lustre: get rid of messing with iovecs

* switch to ->read_iter/->write_iter
* keep a pointer to iov_iter instead of iov/nr_segs
* do not modify iovecs; use iov_iter_truncate()/iov_iter_advance() and
a new primitive - iov_iter_reexpand() (expand previously truncated
iterator) istead.
* (racy) check for lustre VMAs intersecting with iovecs kept for now as
for_each_iov() loop.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

drivers/staging/lustre/lustre/include/lclient.h

@@ -82,16 +82,7 @@ struct ccc_io {
 	/**
 	 * I/O vector information to or from which read/write is going.
 	 */
-	struct iovec *cui_iov;
-	unsigned long cui_nrsegs;
-	/**
-	 * Total iov count for left IO.
-	 */
-	unsigned long cui_tot_nrsegs;
-	/**
-	 * Old length for iov that was truncated partially.
-	 */
-	size_t cui_iov_olen;
+	struct iov_iter *cui_iter;
 	/**
 	 * Total size for the left IO.
 	 */

drivers/staging/lustre/lustre/lclient/lcommon_cl.c

@@ -721,31 +721,12 @@ int ccc_io_one_lock_index(const struct lu_env *env, struct cl_io *io,
 void ccc_io_update_iov(const struct lu_env *env,
 		       struct ccc_io *cio, struct cl_io *io)
 {
-	int i;
 	size_t size = io->u.ci_rw.crw_count;
 
-	cio->cui_iov_olen = 0;
-	if (!cl_is_normalio(env, io) || cio->cui_tot_nrsegs == 0)
+	if (!cl_is_normalio(env, io) || cio->cui_iter == NULL)
 		return;
 
-	for (i = 0; i < cio->cui_tot_nrsegs; i++) {
-		struct iovec *iv = &cio->cui_iov[i];
-
-		if (iv->iov_len < size)
-			size -= iv->iov_len;
-		else {
-			if (iv->iov_len > size) {
-				cio->cui_iov_olen = iv->iov_len;
-				iv->iov_len = size;
-			}
-			break;
-		}
-	}
-
-	cio->cui_nrsegs = i + 1;
-	LASSERTF(cio->cui_tot_nrsegs >= cio->cui_nrsegs,
-		 "tot_nrsegs: %lu, nrsegs: %lu\n",
-		 cio->cui_tot_nrsegs, cio->cui_nrsegs);
+	iov_iter_truncate(cio->cui_iter, size);
 }
 
 int ccc_io_one_lock(const struct lu_env *env, struct cl_io *io,
@@ -776,30 +757,7 @@ void ccc_io_advance(const struct lu_env *env,
 	if (!cl_is_normalio(env, io))
 		return;
 
-	LASSERT(cio->cui_tot_nrsegs >= cio->cui_nrsegs);
-	LASSERT(cio->cui_tot_count  >= nob);
-
-	cio->cui_iov	+= cio->cui_nrsegs;
-	cio->cui_tot_nrsegs -= cio->cui_nrsegs;
-	cio->cui_tot_count  -= nob;
-
-	/* update the iov */
-	if (cio->cui_iov_olen > 0) {
-		struct iovec *iv;
-
-		cio->cui_iov--;
-		cio->cui_tot_nrsegs++;
-		iv = &cio->cui_iov[0];
-		if (io->ci_continue) {
-			iv->iov_base += iv->iov_len;
-			LASSERT(cio->cui_iov_olen > iv->iov_len);
-			iv->iov_len = cio->cui_iov_olen - iv->iov_len;
-		} else {
-			/* restore the iov_len, in case of restart io. */
-			iv->iov_len = cio->cui_iov_olen;
-		}
-		cio->cui_iov_olen = 0;
-	}
+	iov_iter_reexpand(cio->cui_iter, cio->cui_tot_count  -= nob);
 }
 
