break; + default: + pr_debug("fsl-hv: bad ioctl dir=%u type=%u cmd=%u size=%u\n", + _IOC_DIR(cmd), _IOC_TYPE(cmd), _IOC_NR(cmd), + _IOC_SIZE(cmd)); + return -ENOTTY; + } + + return ret; +} + +/* Linked list of processes that have us open */ +static struct list_head db_list; + +/* spinlock for db_list */ +static DEFINE_SPINLOCK(db_list_lock); + +/* The size of the doorbell event queue. This must be a power of two. */ +#define QSIZE 16 + +/* Returns the next head/tail pointer, wrapping around the queue if necessary */ +#define nextp(x) (((x) + 1) & (QSIZE - 1)) + +/* Per-open data structure */ +struct doorbell_queue { + struct list_head list; + spinlock_t lock; + wait_queue_head_t wait; + unsigned int head; + unsigned int tail; + uint32_t q[QSIZE]; +}; + +/* Linked list of ISRs that we registered */ +struct list_head isr_list; + +/* Per-ISR data structure */ +struct doorbell_isr { + struct list_head list; + unsigned int irq; + uint32_t doorbell; /* The doorbell handle */ + uint32_t partition; /* The partition handle, if used */ +}; + +/* + * Add a doorbell to all of the doorbell queues + */ +static void fsl_hv_queue_doorbell(uint32_t doorbell) +{ + struct doorbell_queue *dbq; + unsigned long flags; + + /* Prevent another core from modifying db_list */ + spin_lock_irqsave(&db_list_lock, flags); + + list_for_each_entry(dbq, &db_list, list) { + if (dbq->head != nextp(dbq->tail)) { + dbq->q[dbq->tail] = doorbell; + /* + * This memory barrier eliminates the need to grab + * the spinlock for dbq. + */ + smp_wmb(); + dbq->tail = nextp(dbq->tail); + wake_up_interruptible(&dbq->wait); + } + } + + spin_unlock_irqrestore(&db_list_lock, flags); +} + +/* + * Interrupt handler for all doorbells + * + * We use the same interrupt handler for all doorbells. Whenever a doorbell + * is rung, and we receive an interrupt, we just put the handle for that + * doorbell (passed to us as *data) into all of the queues. + */ +static irqreturn_t fsl_hv_isr(int irq, void *data) +{ + fsl_hv_queue_doorbell((uintptr_t) data); + + return IRQ_HANDLED; +} + +/* + * State change thread function + * + * The state change notification arrives in an interrupt, but we can't call + * blocking_notifier_call_chain() in an interrupt handler. We could call + * atomic_notifier_call_chain(), but that would require the clients' call-back + * function to run in interrupt context. Since we don't want to impose that + * restriction on the clients, we use a threaded IRQ to process the + * notification in kernel context. + */ +static irqreturn_t fsl_hv_state_change_thread(int irq, void *data) +{ + struct doorbell_isr *dbisr = data; + + blocking_notifier_call_chain(&failover_subscribers, dbisr->partition, + NULL); + + return IRQ_HANDLED; +} + +/* + * Interrupt handler for state-change doorbells + */ +static irqreturn_t fsl_hv_state_change_isr(int irq, void *data) +{ + unsigned int status; + struct doorbell_isr *dbisr = data; + int ret; + + /* It's still a doorbell, so add it to all the queues. */ + fsl_hv_queue_doorbell(dbisr->doorbell); + + /* Determine the new state, and if it's stopped, notify the clients. */ + ret = fh_partition_get_status(dbisr->partition, &status); + if (!ret && (status == FH_PARTITION_STOPPED)) + return IRQ_WAKE_THREAD; + + return IRQ_HANDLED; +} + +/* + * Returns a bitmask indicating whether a read will block + */ +static __poll_t fsl_hv_poll(struct file *filp, struct poll_table_struct *p) +{ + struct doorbell_queue *dbq = filp->private_data; + unsigned long flags; + __poll_t mask; + + spin_lock_irqsave(&dbq->lock, flags); + + poll_wait(filp, &dbq->wait, p); + mask = (dbq->head == dbq->tail) ? 