Merge tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-tracegrafted

Pull tracing fixes from Steven Rostedt:

 - Don't destroy user event fields when removal fails

   User event fields are destroyed before the event is removed from
   visibility. But that can fail leaving the still visible event with no
   fields. Move the destroying of the fields to after the event is
   successfully removed from visibility.

 - Initialize function graph state is fork before calling
   copy_exec_state()

   For non-CLONE_VM forks, copy_exec_state() allocates a new
   task_exec_state. If that allocation fails, ftrace_graph_exit_task()
   will free the tasks ret_stack pointer. Since that pointer is still
   using the parent's ret_stack, it mistakenly frees the parent's
   pointer too.

   Call ftrace_graph_init() on the task first which will NULL out the
   new tasks's ret_stack and if the copy fails, it will not free
   anything.

 - Remove FGRAPH_MAX_INDEX

   The macro FGRAPH_MAX_INDEX was added but never used. Remove it.

 - Save ent_size in function graph printing of nested functions

   The function graph tracer needs to look at the next event to see if
   the next event is the return of the current function entry. If it is,
   it prints a single line:

	ktime_get();

   Otherwise it prints it like a nested function:

	tick_nohz_irq_exit() {
	    ktime_get();
	    kcpustat_irq_exit();
	}

   In order to look at the next event, it must save the current event so
   that it has the information to print from it. It saves the event in
   the iterator descriptor called "ent". What it doesn't save is the
   ent_size of the event which is now used to know if the function graph
   arguments are to be printed. The peek doesn't save the size so the
   size used happens to be that of the size of the last event that was
   seen.

   Save the entry event size in the iterator descriptor so that the
   correct size is used.

 - Fix several errors with freeing data in the histogram code

   The histogram code had a lot of leaked or or incorrect accounting
   when failures happen. Correct them.

 - Fix histogram regression of .percent and .graph modifiers

   Up until 6.3 histogram values could have "percent" or "graph"
   modifiers that changed how they were printed. But a change that added
   restricting histograms values from being strings, stack traces and
   other modifiers inadvertently prevented them from using the percent
   and graph modifiers, which were legal use cases for values.

   Put back the percent and graph modifiers.

 - Fix various typos in the comments

 - Set the trace_clock before initializing a histogram with clock
   argument

   The histogram API allows the user to specific which trace clock to
   use via a "clock=" string. The histogram is set up first before the
   clock is checked. If the passed in clock is not valid, it exits
   without fully fixing up the histogram leaving it on the list and a
   use-after-free can trigger.

   Update the clock argument first and if it fails then exit gracefully
   before the histogram trigger is placed on any lists.

 - Restore :mod: trailer after parsing in ftrace_set_clr_event

   The function ftrace_set_clr_event() modifies the parse string and
   needs to put it back to what was passed in. It searches for ":mod:"
   via a strsep() but fails to put back the first ':' in the string.

   Add back the ':' in the passed in string.

 - Take trace_array reference when opening a tracer options file

   The options files are dynamically created and some tracers add their
   own options. When a tracer adds their own list of options, the
   trace_array holding them has an array to hold the list of options for
   each tracer. This array increases in size via a krealloc(), and the
   new entry gets a newly allocated array to hold the options of the new
   tracer being added.

   The element in each entry of the tracer's option array holds a
   pointer back to the trace_array, a pointer to the tracer it is
   associated to, a pointer to the flags of the option.

   The issue is that these arrays are freed when the trace_array is
   freed when its instance it represents is removed from the instances
   directory. There's a race that an open of one of these options files
   can happen when the instance is being removed.

   Add a new helper function to be called by the open function of the
   options file to iterate all existing trace_arrays under a lock and
   find the one that has the given option element in one of it's tracer
   arrays. If found, then update the associated trace_array's reference
   counter to keep it from being freed. If not found, have the open call
   return -ENODEV.

 - Disable interrupts when acquiring the lock in rb_wake_up_waiters()

   The function rb_wake_up_waiters() assumes it will be called in
   interrupt context and does not disable irqs when taking
   cpu_buffer->reader_lock, which can be called in hard interrupt
   context. The issue is in PREEMPT_RT, this function is called in
   thread context leaving this lock open to a deadlock.

   Take the lock with interrupts disabled.

 - Use rcu_assign_pointer() for tmp_ops filter hash

   The tmp_ops used in update_ftrace_direct_mod() assigns its
   filter_hash field directly, but that field is annotated as __rcu and
   sparse complains. Assign it with rcu_assign_pointer()

 - Fix use-after-free in enable_trigger_private_data_free()

   The trace_event_call is accessed through the event_trigger_data's
   trace_event_file pointer to put the trace_event_call on freeing. The
   issue is that the trace_event_file data may have been freed already
   causing a use-after-free. Add a field to the event_trigger_data that
   points directly to the trace_event_call so that it can decrement its
   reference directly without needing to go through the
   trace_event_file.

 - Fix accounting of buffer data remote headers

   trace_buffer_desc_size() and trace_remote_alloc_buffer() undercount
   the number of pages is needed for the asked for size as it doesn't
   take into account the meta data on each page. Add a helper function
   to do the calculation properly and use that in these functions.

