ACPI: bay: new driver adding removable drive bay support
authorKristen Carlson Accardi <kristen.c.accardi@intel.com>
Fri, 20 Oct 2006 21:30:25 +0000 (14:30 -0700)
committerLen Brown <len.brown@intel.com>
Sat, 3 Feb 2007 06:11:30 +0000 (01:11 -0500)
Signed-off-by: Kristen Carlson Accardi <kristen.c.accardi@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Len Brown <len.brown@intel.com>
drivers/acpi/Kconfig
drivers/acpi/Makefile
drivers/acpi/bay.c [new file with mode: 0644]

index f4f000a..2caef09 100644 (file)
@@ -139,6 +139,13 @@ config ACPI_DOCK
        help
          This driver adds support for ACPI controlled docking stations
 
+config ACPI_BAY
+       tristate "Removable Drive Bay"
+       depends on EXPERIMENTAL
+       help
+         This driver adds support for ACPI controlled removable drive
+         bays such as the IBM ultrabay or the Dell Module Bay.
+
 config ACPI_PROCESSOR
        tristate "Processor"
        default y
index bce7ca2..1a73805 100644 (file)
@@ -43,7 +43,8 @@ obj-$(CONFIG_ACPI_BUTTON)     += button.o
 obj-$(CONFIG_ACPI_EC)          += ec.o
 obj-$(CONFIG_ACPI_FAN)         += fan.o
 obj-$(CONFIG_ACPI_DOCK)                += dock.o
-obj-$(CONFIG_ACPI_VIDEO)       += video.o 
+obj-$(CONFIG_ACPI_BAY)         += bay.o
+obj-$(CONFIG_ACPI_VIDEO)       += video.o
 obj-$(CONFIG_ACPI_HOTKEY)      += hotkey.o
 obj-y                          += pci_root.o pci_link.o pci_irq.o pci_bind.o
 obj-$(CONFIG_ACPI_POWER)       += power.o
diff --git a/drivers/acpi/bay.c b/drivers/acpi/bay.c
new file mode 100644 (file)
index 0000000..1cfc0b7
--- /dev/null
@@ -0,0 +1,592 @@
+/*
+ *  bay.c - ACPI removable drive bay driver
+ *
+ *  Copyright (C) 2006 Kristen Carlson Accardi <kristen.c.accardi@intel.com>
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ *
+ *  This program is free software; you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation; either version 2 of the License, or (at
+ *  your option) any later version.
+ *
+ *  This program is distributed in the hope that it will be useful, but
+ *  WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ *  General Public License for more details.
+ *
+ *  You should have received a copy of the GNU General Public License along
+ *  with this program; if not, write to the Free Software Foundation, Inc.,
+ *  59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
+ *
+ * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/types.h>
+#include <linux/notifier.h>
+#include <acpi/acpi_bus.h>
+#include <acpi/acpi_drivers.h>
+#include <linux/proc_fs.h>
+#include <linux/seq_file.h>
+#include <asm/uaccess.h>
+
+#define ACPI_BAY_DRIVER_NAME "ACPI Removable Drive Bay Driver"
+
+ACPI_MODULE_NAME("bay")
+MODULE_AUTHOR("Kristen Carlson Accardi");
+MODULE_DESCRIPTION(ACPI_BAY_DRIVER_NAME);
+MODULE_LICENSE("GPL");
+#define ACPI_BAY_CLASS "bay"
+#define ACPI_BAY_COMPONENT     0x10000000
+#define _COMPONENT ACPI_BAY_COMPONENT
+#define bay_dprintk(h,s) {\
+       char prefix[80] = {'\0'};\
+       struct acpi_buffer buffer = {sizeof(prefix), prefix};\
+       acpi_get_name(h, ACPI_FULL_PATHNAME, &buffer);\
+       printk(KERN_DEBUG PREFIX "%s: %s\n", prefix, s); }
+
+static void bay_notify(acpi_handle handle, u32 event, void *data);
+static int acpi_bay_add(struct acpi_device *device);
+static int acpi_bay_remove(struct acpi_device *device, int type);
+static int acpi_bay_match(struct acpi_device *device,
+                               struct acpi_driver *driver);
+
+static struct acpi_driver acpi_bay_driver = {
+       .name = ACPI_BAY_DRIVER_NAME,
+       .class = ACPI_BAY_CLASS,
+       .ops = {
+               .add = acpi_bay_add,
+               .remove = acpi_bay_remove,
+               .match = acpi_bay_match,
+               },
+};
+
+struct bay {
