vfs: use iovec_iterator for all file read and write, add vfs_read_iovec and vfs_write_iovec

kernel-src/fs/devfs.c

@@ -71,7 +71,7 @@ int devfs_close(vnode_t *node, int flags, cred_t *cred) {
 	return devnode->devops->close(devnode->attr.rdevminor, flags);
 }
 
-int devfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
+int devfs_read(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
 	if (node->type == V_TYPE_DIR)
 		return EISDIR;
 
@@ -85,19 +85,10 @@ int devfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int f
 	if (devnode->devops->read == NULL)
 		return ENODEV;
 
-	// TODO remove this once the whole fs uses iovec
-	iovec_iterator_t iovec_iterator;
-	iovec_t iovec = {
-		.addr = buffer,
-		.len = size
-	};
-
-	iovec_iterator_init(&iovec_iterator, &iovec, 1);
-
-	return devnode->devops->read(devnode->attr.rdevminor, &iovec_iterator, size, offset, flags, readc);
+	return devnode->devops->read(devnode->attr.rdevminor, iovec_iterator, size, offset, flags, readc);
 }
 
-int devfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
+int devfs_write(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
 	if (node->type == V_TYPE_DIR)
 		return EISDIR;
 
@@ -111,16 +102,7 @@ int devfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int
 	if (devnode->devops->write == NULL)
 		return ENODEV;
 
-	// TODO remove this once the whole fs uses iovec
-	iovec_iterator_t iovec_iterator;
-	iovec_t iovec = {
-		.addr = buffer,
-		.len = size
-	};
-
-	iovec_iterator_init(&iovec_iterator, &iovec, 1);
-
-	return devnode->devops->write(devnode->attr.rdevminor, &iovec_iterator, size, offset, flags, writec);
+	return devnode->devops->write(devnode->attr.rdevminor, iovec_iterator, size, offset, flags, writec);
 }
 
 int devfs_lookup(vnode_t *node, char *name, vnode_t **result, cred_t *cred) {
@@ -400,8 +382,16 @@ static int devfs_getdents(vnode_t *node, dent_t *buffer, size_t count, uintmax_t
 static int devfs_getpage(vnode_t *node, uintmax_t offset, struct page_t *page) {
 	// only block devices will have this called
 	__assert(node->type == V_TYPE_BLKDEV);
+	iovec_t iovec = {
+		.addr = MAKE_HHDM(pmm_getpageaddress(page)),
+		.len = PAGE_SIZE
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+
 	size_t readc;
-	int error = VOP_READ(node, MAKE_HHDM(pmm_getpageaddress(page)), PAGE_SIZE, offset, 0, &readc, NULL);
+	int error = VOP_READ(node, &iovec_iterator, PAGE_SIZE, offset, 0, &readc, NULL);
 	if (readc == 0)
 		return ENXIO;
 
@@ -411,8 +401,16 @@ static int devfs_getpage(vnode_t *node, uintmax_t offset, struct page_t *page) {
 static int devfs_putpage(vnode_t *node, uintmax_t offset, struct page_t *page) {
 	// only block devices will have this called
 	__assert(node->type == V_TYPE_BLKDEV);
+	iovec_t iovec = {
+		.addr = MAKE_HHDM(pmm_getpageaddress(page)),
+		.len = PAGE_SIZE
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+
 	size_t writec;
-	int error = VOP_WRITE(node, MAKE_HHDM(pmm_getpageaddress(page)), PAGE_SIZE, offset, 0, &writec, NULL);
+	int error = VOP_WRITE(node, &iovec_iterator, PAGE_SIZE, offset, 0, &writec, NULL);
 	__assert(writec != 0);
 
 	return error;

kernel-src/fs/ext2.c

@@ -749,12 +749,11 @@ static int resizeinode(ext2fs_t *fs, ext2node_t *node, size_t newsize) {
 	return 0;
 }
 
-static int rwblocks(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, uintmax_t index, bool write, bool cache) {
+static int rwblocks_iovec(ext2fs_t *fs, ext2node_t *node, iovec_iterator_t *iovec_iterator, size_t count, uintmax_t index, bool write, bool cache) {
 	for (uintmax_t i = 0; i < count; ++i) {
 		size_t inodesize = INODE_SIZE(&node->inode);
 		__assert(index + i < ROUND_UP(inodesize, fs->blocksize) / fs->blocksize);
 