 /**

drivers/staging/lustre/lustre/llite/file.c

@@ -1105,9 +1105,7 @@ ll_file_io_generic(const struct lu_env *env, struct vvp_io_args *args,
 
 		switch (vio->cui_io_subtype) {
 		case IO_NORMAL:
-			cio->cui_iov = args->u.normal.via_iov;
-			cio->cui_nrsegs = args->u.normal.via_nrsegs;
-			cio->cui_tot_nrsegs = cio->cui_nrsegs;
+			cio->cui_iter = args->u.normal.via_iter;
 			cio->cui_iocb = args->u.normal.via_iocb;
 			if ((iot == CIT_WRITE) &&
 			    !(cio->cui_fd->fd_flags & LL_FILE_GROUP_LOCKED)) {
@@ -1171,56 +1169,23 @@ ll_file_io_generic(const struct lu_env *env, struct vvp_io_args *args,
 	return result;
 }
 
-static ssize_t ll_file_aio_read(struct kiocb *iocb, const struct iovec *iov,
-				unsigned long nr_segs, loff_t pos)
+static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
 {
 	struct lu_env      *env;
 	struct vvp_io_args *args;
-	size_t	      count = 0;
 	ssize_t	     result;
 	int		 refcheck;
 
-	count = iov_length(iov, nr_segs);
-
 	env = cl_env_get(&refcheck);
 	if (IS_ERR(env))
 		return PTR_ERR(env);
 
 	args = vvp_env_args(env, IO_NORMAL);
-	args->u.normal.via_iov = (struct iovec *)iov;
-	args->u.normal.via_nrsegs = nr_segs;
+	args->u.normal.via_iter = to;
 	args->u.normal.via_iocb = iocb;
 
 	result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_READ,
-				    &iocb->ki_pos, count);
-	cl_env_put(env, &refcheck);
-	return result;
-}
-
-static ssize_t ll_file_read(struct file *file, char *buf, size_t count,
-			    loff_t *ppos)
-{
-	struct lu_env *env;
-	struct iovec  *local_iov;
-	struct kiocb  *kiocb;
-	ssize_t	result;
-	int	    refcheck;
-
-	env = cl_env_get(&refcheck);
-	if (IS_ERR(env))
-		return PTR_ERR(env);
-
-	local_iov = &vvp_env_info(env)->vti_local_iov;
-	kiocb = &vvp_env_info(env)->vti_kiocb;
-	local_iov->iov_base = (void __user *)buf;
-	local_iov->iov_len = count;
-	init_sync_kiocb(kiocb, file);
-	kiocb->ki_pos = *ppos;
-	kiocb->ki_nbytes = count;
-
-	result = ll_file_aio_read(kiocb, local_iov, 1, kiocb->ki_pos);
-	*ppos = kiocb->ki_pos;
-
+				    &iocb->ki_pos, iov_iter_count(to));
 	cl_env_put(env, &refcheck);
 	return result;
 }
@@ -1228,12 +1193,10 @@ static ssize_t ll_file_read(struct file *file, char *buf, size_t count,
 /*
  * Write to a file (through the page cache).
  */
-static ssize_t ll_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
-				 unsigned long nr_segs, loff_t pos)
+static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 {
 	struct lu_env      *env;
 	struct vvp_io_args *args;
-	size_t	      count = iov_length(iov, nr_segs);
 	ssize_t	     result;
 	int		 refcheck;
 
@@ -1242,46 +1205,15 @@ static ssize_t ll_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
 		return PTR_ERR(env);
 
 	args = vvp_env_args(env, IO_NORMAL);
-	args->u.normal.via_iov = (struct iovec *)iov;
-	args->u.normal.via_nrsegs = nr_segs;
+	args->u.normal.via_iter = from;
 	args->u.normal.via_iocb = iocb;
 
 	result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_WRITE,
-				  &iocb->ki_pos, count);
+				  &iocb->ki_pos, iov_iter_count(from));
 	cl_env_put(env, &refcheck);
 	return result;
 }
 
