From 53c165e0a6c8a4ff7df316557528fa7a52d20711 Mon Sep 17 00:00:00 2001 From: James Bottomley Date: Mon, 22 Aug 2005 10:06:19 -0500 Subject: [SCSI] correct attribute_container list usage One of the changes in the attribute_container code in the scsi-misc tree was to add a lock to protect the list of devices per container. This, unfortunately, leads to potential scheduling while atomic problems if there's a sleep in the function called by a trigger. The correct solution is to use the kernel klist infrastructure instead which allows lockless traversal of a list. Signed-off-by: James Bottomley --- include/linux/attribute_container.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include') diff --git a/include/linux/attribute_container.h b/include/linux/attribute_container.h index ee83fe64a102..93bfb0beb62a 100644 --- a/include/linux/attribute_container.h +++ b/include/linux/attribute_container.h @@ -11,12 +11,12 @@ #include #include +#include #include struct attribute_container { struct list_head node; - struct list_head containers; - spinlock_t containers_lock; + struct klist containers; struct class *class; struct class_device_attribute **attrs; int (*match)(struct attribute_container *, struct device *); -- cgit v1.2.3 From 61a7afa2c476a3be261cf88a95b0dea0c3bd29d4 Mon Sep 17 00:00:00 2001 From: James Bottomley Date: Tue, 16 Aug 2005 18:27:34 -0500 Subject: [SCSI] embryonic RAID class The idea behind a RAID class is to provide a uniform interface to all RAID subsystems (both hardware and software) in the kernel. To do that, I've made this class a transport class that's entirely subsystem independent (although the matching routines have to match per subsystem, as you'll see looking at the code). I put it in the scsi subdirectory purely because I needed somewhere to play with it, but it's not a scsi specific module. I used a fusion raid card as the test bed for this; with that kind of card, this is the type of class output you get: jejb@titanic> ls -l /sys/class/raid_devices/20\:0\:0\:0/ total 0 lrwxrwxrwx 1 root root 0 Aug 16 17:21 component-0 -> ../../../devices/pci0000:80/0000:80:04.0/host20/target20:1:0/20:1:0:0/ lrwxrwxrwx 1 root root 0 Aug 16 17:21 component-1 -> ../../../devices/pci0000:80/0000:80:04.0/host20/target20:1:1/20:1:1:0/ lrwxrwxrwx 1 root root 0 Aug 16 17:21 device -> ../../../devices/pci0000:80/0000:80:04.0/host20/target20:0:0/20:0:0:0/ -r--r--r-- 1 root root 16384 Aug 16 17:21 level -r--r--r-- 1 root root 16384 Aug 16 17:21 resync -r--r--r-- 1 root root 16384 Aug 16 17:21 state So it's really simple: for a SCSI device representing a hardware raid, it shows the raid level, the array state, the resync % complete (if the state is resyncing) and the underlying components of the RAID (these are exposed in fusion on the virtual channel 1). As you can see, this type of information can be exported by almost anything, including software raid. The more difficult trick, of course, is going to be getting it to perform configuration type actions with writable attributes. Signed-off-by: James Bottomley --- drivers/scsi/Kconfig | 6 ++ drivers/scsi/Makefile | 2 + drivers/scsi/raid_class.c | 250 +++++++++++++++++++++++++++++++++++++++++++++ include/linux/raid_class.h | 59 +++++++++++ 4 files changed, 317 insertions(+) create mode 100644 drivers/scsi/raid_class.c create mode 100644 include/linux/raid_class.h (limited to 'include') diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig index 96df148ed969..68adc3cc8ad2 100644 --- a/drivers/scsi/Kconfig +++ b/drivers/scsi/Kconfig @@ -1,5 +1,11 @@ menu "SCSI device support" +config RAID_ATTRS + tristate "RAID Transport Class" + default n + ---help--- + Provides RAID + config SCSI tristate "SCSI device support" ---help--- diff --git a/drivers/scsi/Makefile b/drivers/scsi/Makefile index 3746fb9fa2f5..85f9e6bb34b9 100644 --- a/drivers/scsi/Makefile +++ b/drivers/scsi/Makefile @@ -22,6 +22,8 @@ subdir-$(CONFIG_PCMCIA) += pcmcia obj-$(CONFIG_SCSI) += scsi_mod.o +obj-$(CONFIG_RAID_ATTRS) += raid_class.o + # --- NOTE ORDERING HERE --- # For kernel non-modular link, transport attributes need to # be initialised before drivers diff --git a/drivers/scsi/raid_class.c b/drivers/scsi/raid_class.c new file mode 100644 index 000000000000..f1ea5027865f --- /dev/null +++ b/drivers/scsi/raid_class.c @@ -0,0 +1,250 @@ +/* + * RAID Attributes + */ +#include +#include +#include +#include +#include +#include + +#define RAID_NUM_ATTRS 3 + +struct raid_internal { + struct raid_template r; + struct raid_function_template *f; + /* The actual attributes */ + struct class_device_attribute private_attrs[RAID_NUM_ATTRS]; + /* The array of null terminated pointers to attributes + * needed by scsi_sysfs.c */ + struct class_device_attribute *attrs[RAID_NUM_ATTRS + 1]; +}; + +struct raid_component { + struct list_head node; + struct device *dev; + int num; +}; + +#define to_raid_internal(tmpl) container_of(tmpl, struct raid_internal, r) + +#define tc_to_raid_internal(tcont) ({ \ + struct raid_template *r = \ + container_of(tcont, struct raid_template, raid_attrs); \ + to_raid_internal(r); \ +}) + +#define ac_to_raid_internal(acont) ({ \ + struct transport_container *tc = \ + container_of(acont, struct transport_container, ac); \ + tc_to_raid_internal(tc); \ +}) + +#define class_device_to_raid_internal(cdev) ({ \ + struct attribute_container *ac = \ + attribute_container_classdev_to_container(cdev); \ + ac_to_raid_internal(ac); \ +}) + + +static int raid_match(struct attribute_container *cont, struct device *dev) +{ + /* We have to look for every subsystem that could house + * emulated RAID devices, so start with SCSI */ + struct raid_internal *i = ac_to_raid_internal(cont); + + if (scsi_is_sdev_device(dev)) { + struct scsi_device *sdev = to_scsi_device(dev); + + if (i->f->cookie != sdev->host->hostt) + return 0; + + return i->f->is_raid(dev); + } + /* FIXME: look at other subsystems too */ + return 0; +} + +static int raid_setup(struct transport_container *tc, struct device *dev, + struct class_device *cdev) +{ + struct raid_data *rd; + + BUG_ON(class_get_devdata(cdev)); + + rd = kmalloc(sizeof(*rd), GFP_KERNEL); + if (!rd) + return -ENOMEM; + + memset(rd, 0, sizeof(*rd)); + INIT_LIST_HEAD(&rd->component_list); + class_set_devdata(cdev, rd); + + return 0; +} + +static int raid_remove(struct transport_container *tc, struct device *dev, + struct class_device *cdev) +{ + struct raid_data *rd = class_get_devdata(cdev); + struct raid_component *rc, *next; + class_set_devdata(cdev, NULL); + list_for_each_entry_safe(rc, next, &rd->component_list, node) { + char buf[40]; + snprintf(buf, sizeof(buf), "component-%d", rc->num); + list_del(&rc->node); + sysfs_remove_link(&cdev->kobj, buf); + kfree(rc); + } + kfree(class_get_devdata(cdev)); + return 0; +} + +static DECLARE_TRANSPORT_CLASS(raid_class, + "raid_devices", + raid_setup, + raid_remove, + NULL); + +static struct { + enum raid_state value; + char *name; +} raid_states[] = { + { RAID_ACTIVE, "active" }, + { RAID_DEGRADED, "degraded" }, + { RAID_RESYNCING, "resyncing" }, + { RAID_OFFLINE, "offline" }, +}; + +static const char *raid_state_name(enum raid_state state) +{ + int i; + char *name = NULL; + + for (i = 0; i < sizeof(raid_states)/sizeof(raid_states[0]); i++) { + if (raid_states[i].value == state) { + name = raid_states[i].name; + break; + } + } + return name; +} + + +#define raid_attr_show_internal(attr, fmt, var, code) \ +static ssize_t raid_show_##attr(struct class_device *cdev, char *buf) \ +{ \ + struct raid_data *rd = class_get_devdata(cdev); \ + code \ + return snprintf(buf, 20, #fmt "\n", var); \ +} + +#define raid_attr_ro_states(attr, states, code) \ +raid_attr_show_internal(attr, %s, name, \ + const char *name; \ + code \ + name = raid_##states##_name(rd->attr); \ +) \ +static CLASS_DEVICE_ATTR(attr, S_IRUGO, raid_show_##attr, NULL) + + +#define raid_attr_ro_internal(attr, code) \ +raid_attr_show_internal(attr, %d, rd->attr, code) \ +static CLASS_DEVICE_ATTR(attr, S_IRUGO, raid_show_##attr, NULL) + +#define ATTR_CODE(attr) \ + struct raid_internal *i = class_device_to_raid_internal(cdev); \ + if (i->f->get_##attr) \ + i->f->get_##attr(cdev->dev); + +#define raid_attr_ro(attr) raid_attr_ro_internal(attr, ) +#define raid_attr_ro_fn(attr) raid_attr_ro_internal(attr, ATTR_CODE(attr)) +#define raid_attr_ro_state(attr) raid_attr_ro_states(attr, attr, ATTR_CODE(attr)) + +raid_attr_ro(level); +raid_attr_ro_fn(resync); +raid_attr_ro_state(state); + +void raid_component_add(struct raid_template *r,struct device *raid_dev, + struct device *component_dev) +{ + struct class_device *cdev = + attribute_container_find_class_device(&r->raid_attrs.ac, + raid_dev); + struct raid_component *rc; + struct raid_data *rd = class_get_devdata(cdev); + char buf[40]; + + rc = kmalloc(sizeof(*rc), GFP_KERNEL); + if (!rc) + return; + + INIT_LIST_HEAD(&rc->node); + rc->dev = component_dev; + rc->num = rd->component_count++; + + snprintf(buf, sizeof(buf), "component-%d", rc->num); + list_add_tail(&rc->node, &rd->component_list); + sysfs_create_link(&cdev->kobj, &component_dev->kobj, buf); +} +EXPORT_SYMBOL(raid_component_add); + +struct raid_template * +raid_class_attach(struct raid_function_template *ft) +{ + struct raid_internal *i = kmalloc(sizeof(struct raid_internal), + GFP_KERNEL); + int count = 0; + + if (unlikely(!i)) + return NULL; + + memset(i, 0, sizeof(*i)); + + i->f = ft; + + i->r.raid_attrs.ac.class = &raid_class.class; + i->r.raid_attrs.ac.match = raid_match; + i->r.raid_attrs.ac.attrs = &i->attrs[0]; + + attribute_container_register(&i->r.raid_attrs.ac); + + i->attrs[count++] = &class_device_attr_level; + i->attrs[count++] = &class_device_attr_resync; + i->attrs[count++] = &class_device_attr_state; + + i->attrs[count] = NULL; + BUG_ON(count > RAID_NUM_ATTRS); + + return &i->r; +} +EXPORT_SYMBOL(raid_class_attach); + +void +raid_class_release(struct raid_template *r) +{ + struct raid_internal *i = to_raid_internal(r); + + attribute_container_unregister(&i->r.raid_attrs.ac); + + kfree(i); +} +EXPORT_SYMBOL(raid_class_release); + +static __init int raid_init(void) +{ + return transport_class_register(&raid_class); +} + +static __exit void raid_exit(void) +{ + transport_class_unregister(&raid_class); +} + +MODULE_AUTHOR("James Bottomley"); +MODULE_DESCRIPTION("RAID device class"); +MODULE_LICENSE("GPL"); + +module_init(raid_init); +module_exit(raid_exit); + diff --git a/include/linux/raid_class.h b/include/linux/raid_class.h new file mode 100644 index 000000000000..a71123c28272 --- /dev/null +++ b/include/linux/raid_class.h @@ -0,0 +1,59 @@ +/* + */ +#include + +struct raid_template { + struct transport_container raid_attrs; +}; + +struct