-		void *bufferp = (void *)((uintptr_t)buffer + i * fs->blocksize);
 		blockptr_t block;
 		int e = getinodeblock(fs, node, index + i, &block);
 		if (e)
@@ -775,14 +774,14 @@ static int rwblocks(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count,
 
 			block = newblock;
 		} else if (block == 0 && write == false) {
-			memset(bufferp, 0, fs->blocksize);
+			iovec_iterator_memset(iovec_iterator, 0, fs->blocksize);
 			continue;
 		}
 
 		size_t donecount;
 		e = write ?
-			vfs_write(fs->backing, bufferp, fs->blocksize, BLOCK_GETDISKOFFSET(fs, block), &donecount, cache ? 0 : V_FFLAGS_NOCACHE) :
-			vfs_read(fs->backing, bufferp, fs->blocksize, BLOCK_GETDISKOFFSET(fs, block), &donecount, cache ? 0 : V_FFLAGS_NOCACHE);
+			vfs_write_iovec(fs->backing, iovec_iterator, fs->blocksize, BLOCK_GETDISKOFFSET(fs, block), &donecount, cache ? 0 : V_FFLAGS_NOCACHE) :
+			vfs_read_iovec(fs->backing, iovec_iterator, fs->blocksize, BLOCK_GETDISKOFFSET(fs, block), &donecount, cache ? 0 : V_FFLAGS_NOCACHE);
 
 		if (e)
 			return e;
@@ -792,7 +791,7 @@ static int rwblocks(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count,
 	return 0;
 }
 
-static int rwblock(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, uintmax_t offset, uintmax_t index, bool write, bool cache) {
+static int rwblock_iovec(ext2fs_t *fs, ext2node_t *node, iovec_iterator_t *iovec_iterator, size_t count, uintmax_t offset, uintmax_t index, bool write, bool cache) {
 	__assert(offset + count <= fs->blocksize);
 	size_t inodesize = INODE_SIZE(&node->inode);
 	__assert(index < ROUND_UP(inodesize, fs->blocksize) / fs->blocksize);
@@ -818,14 +817,14 @@ static int rwblock(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, u
 
 		block = newblock;
 	} else if (block == 0 && write == false) {
-		memset(buffer, 0, count);
+		iovec_iterator_memset(iovec_iterator, 0, count);
 		return 0;
 	}
 
 	size_t donecount;
 	e = write ?
-		vfs_write(fs->backing, buffer, count, BLOCK_GETDISKOFFSET(fs, block) + offset, &donecount, cache ? 0 : V_FFLAGS_NOCACHE) :
-		vfs_read(fs->backing, buffer, count, BLOCK_GETDISKOFFSET(fs, block) + offset, &donecount, cache ? 0 : V_FFLAGS_NOCACHE);
+		vfs_write_iovec(fs->backing, iovec_iterator, count, BLOCK_GETDISKOFFSET(fs, block) + offset, &donecount, cache ? 0 : V_FFLAGS_NOCACHE) :
+		vfs_read_iovec(fs->backing, iovec_iterator, count, BLOCK_GETDISKOFFSET(fs, block) + offset, &donecount, cache ? 0 : V_FFLAGS_NOCACHE);
 
 	if (e)
 		return e;
@@ -834,7 +833,7 @@ static int rwblock(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, u
 	return e;
 }
 
-static int rwbytes(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, uintmax_t offset, bool write, bool cache) {
+static int rwbytes_iovec(ext2fs_t *fs, ext2node_t *node, iovec_iterator_t *iovec_iterator, size_t count, uintmax_t offset, bool write, bool cache) {
 	uintmax_t index = offset / fs->blocksize;
 	uintmax_t startoffset = offset % fs->blocksize;
 
@@ -842,11 +841,10 @@ static int rwbytes(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, u
 	if (startoffset) {
 		size_t blockremaining = fs->blocksize - startoffset;
 		size_t docount = min(blockremaining, count);
-		int e = rwblock(fs, node, buffer, docount, startoffset, index, write, cache);
+		int e = rwblock_iovec(fs, node, iovec_iterator, docount, startoffset, index, write, cache);
 		if (e)
 			return e;
 