 - Catch nr_page_va overflow in ring_buffer_desc sizing

   The number of pages per remote ring buffer is capped by
   ring_buffer_desc::nr_page_va (32 bits). A buffer_size large enough to
   overflow that field would silently allocate a descriptor smaller than
   what was asked for.

 - Do not resize the subbuf order if any per_cpu buffer is disabled

   The mmapping of ring buffers disables resizing the subbuffers, but it
   is done per-cpu whereas the subbuf size change is done for all the
   per_cpu buffers under the buffer->mutex. It could change the size of
   some while the mapping is happening on others. Have the resize of the
   subbuf order check all the per_cpu buffers under the lock to see if
   any of them is disabled before starting and causing an inconsistency
   between buffers that are being mapped.

* tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (25 commits)
  ring-buffer: Check resize_disabled before publishing the new subbuf order
  tracing/remotes: Catch nr_page_va overflow in ring_buffer_desc sizing
  tracing/remotes: Account for ring buffer page header in size calculation
  tracing: Don't dereference trace_event_file in deferred trigger free
  ftrace: Use rcu_assign_pointer() for tmp_ops filter hash
  ring-buffer: Acquire the lock with irqsave in rb_wake_up_waiters()
  tracing: Take trace_array reference when opening a tracer options file
  tracing: Fix ring_buffer_read_page_size() kernel-doc
  tracing: Restore :mod: trailer after parsing in ftrace_set_clr_event()
  tracing: Fix memory corruption from a "STACKTRACE" histogram key
  tracing: Fix memory corruption from the histogram stacktrace modifier
  tracing: Undo the registration when enabling the histogram trigger fails
  tracing: Take the reference before publishing the named histogram trigger
  tracing: Set the trace clock before registering the histogram trigger
  tracing: Fix typo "preceeded" in comment
  tracing: Fix typo "availabe" in comment
  tracing: Let histogram values keep the percent and graph modifiers
  tracing: Keep the entry count when the histogram stats allocation fails
  tracing: Free histogram the field rejected for a bad modifier
  tracing: Free histogram the var ref when its initialization fails
  ...