0 : (EPOLLIN | EPOLLRDNORM); + + spin_unlock_irqrestore(&dbq->lock, flags); + + return mask; +} + +/* + * Return the handles for any incoming doorbells + * + * If there are doorbell handles in the queue for this open instance, then + * return them to the caller as an array of 32-bit integers. Otherwise, + * block until there is at least one handle to return. + */ +static ssize_t fsl_hv_read(struct file *filp, char __user *buf, size_t len, + loff_t *off) +{ + struct doorbell_queue *dbq = filp->private_data; + uint32_t __user *p = (uint32_t __user *) buf; /* for put_user() */ + unsigned long flags; + ssize_t count = 0; + + /* Make sure we stop when the user buffer is full. */ + while (len >= sizeof(uint32_t)) { + uint32_t dbell; /* Local copy of doorbell queue data */ + + spin_lock_irqsave(&dbq->lock, flags); + + /* + * If the queue is empty, then either we're done or we need + * to block. If the application specified O_NONBLOCK, then + * we return the appropriate error code. + */ + if (dbq->head == dbq->tail) { + spin_unlock_irqrestore(&dbq->lock, flags); + if (count) + break; + if (filp->f_flags & O_NONBLOCK) + return -EAGAIN; + if (wait_event_interruptible(dbq->wait, + dbq->head != dbq->tail)) + return -ERESTARTSYS; + continue; + } + + /* + * Even though we have an smp_wmb() in the ISR, the core + * might speculatively execute the "dbell = ..." below while + * it's evaluating the if-statement above. In that case, the + * value put into dbell could be stale if the core accepts the + * speculation. To prevent that, we need a read memory barrier + * here as well. + */ + smp_rmb(); + + /* Copy the data to a temporary local buffer, because + * we can't call copy_to_user() from inside a spinlock + */ + dbell = dbq->q[dbq->head]; + dbq->head = nextp(dbq->head); + + spin_unlock_irqrestore(&dbq->lock, flags); + + if (put_user(dbell, p)) + return -EFAULT; + p++; + count += sizeof(uint32_t); + len -= sizeof(uint32_t); + } + + return count; +} + +/* + * Open the driver and prepare for reading doorbells. + * + * Every time an application opens the driver, we create a doorbell queue + * for that file handle. This queue is used for any incoming doorbells. + */ +static int fsl_hv_open(struct inode *inode, struct file *filp) +{ + struct doorbell_queue *dbq; + unsigned long flags; + + dbq = kzalloc(sizeof(struct doorbell_queue), GFP_KERNEL); + if (!dbq) { + pr_err("fsl-hv: out of memory\n"); + return -ENOMEM; + } + + spin_lock_init(&dbq->lock); + init_waitqueue_head(&dbq->wait); + + spin_lock_irqsave(&db_list_lock, flags); + list_add(&dbq->list, &db_list); + spin_unlock_irqrestore(&db_list_lock, flags); + + filp->private_data = dbq; + + return 0; +} + +/* + * Close the driver + */ +static int fsl_hv_close(struct inode *inode, struct file *filp) +{ + struct doorbell_queue *dbq = filp->private_data; + unsigned long flags; + + spin_lock_irqsave(&db_list_lock, flags); + list_del(&dbq->list); + spin_unlock_irqrestore(&db_list_lock, flags); + + kfree(dbq); + + return 0; +} + +static const struct file_operations fsl_hv_fops = { + .owner = THIS_MODULE, + .open = fsl_hv_open, + .release = fsl_hv_close, + .poll = fsl_hv_poll, + .read = fsl_hv_read, + .unlocked_ioctl = fsl_hv_ioctl, + .compat_ioctl = compat_ptr_ioctl, +}; + +static struct miscdevice fsl_hv_misc_dev = { + MISC_DYNAMIC_MINOR, + "fsl-hv", + &fsl_hv_fops +}; + +static irqreturn_t fsl_hv_shutdown_isr(int irq, void *data) +{ + orderly_poweroff(false); + + return IRQ_HANDLED; +} + +/* + * Returns the handle of the parent of the given node + * + * The handle is the value of the 'hv-handle' property + */ +static int get_parent_handle(struct device_node *np) +{ + struct device_node *parent; + const uint32_t *prop; + uint32_t handle; + int len; + + parent = of_get_parent(np); + if (!parent) + /* It's not really possible for this to fail */ + return -ENODEV; + + /* + * The proper name for the handle property is "hv-handle", but some + * older versions of the hypervisor used "reg". + */ + prop = of_get_property(parent, "hv-handle", &len); + if (!prop) + prop = of_get_property(parent, "reg", &len); + + if (!prop || (len != sizeof(uint32_t))) { + /* This can happen only if the node is malformed */ + of_node_put(parent); + return -ENODEV; + } + + handle = be32_to_cpup(prop); + of_node_put(parent); + + return handle; +} + +/* + * Register a callback for failover events + * + * This function is called by device drivers to register their callback + * functions for fail-over events. + */ +int fsl_hv_failover_register(struct notifier_block *nb) +{ + return blocking_notifier_chain_register(&failover_subscribers, nb); +} +EXPORT_SYMBOL(fsl_hv_failover_register); + +/* + * Unregister