-		buffer = (void *)((uintptr_t)buffer + docount);
 		count -= docount;
 		index += 1;
 	}
@@ -854,17 +852,28 @@ static int rwbytes(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, u
 	// r/w all middle blocks
 	size_t blocks = count / fs->blocksize;
 	if (blocks) {
-		int e = rwblocks(fs, node, buffer, blocks, index, write, cache);
+		int e = rwblocks_iovec(fs, node, iovec_iterator, blocks, index, write, cache);
 		if (e)
 			return e;
 
 		count -= blocks * fs->blocksize;
-		buffer = (void *)((uintptr_t)buffer + blocks * fs->blocksize);
 		index += blocks;
 	}
 
 	// r/w remaining block if there's any data left and return
-	return count ? rwblock(fs, node, buffer, count, 0, index, write, cache) : 0;
+	return count ? rwblock_iovec(fs, node, iovec_iterator, count, 0, index, write, cache) : 0;
+}
+
+static int rwbytes(ext2fs_t *fs, ext2node_t *node, void *buffer, size_t count, uintmax_t offset, bool write, bool cache) {
+	iovec_t iovec = {
+		.addr = buffer,
+		.len = count
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+
+	return rwbytes_iovec(fs, node, &iovec_iterator, count, offset, write, cache);
 }
 
 #define BUFFER_MAP_FLAGS (ARCH_MMU_FLAGS_READ | ARCH_MMU_FLAGS_WRITE | ARCH_MMU_FLAGS_NOEXEC)
@@ -1317,9 +1326,7 @@ static int ext2_readlink(vnode_t *vnode, char **link, cred_t *cred) {
 	return err;
 }
 
-// this *should*, by the rules of the vnode ops, accept a user buffer. however, this will never be the case. all I/O will
-// go through the page cache
-static int ext2_read(vnode_t *vnode, void *buffer, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
+static int ext2_read(vnode_t *vnode, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
 	if (vnode->type == V_TYPE_DIR)
 		return EISDIR;
 
@@ -1350,7 +1357,7 @@ static int ext2_read(vnode_t *vnode, void *buffer, size_t size, uintmax_t offset
 	if (size == 0)
 		goto cleanup;
 
-	err = rwbytes(fs, node, buffer, size, offset, false, false);
+	err = rwbytes_iovec(fs, node, iovec_iterator, size, offset, false, false);
 
 	*readc = err ? -1 : size;
 
@@ -1360,7 +1367,7 @@ static int ext2_read(vnode_t *vnode, void *buffer, size_t size, uintmax_t offset
 
 // this *should*, by the rules of the vnode ops, accept a user buffer. however, this will never be the case. all I/O will
 // go through the page cache
-int ext2_write(vnode_t *vnode, void *buffer, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
+int ext2_write(vnode_t *vnode, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
 	if (vnode->type == V_TYPE_DIR)
 		return EISDIR;
 
@@ -1386,7 +1393,7 @@ int ext2_write(vnode_t *vnode, void *buffer, size_t size, uintmax_t offset, int
 		size = min(endoffset, inodesize) - offset;
 	}
 
-	err = rwbytes(fs, node, buffer, size, offset, true, false);
+	err = rwbytes(fs, node, iovec_iterator, size, offset, true, false);
 	*writec = err ? -1 : size;
 	if (err)
 		goto cleanup;
@@ -1695,7 +1702,15 @@ static int ext2_getpage(vnode_t *node, uintmax_t offset, struct page_t *page) {
 	__assert(node->type == V_TYPE_REGULAR);
 	size_t readc = PAGE_SIZE;
 	void *addr = MAKE_HHDM(pmm_getpageaddress(page));
-	int error = VOP_READ(node, addr, PAGE_SIZE, offset, 0, &readc, NULL);
+
+	iovec_t iovec = {
+		.addr = addr,
+		.len = PAGE_SIZE
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+	int error = VOP_READ(node, &iovec_iterator, PAGE_SIZE, offset, 0, &readc, NULL);
 	if (readc == 0)
 		return ENXIO;
 	else if (readc != PAGE_SIZE)
@@ -1708,7 +1723,14 @@ static int ext2_putpage(vnode_t *node, uintmax_t offset, struct page_t *page) {
 	// only regular files get cached
 	__assert(node->type == V_TYPE_REGULAR);
 	size_t writec;
-	int error = VOP_WRITE(node, MAKE_HHDM(pmm_getpageaddress(page)), PAGE_SIZE, offset, 0, &writec, NULL);
+	iovec_t iovec = {
+		.addr = MAKE_HHDM(pmm_getpageaddress(page)),
+		.len = PAGE_SIZE
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+	int error = VOP_WRITE(node, &iovec_iterator, PAGE_SIZE, offset, 0, &writec, NULL);
 
