sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

alb_break_llc() decides whether to break LLC preference during active
load balance. It does so by testing that every runnable fair task on the
source rq prefers its LLC:

	env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable

But the two counters cover different sets. nr_pref_llc_running is updated
in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued,
so it follows queued tasks. h_nr_runnable is updated in set_delayed()/
clear_delayed() and drops delay-dequeued tasks.

So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted
in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks,
alb_break_llc() returns false, and active balance is free to pull a task
off its preferred LLC. Active balance only moves runnable tasks, and this
is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB
skips the per-task test in can_migrate_task(). The runnable set is the one
we want.

Fix it on the counter side. A task should be counted in
nr_pref_llc_running exactly while it is both queued on its preferred LLC
(pref_llc_queued) and runnable (!sched_delayed). Define that membership
once in task_pref_llc_runnable(), and adjust the counter only through
pref_llc_running_inc()/pref_llc_running_dec() from the four sites that
change either input: account_llc_enqueue(), account_llc_dequeue(),
set_delayed() and clear_delayed(). Gating every update on the same
predicate keeps the delay, wake and dequeue paths from double-counting
or underflowing; see the comments at those sites for the ordering.

nr_llc_running and sd->llc_counts are not touched and stay on queued
semantics.

Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC")
Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/
Reported-by: Zhan Xusheng <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 286 insertions, 0 deletions

