#include <linux/highmem.h>
#include <linux/workqueue.h>
#include <linux/security.h>
+#include <linux/rcuref.h>
#include <asm/kmap_types.h>
#include <asm/uaccess.h>
#define dprintk(x...) do { ; } while (0)
#endif
-static long aio_run = 0; /* for testing only */
-static long aio_wakeups = 0; /* for testing only */
-
/*------ sysctl variables----*/
atomic_t aio_nr = ATOMIC_INIT(0); /* current system wide number of aio requests */
unsigned aio_max_nr = 0x10000; /* system wide maximum number of aio requests */
static LIST_HEAD(fput_head);
static void aio_kick_handler(void *);
+static void aio_queue_work(struct kioctx *);
/* aio_setup
* Creates the slab caches used by the aio routines, panic on
req->ki_ctx = ctx;
req->ki_cancel = NULL;
req->ki_retry = NULL;
- req->ki_obj.user = NULL;
req->ki_dtor = NULL;
req->private = NULL;
INIT_LIST_HEAD(&req->ki_run_list);
{
if (req->ki_dtor)
req->ki_dtor(req);
- req->ki_ctx = NULL;
- req->ki_filp = NULL;
- req->ki_obj.user = NULL;
- req->ki_dtor = NULL;
- req->private = NULL;
kmem_cache_free(kiocb_cachep, req);
ctx->reqs_active--;
/* Must be done under the lock to serialise against cancellation.
* Call this aio_fput as it duplicates fput via the fput_work.
*/
- if (unlikely(atomic_dec_and_test(&req->ki_filp->f_count))) {
+ if (unlikely(rcuref_dec_and_test(&req->ki_filp->f_count))) {
get_ioctx(ctx);
spin_lock(&fput_lock);
list_add(&req->ki_list, &fput_head);
return ioctx;
}
+static int lock_kiocb_action(void *param)
+{
+ schedule();
+ return 0;
+}
+
+static inline void lock_kiocb(struct kiocb *iocb)
+{
+ wait_on_bit_lock(&iocb->ki_flags, KIF_LOCKED, lock_kiocb_action,
+ TASK_UNINTERRUPTIBLE);
+}
+
+static inline void unlock_kiocb(struct kiocb *iocb)
+{
+ kiocbClearLocked(iocb);
+ smp_mb__after_clear_bit();
+ wake_up_bit(&iocb->ki_flags, KIF_LOCKED);
+}
+
/*
* use_mm
* Makes the calling kernel thread take on the specified
atomic_inc(&mm->mm_count);
tsk->mm = mm;
tsk->active_mm = mm;
+ /*
+ * Note that on UML this *requires* PF_BORROWED_MM to be set, otherwise
+ * it won't work. Update it accordingly if you change it here
+ */
activate_mm(active_mm, mm);
task_unlock(tsk);
if (list_empty(&iocb->ki_run_list)) {
list_add_tail(&iocb->ki_run_list,
&ctx->run_list);
- iocb->ki_queued++;
return 1;
}
return 0;
}
if (!(iocb->ki_retried & 0xff)) {
- pr_debug("%ld retry: %d of %d (kick %ld, Q %ld run %ld, wake %ld)\n",
- iocb->ki_retried,
- iocb->ki_nbytes - iocb->ki_left, iocb->ki_nbytes,
- iocb->ki_kicked, iocb->ki_queued, aio_run, aio_wakeups);
+ pr_debug("%ld retry: %d of %d\n", iocb->ki_retried,
+ iocb->ki_nbytes - iocb->ki_left, iocb->ki_nbytes);
}
if (!(retry = iocb->ki_retry)) {
ret = retry(iocb);
current->io_wait = NULL;
- if (-EIOCBRETRY != ret) {
- if (-EIOCBQUEUED != ret) {
- BUG_ON(!list_empty(&iocb->ki_wait.task_list));
- aio_complete(iocb, ret, 0);
- /* must not access the iocb after this */
- }
- } else {
- /*
- * Issue an additional retry to avoid waiting forever if
- * no waits were queued (e.g. in case of a short read).
- */
- if (list_empty(&iocb->ki_wait.task_list))
- kiocbSetKicked(iocb);
+ if (ret != -EIOCBRETRY && ret != -EIOCBQUEUED) {
+ BUG_ON(!list_empty(&iocb->ki_wait.task_list));
+ aio_complete(iocb, ret, 0);
}
out:
spin_lock_irq(&ctx->ctx_lock);
* has already been kicked */
if (kiocbIsKicked(iocb)) {
__queue_kicked_iocb(iocb);
+
+ /*
+ * __queue_kicked_iocb will always return 1 here, because
+ * iocb->ki_run_list is empty at this point so it should
+ * be safe to unconditionally queue the context into the
+ * work queue.