a callback for failover events + */ +int fsl_hv_failover_unregister(struct notifier_block *nb) +{ + return blocking_notifier_chain_unregister(&failover_subscribers, nb); +} +EXPORT_SYMBOL(fsl_hv_failover_unregister); + +/* + * Return TRUE if we're running under FSL hypervisor + * + * This function checks to see if we're running under the Freescale + * hypervisor, and returns zero if we're not, or non-zero if we are. + * + * First, it checks if MSR[GS]==1, which means we're running under some + * hypervisor. Then it checks if there is a hypervisor node in the device + * tree. Currently, that means there needs to be a node in the root called + * "hypervisor" and which has a property named "fsl,hv-version". + */ +static int has_fsl_hypervisor(void) +{ + struct device_node *node; + int ret; + + node = of_find_node_by_path("/hypervisor"); + if (!node) + return 0; + + ret = of_find_property(node, "fsl,hv-version", NULL) != NULL; + + of_node_put(node); + + return ret; +} + +/* + * Freescale hypervisor management driver init + * + * This function is called when this module is loaded. + * + * Register ourselves as a miscellaneous driver. This will register the + * fops structure and create the right sysfs entries for udev. + */ +static int __init fsl_hypervisor_init(void) +{ + struct device_node *np; + struct doorbell_isr *dbisr, *n; + int ret; + + pr_info("Freescale hypervisor management driver\n"); + + if (!has_fsl_hypervisor()) { + pr_info("fsl-hv: no hypervisor found\n"); + return -ENODEV; + } + + ret = misc_register(&fsl_hv_misc_dev); + if (ret) { + pr_err("fsl-hv: cannot register device\n"); + return ret; + } + + INIT_LIST_HEAD(&db_list); + INIT_LIST_HEAD(&isr_list); + + for_each_compatible_node(np, NULL, "epapr,hv-receive-doorbell") { + unsigned int irq; + const uint32_t *handle; + + handle = of_get_property(np, "interrupts", NULL); + irq = irq_of_parse_and_map(np, 0); + if (!handle || !irq) { + pr_err("fsl-hv: no 'interrupts' property in %pOF node\n", + np); + continue; + } + + dbisr = kzalloc(sizeof(*dbisr), GFP_KERNEL); + if (!dbisr) + goto out_of_memory; + + dbisr->irq = irq; + dbisr->doorbell = be32_to_cpup(handle); + + if (of_device_is_compatible(np, "fsl,hv-shutdown-doorbell")) { + /* The shutdown doorbell gets its own ISR */ + ret = request_irq(irq, fsl_hv_shutdown_isr, 0, + np->name, NULL); + } else if (of_device_is_compatible(np, + "fsl,hv-state-change-doorbell")) { + /* + * The state change doorbell triggers a notification if + * the state of the managed partition changes to + * "stopped". We need a separate interrupt handler for + * that, and we also need to know the handle of the + * target partition, not just the handle of the + * doorbell. + */ + dbisr->partition = ret = get_parent_handle(np); + if (ret < 0) { + pr_err("fsl-hv: node %pOF has missing or " + "malformed parent\n", np); + kfree(dbisr); + continue; + } + ret = request_threaded_irq(irq, fsl_hv_state_change_isr, + fsl_hv_state_change_thread, + 0, np->name, dbisr); + } else + ret = request_irq(irq, fsl_hv_isr, 0, np->name, dbisr); + + if (ret < 0) { + pr_err("fsl-hv: could not request irq %u for node %pOF\n", + irq, np); + kfree(dbisr); + continue; + } + + list_add(&dbisr->list, &isr_list); + + pr_info("fsl-hv: registered handler for doorbell %u\n", + dbisr->doorbell); + } + + return 0; + +out_of_memory: + list_for_each_entry_safe(dbisr, n, &isr_list, list) { + free_irq(dbisr->irq, dbisr); + list_del(&dbisr->list); + kfree(dbisr); + } + + misc_deregister(&fsl_hv_misc_dev); + + return -ENOMEM; +} + +/* + * Freescale hypervisor management driver termination + * + * This function is called when this driver is unloaded. + */ +static void __exit fsl_hypervisor_exit(void) +{ + struct doorbell_isr *dbisr, *n; + + list_for_each_entry_safe(dbisr, n, &isr_list, list) { + free_irq(dbisr->irq, dbisr); + list_del(&dbisr->list); + kfree(dbisr); + } + + misc_deregister(&fsl_hv_misc_dev); +} + +module_init(fsl_hypervisor_init); +module_exit(fsl_hypervisor_exit); + +MODULE_AUTHOR("Timur Tabi "); +MODULE_DESCRIPTION("Freescale hypervisor management driver"); +MODULE_LICENSE("GPL v2"); -- cgit v1.2.3