#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/pagemap.h>
+#include <linux/task_io_accounting_ops.h>
#include <linux/bio.h>
#include <linux/wait.h>
#include <linux/err.h>
NULL); /* vmas */
up_read(¤t->mm->mmap_sem);
- if (ret < 0 && dio->blocks_available && (dio->rw == WRITE)) {
+ if (ret < 0 && dio->blocks_available && (dio->rw & WRITE)) {
struct page *page = ZERO_PAGE(dio->curr_user_address);
/*
* A memory fault, but the filesystem has some outstanding
return dio->pages[dio->head++];
}
-/*
- * Called when all DIO BIO I/O has been completed - let the filesystem
- * know, if it registered an interest earlier via get_block. Pass the
- * private field of the map buffer_head so that filesystems can use it
- * to hold additional state between get_block calls and dio_complete.
+/**
+ * dio_complete() - called when all DIO BIO I/O has been completed
+ * @offset: the byte offset in the file of the completed operation
+ *
+ * This releases locks as dictated by the locking type, lets interested parties
+ * know that a DIO operation has completed, and calculates the resulting return
+ * code for the operation.
+ *
+ * It lets the filesystem know if it registered an interest earlier via
+ * get_block. Pass the private field of the map buffer_head so that
+ * filesystems can use it to hold additional state between get_block calls and
+ * dio_complete.
*/
-static void dio_complete(struct dio *dio, loff_t offset, ssize_t bytes)
+static int dio_complete(struct dio *dio, loff_t offset, int ret)
{
+ ssize_t transferred = 0;
+
+ if (dio->result) {
+ transferred = dio->result;
+
+ /* Check for short read case */
+ if ((dio->rw == READ) && ((offset + transferred) > dio->i_size))
+ transferred = dio->i_size - offset;
+ }
+
if (dio->end_io && dio->result)
- dio->end_io(dio->iocb, offset, bytes, dio->map_bh.b_private);
+ dio->end_io(dio->iocb, offset, transferred,
+ dio->map_bh.b_private);
if (dio->lock_type == DIO_LOCKING)
- up_read(&dio->inode->i_alloc_sem);
+ /* lockdep: non-owner release */
+ up_read_non_owner(&dio->inode->i_alloc_sem);
+
+ if (ret == 0)
+ ret = dio->page_errors;
+ if (ret == 0)
+ ret = dio->io_error;
+ if (ret == 0)
+ ret = transferred;
+
+ return ret;
}
/*
spin_lock_irqsave(&dio->bio_lock, flags);
if (dio->bio_count == 1) {
if (dio->is_async) {
- ssize_t transferred;
- loff_t offset;
+ int ret;
/*
* Last reference to the dio is going away.
*/
spin_unlock_irqrestore(&dio->bio_lock, flags);
- /* Check for short read case */
- transferred = dio->result;
- offset = dio->iocb->ki_pos;
-
- if ((dio->rw == READ) &&
- ((offset + transferred) > dio->i_size))
- transferred = dio->i_size - offset;
-
- /* check for error in completion path */
- if (dio->io_error)
- transferred = dio->io_error;
-
- dio_complete(dio, offset, transferred);
+ ret = dio_complete(dio, dio->iocb->ki_pos, 0);
/* Complete AIO later if falling back to buffered i/o */
if (dio->result == dio->size ||
((dio->rw == READ) && dio->result)) {
- aio_complete(dio->iocb, transferred, 0);
+ aio_complete(dio->iocb, ret, 0);
kfree(dio);
return;
} else {
/*
* Wait on and process all in-flight BIOs.
*/
-static int dio_await_completion(struct dio *dio)
+static void dio_await_completion(struct dio *dio)
{
- int ret = 0;
-
if (dio->bio)
dio_bio_submit(dio);
*/
while (dio->bio_count) {
struct bio *bio = dio_await_one(dio);
- int ret2;
-
- ret2 = dio_bio_complete(dio, bio);
- if (ret == 0)
- ret = ret2;
+ /* io errors are propogated through dio->io_error */
+ dio_bio_complete(dio, bio);
}
- return ret;
}
/*
map_bh->b_state = 0;
map_bh->b_size = fs_count << dio->inode->i_blkbits;
- create = dio->rw == WRITE;
+ create = dio->rw & WRITE;
if (dio->lock_type == DIO_LOCKING) {
if (dio->block_in_file < (i_size_read(dio->inode) >>
dio->blkbits))
{
int ret = 0;
+ if (dio->rw & WRITE) {
+ /*
+ * Read accounting is performed in submit_bio()
+ */
+ task_io_account_write(len);
+ }
+
/*
* Can we just grow the current page's presence in the dio?
*/
loff_t i_size_aligned;
/* AKPM: eargh, -ENOTBLK is a hack */
- if (dio->rw == WRITE) {
+ if (dio->rw & WRITE) {
page_cache_release(page);
return -ENOTBLK;
}
}
} /* end iovec loop */
- if (ret == -ENOTBLK && rw == WRITE) {
+ if (ret == -ENOTBLK && (rw & WRITE)) {
/*
* The remaining part of the request will be
* be handled by buffered I/O when we return
if (dio->is_async) {
int should_wait = 0;
- if (dio->result < dio->size && rw == WRITE) {
+ if (dio->result < dio->size && (rw & WRITE)) {
dio->waiter = current;
should_wait = 1;
}
kfree(dio);
}
} else {
- ssize_t transferred = 0;
-
finished_one_bio(dio);
- ret2 = dio_await_completion(dio);
- if (ret == 0)
- ret = ret2;
- if (ret == 0)
- ret = dio->page_errors;
- if (dio->result) {
- loff_t i_size = i_size_read(inode);
+ dio_await_completion(dio);
- transferred = dio->result;
- /*
- * Adjust the return value if the read crossed a
- * non-block-aligned EOF.
- */
- if (rw == READ && (offset + transferred > i_size))
- transferred = i_size - offset;
- }
- dio_complete(dio, offset, transferred);
- if (ret == 0)
- ret = transferred;
+ ret = dio_complete(dio, offset, ret);
/* We could have also come here on an AIO file extend */
- if (!is_sync_kiocb(iocb) && rw == WRITE &&
+ if (!is_sync_kiocb(iocb) && (rw & WRITE) &&
ret >= 0 && dio->result == dio->size)
/*
* For AIO writes where we have completed the
int acquire_i_mutex = 0;
if (rw & WRITE)
- current->flags |= PF_SYNCWRITE;
+ rw = WRITE_SYNC;
if (bdev)
bdev_blkbits = blksize_bits(bdev_hardsect_size(bdev));
}
if (dio_lock_type == DIO_LOCKING)
- down_read(&inode->i_alloc_sem);
+ /* lockdep: not the owner will release it */
+ down_read_non_owner(&inode->i_alloc_sem);
}
/*
* even for AIO, we need to wait for i/o to complete before
* returning in this case.
*/
- dio->is_async = !is_sync_kiocb(iocb) && !((rw == WRITE) &&
+ dio->is_async = !is_sync_kiocb(iocb) && !((rw & WRITE) &&
(end > i_size_read(inode)));
retval = direct_io_worker(rw, iocb, inode, iov, offset,
mutex_unlock(&inode->i_mutex);
else if (acquire_i_mutex)
mutex_lock(&inode->i_mutex);
- if (rw & WRITE)
- current->flags &= ~PF_SYNCWRITE;
return retval;
}
EXPORT_SYMBOL(__blockdev_direct_IO);