+       acpi_handle handle;
+       char *name;
+       struct list_head list;
+       struct proc_dir_entry *proc;
+};
+
+LIST_HEAD(drive_bays);
+
+static struct proc_dir_entry *acpi_bay_dir;
+
+/*****************************************************************************
+ *                         Drive Bay functions                               *
+ *****************************************************************************/
+/**
+ * is_ejectable - see if a device is ejectable
+ * @handle: acpi handle of the device
+ *
+ * If an acpi object has a _EJ0 method, then it is ejectable
+ */
+static int is_ejectable(acpi_handle handle)
+{
+       acpi_status status;
+       acpi_handle tmp;
+
+       status = acpi_get_handle(handle, "_EJ0", &tmp);
+       if (ACPI_FAILURE(status))
+               return 0;
+       return 1;
+}
+
+/**
+ * bay_present - see if the bay device is present
+ * @bay: the drive bay
+ *
+ * execute the _STA method.
+ */
+static int bay_present(struct bay *bay)
+{
+       unsigned long sta;
+       acpi_status status;
+
+       if (bay) {
+               status = acpi_evaluate_integer(bay->handle, "_STA", NULL, &sta);
+               if (ACPI_SUCCESS(status) && sta)
+                       return 1;
+       }
+       return 0;
+}
+
+/**
+ * eject_device - respond to an eject request
+ * @handle - the device to eject
+ *
+ * Call this devices _EJ0 method.
+ */
+static void eject_device(acpi_handle handle)
+{
+       struct acpi_object_list arg_list;
+       union acpi_object arg;
+
+       bay_dprintk(handle, "Ejecting device");
+
+       arg_list.count = 1;
+       arg_list.pointer = &arg;
+       arg.type = ACPI_TYPE_INTEGER;
+       arg.integer.value = 1;
+
+       if (ACPI_FAILURE(acpi_evaluate_object(handle, "_EJ0",
+                                             &arg_list, NULL)))
+               pr_debug("Failed to evaluate _EJ0!\n");
+}
+
+
+/**
+ * is_ata - see if a device is an ata device
+ * @handle: acpi handle of the device
+ *
+ * If an acpi object has one of 4 ATA ACPI methods defined,
+ * then it is an ATA device
+ */
+static int is_ata(acpi_handle handle)
+{
+       acpi_handle tmp;
+
+       if ((ACPI_SUCCESS(acpi_get_handle(handle, "_GTF", &tmp))) ||
+          (ACPI_SUCCESS(acpi_get_handle(handle, "_GTM", &tmp))) ||
+          (ACPI_SUCCESS(acpi_get_handle(handle, "_STM", &tmp))) ||
+          (ACPI_SUCCESS(acpi_get_handle(handle, "_SDD", &tmp))))
+               return 1;
+
+       return 0;
+}
+
+/**
+ * parent_is_ata(acpi_handle handle)
+ *
+ */
+static int parent_is_ata(acpi_handle handle)
+{
+       acpi_handle phandle;
+
+       if (acpi_get_parent(handle, &phandle))
+               return 0;
+
+       return is_ata(phandle);
+}
+
+/**
+ * is_ejectable_bay - see if a device is an ejectable drive bay
+ * @handle: acpi handle of the device
+ *
+ * If an acpi object is ejectable and has one of the ACPI ATA
+ * methods defined, then we can safely call it an ejectable
+ * drive bay
+ */
+static int is_ejectable_bay(acpi_handle handle)
+{
+       if ((is_ata(handle) || parent_is_ata(handle)) && is_ejectable(handle))
+               return 1;
+       return 0;
+}
+
+/**
+ * eject_removable_drive - try to eject this drive
+ * @dev : the device structure of the drive
+ *
+ * If a device is a removable drive that requires an _EJ0 method
+ * to be executed in order to safely remove from the system, do
+ * it.  ATM - always returns success
+ */
+int eject_removable_drive(struct device *dev)
+{
+       acpi_handle handle = DEVICE_ACPI_HANDLE(dev);
+
+       if (handle) {
+               bay_dprintk(handle, "Got device handle");
+               if (is_ejectable_bay(handle))
+                       eject_device(handle);
+       } else {
+               printk("No acpi handle for device\n");