diff --git a/Documentation/userspace-api/vduse.rst b/Documentation/userspace-api/vduse.rst
new file mode 100644
index 000000000..81479d47c
--- /dev/null
+++ b/Documentation/userspace-api/vduse.rst
@@ -0,0 +1,286 @@
+==================================
+VDUSE - "vDPA Device in Userspace"
+==================================
+
+vDPA (virtio data path acceleration) device is a device that uses a
+datapath which complies with the virtio specifications with vendor
+specific control path. vDPA devices can be both physically located on
+the hardware or emulated by software. VDUSE is a framework that makes it
+possible to implement software-emulated vDPA devices in userspace. And
+to make the device emulation more secure, the emulated vDPA device's
+control path is handled in the kernel and only the data path is
+implemented in the userspace.
+
+Note that only virtio block device is supported by VDUSE framework now,
+which can reduce security risks when the userspace process that implements
+the data path is run by an unprivileged user. The support for other device
+types can be added after the security issue of corresponding device driver
+is clarified or fixed in the future.
+
+Create/Destroy VDUSE devices
+----------------------------
+
+VDUSE devices are created as follows:
+
+1. Create a new VDUSE instance with ioctl(VDUSE_CREATE_DEV) on
+   /dev/vduse/control.
+
+2. Setup each virtqueue with ioctl(VDUSE_VQ_SETUP) on /dev/vduse/$NAME.
+
+3. Begin processing VDUSE messages from /dev/vduse/$NAME. The first
+   messages will arrive while attaching the VDUSE instance to vDPA bus.
+
+4. Send the VDPA_CMD_DEV_NEW netlink message to attach the VDUSE
+   instance to vDPA bus.
+
+VDUSE devices are destroyed as follows:
+
+1. Send the VDPA_CMD_DEV_DEL netlink message to detach the VDUSE
+   instance from vDPA bus.
+
+2. Close the file descriptor referring to /dev/vduse/$NAME.
+
+3. Destroy the VDUSE instance with ioctl(VDUSE_DESTROY_DEV) on
+   /dev/vduse/control.
+
+The netlink messages can be sent via vdpa tool in iproute2 or use the
+below sample codes:
+
+.. code-block:: c
+
+	static int netlink_add_vduse(const char *name, enum vdpa_command cmd)
+	{
+		struct nl_sock *nlsock;
+		struct nl_msg *msg;
+		int famid;
+
+		nlsock = nl_socket_alloc();
+		if (!nlsock)
+			return -ENOMEM;
+
+		if (genl_connect(nlsock))
+			goto free_sock;
+
+		famid = genl_ctrl_resolve(nlsock, VDPA_GENL_NAME);
+		if (famid < 0)
+			goto close_sock;
+
+		msg = nlmsg_alloc();
+		if (!msg)
+			goto close_sock;
+
+		if (!genlmsg_put(msg, NL_AUTO_PORT, NL_AUTO_SEQ, famid, 0, 0, cmd, 0))
+			goto nla_put_failure;
+
+		NLA_PUT_STRING(msg, VDPA_ATTR_DEV_NAME, name);
+		if (cmd == VDPA_CMD_DEV_NEW)
+			NLA_PUT_STRING(msg, VDPA_ATTR_MGMTDEV_DEV_NAME, "vduse");
+
+		if (nl_send_sync(nlsock, msg))
+			goto close_sock;
+
+		nl_close(nlsock);
+		nl_socket_free(nlsock);
+
+		return 0;
+	nla_put_failure:
+		nlmsg_free(msg);
+	close_sock:
+		nl_close(nlsock);
+	free_sock:
+		nl_socket_free(nlsock);
+		return -1;
+	}
+
+How VDUSE works
+---------------
+
+As mentioned above, a VDUSE device is created by ioctl(VDUSE_CREATE_DEV) on
+/dev/vduse/control. With this ioctl, userspace can specify some basic configuration
+such as device name (uniquely identify a VDUSE device), virtio features, virtio
+configuration space, the number of virtqueues and so on for this emulated device.
+Then a char device interface (/dev/vduse/$NAME) is exported to userspace for device
+emulation. Userspace can use the VDUSE_VQ_SETUP ioctl on /dev/vduse/$NAME to
+add per-virtqueue configuration such as the max size of virtqueue to the device.
+
+After the initialization, the VDUSE device can be attached to vDPA bus via
+the VDPA_CMD_DEV_NEW netlink message. Userspace needs to read()/write() on
+/dev/vduse/$NAME to receive/reply some control messages from/to VDUSE kernel
+module as follows:
+
+.. code-block:: c
+
+	static int vduse_message_handler(int dev_fd)
+	{
+		int len;
+		struct vduse_dev_request req;
+		struct vduse_dev_response resp;
+
+		len = read(dev_fd, &req, sizeof(req));
+		if (len != sizeof(req))
+			return -1;
+
+		resp.request_id = req.request_id;
+
+		switch (req.type) {
+
+		/* handle different types of messages */
+
+		}
+
+		len = write(dev_fd, &resp, sizeof(resp));
+		if (len != sizeof(resp))
+			return -1;
+
+		return 0;
+	}
+
+There are now three types of messages introduced by VDUSE framework:
+
+- VDUSE_GET_VQ_STATE: Get the state for virtqueue, userspace should return
+  avail index for split virtqueue or the device/driver ring wrap counters and
+  the avail and used index for packed virtqueue.
+
+- VDUSE_SET_STATUS: Set the device status, userspace should follow