1 files changed, 932 insertions, 0 deletions

diff --git a/drivers/misc/hpilo.c b/drivers/misc/hpilo.c
new file mode 100644
index 000000000..8988858c2
--- /dev/null
+++ b/drivers/misc/hpilo.c
@@ -0,0 +1,932 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Driver for the HP iLO management processor.
+ *
+ * Copyright (C) 2008 Hewlett-Packard Development Company, L.P.
+ *	David Altobelli <david.altobelli@hpe.com>
+ */
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/pci.h>
+#include <linux/interrupt.h>
+#include <linux/ioport.h>
+#include <linux/device.h>
+#include <linux/file.h>
+#include <linux/cdev.h>
+#include <linux/sched.h>
+#include <linux/spinlock.h>
+#include <linux/delay.h>
+#include <linux/uaccess.h>
+#include <linux/io.h>
+#include <linux/wait.h>
+#include <linux/poll.h>
+#include <linux/slab.h>
+#include "hpilo.h"
+
+static const struct class ilo_class = {
+	.name = "iLO",
+};
+static unsigned int ilo_major;
+static unsigned int max_ccb = 16;
+static char ilo_hwdev[MAX_ILO_DEV];
+static const struct pci_device_id ilo_blacklist[] = {
+	/* auxiliary iLO */
+	{ PCI_DEVICE_SUB(PCI_VENDOR_ID_HP, 0x3307, PCI_VENDOR_ID_HP, 0x1979) },
+	/* CL */
+	{ PCI_DEVICE_SUB(PCI_VENDOR_ID_HP, 0x3307, PCI_VENDOR_ID_HP_3PAR, 0x0289) },
+	{ }
+};
+
+static inline int get_entry_id(int entry)
+{
+	return (entry & ENTRY_MASK_DESCRIPTOR) >> ENTRY_BITPOS_DESCRIPTOR;
+}
+
+static inline int get_entry_len(int entry)
+{
+	return ((entry & ENTRY_MASK_QWORDS) >> ENTRY_BITPOS_QWORDS) << 3;
+}
+
+static inline int mk_entry(int id, int len)
+{
+	int qlen = len & 7 ? (len >> 3) + 1 : len >> 3;
+	return id << ENTRY_BITPOS_DESCRIPTOR | qlen << ENTRY_BITPOS_QWORDS;
+}
+
+static inline int desc_mem_sz(int nr_entry)
+{
+	return nr_entry << L2_QENTRY_SZ;
+}
+
+/*
+ * FIFO queues, shared with hardware.
+ *
+ * If a queue has empty slots, an entry is added to the queue tail,
+ * and that entry is marked as occupied.
+ * Entries can be dequeued from the head of the list, when the device
+ * has marked the entry as consumed.
+ *
+ * Returns true on successful queue/dequeue, false on failure.
+ */
+static int fifo_enqueue(struct ilo_hwinfo *hw, char *fifobar, int entry)
+{
+	struct fifo *fifo_q = FIFOBARTOHANDLE(fifobar);
+	unsigned long flags;
+	int ret = 0;
+
+	spin_lock_irqsave(&hw->fifo_lock, flags);
+	if (!(fifo_q->fifobar[(fifo_q->tail + 1) & fifo_q->imask]
+	      & ENTRY_MASK_O)) {
+		fifo_q->fifobar[fifo_q->tail & fifo_q->imask] |=
+				(entry & ENTRY_MASK_NOSTATE) | fifo_q->merge;
+		fifo_q->tail += 1;
+		ret = 1;
+	}
+	spin_unlock_irqrestore(&hw->fifo_lock, flags);
+
+	return ret;
+}
+
+static int fifo_dequeue(struct ilo_hwinfo *hw, char *fifobar, int *entry)
+{
+	struct fifo *fifo_q = FIFOBARTOHANDLE(fifobar);
+	unsigned long flags;
+	int ret = 0;
+	u64 c;
+
+	spin_lock_irqsave(&hw->fifo_lock, flags);
+	c = fifo_q->fifobar[fifo_q->head & fifo_q->imask];
+	if (c & ENTRY_MASK_C) {
+		if (entry)
+			*entry = c & ENTRY_MASK_NOSTATE;
+
+		fifo_q->fifobar[fifo_q->head & fifo_q->imask] =
+							(c | ENTRY_MASK) + 1;
+		fifo_q->head += 1;
+		ret = 1;
+	}
+	spin_unlock_irqrestore(&hw->fifo_lock, flags);
+
+	return ret;
+}
+
+static int fifo_check_recv(struct ilo_hwinfo *hw, char *fifobar)
+{
+	struct fifo *fifo_q = FIFOBARTOHANDLE(fifobar);
+	unsigned long flags;
+	int ret = 0;
+	u64 c;
+
+	spin_lock_irqsave(&hw->fifo_lock, flags);
+	c = fifo_q->fifobar[fifo_q->head & fifo_q->imask];
+	if (c & ENTRY_MASK_C)
+		ret = 1;
+	spin_unlock_irqrestore(&hw->fifo_lock, flags);
+
+	return ret;
+}
+
+static int ilo_pkt_enqueue(struct ilo_hwinfo *hw, struct ccb *ccb,
+			   int dir, int id, int len)
+{
+	char *fifobar;
+	int entry;
+
+	if (dir == SENDQ)
+		fifobar = ccb->ccb_u1.send_fifobar;
+	else
+		fifobar = ccb->ccb_u3.recv_fifobar;
+
+	entry = mk_entry(id, len);
+	return fifo_enqueue(hw, fifobar, entry);
+}
+
+static int ilo_pkt_dequeue(struct ilo_hwinfo *hw, struct ccb *ccb,