+       }
+
+       /* should I return an error code? */
+       return 0;
+}
+EXPORT_SYMBOL_GPL(eject_removable_drive);
+
+static int acpi_bay_add(struct acpi_device *device)
+{
+       bay_dprintk(device->handle, "adding bay device");
+       strcpy(acpi_device_name(device), "Dockable Bay");
+       strcpy(acpi_device_class(device), "bay");
+       return 0;
+}
+
+static int acpi_bay_status_seq_show(struct seq_file *seq, void *offset)
+{
+       struct bay *bay = (struct bay *)seq->private;
+
+       if (!bay)
+               return 0;
+
+       if (bay_present(bay))
+               seq_printf(seq, "present\n");
+       else
+               seq_printf(seq, "removed\n");
+
+       return 0;
+}
+
+static ssize_t
+acpi_bay_write_eject(struct file *file,
+                     const char __user * buffer,
+                     size_t count, loff_t * data)
+{
+       struct seq_file *m = (struct seq_file *)file->private_data;
+       struct bay *bay = (struct bay *)m->private;
+       char str[12] = { 0 };
+       u32 state = 0;
+
+       /* FIXME - our only valid value here is 1 */
+       if (!bay || count + 1 > sizeof str)
+               return -EINVAL;
+
+       if (copy_from_user(str, buffer, count))
+               return -EFAULT;
+
+       str[count] = 0;
+       state = simple_strtoul(str, NULL, 0);
+       if (state)
+               eject_device(bay->handle);
+
+       return count;
+}
+
+static int
+acpi_bay_status_open_fs(struct inode *inode, struct file *file)
+{
+       return single_open(file, acpi_bay_status_seq_show,
+                          PDE(inode)->data);
+}
+
+static int
+acpi_bay_eject_open_fs(struct inode *inode, struct file *file)
+{
+       return single_open(file, acpi_bay_status_seq_show,
+                          PDE(inode)->data);
+}
+
+static struct file_operations acpi_bay_status_fops = {
+       .open = acpi_bay_status_open_fs,
+       .read = seq_read,
+       .llseek = seq_lseek,
+       .release = single_release,
+};
+
+static struct file_operations acpi_bay_eject_fops = {
+       .open = acpi_bay_eject_open_fs,
+       .read = seq_read,
+       .write = acpi_bay_write_eject,
+       .llseek = seq_lseek,
+       .release = single_release,
+};
+#if 0
+static struct file_operations acpi_bay_insert_fops = {
+       .open = acpi_bay_insert_open_fs,
+       .read = seq_read,
+       .llseek = seq_lseek,
+       .release = single_release,
+};
+#endif
+static int acpi_bay_add_fs(struct bay *bay)
+{
+       struct proc_dir_entry *entry = NULL;
+
+       if (!bay)
+               return -EINVAL;
+
+       /*
+        * create a proc entry for this device
+        * we need to do this a little bit differently than normal
+        * acpi device drivers because our device may not be present
+        * at the moment, and therefore we have no acpi_device struct
+        */
+
+       bay->proc = proc_mkdir(bay->name, acpi_bay_dir);
+
+       /* 'status' [R] */
+       entry = create_proc_entry("status",
+                                 S_IRUGO, bay->proc);
+       if (!entry)
+               return -EIO;
+       else {
+               entry->proc_fops = &acpi_bay_status_fops;
+               entry->data = bay;
+               entry->owner = THIS_MODULE;
+       }
+       /* 'eject' [W] */
+       entry = create_proc_entry("eject",
+                                 S_IWUGO, bay->proc);
+       if (!entry)
+               return -EIO;
+       else {
+               entry->proc_fops = &acpi_bay_eject_fops;
+               entry->data = bay;
+               entry->owner = THIS_MODULE;
+       }
+#if 0
+       /* 'insert' [W] */
+       entry = create_proc_entry("insert",
+                                 S_IWUGO, bay->proc);
+       if (!entry)
+               return -EIO;