 	// its possible that writec is zero here. the condition where this is possible is when
 	// the file is truncated after the check in the page sync function but before it actually is

kernel-src/fs/pipefs.c

@@ -130,8 +130,7 @@ static int internalpoll(vnode_t *node, polldata_t *data, int events) {
 	return revents;
 }
 
-// buffer can possibly be a user address
-int pipefs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
+int pipefs_read(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
 	pipenode_t *pipenode = (pipenode_t *)node;
 	INTERNAL_LOCK(node);
 
@@ -171,7 +170,7 @@ int pipefs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int
 		INTERNAL_LOCK(node);
 	}
 
-	*readc = ringbuffer_read(&pipenode->data, buffer, size);
+	*readc = iovec_iterator_read_from_ringbuffer(iovec_iterator, &pipenode->data, size);
 	if (*readc == RINGBUFFER_USER_COPY_FAILED)
 		error = EFAULT;
 
@@ -184,7 +183,7 @@ int pipefs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int
 	return error;
 }
 
-static int writetopipe(pipenode_t *pipenode, void *buffer, size_t size, size_t *writec) {
+static int writetopipe(pipenode_t *pipenode, iovec_iterator_t *iovec_iterator, size_t size, size_t *writec) {
 	if (pipenode->readers == 0) {
 		if (current_thread()->proc)
 			signal_signalproc(current_thread()->proc, SIGPIPE);
@@ -192,7 +191,7 @@ static int writetopipe(pipenode_t *pipenode, void *buffer, size_t size, size_t *
 		return EPIPE;
 	}
 
-	*writec = ringbuffer_write(&pipenode->data, buffer, size);
+	*writec = iovec_iterator_write_to_ringbuffer(iovec_iterator, &pipenode->data, size);
 	if (*writec == RINGBUFFER_USER_COPY_FAILED)
 		return EFAULT;
 
@@ -201,7 +200,7 @@ static int writetopipe(pipenode_t *pipenode, void *buffer, size_t size, size_t *
 	return 0;
 }
 
-int pipefs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
+int pipefs_write(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
 	pipenode_t *pipenode = (pipenode_t *)node;
 	INTERNAL_LOCK(node);
 
@@ -247,7 +246,7 @@ int pipefs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int
 		if (nonblock == false && atomic == false && freebytes) {
 			size_t tmp;
 			size_t writecount = min(freebytes, size - case3written);
-			error = writetopipe(pipenode, (void *)((uintptr_t)buffer + case3written), writecount, &tmp);
+			error = writetopipe(pipenode, iovec_iterator, writecount, &tmp);
 			if (error)
 				goto leave;
 
@@ -290,7 +289,7 @@ int pipefs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int
 		INTERNAL_LOCK(node);
 	}
 
-	error = writetopipe(pipenode, buffer, size, writec);
+	error = writetopipe(pipenode, iovec_iterator, size, writec);
 
 	leave:
 	INTERNAL_UNLOCK(node);

kernel-src/fs/sockfs.c

@@ -24,14 +24,42 @@ int sockfs_close(vnode_t *node, int flags, cred_t *cred) {
 	return ENOSYS; // not needed
 }
 
-int sockfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
+int sockfs_read(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *bytes_read, cred_t *cred) {
 	socket_t *socket = SOCKFS_SOCKET_FROM_NODE(node);
-	return socket_read(socket, buffer, size, flags, readc);
+	sockdesc_t desc = {
+		.addr = NULL,
+		.iovec_iterator = iovec_iterator,
+		.count = size,
+		.flags = flags,
+		.donecount = 0,
+		.ctrl = NULL,
+		.ctrllen = 0
+	};
+
+	int e = socket->ops->recv(socket, &desc);
+
+	*bytes_read = desc.donecount;
+
+	return e;
 }
 