+			   int dir, int *id, int *len, void **pkt)
+{
+	char *fifobar, *desc;
+	int entry = 0, pkt_id = 0;
+	int ret;
+
+	if (dir == SENDQ) {
+		fifobar = ccb->ccb_u1.send_fifobar;
+		desc = ccb->ccb_u2.send_desc;
+	} else {
+		fifobar = ccb->ccb_u3.recv_fifobar;
+		desc = ccb->ccb_u4.recv_desc;
+	}
+
+	ret = fifo_dequeue(hw, fifobar, &entry);
+	if (ret) {
+		int pkt_len;
+
+		pkt_id = get_entry_id(entry);
+		pkt_len = get_entry_len(entry);
+		if (pkt_id >= NR_QENTRY || pkt_len > desc_mem_sz(1))
+			return 0;
+		if (id)
+			*id = pkt_id;
+		if (len)
+			*len = pkt_len;
+		if (pkt)
+			*pkt = (void *)(desc + desc_mem_sz(pkt_id));
+	}
+
+	return ret;
+}
+
+static int ilo_pkt_recv(struct ilo_hwinfo *hw, struct ccb *ccb)
+{
+	char *fifobar = ccb->ccb_u3.recv_fifobar;
+
+	return fifo_check_recv(hw, fifobar);
+}
+
+static inline void doorbell_set(struct ccb *ccb)
+{
+	iowrite8(1, ccb->ccb_u5.db_base);
+}
+
+static inline void doorbell_clr(struct ccb *ccb)
+{
+	iowrite8(2, ccb->ccb_u5.db_base);
+}
+
+static inline int ctrl_set(int l2sz, int idxmask, int desclim)
+{
+	int active = 0, go = 1;
+	return l2sz << CTRL_BITPOS_L2SZ |
+	       idxmask << CTRL_BITPOS_FIFOINDEXMASK |
+	       desclim << CTRL_BITPOS_DESCLIMIT |
+	       active << CTRL_BITPOS_A |
+	       go << CTRL_BITPOS_G;
+}
+
+static void ctrl_setup(struct ccb *ccb, int nr_desc, int l2desc_sz)
+{
+	/* for simplicity, use the same parameters for send and recv ctrls */
+	ccb->send_ctrl = ctrl_set(l2desc_sz, nr_desc-1, nr_desc-1);
+	ccb->recv_ctrl = ctrl_set(l2desc_sz, nr_desc-1, nr_desc-1);
+}
+
+static inline int fifo_sz(int nr_entry)
+{
+	/* size of a fifo is determined by the number of entries it contains */
+	return nr_entry * sizeof(u64) + FIFOHANDLESIZE;
+}
+
+static void fifo_setup(void *base_addr, int nr_entry)
+{
+	struct fifo *fifo_q = base_addr;
+	int i;
+
+	/* set up an empty fifo */
+	fifo_q->head = 0;
+	fifo_q->tail = 0;
+	fifo_q->reset = 0;
+	fifo_q->nrents = nr_entry;
+	fifo_q->imask = nr_entry - 1;
+	fifo_q->merge = ENTRY_MASK_O;
+
+	for (i = 0; i < nr_entry; i++)
+		fifo_q->fifobar[i] = 0;
+}
+
+static void ilo_ccb_close(struct pci_dev *pdev, struct ccb_data *data)
+{
+	struct ccb *driver_ccb = &data->driver_ccb;
+	struct ccb __iomem *device_ccb = data->mapped_ccb;
+	int retries;
+
+	/* complicated dance to tell the hw we are stopping */
+	doorbell_clr(driver_ccb);
+	iowrite32(ioread32(&device_ccb->send_ctrl) & ~(1 << CTRL_BITPOS_G),
+		  &device_ccb->send_ctrl);
+	iowrite32(ioread32(&device_ccb->recv_ctrl) & ~(1 << CTRL_BITPOS_G),
+		  &device_ccb->recv_ctrl);
+
+	/* give iLO some time to process stop request */
+	for (retries = MAX_WAIT; retries > 0; retries--) {
+		doorbell_set(driver_ccb);
+		udelay(WAIT_TIME);
+		if (!(ioread32(&device_ccb->send_ctrl) & (1 << CTRL_BITPOS_A))
+		    &&
+		    !(ioread32(&device_ccb->recv_ctrl) & (1 << CTRL_BITPOS_A)))
+			break;
+	}
+	if (retries == 0)
+		dev_err(&pdev->dev, "Closing, but controller still active\n");
+
+	/* clear the hw ccb */
+	memset_io(device_ccb, 0, sizeof(struct ccb));
+
+	/* free resources used to back send/recv queues */
+	dma_free_coherent(&pdev->dev, data->dma_size, data->dma_va,
+			  data->dma_pa);
+}
+
+static int ilo_ccb_setup(struct ilo_hwinfo *hw, struct ccb_data *data, int slot)
+{
+	char *dma_va;
+	dma_addr_t dma_pa;
+	struct ccb *driver_ccb, *ilo_ccb;
+
+	driver_ccb = &data->driver_ccb;
+	ilo_ccb = &data->ilo_ccb;
+
+	data->dma_size = 2 * fifo_sz(NR_QENTRY) +
+			 2 * desc_mem_sz(NR_QENTRY) +
+			 ILO_START_ALIGN + ILO_CACHE_SZ;
+
+	data->dma_va = dma_alloc_coherent(&hw->ilo_dev->dev, data->dma_size,
+					  &data->dma_pa, GFP_ATOMIC);
+	if (!data->dma_va)
+		return -ENOMEM;
+
+	dma_va = (char *)data->dma_va;
+	dma_pa = data->dma_pa;
+
+	dma_va = (char *)roundup((unsigned long)dma_va, ILO_START_ALIGN);
+	dma_pa = roundup(dma_pa, ILO_START_ALIGN);
+
+	/*
+	 * Create two ccb's, one with virt addrs, one with phys addrs.