+       else {
+               entry->proc_fops = &acpi_bay_insert_fops;
+               entry->data = bay;
+               entry->owner = THIS_MODULE;
+       }
+#endif
+       return 0;
+}
+
+static void acpi_bay_remove_fs(struct bay *bay)
+{
+       if (!bay)
+               return;
+
+       if (bay->proc) {
+               remove_proc_entry("status", bay->proc);
+               remove_proc_entry("eject", bay->proc);
+#if 0
+               remove_proc_entry("insert", bay->proc);
+#endif
+               remove_proc_entry(bay->name, acpi_bay_dir);
+               bay->proc = NULL;
+       }
+}
+
+static int bay_is_dock_device(acpi_handle handle)
+{
+       acpi_handle parent;
+
+       acpi_get_parent(handle, &parent);
+
+       /* if the device or it's parent is dependent on the
+        * dock, then we are a dock device
+        */
+       return (is_dock_device(handle) || is_dock_device(parent));
+}
+
+static int bay_add(acpi_handle handle)
+{
+       acpi_status status;
+       struct bay *new_bay;
+       struct acpi_buffer nbuffer = {ACPI_ALLOCATE_BUFFER, NULL};
+       acpi_get_name(handle, ACPI_FULL_PATHNAME, &nbuffer);
+
+       bay_dprintk(handle, "Adding notify handler");
+
+       /*
+        * if this is the first bay device found, make the root
+        * proc entry
+        */
+       if (acpi_bay_dir == NULL)
+               acpi_bay_dir = proc_mkdir(ACPI_BAY_CLASS, acpi_root_dir);
+
+       /*
+        * Initialize bay device structure
+        */
+       new_bay = kmalloc(GFP_ATOMIC, sizeof(*new_bay));
+       INIT_LIST_HEAD(&new_bay->list);
+       new_bay->handle = handle;
+       new_bay->name = (char *)nbuffer.pointer;
+       list_add(&new_bay->list, &drive_bays);
+       acpi_bay_add_fs(new_bay);
+
+       /* register for events on this device */
+       status = acpi_install_notify_handler(handle, ACPI_SYSTEM_NOTIFY,
+                       bay_notify, new_bay);
+       if (ACPI_FAILURE(status)) {
+               printk(KERN_ERR PREFIX "Error installing bay notify handler\n");
+       }
+
+       /* if we are on a dock station, we should register for dock
+        * notifications.
+        */
+       if (bay_is_dock_device(handle)) {
+               bay_dprintk(handle, "Is dependent on dock\n");
+               register_hotplug_dock_device(handle, bay_notify, new_bay);
+       }
+       printk(KERN_INFO PREFIX "Bay [%s] Added\n", new_bay->name);
+       return 0;
+}
+
+static int acpi_bay_remove(struct acpi_device *device, int type)
+{
+       /*** FIXME: do something here */
+       return 0;
+}
+
+static int acpi_bay_match(struct acpi_device *device,
+                               struct acpi_driver *driver)
+{
+       if (!device || !driver)
+               return -EINVAL;
+
+       if (is_ejectable_bay(device->handle)) {
+               bay_dprintk(device->handle, "matching bay device");
+               return 0;
+       }
+
+       return -ENODEV;
+}
+
+/**
+ * bay_create_acpi_device - add new devices to acpi
+ * @handle - handle of the device to add
+ *
+ *  This function will create a new acpi_device for the given
+ *  handle if one does not exist already.  This should cause
+ *  acpi to scan for drivers for the given devices, and call
+ *  matching driver's add routine.
+ *
+ *  Returns a pointer to the acpi_device corresponding to the handle.
+ */
+static struct acpi_device * bay_create_acpi_device(acpi_handle handle)
+{
+       struct acpi_device *device = NULL;
+       struct acpi_device *parent_device;
+       acpi_handle parent;
+       int ret;
+
+       bay_dprintk(handle, "Trying to get device");
+       if (acpi_bus_get_device(handle, &device)) {
+               /*
+                * no device created for this object,
+                * so we should create one.