raid_function_template { + void *cookie; + int (*is_raid)(struct device *); + void (*get_resync)(struct device *); + void (*get_state)(struct device *); +}; + +enum raid_state { + RAID_ACTIVE = 1, + RAID_DEGRADED, + RAID_RESYNCING, + RAID_OFFLINE, +}; + +struct raid_data { + struct list_head component_list; + int component_count; + int level; + enum raid_state state; + int resync; +}; + +#define DEFINE_RAID_ATTRIBUTE(type, attr) \ +static inline void \ +raid_set_##attr(struct raid_template *r, struct device *dev, type value) { \ + struct class_device *cdev = \ + attribute_container_find_class_device(&r->raid_attrs.ac, dev);\ + struct raid_data *rd; \ + BUG_ON(!cdev); \ + rd = class_get_devdata(cdev); \ + rd->attr = value; \ +} \ +static inline type \ +raid_get_##attr(struct raid_template *r, struct device *dev) { \ + struct class_device *cdev = \ + attribute_container_find_class_device(&r->raid_attrs.ac, dev);\ + struct raid_data *rd; \ + BUG_ON(!cdev); \ + rd = class_get_devdata(cdev); \ + return rd->attr; \ +} + +DEFINE_RAID_ATTRIBUTE(int, level) +DEFINE_RAID_ATTRIBUTE(int, resync) +DEFINE_RAID_ATTRIBUTE(enum raid_state, state) + +struct raid_template *raid_class_attach(struct raid_function_template *); +void raid_class_release(struct raid_template *); + +void raid_component_add(struct raid_template *, struct device *, + struct device *); -- cgit v1.2.3 From c5478def7a3a2dba9ceda452c2aa3539514d30a9 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 6 Sep 2005 14:04:26 +0200 Subject: [SCSI] switch EH thread startup to the kthread API Signed-off-by: James Bottomley --- drivers/scsi/hosts.c | 23 ++++++++--------------- drivers/scsi/scsi_error.c | 29 ++--------------------------- include/scsi/scsi_host.h | 2 -- 3 files changed, 10 insertions(+), 44 deletions(-) (limited to 'include') diff --git a/drivers/scsi/hosts.c b/drivers/scsi/hosts.c index 8640ad1c17e2..85503fad789a 100644 --- a/drivers/scsi/hosts.c +++ b/drivers/scsi/hosts.c @@ -24,6 +24,7 @@ #include #include #include +#include #include #include #include @@ -225,15 +226,8 @@ static void scsi_host_dev_release(struct device *dev) struct Scsi_Host *shost = dev_to_shost(dev); struct device *parent = dev->parent; - if (shost->ehandler) { - DECLARE_COMPLETION(sem); - shost->eh_notify = &sem; - shost->eh_kill = 1; - up(shost->eh_wait); - wait_for_completion(&sem); - shost->eh_notify = NULL; - } - + if (shost->ehandler) + kthread_stop(shost->ehandler); if (shost->work_q) destroy_workqueue(shost->work_q); @@ -263,7 +257,6 @@ struct Scsi_Host *scsi_host_alloc(struct scsi_host_template *sht, int privsize) { struct Scsi_Host *shost; int gfp_mask = GFP_KERNEL, rval; - DECLARE_COMPLETION(complete); if (sht->unchecked_isa_dma && privsize) gfp_mask |= __GFP_DMA; @@ -369,12 +362,12 @@ struct Scsi_Host *scsi_host_alloc(struct scsi_host_template *sht, int privsize) snprintf(shost->shost_classdev.class_id, BUS_ID_SIZE, "host%d", shost->host_no); - shost->eh_notify = &complete; - rval = kernel_thread(scsi_error_handler, shost, 0); - if (rval < 0) + shost->ehandler = kthread_run(scsi_error_handler, shost, + "scsi_eh_%d", shost->host_no); + if (IS_ERR(shost->ehandler)) { + rval = PTR_ERR(shost->ehandler); goto fail_destroy_freelist; - wait_for_completion(&complete); - shost->eh_notify = NULL; + } scsi_proc_hostdir_add(shost->hostt); return shost; diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c index 688bce740786..ebe74ccb518a 100644 --- a/drivers/scsi/scsi_error.c +++ b/drivers/scsi/scsi_error.c @@ -20,6 +20,7 @@ #include #include #include +#include #include #include #include @@ -1585,16 +1586,8 @@ int scsi_error_handler(void *data) int rtn; DECLARE_MUTEX_LOCKED(sem); - /* - * Flush resources - */ - - daemonize("scsi_eh_%d", shost->host_no); - current->flags |= PF_NOFREEZE; - shost->eh_wait = &sem; - shost->ehandler = current; /* * Wake up the thread that created us. @@ -1602,8 +1595,6 @@ int scsi_error_handler(void *data) SCSI_LOG_ERROR_RECOVERY(3, printk("Wake up parent of" " scsi_eh_%d\n",shost->host_no)); - complete(shost->eh_notify); - while (1) { /* * If we get a signal, it means we are supposed to go @@ -1624,7 +1615,7 @@ int scsi_error_handler(void *data) * semaphores isn't unreasonable. */ down_interruptible(&sem); - if (shost->eh_kill) + if (kthread_should_stop()) break; SCSI_LOG_ERROR_RECOVERY(1, printk("Error handler" @@ -1663,22 +1654,6 @@ int scsi_error_handler(void *data) * Make sure that nobody tries to wake us up again. */ shost->eh_wait = NULL; - - /* - * Knock this down too. From this point on, the host is flying - * without a pilot. If this is because the module is being unloaded, - * that's fine. If the user sent a signal to this thing, we are - * potentially in real danger. - */ - shost->eh_active = 0; - shost->ehandler = NULL; - - /* - * If anyone is waiting for us to exit (i.e. someone trying to unload - * a driver), then wake up that process to let them know we are on - * the way out the door. - */ - complete_and_exit(shost->eh_notify, 0); return 0; } diff --git a/include/scsi/scsi_host.h b/include/scsi/scsi_host.h index ac1b6125e3ae..916144be208b 100644 --- a/include/scsi/scsi_host.h +++ b/include/scsi/scsi_host.h @@ -467,12 +467,10 @@ struct Scsi_Host { struct task_struct * ehandler; /* Error recovery thread. */ struct semaphore * eh_wait; /* The error recovery thread waits on this. */ - struct completion * eh_notify; /* wait for eh to begin or end */ struct semaphore * eh_action; /* Wait for specific actions on the host. */ unsigned int eh_active:1; /* Indicates the eh thread is awake and active if this is true. */ - unsigned int eh_kill:1; /* set when killing the eh thread */ wait_queue_head_t host_wait; struct scsi_host_template *hostt; struct scsi_transport_template *transportt; -- cgit v1.2.3 From fe1b2d544d71300f8e2d151c3c77a130d13a58be Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 6 Sep 2005 14:15:37 +0200 Subject: [SCSI] unexport scsi_add_timer/scsi_delete_timer Signed-off-by: James Bottomley --- Documentation/scsi/scsi_mid_low_api.txt | 41 --------------------------------- drivers/scsi/scsi_error.c | 2 -- drivers/scsi/scsi_priv.h | 3 +++ include/scsi/scsi_eh.h | 3 --- 4 files changed, 3 insertions(+), 46 deletions(-) (limited to 'include') diff --git a/Documentation/scsi/scsi_mid_low_api.txt b/Documentation/scsi/scsi_mid_low_api.txt index 7536823c0cb1..44df89c9c049 100644 --- a/Documentation/scsi/scsi_mid_low_api.txt +++ b/Documentation/scsi/scsi_mid_low_api.txt @@ -373,13 +373,11 @@ Summary: scsi_activate_tcq - turn on tag command queueing scsi_add_device - creates new scsi device (lu) instance scsi_add_host - perform sysfs registration and SCSI bus scan. - scsi_add_timer - (re-)start timer on a SCSI command. scsi_adjust_queue_depth - change the queue depth on a SCSI device scsi_assign_lock - replace default host_lock with given lock scsi_bios_ptable - return copy of block device's partition table scsi_block_requests - prevent further commands being queued to given host scsi_deactivate_tcq - turn off tag command queueing - scsi_delete_timer - cancel timer on a SCSI command. scsi_host_alloc - return a new