-int sockfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
+int sockfs_write(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *bytes_written, cred_t *cred) {
 	socket_t *socket = SOCKFS_SOCKET_FROM_NODE(node);
-	return socket_write(socket, buffer, size, flags, writec);
+	sockdesc_t desc = {
+		.addr = NULL,
+		.iovec_iterator = iovec_iterator,
+		.count = size,
+		.flags = flags,
+		.donecount = 0,
+		.ctrl = NULL,
+		.ctrllen = 0
+	};
+
+	int e = socket->ops->send(socket, &desc);
+
+	*bytes_written = desc.donecount;
+
+	return e;
 }
 
 int sockfs_poll(vnode_t *node, polldata_t *data, int events) {

kernel-src/fs/tmpfs.c

@@ -220,7 +220,7 @@ static int tmpfs_root(vfs_t *vfs, vnode_t **vnode) {
 	return 0;
 }
 
-static int tmpfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
+static int tmpfs_read(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred) {
 	if (node->type == V_TYPE_DIR)
 		return EISDIR;
 
@@ -230,7 +230,7 @@ static int tmpfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset
 	__assert(!"tmpfs_read is handled by the page cache");
 }
 
-static int tmpfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
+static int tmpfs_write(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred) {
 	if (node->type == V_TYPE_DIR)
 		return EISDIR;
 

kernel-src/fs/vfs.c

@@ -229,8 +229,7 @@ static int writenocache(vnode_t *node, page_t *page, uintmax_t pageoffset) {
 	return e;
 }
 
-// buffer can be a user address
-int vfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t *written, int flags) {
+int vfs_write_iovec(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, size_t *written, int flags) {
 	int err = 0;
 	if (vfs_iscacheable(node)) {
 		*written = 0;
@@ -292,7 +291,7 @@ int vfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t
 			size_t writesize = min(PAGE_SIZE - startoffset, size);
 			void *address = MAKE_HHDM(pmm_getpageaddress(page));
 
-			err = USERCOPY_POSSIBLY_FROM_USER((void *)((uintptr_t)address + startoffset), buffer, writesize);
+			err = iovec_iterator_copy_to_buffer(iovec_iterator, (void *)((uintptr_t)address + startoffset), writesize);
 			if (err) {
 				pmm_release(FROM_HHDM(address));
 				goto leave;
@@ -320,7 +319,7 @@ int vfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t
 			size_t writesize = min(PAGE_SIZE, size - *written);
 			void *address = MAKE_HHDM(pmm_getpageaddress(page));
 
-			err = USERCOPY_POSSIBLY_FROM_USER(address, (void *)((uintptr_t)buffer + *written), writesize);
+			err = iovec_iterator_copy_to_buffer(iovec_iterator, address, writesize);
 			if (err) {
 				pmm_release(FROM_HHDM(address));
 				goto leave;
@@ -343,14 +342,14 @@ int vfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t
 	} else {
 		// special file, just write as its not being cached
 		VOP_LOCK(node);
-		err = VOP_WRITE(node, buffer, size, offset, flags, written, getcred());
+		err = VOP_WRITE(node, iovec_iterator, size, offset, flags, written, getcred());
 		VOP_UNLOCK(node);
 	}
 
 	return err;
 }
 
-int vfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t *bytesread, int flags) {
+int vfs_read_iovec(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, size_t *bytesread, int flags) {
 	int err = 0;
 	if (vfs_iscacheable(node)) {
 		*bytesread = 0;
@@ -406,7 +405,7 @@ int vfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t
 			size_t readsize = min(PAGE_SIZE - startoffset, size);
 			void *address = MAKE_HHDM(pmm_getpageaddress(page));
 
-			err = USERCOPY_POSSIBLY_TO_USER(buffer, (void *)((uintptr_t)address + startoffset), readsize);
+			err = iovec_iterator_copy_from_buffer(iovec_iterator, (void *)((uintptr_t)address + startoffset), readsize);
 			if (err) {
 				pmm_release(FROM_HHDM(address));
 				goto leave;
@@ -433,7 +432,7 @@ int vfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t
 			size_t readsize = min(PAGE_SIZE, size - *bytesread);
 			void *address = MAKE_HHDM(pmm_getpageaddress(page));
 