+	 * Copy the phys addr ccb to device shared mem.
+	 */
+	ctrl_setup(driver_ccb, NR_QENTRY, L2_QENTRY_SZ);
+	ctrl_setup(ilo_ccb, NR_QENTRY, L2_QENTRY_SZ);
+
+	fifo_setup(dma_va, NR_QENTRY);
+	driver_ccb->ccb_u1.send_fifobar = dma_va + FIFOHANDLESIZE;
+	ilo_ccb->ccb_u1.send_fifobar_pa = dma_pa + FIFOHANDLESIZE;
+	dma_va += fifo_sz(NR_QENTRY);
+	dma_pa += fifo_sz(NR_QENTRY);
+
+	dma_va = (char *)roundup((unsigned long)dma_va, ILO_CACHE_SZ);
+	dma_pa = roundup(dma_pa, ILO_CACHE_SZ);
+
+	fifo_setup(dma_va, NR_QENTRY);
+	driver_ccb->ccb_u3.recv_fifobar = dma_va + FIFOHANDLESIZE;
+	ilo_ccb->ccb_u3.recv_fifobar_pa = dma_pa + FIFOHANDLESIZE;
+	dma_va += fifo_sz(NR_QENTRY);
+	dma_pa += fifo_sz(NR_QENTRY);
+
+	driver_ccb->ccb_u2.send_desc = dma_va;
+	ilo_ccb->ccb_u2.send_desc_pa = dma_pa;
+	dma_pa += desc_mem_sz(NR_QENTRY);
+	dma_va += desc_mem_sz(NR_QENTRY);
+
+	driver_ccb->ccb_u4.recv_desc = dma_va;
+	ilo_ccb->ccb_u4.recv_desc_pa = dma_pa;
+
+	driver_ccb->channel = slot;
+	ilo_ccb->channel = slot;
+
+	driver_ccb->ccb_u5.db_base = hw->db_vaddr + (slot << L2_DB_SIZE);
+	ilo_ccb->ccb_u5.db_base = NULL; /* hw ccb's doorbell is not used */
+
+	return 0;
+}
+
+static void ilo_ccb_open(struct ilo_hwinfo *hw, struct ccb_data *data, int slot)
+{
+	int pkt_id, pkt_sz;
+	struct ccb *driver_ccb = &data->driver_ccb;
+
+	/* copy the ccb with physical addrs to device memory */
+	data->mapped_ccb = (struct ccb __iomem *)
+				(hw->ram_vaddr + (slot * ILOHW_CCB_SZ));
+	memcpy_toio(data->mapped_ccb, &data->ilo_ccb, sizeof(struct ccb));
+
+	/* put packets on the send and receive queues */
+	pkt_sz = 0;
+	for (pkt_id = 0; pkt_id < NR_QENTRY; pkt_id++) {
+		ilo_pkt_enqueue(hw, driver_ccb, SENDQ, pkt_id, pkt_sz);
+		doorbell_set(driver_ccb);
+	}
+
+	pkt_sz = desc_mem_sz(1);
+	for (pkt_id = 0; pkt_id < NR_QENTRY; pkt_id++)
+		ilo_pkt_enqueue(hw, driver_ccb, RECVQ, pkt_id, pkt_sz);
+
+	/* the ccb is ready to use */
+	doorbell_clr(driver_ccb);
+}
+
+static int ilo_ccb_verify(struct ilo_hwinfo *hw, struct ccb_data *data)
+{
+	int pkt_id, i;
+	struct ccb *driver_ccb = &data->driver_ccb;
+
+	/* make sure iLO is really handling requests */
+	for (i = MAX_WAIT; i > 0; i--) {
+		if (ilo_pkt_dequeue(hw, driver_ccb, SENDQ, &pkt_id, NULL, NULL))
+			break;
+		udelay(WAIT_TIME);
+	}
+
+	if (i == 0) {
+		dev_err(&hw->ilo_dev->dev, "Open could not dequeue a packet\n");
+		return -EBUSY;
+	}
+
+	ilo_pkt_enqueue(hw, driver_ccb, SENDQ, pkt_id, 0);
+	doorbell_set(driver_ccb);
+	return 0;
+}
+
+static inline int is_channel_reset(struct ccb *ccb)
+{
+	/* check for this particular channel needing a reset */
+	return FIFOBARTOHANDLE(ccb->ccb_u1.send_fifobar)->reset;
+}
+
+static inline void set_channel_reset(struct ccb *ccb)
+{
+	/* set a flag indicating this channel needs a reset */
+	FIFOBARTOHANDLE(ccb->ccb_u1.send_fifobar)->reset = 1;
+}
+
+static inline int get_device_outbound(struct ilo_hwinfo *hw)
+{
+	return ioread32(&hw->mmio_vaddr[DB_OUT]);
+}
+
+static inline int is_db_reset(int db_out)
+{
+	return db_out & (1 << DB_RESET);
+}
+
+static inline void clear_pending_db(struct ilo_hwinfo *hw, int clr)
+{
+	iowrite32(clr, &hw->mmio_vaddr[DB_OUT]);
+}
+
+static inline void clear_device(struct ilo_hwinfo *hw)
+{
+	/* clear the device (reset bits, pending channel entries) */
+	clear_pending_db(hw, -1);
+}
+
+static inline void ilo_enable_interrupts(struct ilo_hwinfo *hw)
+{
+	iowrite8(ioread8(&hw->mmio_vaddr[DB_IRQ]) | 1, &hw->mmio_vaddr[DB_IRQ]);
+}
+
+static inline void ilo_disable_interrupts(struct ilo_hwinfo *hw)
+{
+	iowrite8(ioread8(&hw->mmio_vaddr[DB_IRQ]) & ~1,
+		 &hw->mmio_vaddr[DB_IRQ]);
+}
+
+static void ilo_set_reset(struct ilo_hwinfo *hw)
+{
+	int slot;
+
+	/*
+	 * Mapped memory is zeroed on ilo reset, so set a per ccb flag
+	 * to indicate that this ccb needs to be closed and reopened.
+	 */
+	for (slot = 0; slot < max_ccb; slot++) {
+		if (!hw->ccb_alloc[slot])
+			continue;
+		set_channel_reset(&hw->ccb_alloc[slot]->driver_ccb);