-static ssize_t ll_file_write(struct file *file, const char *buf, size_t count,
-			     loff_t *ppos)
-{
-	struct lu_env *env;
-	struct iovec  *local_iov;
-	struct kiocb  *kiocb;
-	ssize_t	result;
-	int	    refcheck;
-
-	env = cl_env_get(&refcheck);
-	if (IS_ERR(env))
-		return PTR_ERR(env);
-
-	local_iov = &vvp_env_info(env)->vti_local_iov;
-	kiocb = &vvp_env_info(env)->vti_kiocb;
-	local_iov->iov_base = (void __user *)buf;
-	local_iov->iov_len = count;
-	init_sync_kiocb(kiocb, file);
-	kiocb->ki_pos = *ppos;
-	kiocb->ki_nbytes = count;
-
-	result = ll_file_aio_write(kiocb, local_iov, 1, kiocb->ki_pos);
-	*ppos = kiocb->ki_pos;
-
-	cl_env_put(env, &refcheck);
-	return result;
-}
-
-
-
 /*
  * Send file content (through pagecache) somewhere with helper
  */
@@ -3133,10 +3065,10 @@ int ll_inode_permission(struct inode *inode, int mask)
 
 /* -o localflock - only provides locally consistent flock locks */
 struct file_operations ll_file_operations = {
-	.read	   = ll_file_read,
-	.aio_read = ll_file_aio_read,
-	.write	  = ll_file_write,
-	.aio_write = ll_file_aio_write,
+	.read	   = new_sync_read,
+	.read_iter = ll_file_read_iter,
+	.write	  = new_sync_write,
+	.write_iter = ll_file_write_iter,
 	.unlocked_ioctl = ll_file_ioctl,
 	.open	   = ll_file_open,
 	.release	= ll_file_release,
@@ -3148,10 +3080,10 @@ struct file_operations ll_file_operations = {
 };
 
 struct file_operations ll_file_operations_flock = {
-	.read	   = ll_file_read,
-	.aio_read    = ll_file_aio_read,
-	.write	  = ll_file_write,
-	.aio_write   = ll_file_aio_write,
+	.read	   = new_sync_read,
+	.read_iter    = ll_file_read_iter,
+	.write	  = new_sync_write,
+	.write_iter   = ll_file_write_iter,
 	.unlocked_ioctl = ll_file_ioctl,
 	.open	   = ll_file_open,
 	.release	= ll_file_release,
@@ -3166,10 +3098,10 @@ struct file_operations ll_file_operations_flock = {
 
 /* These are for -o noflock - to return ENOSYS on flock calls */
 struct file_operations ll_file_operations_noflock = {
-	.read	   = ll_file_read,
-	.aio_read    = ll_file_aio_read,
-	.write	  = ll_file_write,
-	.aio_write   = ll_file_aio_write,
+	.read	   = new_sync_read,
+	.read_iter    = ll_file_read_iter,
+	.write	  = new_sync_write,
+	.write_iter   = ll_file_write_iter,
 	.unlocked_ioctl = ll_file_ioctl,
 	.open	   = ll_file_open,
 	.release	= ll_file_release,

drivers/staging/lustre/lustre/llite/llite_internal.h

@@ -974,8 +974,7 @@ struct vvp_io_args {
 	union {
 		struct {
 			struct kiocb      *via_iocb;
-			struct iovec      *via_iov;
-			unsigned long      via_nrsegs;
+			struct iov_iter   *via_iter;
 		} normal;
 		struct {
 			struct pipe_inode_info  *via_pipe;

drivers/staging/lustre/lustre/llite/rw.c

@@ -157,8 +157,7 @@ static struct ll_cl_context *ll_cl_init(struct file *file,
 		result = cl_io_rw_init(env, io, CIT_WRITE, pos, PAGE_CACHE_SIZE);
 		if (result == 0) {
 			cio->cui_fd = LUSTRE_FPRIVATE(file);
-			cio->cui_iov = NULL;
-			cio->cui_nrsegs = 0;
+			cio->cui_iter = NULL;
 			result = cl_io_iter_init(env, io);
 			if (result == 0) {
 				result = cl_io_lock(env, io);

drivers/staging/lustre/lustre/llite/vvp_io.c

@@ -211,27 +211,26 @@ static int vvp_mmap_locks(const struct lu_env *env,
 	struct cl_lock_descr   *descr = &cti->cti_descr;
 	ldlm_policy_data_t      policy;
 	unsigned long	   addr;
-	unsigned long	   seg;
 	ssize_t		 count;
 	int		     result;
+	struct iov_iter i;
+	struct iovec iov;
 