-			err = USERCOPY_POSSIBLY_TO_USER((void *)((uintptr_t)buffer + *bytesread), address, readsize);
+			err = iovec_iterator_copy_from_buffer(iovec_iterator, address, readsize);
 			if (err) {
 				pmm_release(FROM_HHDM(address));
 				goto leave;
@@ -451,12 +450,36 @@ int vfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t
 	} else {
 		// special file, just read as size doesn't matter
 		VOP_LOCK(node);
-		err = VOP_READ(node, buffer, size, offset, flags, bytesread, getcred());
+		err = VOP_READ(node, iovec_iterator, size, offset, flags, bytesread, getcred());
 		VOP_UNLOCK(node);
 	}
 	return err;
 }
 
+int vfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t *written, int flags) {
+	iovec_t iovec = {
+		.addr = buffer,
+		.len = size
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+
+	return vfs_write_iovec(node, &iovec_iterator, size, offset, written, flags);
+}
+
+int vfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t *bytes_read, int flags) {
+	iovec_t iovec = {
+		.addr = buffer,
+		.len = size
+	};
+
+	iovec_iterator_t iovec_iterator;
+	iovec_iterator_init(&iovec_iterator, &iovec, 1);
+
+	return vfs_read_iovec(node, &iovec_iterator, size, offset, bytes_read, flags);
+}
+
 // if type is V_TYPE_LINK, a symlink is made
 // in that case, destref is ignored, destpath is the link value and attr points to the attributes of the symlink
 // if type is V_TYPE_REGULAR, a hardlink is made.

kernel-src/include/kernel/sock.h

@@ -39,7 +39,7 @@ typedef struct {
 
 typedef struct {
 	sockaddr_t *addr;
-	iovec_iterator_t iovec_iterator;
+	iovec_iterator_t *iovec_iterator;
 	size_t count;
 	uintmax_t flags;
 	size_t donecount;
@@ -80,8 +80,11 @@ typedef struct {
 #define SOCKFS_SOCKET_FROM_NODE(nodep) (((socketnode_t *)(nodep))->socket)
 
 static inline int socket_read(socket_t *socket, void *buffer, size_t size, uintmax_t flags, size_t *bytes_read) {
+	iovec_iterator_t iovec_iterator;
+
 	sockdesc_t desc = {
 		.addr = NULL,
+		.iovec_iterator = &iovec_iterator,
 		.count = size,
 		.flags = flags,
 		.donecount = 0,
@@ -94,7 +97,7 @@ static inline int socket_read(socket_t *socket, void *buffer, size_t size, uintm
 		.len = size
 	};
 
-	iovec_iterator_init(&desc.iovec_iterator, &iovec, 1);
+	iovec_iterator_init(desc.iovec_iterator, &iovec, 1);
 
 	int e = socket->ops->recv(socket, &desc);
 
@@ -104,8 +107,11 @@ static inline int socket_read(socket_t *socket, void *buffer, size_t size, uintm
 }
 
 static inline int socket_write(socket_t *socket, void *buffer, size_t size, uintmax_t flags, size_t *bytes_written) {
+	iovec_iterator_t iovec_iterator;
+
 	sockdesc_t desc = {
 		.addr = NULL,
+		.iovec_iterator = &iovec_iterator,
 		.count = size,
 		.flags = flags,
 		.donecount = 0,
@@ -118,7 +124,7 @@ static inline int socket_write(socket_t *socket, void *buffer, size_t size, uint
 		.len = size
 	};
 
-	iovec_iterator_init(&desc.iovec_iterator, &iovec, 1);
+	iovec_iterator_init(desc.iovec_iterator, &iovec, 1);
 
 	int e = socket->ops->send(socket, &desc);
 

kernel-src/include/kernel/vfs.h

@@ -118,8 +118,8 @@ struct polldata;
 typedef struct vops_t {
 	int (*open)(vnode_t **node, int flags, cred_t *cred);
 	int (*close)(vnode_t *node, int flags, cred_t *cred);
-	int (*read)(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred);
-	int (*write)(vnode_t *node, void *buffer, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred);
+	int (*read)(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *readc, cred_t *cred);
+	int (*write)(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, int flags, size_t *writec, cred_t *cred);
 	int (*lookup)(vnode_t *node, char *name, vnode_t **result, cred_t *cred);
 	int (*create)(vnode_t *parent, char *name, vattr_t *attr, int type, vnode_t **result, cred_t *cred);
 	int (*getattr)(vnode_t *node, vattr_t *attr, cred_t *cred);
@@ -214,6 +214,8 @@ int vfs_register(vfsops_t *vfsops, char *name);
 