+	}
+}
+
+static ssize_t ilo_read(struct file *fp, char __user *buf,
+			size_t len, loff_t *off)
+{
+	int err, found, cnt, pkt_id, pkt_len;
+	struct ccb_data *data = fp->private_data;
+	struct ccb *driver_ccb = &data->driver_ccb;
+	struct ilo_hwinfo *hw = data->ilo_hw;
+	void *pkt;
+
+	if (is_channel_reset(driver_ccb)) {
+		/*
+		 * If the device has been reset, applications
+		 * need to close and reopen all ccbs.
+		 */
+		return -ENODEV;
+	}
+
+	/*
+	 * This function is to be called when data is expected
+	 * in the channel, and will return an error if no packet is found
+	 * during the loop below.  The sleep/retry logic is to allow
+	 * applications to call read() immediately post write(),
+	 * and give iLO some time to process the sent packet.
+	 */
+	cnt = 20;
+	do {
+		/* look for a received packet */
+		found = ilo_pkt_dequeue(hw, driver_ccb, RECVQ, &pkt_id,
+					&pkt_len, &pkt);
+		if (found)
+			break;
+		cnt--;
+		msleep(100);
+	} while (!found && cnt);
+
+	if (!found)
+		return -EAGAIN;
+
+	/* only copy the length of the received packet */
+	if (pkt_len < len)
+		len = pkt_len;
+
+	err = copy_to_user(buf, pkt, len);
+
+	/* return the received packet to the queue */
+	ilo_pkt_enqueue(hw, driver_ccb, RECVQ, pkt_id, desc_mem_sz(1));
+
+	return err ? -EFAULT : len;
+}
+
+static ssize_t ilo_write(struct file *fp, const char __user *buf,
+			 size_t len, loff_t *off)
+{
+	int err, pkt_id, pkt_len;
+	struct ccb_data *data = fp->private_data;
+	struct ccb *driver_ccb = &data->driver_ccb;
+	struct ilo_hwinfo *hw = data->ilo_hw;
+	void *pkt;
+
+	if (is_channel_reset(driver_ccb))
+		return -ENODEV;
+
+	/* get a packet to send the user command */
+	if (!ilo_pkt_dequeue(hw, driver_ccb, SENDQ, &pkt_id, &pkt_len, &pkt))
+		return -EBUSY;
+
+	/* limit the length to the length of the packet */
+	if (pkt_len < len)
+		len = pkt_len;
+
+	/* on failure, set the len to 0 to return empty packet to the device */
+	err = copy_from_user(pkt, buf, len);
+	if (err)
+		len = 0;
+
+	/* send the packet */
+	ilo_pkt_enqueue(hw, driver_ccb, SENDQ, pkt_id, len);
+	doorbell_set(driver_ccb);
+
+	return err ? -EFAULT : len;
+}
+
+static __poll_t ilo_poll(struct file *fp, poll_table *wait)
+{
+	struct ccb_data *data = fp->private_data;
+	struct ccb *driver_ccb = &data->driver_ccb;
+
+	poll_wait(fp, &data->ccb_waitq, wait);
+
+	if (is_channel_reset(driver_ccb))
+		return EPOLLERR;
+	else if (ilo_pkt_recv(data->ilo_hw, driver_ccb))
+		return EPOLLIN | EPOLLRDNORM;
+
+	return 0;
+}
+
+static int ilo_close(struct inode *ip, struct file *fp)
+{
+	int slot;
+	struct ccb_data *data;
+	struct ilo_hwinfo *hw;
+	unsigned long flags;
+
+	slot = iminor(ip) % max_ccb;
+	hw = container_of(ip->i_cdev, struct ilo_hwinfo, cdev);
+
+	spin_lock(&hw->open_lock);
+
+	if (hw->ccb_alloc[slot]->ccb_cnt == 1) {
+
+		data = fp->private_data;
+
+		spin_lock_irqsave(&hw->alloc_lock, flags);
+		hw->ccb_alloc[slot] = NULL;
+		spin_unlock_irqrestore(&hw->alloc_lock, flags);
+
+		ilo_ccb_close(hw->ilo_dev, data);
+
+		kfree(data);
+	} else
+		hw->ccb_alloc[slot]->ccb_cnt--;
+
+	spin_unlock(&hw->open_lock);
+
+	return 0;
+}
+
+static int ilo_open(struct inode *ip, struct file *fp)
+{
+	int slot, error;
+	struct ccb_data *data;
+	struct ilo_hwinfo *hw;
+	unsigned long flags;
+
+	slot = iminor(ip) % max_ccb;
+	hw = container_of(ip->i_cdev, struct ilo_hwinfo, cdev);
+
+	/* new ccb allocation */
+	data = kzalloc_obj(*data);
+	if (!data)
+		return -ENOMEM;
+
+	spin_lock(&hw->open_lock);
+
+	/* each fd private_data holds sw/hw view of ccb */
+	if (hw->ccb_alloc[slot] == NULL) {
+		/* create a channel control block for this minor */
+		error = ilo_ccb_setup(hw, data, slot);
+		if (error) {
+			kfree(data);
+			goto out;
+		}
+
+		data->ccb_cnt = 1;
+		data->ccb_excl = fp->f_flags & O_EXCL;
+		data->ilo_hw = hw;
+		init_waitqueue_head(&data->ccb_waitq);
+
+		/* write the ccb to hw */