+ */
+ aio_queue_work(ctx);
}
}
return ret;
static int __aio_run_iocbs(struct kioctx *ctx)
{
struct kiocb *iocb;
- int count = 0;
LIST_HEAD(run_list);
list_splice_init(&ctx->run_list, &run_list);
* Hold an extra reference while retrying i/o.
*/
iocb->ki_users++; /* grab extra reference */
+ lock_kiocb(iocb);
aio_run_iocb(iocb);
+ unlock_kiocb(iocb);
if (__aio_put_req(ctx, iocb)) /* drop extra ref */
put_ioctx(ctx);
- count++;
}
- aio_run++;
if (!list_empty(&ctx->run_list))
return 1;
return 0;
* and if required activate the aio work queue to process
* it
*/
-static void queue_kicked_iocb(struct kiocb *iocb)
+static void try_queue_kicked_iocb(struct kiocb *iocb)
{
struct kioctx *ctx = iocb->ki_ctx;
unsigned long flags;
int run = 0;
- WARN_ON((!list_empty(&iocb->ki_wait.task_list)));
+ /* We're supposed to be the only path putting the iocb back on the run
+ * list. If we find that the iocb is *back* on a wait queue already
+ * than retry has happened before we could queue the iocb. This also
+ * means that the retry could have completed and freed our iocb, no
+ * good. */
+ BUG_ON((!list_empty(&iocb->ki_wait.task_list)));
spin_lock_irqsave(&ctx->ctx_lock, flags);
- run = __queue_kicked_iocb(iocb);
+ /* set this inside the lock so that we can't race with aio_run_iocb()
+ * testing it and putting the iocb on the run list under the lock */
+ if (!kiocbTryKick(iocb))
+ run = __queue_kicked_iocb(iocb);
spin_unlock_irqrestore(&ctx->ctx_lock, flags);
- if (run) {
+ if (run)
aio_queue_work(ctx);
- aio_wakeups++;
- }
}
/*
return;
}
- iocb->ki_kicked++;
- /* If its already kicked we shouldn't queue it again */
- if (!kiocbTryKick(iocb)) {
- queue_kicked_iocb(iocb);
- }
+ try_queue_kicked_iocb(iocb);
}
EXPORT_SYMBOL(kick_iocb);
tail = info->tail;
event = aio_ring_event(info, tail, KM_IRQ0);
- tail = (tail + 1) % info->nr;
+ if (++tail >= info->nr)
+ tail = 0;
event->obj = (u64)(unsigned long)iocb->ki_obj.user;
event->data = iocb->ki_user_data;
pr_debug("added to ring %p at [%lu]\n", iocb, tail);
- pr_debug("%ld retries: %d of %d (kicked %ld, Q %ld run %ld wake %ld)\n",
- iocb->ki_retried,
- iocb->ki_nbytes - iocb->ki_left, iocb->ki_nbytes,
- iocb->ki_kicked, iocb->ki_queued, aio_run, aio_wakeups);
+ pr_debug("%ld retries: %d of %d\n", iocb->ki_retried,
+ iocb->ki_nbytes - iocb->ki_left, iocb->ki_nbytes);
put_rq:
/* everything turned out well, dispose of the aiocb. */
ret = __aio_put_req(ctx, iocb);
int i = 0;
struct io_event ent;
struct aio_timeout to;
- int event_loop = 0; /* testing only */
int retry = 0;
/* needed to zero any padding within an entry (there shouldn't be
if (to.timed_out) /* Only check after read evt */
break;
schedule();
- event_loop++;
if (signal_pending(tsk)) {
ret = -EINTR;
break;
if (timeout)
clear_timeout(&to);
out:
- pr_debug("event loop executed %d times\n", event_loop);
- pr_debug("aio_run %ld\n", aio_run);
- pr_debug("aio_wakeups %ld\n", aio_wakeups);
return i ? i : ret;
}
}
/*
- * Default retry method for aio_read (also used for first time submit)
- * Responsible for updating iocb state as retries progress
+ * aio_p{read,write} are the default ki_retry methods for
+ * IO_CMD_P{READ,WRITE}. They maintains kiocb retry state around potentially
+ * multiple calls to f_op->aio_read(). They loop around partial progress
+ * instead of returning -EIOCBRETRY because they don't have the means to call
+ * kick_iocb().