+  the virtio spec: https://docs.oasis-open.org/virtio/virtio/v1.1/virtio-v1.1.html
+  to process this message. For example, fail to set the FEATURES_OK device
+  status bit if the device can not accept the negotiated virtio features
+  get from the VDUSE_DEV_GET_FEATURES ioctl.
+
+- VDUSE_UPDATE_IOTLB: Notify userspace to update the memory mapping for specified
+  IOVA range, userspace should firstly remove the old mapping, then setup the new
+  mapping via the VDUSE_IOTLB_GET_FD ioctl.
+
+After DRIVER_OK status bit is set via the VDUSE_SET_STATUS message, userspace is
+able to start the dataplane processing as follows:
+
+1. Get the specified virtqueue's information with the VDUSE_VQ_GET_INFO ioctl,
+   including the size, the IOVAs of descriptor table, available ring and used ring,
+   the state and the ready status.
+
+2. Pass the above IOVAs to the VDUSE_IOTLB_GET_FD ioctl so that those IOVA regions
+   can be mapped into userspace. Some sample codes is shown below:
+
+.. code-block:: c
+
+	static int perm_to_prot(uint8_t perm)
+	{
+		int prot = 0;
+
+		switch (perm) {
+		case VDUSE_ACCESS_WO:
+			prot |= PROT_WRITE;
+			break;
+		case VDUSE_ACCESS_RO:
+			prot |= PROT_READ;
+			break;
+		case VDUSE_ACCESS_RW:
+			prot |= PROT_READ | PROT_WRITE;
+			break;
+		}
+
+		return prot;
+	}
+
+	static void *iova_to_va(int dev_fd, uint64_t iova, uint64_t *len)
+	{
+		int fd;
+		void *addr;
+		size_t size;
+		struct vduse_iotlb_entry entry;
+
+		entry.start = iova;
+		entry.last = iova;
+
+		/*
+		 * Find the first IOVA region that overlaps with the specified
+		 * range [start, last] and return the corresponding file descriptor.
+		 */
+		fd = ioctl(dev_fd, VDUSE_IOTLB_GET_FD, &entry);
+		if (fd < 0)
+			return NULL;
+
+		size = entry.last - entry.start + 1;
+		*len = entry.last - iova + 1;
+		addr = mmap(0, size, perm_to_prot(entry.perm), MAP_SHARED,
+			    fd, entry.offset);
+		close(fd);
+		if (addr == MAP_FAILED)
+			return NULL;
+
+		/*
+		 * Using some data structures such as linked list to store
+		 * the iotlb mapping. The munmap(2) should be called for the
+		 * cached mapping when the corresponding VDUSE_UPDATE_IOTLB
+		 * message is received or the device is reset.
+		 */
+
+		return addr + iova - entry.start;
+	}
+
+3. Setup the kick eventfd for the specified virtqueues with the VDUSE_VQ_SETUP_KICKFD
+   ioctl. The kick eventfd is used by VDUSE kernel module to notify userspace to
+   consume the available ring. This is optional since userspace can choose to poll the
+   available ring instead.
+
+4. Listen to the kick eventfd (optional) and consume the available ring. The buffer
+   described by the descriptors in the descriptor table should be also mapped into
+   userspace via the VDUSE_IOTLB_GET_FD ioctl before accessing.
+
+5. Inject an interrupt for specific virtqueue with the VDUSE_INJECT_VQ_IRQ ioctl
+   after the used ring is filled.
+
+Enabling ASID (API version 1)
+------------------------------
+
+VDUSE supports per-address-space identifiers (ASIDs) starting with API
+version 1. Set it up with ioctl(VDUSE_SET_API_VERSION) on `/dev/vduse/control`
+and pass `VDUSE_API_VERSION_1` before creating a new VDUSE instance with
+ioctl(VDUSE_CREATE_DEV).
+
+Afterwards, you can use the member asid of ioctl(VDUSE_VQ_SETUP) argument to
+select the address space of the IOTLB you are querying.  The driver could
+change the address space of any virtqueue group by using the
+VDUSE_SET_VQ_GROUP_ASID VDUSE message type, and the VDUSE instance needs to
+reply with VDUSE_REQ_RESULT_OK if it was possible to change it.
+
+Similarly, you can use ioctl(VDUSE_IOTLB_GET_FD2) to obtain the file descriptor
+describing an IOVA region of a specific ASID. Example usage:
+
+.. code-block:: c
+
+	static void *iova_to_va(int dev_fd, uint32_t asid, uint64_t iova,
+	                        uint64_t *len)
+	{
+		int fd;
+		void *addr;
+		size_t size;
+		struct vduse_iotlb_entry_v2 entry = { 0 };
+
+		entry.v1.start = iova;
+		entry.v1.last = iova;
+		entry.asid = asid;
+
+		fd = ioctl(dev_fd, VDUSE_IOTLB_GET_FD2, &entry);
+		if (fd < 0)
+			return NULL;
+
+		size = entry.v1.last - entry.v1.start + 1;
+		*len = entry.v1.last - iova + 1;
+		addr = mmap(0, size, perm_to_prot(entry.v1.perm), MAP_SHARED,
+			    fd, entry.v1.offset);
+		close(fd);
+		if (addr == MAP_FAILED)
+			return NULL;
+
+		/*
+		 * Using some data structures such as linked list to store
+		 * the iotlb mapping. The munmap(2) should be called for the
+		 * cached mapping when the corresponding VDUSE_UPDATE_IOTLB
+		 * message is received or the device is reset.
+		 */
+
+		return addr + iova - entry.v1.start;
+	}
+
+For more details on the uAPI, please see include/uapi/linux/vduse.h.