+		spin_lock_irqsave(&hw->alloc_lock, flags);
+		ilo_ccb_open(hw, data, slot);
+		hw->ccb_alloc[slot] = data;
+		spin_unlock_irqrestore(&hw->alloc_lock, flags);
+
+		/* make sure the channel is functional */
+		error = ilo_ccb_verify(hw, data);
+		if (error) {
+
+			spin_lock_irqsave(&hw->alloc_lock, flags);
+			hw->ccb_alloc[slot] = NULL;
+			spin_unlock_irqrestore(&hw->alloc_lock, flags);
+
+			ilo_ccb_close(hw->ilo_dev, data);
+
+			kfree(data);
+			goto out;
+		}
+
+	} else {
+		kfree(data);
+		if (fp->f_flags & O_EXCL || hw->ccb_alloc[slot]->ccb_excl) {
+			/*
+			 * The channel exists, and either this open
+			 * or a previous open of this channel wants
+			 * exclusive access.
+			 */
+			error = -EBUSY;
+		} else {
+			hw->ccb_alloc[slot]->ccb_cnt++;
+			error = 0;
+		}
+	}
+out:
+	spin_unlock(&hw->open_lock);
+
+	if (!error)
+		fp->private_data = hw->ccb_alloc[slot];
+
+	return error;
+}
+
+static const struct file_operations ilo_fops = {
+	.owner		= THIS_MODULE,
+	.read		= ilo_read,
+	.write		= ilo_write,
+	.poll		= ilo_poll,
+	.open 		= ilo_open,
+	.release 	= ilo_close,
+	.llseek		= noop_llseek,
+};
+
+static irqreturn_t ilo_isr(int irq, void *data)
+{
+	struct ilo_hwinfo *hw = data;
+	int pending, i;
+
+	spin_lock(&hw->alloc_lock);
+
+	/* check for ccbs which have data */
+	pending = get_device_outbound(hw);
+	if (!pending) {
+		spin_unlock(&hw->alloc_lock);
+		return IRQ_NONE;
+	}
+
+	if (is_db_reset(pending)) {
+		/* wake up all ccbs if the device was reset */
+		pending = -1;
+		ilo_set_reset(hw);
+	}
+
+	for (i = 0; i < max_ccb; i++) {
+		if (!hw->ccb_alloc[i])
+			continue;
+		if (pending & (1 << i))
+			wake_up_interruptible(&hw->ccb_alloc[i]->ccb_waitq);
+	}
+
+	/* clear the device of the channels that have been handled */
+	clear_pending_db(hw, pending);
+
+	spin_unlock(&hw->alloc_lock);
+
+	return IRQ_HANDLED;
+}
+
+static void ilo_unmap_device(struct pci_dev *pdev, struct ilo_hwinfo *hw)
+{
+	pci_iounmap(pdev, hw->db_vaddr);
+	pci_iounmap(pdev, hw->ram_vaddr);
+	pci_iounmap(pdev, hw->mmio_vaddr);
+}
+
+static int ilo_map_device(struct pci_dev *pdev, struct ilo_hwinfo *hw)
+{
+	int bar;
+	unsigned long off;
+	u8 pci_rev_id;
+	int rc;
+
+	/* map the memory mapped i/o registers */
+	hw->mmio_vaddr = pci_iomap(pdev, 1, 0);
+	if (hw->mmio_vaddr == NULL) {
+		dev_err(&pdev->dev, "Error mapping mmio\n");
+		goto out;
+	}
+
+	/* map the adapter shared memory region */
+	rc = pci_read_config_byte(pdev, PCI_REVISION_ID, &pci_rev_id);
+	if (rc != 0) {
+		dev_err(&pdev->dev, "Error reading PCI rev id: %d\n", rc);
+		goto out;
+	}
+
+	if (pci_rev_id >= PCI_REV_ID_NECHES) {
+		bar = 5;
+		/* Last 8k is reserved for CCBs */
+		off = pci_resource_len(pdev, bar) - 0x2000;
+	} else {
+		bar = 2;
+		off = 0;
+	}
+	hw->ram_vaddr = pci_iomap_range(pdev, bar, off, max_ccb * ILOHW_CCB_SZ);
+	if (hw->ram_vaddr == NULL) {
+		dev_err(&pdev->dev, "Error mapping shared mem\n");
+		goto mmio_free;
+	}
+
+	/* map the doorbell aperture */
+	hw->db_vaddr = pci_iomap(pdev, 3, max_ccb * ONE_DB_SIZE);
+	if (hw->db_vaddr == NULL) {
+		dev_err(&pdev->dev, "Error mapping doorbell\n");
+		goto ram_free;
+	}
+
+	return 0;
+ram_free:
+	pci_iounmap(pdev, hw->ram_vaddr);
+mmio_free:
+	pci_iounmap(pdev, hw->mmio_vaddr);
+out:
+	return -ENOMEM;
+}
+
+static void ilo_remove(struct pci_dev *pdev)
+{
+	int i, minor;
+	struct ilo_hwinfo *ilo_hw = pci_get_drvdata(pdev);
+
+	if (!ilo_hw)
+		return;
+
+	clear_device(ilo_hw);
+
+	minor = MINOR(ilo_hw->cdev.dev);
+	for (i = minor; i < minor + max_ccb; i++)
+		device_destroy(&ilo_class, MKDEV(ilo_major, i));
+
+	cdev_del(&ilo_hw->cdev);
+	ilo_disable_interrupts(ilo_hw);
+	free_irq(pdev->irq, ilo_hw);
+	ilo_unmap_device(pdev, ilo_hw);
+	pci_release_regions(pdev);
+	/*
+	 * pci_disable_device(pdev) used to be here. But this PCI device has
+	 * two functions with interrupt lines connected to a single pin. The
+	 * other one is a USB host controller. So when we disable the PIN here