*/
static ssize_t aio_pread(struct kiocb *iocb)
{
struct inode *inode = mapping->host;
ssize_t ret = 0;
- ret = file->f_op->aio_read(iocb, iocb->ki_buf,
- iocb->ki_left, iocb->ki_pos);
+ do {
+ ret = file->f_op->aio_read(iocb, iocb->ki_buf,
+ iocb->ki_left, iocb->ki_pos);
+ /*
+ * Can't just depend on iocb->ki_left to determine
+ * whether we are done. This may have been a short read.
+ */
+ if (ret > 0) {
+ iocb->ki_buf += ret;
+ iocb->ki_left -= ret;
+ }
- /*
- * Can't just depend on iocb->ki_left to determine
- * whether we are done. This may have been a short read.
- */
- if (ret > 0) {
- iocb->ki_buf += ret;
- iocb->ki_left -= ret;
/*
- * For pipes and sockets we return once we have
- * some data; for regular files we retry till we
- * complete the entire read or find that we can't
- * read any more data (e.g short reads).
+ * For pipes and sockets we return once we have some data; for
+ * regular files we retry till we complete the entire read or
+ * find that we can't read any more data (e.g short reads).
*/
- if (!S_ISFIFO(inode->i_mode) && !S_ISSOCK(inode->i_mode))
- ret = -EIOCBRETRY;
- }
+ } while (ret > 0 && iocb->ki_left > 0 &&
+ !S_ISFIFO(inode->i_mode) && !S_ISSOCK(inode->i_mode));
/* This means we must have transferred all that we could */
/* No need to retry anymore */
return ret;
}
-/*
- * Default retry method for aio_write (also used for first time submit)
- * Responsible for updating iocb state as retries progress
- */
+/* see aio_pread() */
static ssize_t aio_pwrite(struct kiocb *iocb)
{
struct file *file = iocb->ki_filp;
ssize_t ret = 0;
- ret = file->f_op->aio_write(iocb, iocb->ki_buf,
- iocb->ki_left, iocb->ki_pos);
-
- if (ret > 0) {
- iocb->ki_buf += ret;
- iocb->ki_left -= ret;
-
- ret = -EIOCBRETRY;
- }
+ do {
+ ret = file->f_op->aio_write(iocb, iocb->ki_buf,
+ iocb->ki_left, iocb->ki_pos);
+ if (ret > 0) {
+ iocb->ki_buf += ret;
+ iocb->ki_left -= ret;
+ }
+ } while (ret > 0 && iocb->ki_left > 0);
- /* This means we must have transferred all that we could */
- /* No need to retry anymore */
if ((ret == 0) || (iocb->ki_left == 0))
ret = iocb->ki_nbytes - iocb->ki_left;
}
req->ki_filp = file;
- iocb->aio_key = req->ki_key;
- ret = put_user(iocb->aio_key, &user_iocb->aio_key);
+ ret = put_user(req->ki_key, &user_iocb->aio_key);
if (unlikely(ret)) {
dprintk("EFAULT: aio_key\n");
goto out_put_req;
req->ki_opcode = iocb->aio_lio_opcode;
init_waitqueue_func_entry(&req->ki_wait, aio_wake_function);
INIT_LIST_HEAD(&req->ki_wait.task_list);
- req->ki_run_list.next = req->ki_run_list.prev = NULL;
- req->ki_retry = NULL;
req->ki_retried = 0;
- req->ki_kicked = 0;
- req->ki_queued = 0;
- aio_run = 0;
- aio_wakeups = 0;
ret = aio_setup_iocb(req);
goto out_put_req;
spin_lock_irq(&ctx->ctx_lock);
- list_add_tail(&req->ki_run_list, &ctx->run_list);
- /* drain the run list */
- while (__aio_run_iocbs(ctx))
- ;
+ aio_run_iocb(req);
+ unlock_kiocb(req);
+ if (!list_empty(&ctx->run_list)) {
+ /* drain the run list */
+ while (__aio_run_iocbs(ctx))
+ ;
+ }
spin_unlock_irq(&ctx->ctx_lock);
aio_put_req(req); /* drop extra ref to req */
return 0;
if (NULL != cancel) {
struct io_event tmp;
pr_debug("calling cancel\n");
+ lock_kiocb(kiocb);
memset(&tmp, 0, sizeof(tmp));
tmp.obj = (u64)(unsigned long)kiocb->ki_obj.user;
tmp.data = kiocb->ki_user_data;
if (copy_to_user(result, &tmp, sizeof(tmp)))
ret = -EFAULT;
}
+ unlock_kiocb(kiocb);
} else
- printk(KERN_DEBUG "iocb has no cancel operation\n");
+ ret = -EINVAL;
put_ioctx(ctx);
git.openpandora.org / pandora-kernel.git / blobdiff