 int vfs_open(vnode_t *ref, char *path, int flags, vnode_t **result);
 int vfs_close(vnode_t *node, int flags);
+int vfs_write_iovec(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, size_t *written, int flags);
+int vfs_read_iovec(vnode_t *node, iovec_iterator_t *iovec_iterator, size_t size, uintmax_t offset, size_t *bytesread, int flags);
 int vfs_write(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t *written, int flags);
 int vfs_read(vnode_t *node, void *buffer, size_t size, uintmax_t offset, size_t *bytesread, int flags);
 int vfs_create(vnode_t *ref, char *path, vattr_t *attr, int type, vnode_t **node);

kernel-src/io/net/localsocket.c

@@ -363,7 +363,7 @@ static int localsock_send(socket_t *socket, sockdesc_t *sockdesc) {
 		goto leave;
 	}
 
-	sockdesc->donecount = iovec_iterator_write_to_ringbuffer(&sockdesc->iovec_iterator, &peer->ringbuffer, sockdesc->count);
+	sockdesc->donecount = iovec_iterator_write_to_ringbuffer(sockdesc->iovec_iterator, &peer->ringbuffer, sockdesc->count);
 	if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 		error = EFAULT;
 		goto leave;
@@ -467,13 +467,13 @@ static int localsock_recv(socket_t *socket, sockdesc_t *sockdesc) {
 	}
 
 	if (flags & SOCKET_RECV_FLAGS_PEEK) {
-		sockdesc->donecount = iovec_iterator_peek_from_ringbuffer(&sockdesc->iovec_iterator, &localsocket->ringbuffer, 0, recvcount);
+		sockdesc->donecount = iovec_iterator_peek_from_ringbuffer(sockdesc->iovec_iterator, &localsocket->ringbuffer, 0, recvcount);
 		if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 			error = EFAULT;
 			goto leave;
 		}
 	} else {
-		sockdesc->donecount = iovec_iterator_read_from_ringbuffer(&sockdesc->iovec_iterator, &localsocket->ringbuffer, recvcount);
+		sockdesc->donecount = iovec_iterator_read_from_ringbuffer(sockdesc->iovec_iterator, &localsocket->ringbuffer, recvcount);
 		if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 			error = EPIPE;
 			goto leave;

kernel-src/io/net/tcp.c

@@ -1305,7 +1305,7 @@ static int tcp_send(socket_t *socket, sockdesc_t *sockdesc) {
 	size_t writesize = min(TCB_RINGBUFFER_SIZE - RINGBUFFER_DATACOUNT(&tcb->transmitbuffer), sockdesc->count);
 	__assert(writesize);
 
-	sockdesc->donecount = iovec_iterator_write_to_ringbuffer(&sockdesc->iovec_iterator, &tcb->transmitbuffer, writesize);
+	sockdesc->donecount = iovec_iterator_write_to_ringbuffer(sockdesc->iovec_iterator, &tcb->transmitbuffer, writesize);
 	if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 		error = EFAULT;
 		goto leave;
@@ -1386,13 +1386,13 @@ static int tcp_recv(socket_t *socket, sockdesc_t *sockdesc) {
 
 	size_t copycount = min(RINGBUFFER_DATACOUNT(&tcb->receivebuffer), sockdesc->count);
 	if (flags & SOCKET_RECV_FLAGS_PEEK) {
-		sockdesc->donecount = iovec_iterator_peek_from_ringbuffer(&sockdesc->iovec_iterator, &tcb->receivebuffer, 0, copycount);
+		sockdesc->donecount = iovec_iterator_peek_from_ringbuffer(sockdesc->iovec_iterator, &tcb->receivebuffer, 0, copycount);
 		if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 			error = EFAULT;
 			goto cleanup;
 		}
 	} else {
-		sockdesc->donecount = iovec_iterator_read_from_ringbuffer(&sockdesc->iovec_iterator, &tcb->receivebuffer, copycount);
+		sockdesc->donecount = iovec_iterator_read_from_ringbuffer(sockdesc->iovec_iterator, &tcb->receivebuffer, copycount);
 		if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 			error = EFAULT;
 			goto cleanup;

kernel-src/io/net/udp.c

@@ -203,7 +203,7 @@ static int udp_send(socket_t *socket, sockdesc_t *sockdesc) {
 		socket->state = SOCKET_STATE_BOUND;
 	}
 