 	LASSERT(io->ci_type == CIT_READ || io->ci_type == CIT_WRITE);
 
 	if (!cl_is_normalio(env, io))
 		return 0;
 
-	if (vio->cui_iov == NULL) /* nfs or loop back device write */
+	if (vio->cui_iter == NULL) /* nfs or loop back device write */
 		return 0;
 
 	/* No MM (e.g. NFS)? No vmas too. */
 	if (mm == NULL)
 		return 0;
 
-	for (seg = 0; seg < vio->cui_nrsegs; seg++) {
-		const struct iovec *iv = &vio->cui_iov[seg];
-
-		addr = (unsigned long)iv->iov_base;
-		count = iv->iov_len;
+	iov_for_each(iov, i, *(vio->cui_iter)) {
+		addr = (unsigned long)iov.iov_base;
+		count = iov.iov_len;
 		if (count == 0)
 			continue;
 
@@ -527,9 +526,7 @@ static int vvp_io_read_start(const struct lu_env *env,
 	switch (vio->cui_io_subtype) {
 	case IO_NORMAL:
 		LASSERT(cio->cui_iocb->ki_pos == pos);
-		result = generic_file_aio_read(cio->cui_iocb,
-					       cio->cui_iov, cio->cui_nrsegs,
-					       cio->cui_iocb->ki_pos);
+		result = generic_file_read_iter(cio->cui_iocb, cio->cui_iter);
 		break;
 	case IO_SPLICE:
 		result = generic_file_splice_read(file, &pos,
@@ -595,12 +592,11 @@ static int vvp_io_write_start(const struct lu_env *env,
 
 	CDEBUG(D_VFSTRACE, "write: [%lli, %lli)\n", pos, pos + (long long)cnt);
 
-	if (cio->cui_iov == NULL) /* from a temp io in ll_cl_init(). */
+	if (cio->cui_iter == NULL) /* from a temp io in ll_cl_init(). */
 		result = 0;
 	else
-		result = generic_file_aio_write(cio->cui_iocb,
-						cio->cui_iov, cio->cui_nrsegs,
-						cio->cui_iocb->ki_pos);
+		result = generic_file_write_iter(cio->cui_iocb, cio->cui_iter);
+
 	if (result > 0) {
 		if (result < cnt)
 			io->ci_continue = 0;
@@ -1162,10 +1158,9 @@ int vvp_io_init(const struct lu_env *env, struct cl_object *obj,
 		 *  results."  -- Single Unix Spec */
 		if (count == 0)
 			result = 1;
-		else {
+		else
 			cio->cui_tot_count = count;
-			cio->cui_tot_nrsegs = 0;
-		}
+
 		/* for read/write, we store the jobid in the inode, and
 		 * it'll be fetched by osc when building RPC.
 		 *

include/linux/uio.h

@@ -90,6 +90,15 @@ static inline void iov_iter_truncate(struct iov_iter *i, size_t count)
 		i->count = count;
 }
 
+/*
+ * reexpand a previously truncated iterator; count must be no more than how much
+ * we had shrunk it.
+ */
+static inline void iov_iter_reexpand(struct iov_iter *i, size_t count)
+{
+	i->count = count;
+}
+
 int memcpy_fromiovec(unsigned char *kdata, struct iovec *iov, int len);
 int memcpy_toiovec(struct iovec *iov, unsigned char *kdata, int len);