+	 * e.g. by rmmod hpilo, the controller stops working. It is because
+	 * the interrupt link is disabled in ACPI since it is not refcounted
+	 * yet. See acpi_pci_link_free_irq called from acpi_pci_irq_disable.
+	 */
+	kfree(ilo_hw);
+	ilo_hwdev[(minor / max_ccb)] = 0;
+}
+
+static int ilo_probe(struct pci_dev *pdev,
+			       const struct pci_device_id *ent)
+{
+	int devnum, slot, start, error = 0;
+	struct ilo_hwinfo *ilo_hw;
+
+	if (pci_match_id(ilo_blacklist, pdev)) {
+		dev_dbg(&pdev->dev, "Not supported on this device\n");
+		return -ENODEV;
+	}
+
+	if (max_ccb > MAX_CCB)
+		max_ccb = MAX_CCB;
+	else if (max_ccb < MIN_CCB)
+		max_ccb = MIN_CCB;
+
+	/* find a free range for device files */
+	for (devnum = 0; devnum < MAX_ILO_DEV; devnum++) {
+		if (ilo_hwdev[devnum] == 0) {
+			ilo_hwdev[devnum] = 1;
+			break;
+		}
+	}
+
+	if (devnum == MAX_ILO_DEV) {
+		dev_err(&pdev->dev, "Error finding free device\n");
+		return -ENODEV;
+	}
+
+	/* track global allocations for this device */
+	error = -ENOMEM;
+	ilo_hw = kzalloc_obj(*ilo_hw);
+	if (!ilo_hw)
+		goto out;
+
+	ilo_hw->ilo_dev = pdev;
+	spin_lock_init(&ilo_hw->alloc_lock);
+	spin_lock_init(&ilo_hw->fifo_lock);
+	spin_lock_init(&ilo_hw->open_lock);
+
+	error = pci_enable_device(pdev);
+	if (error)
+		goto free;
+
+	pci_set_master(pdev);
+
+	error = pci_request_regions(pdev, ILO_NAME);
+	if (error)
+		goto disable;
+
+	error = ilo_map_device(pdev, ilo_hw);
+	if (error)
+		goto free_regions;
+
+	pci_set_drvdata(pdev, ilo_hw);
+	clear_device(ilo_hw);
+
+	error = request_irq(pdev->irq, ilo_isr, IRQF_SHARED, "hpilo", ilo_hw);
+	if (error)
+		goto unmap;
+
+	ilo_enable_interrupts(ilo_hw);
+
+	cdev_init(&ilo_hw->cdev, &ilo_fops);
+	ilo_hw->cdev.owner = THIS_MODULE;
+	start = devnum * max_ccb;
+	error = cdev_add(&ilo_hw->cdev, MKDEV(ilo_major, start), max_ccb);
+	if (error) {
+		dev_err(&pdev->dev, "Could not add cdev\n");
+		goto remove_isr;
+	}
+
+	for (slot = 0; slot < max_ccb; slot++) {
+		struct device *dev;
+		dev = device_create(&ilo_class, &pdev->dev,
+				    MKDEV(ilo_major, start + slot), NULL,
+				    "hpilo!d%dccb%d", devnum, slot);
+		if (IS_ERR(dev))
+			dev_err(&pdev->dev, "Could not create files\n");
+	}
+
+	return 0;
+remove_isr:
+	ilo_disable_interrupts(ilo_hw);
+	free_irq(pdev->irq, ilo_hw);
+unmap:
+	ilo_unmap_device(pdev, ilo_hw);
+free_regions:
+	pci_release_regions(pdev);
+disable:
+/*	pci_disable_device(pdev);  see comment in ilo_remove */
+free:
+	kfree(ilo_hw);
+out:
+	ilo_hwdev[devnum] = 0;
+	return error;
+}
+
+static const struct pci_device_id ilo_devices[] = {
+	{ PCI_DEVICE(PCI_VENDOR_ID_COMPAQ, 0xB204) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_HP, 0x3307) },
+	{ }
+};
+MODULE_DEVICE_TABLE(pci, ilo_devices);
+
+static struct pci_driver ilo_driver = {
+	.name 	  = ILO_NAME,
+	.id_table = ilo_devices,
+	.probe 	  = ilo_probe,
+	.remove   = ilo_remove,
+};
+
+static int __init ilo_init(void)
+{
+	int error;
+	dev_t dev;
+
+	error = class_register(&ilo_class);
+	if (error)
+		goto out;
+
+	error = alloc_chrdev_region(&dev, 0, MAX_OPEN, ILO_NAME);
+	if (error)
+		goto class_destroy;
+
+	ilo_major = MAJOR(dev);
+
+	error =	pci_register_driver(&ilo_driver);
+	if (error)
+		goto chr_remove;
+
+	return 0;
+chr_remove:
+	unregister_chrdev_region(dev, MAX_OPEN);
+class_destroy:
+	class_unregister(&ilo_class);
+out:
+	return error;
+}
+
+static void __exit ilo_exit(void)
+{
+	pci_unregister_driver(&ilo_driver);
+	unregister_chrdev_region(MKDEV(ilo_major, 0), MAX_OPEN);
+	class_unregister(&ilo_class);
+}
+
+MODULE_VERSION("1.5.0");
+MODULE_ALIAS(ILO_NAME);
+MODULE_DESCRIPTION(ILO_NAME);
+MODULE_AUTHOR("David Altobelli <david.altobelli@hpe.com>");
+MODULE_LICENSE("GPL v2");
+
+module_param(max_ccb, uint, 0444);
+MODULE_PARM_DESC(max_ccb, "Maximum number of HP iLO channels to attach (8-24)(default=16)");
+
+module_init(ilo_init);
+module_exit(ilo_exit);