-	e = udp_sendpacket(&sockdesc->iovec_iterator, sockdesc->count, sockdesc->addr ? sockdesc->addr->ipv4addr.addr : udpsocket->peeraddress,
+	e = udp_sendpacket(sockdesc->iovec_iterator, sockdesc->count, sockdesc->addr ? sockdesc->addr->ipv4addr.addr : udpsocket->peeraddress,
 			udpsocket->port, sockdesc->addr ? sockdesc->addr->ipv4addr.port : udpsocket->peerport, udpsocket->socket.netdev);
 
 	sockdesc->donecount = sockdesc->count;
@@ -274,7 +274,7 @@ static int udp_recv(socket_t *socket, sockdesc_t *sockdesc) {
 
 	copycount = min(header.length, sockdesc->count);
 
-	sockdesc->donecount = iovec_iterator_peek_from_ringbuffer(&sockdesc->iovec_iterator, &udpsocket->ringbuffer, sizeof(header), copycount);
+	sockdesc->donecount = iovec_iterator_peek_from_ringbuffer(sockdesc->iovec_iterator, &udpsocket->ringbuffer, sizeof(header), copycount);
 	if (sockdesc->donecount == RINGBUFFER_USER_COPY_FAILED) {
 		e = EFAULT;
 		goto cleanup;

kernel-src/sys/syscalls/recvmsg.c

@@ -55,8 +55,10 @@ syscallret_t syscall_recvmsg(context_t *, int fd, msghdr_t *umsghdr, int flags)
 	if (flags & MSG_WAITALL)
 		recvflags |= SOCKET_RECV_FLAGS_WAITALL;
 
+	iovec_iterator_t iovec_iterator;
 	sockdesc_t desc = {
 		.addr = &sockaddr,
+		.iovec_iterator = &iovec_iterator,
 		.count = buffersize,
 		.flags = fileflagstovnodeflags(file->flags) | recvflags,
 		.donecount = 0,
@@ -65,7 +67,7 @@ syscallret_t syscall_recvmsg(context_t *, int fd, msghdr_t *umsghdr, int flags)
 		.ctrldone = 0
 	};
 
-	iovec_iterator_init(&desc.iovec_iterator, msghdr.iov, msghdr.iovcount);
+	iovec_iterator_init(desc.iovec_iterator, msghdr.iov, msghdr.iovcount);
 
 	ret.errno = socket->ops->recv(socket, &desc);
 	if (ret.errno)

kernel-src/sys/syscalls/sendmsg.c

@@ -49,8 +49,10 @@ syscallret_t syscall_sendmsg(context_t *, int fd, msghdr_t *umsghdr, int flags)
 
 	socket_t *socket = SOCKFS_SOCKET_FROM_NODE(file->vnode);
 
+	iovec_iterator_t iovec_iterator;
 	sockdesc_t desc = {
 		.addr = msghdr.addr ? &sockaddr : NULL,
+		.iovec_iterator = &iovec_iterator,
 		.count = buffersize,
 		.flags = fileflagstovnodeflags(file->flags) | ((flags & MSG_NOSIGNAL) ? SOCKET_SEND_FLAGS_NOSIGNAL : 0),
 		.donecount = 0,
@@ -59,7 +61,7 @@ syscallret_t syscall_sendmsg(context_t *, int fd, msghdr_t *umsghdr, int flags)
 		.ctrldone = 0
 	};
 
-	iovec_iterator_init(&desc.iovec_iterator, msghdr.iov, msghdr.iovcount);
+	iovec_iterator_init(desc.iovec_iterator, msghdr.iov, msghdr.iovcount);
 
 	ret.errno = socket->ops->send(socket, &desc);
 	ret.ret = ret.errno ? -1 : desc.donecount;