+                */
+               bay_dprintk(handle, "No device for handle");
+               acpi_get_parent(handle, &parent);
+               if (acpi_bus_get_device(parent, &parent_device))
+                       parent_device = NULL;
+
+               ret = acpi_bus_add(&device, parent_device, handle,
+                       ACPI_BUS_TYPE_DEVICE);
+               if (ret) {
+                       pr_debug("error adding bus, %x\n",
+                               -ret);
+                       return NULL;
+               }
+       }
+       return device;
+}
+
+/**
+ * bay_notify - act upon an acpi bay notification
+ * @handle: the bay handle
+ * @event: the acpi event
+ * @data: our driver data struct
+ *
+ */
+static void bay_notify(acpi_handle handle, u32 event, void *data)
+{
+       struct acpi_device *dev;
+       struct bay *bay = data;
+
+       bay_dprintk(handle, "Bay event");
+
+       switch(event) {
+       case ACPI_NOTIFY_BUS_CHECK:
+               printk("Bus Check\n");
+       case ACPI_NOTIFY_DEVICE_CHECK:
+               printk("Device Check\n");
+               dev = bay_create_acpi_device(handle);
+               if (dev)
+                       acpi_bus_generate_event(dev, event, 0);
+               else
+                       printk("No device for generating event\n");
+               /* wouldn't it be a good idea to just rescan SATA
+                * right here?
+                */
+               break;
+       case ACPI_NOTIFY_EJECT_REQUEST:
+               printk("Eject request\n");
+               dev = bay_create_acpi_device(handle);
+               if (dev)
+                       acpi_bus_generate_event(dev, event, 0);
+               else
+                       printk("No device for generating eventn");
+
+               /* wouldn't it be a good idea to just call the
+                * eject_device here if we were a SATA device?
+                */
+               break;
+       default:
+               printk("unknown event %d\n", event);
+       }
+}
+
+static acpi_status
+find_bay(acpi_handle handle, u32 lvl, void *context, void **rv)
+{
+       int *count = (int *)context;
+
+       /*
+        * there could be more than one ejectable bay.
+        * so, just return AE_OK always so that every object
+        * will be checked.
+        */
+       if (is_ejectable_bay(handle)) {
+               bay_dprintk(handle, "found ejectable bay");
+               bay_add(handle);
+               (*count)++;
+       }
+       return AE_OK;
+}
+
+static int __init bay_init(void)
+{
+       int bays = 0;
+
+       acpi_bay_dir = NULL;
+       INIT_LIST_HEAD(&drive_bays);
+
+       /* look for dockable drive bays */
+       acpi_walk_namespace(ACPI_TYPE_DEVICE, ACPI_ROOT_OBJECT,
+               ACPI_UINT32_MAX, find_bay, &bays, NULL);
+
+       if (bays) {
+               if ((acpi_bus_register_driver(&acpi_bay_driver) < 0)) {
+                       printk(KERN_ERR "Unable to register bay driver\n");
+                       if (acpi_bay_dir)
+                               remove_proc_entry(ACPI_BAY_CLASS,
+                                       acpi_root_dir);
+               }
+       }
+
+       if (!bays)
+               return -ENODEV;
+
+       return 0;
+}
+
+static void __exit bay_exit(void)
+{
+       struct bay *bay, *tmp;
+
+       list_for_each_entry_safe(bay, tmp, &drive_bays, list) {
+               if (is_dock_device(bay->handle))
+                       unregister_hotplug_dock_device(bay->handle);
+               acpi_bay_remove_fs(bay);
+               acpi_remove_notify_handler(bay->handle, ACPI_SYSTEM_NOTIFY,
+                       bay_notify);
+               kfree(bay->name);
+               kfree(bay);
+       }
+
+       if (acpi_bay_dir)
+               remove_proc_entry(ACPI_BAY_CLASS, acpi_root_dir);
+
+       acpi_bus_unregister_driver(&acpi_bay_driver);
+}
+
+postcore_initcall(bay_init);
+module_exit(bay_exit);
+