scsi_host instance whose refcount==1 scsi_host_get - increments Scsi_Host instance's refcount scsi_host_put - decrements Scsi_Host instance's refcount (free if 0) @@ -457,27 +455,6 @@ struct scsi_device * scsi_add_device(struct Scsi_Host *shost, int scsi_add_host(struct Scsi_Host *shost, struct device * dev) -/** - * scsi_add_timer - (re-)start timer on a SCSI command. - * @scmd: pointer to scsi command instance - * @timeout: duration of timeout in "jiffies" - * @complete: pointer to function to call if timeout expires - * - * Returns nothing - * - * Might block: no - * - * Notes: Each scsi command has its own timer, and as it is added - * to the queue, we set up the timer. When the command completes, - * we cancel the timer. An LLD can use this function to change - * the existing timeout value. - * - * Defined in: drivers/scsi/scsi_error.c - **/ -void scsi_add_timer(struct scsi_cmnd *scmd, int timeout, - void (*complete)(struct scsi_cmnd *)) - - /** * scsi_adjust_queue_depth - allow LLD to change queue depth on a SCSI device * @sdev: pointer to SCSI device to change queue depth on @@ -565,24 +542,6 @@ void scsi_block_requests(struct Scsi_Host * shost) void scsi_deactivate_tcq(struct scsi_device *sdev, int depth) -/** - * scsi_delete_timer - cancel timer on a SCSI command. - * @scmd: pointer to scsi command instance - * - * Returns 1 if able to cancel timer else 0 (i.e. too late or already - * cancelled). - * - * Might block: no [may in the future if it invokes del_timer_sync()] - * - * Notes: All commands issued by upper levels already have a timeout - * associated with them. An LLD can use this function to cancel the - * timer. - * - * Defined in: drivers/scsi/scsi_error.c - **/ -int scsi_delete_timer(struct scsi_cmnd *scmd) - - /** * scsi_host_alloc - create a scsi host adapter instance and perform basic * initialization. diff --git a/drivers/scsi/scsi_error.c b/drivers/scsi/scsi_error.c index ebe74ccb518a..ae28bcb79242 100644 --- a/drivers/scsi/scsi_error.c +++ b/drivers/scsi/scsi_error.c @@ -116,7 +116,6 @@ void scsi_add_timer(struct scsi_cmnd *scmd, int timeout, add_timer(&scmd->eh_timeout); } -EXPORT_SYMBOL(scsi_add_timer); /** * scsi_delete_timer - Delete/cancel timer for a given function. @@ -144,7 +143,6 @@ int scsi_delete_timer(struct scsi_cmnd *scmd) return rtn; } -EXPORT_SYMBOL(scsi_delete_timer); /** * scsi_times_out - Timeout function for normal scsi commands. diff --git a/drivers/scsi/scsi_priv.h b/drivers/scsi/scsi_priv.h index d30d7f4e63ec..ee6de1768e53 100644 --- a/drivers/scsi/scsi_priv.h +++ b/drivers/scsi/scsi_priv.h @@ -63,6 +63,9 @@ extern int __init scsi_init_devinfo(void); extern void scsi_exit_devinfo(void); /* scsi_error.c */ +extern void scsi_add_timer(struct scsi_cmnd *, int, + void (*)(struct scsi_cmnd *)); +extern int scsi_delete_timer(struct scsi_cmnd *); extern void scsi_times_out(struct scsi_cmnd *cmd); extern int scsi_error_handler(void *host); extern int scsi_decide_disposition(struct scsi_cmnd *cmd); diff --git a/include/scsi/scsi_eh.h b/include/scsi/scsi_eh.h index 80557f879e3e..4b71095be681 100644 --- a/include/scsi/scsi_eh.h +++ b/include/scsi/scsi_eh.h @@ -27,9 +27,6 @@ struct scsi_sense_hdr { /* See SPC-3 section 4.5 */ }; -extern void scsi_add_timer(struct scsi_cmnd *, int, - void (*)(struct scsi_cmnd *)); -extern int scsi_delete_timer(struct scsi_cmnd *); extern void scsi_report_bus_reset(struct Scsi_Host *, int); extern void scsi_report_device_reset(struct Scsi_Host *, int, int); extern int scsi_block_when_processing_errors(struct scsi_device *); -- cgit v1.2.3