linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Matthew Wilcox <matthew.r.wilcox@intel.com>
To: linux-fsdevel@vger.kernel.org, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Cc: Matthew Wilcox <willy@linux.intel.com>
Subject: [PATCH v8 07/22] Add copy_to_iter(), copy_from_iter() and iov_iter_zero()
Date: Tue, 22 Jul 2014 15:47:55 -0400	[thread overview]
Message-ID: <02a76f478c739ad87d7d5354b0e63818ebea8541.1406058387.git.matthew.r.wilcox@intel.com> (raw)
In-Reply-To: <cover.1406058387.git.matthew.r.wilcox@intel.com>
In-Reply-To: <cover.1406058387.git.matthew.r.wilcox@intel.com>

From: Matthew Wilcox <willy@linux.intel.com>

For DAX, we want to be able to copy between iovecs and kernel addresses
that don't necessarily have a struct page.  This is a fairly simple
rearrangement for bvec iters to kmap the pages outside and pass them in,
but for user iovecs it gets more complicated because we might try various
different ways to kmap the memory.  Duplicating the existing logic works
out best in this case.

We need to be able to write zeroes to an iovec for reads from unwritten
ranges in a file.  This is performed by the new iov_iter_zero() function,
again patterned after the existing code that handles iovec iterators.

Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
---
 include/linux/uio.h |   3 +
 mm/iov_iter.c       | 237 ++++++++++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 226 insertions(+), 14 deletions(-)

diff --git a/include/linux/uio.h b/include/linux/uio.h
index 09a7cff..86fb3e9 100644
--- a/include/linux/uio.h
+++ b/include/linux/uio.h
@@ -80,6 +80,9 @@ size_t copy_page_to_iter(struct page *page, size_t offset, size_t bytes,
 			 struct iov_iter *i);
 size_t copy_page_from_iter(struct page *page, size_t offset, size_t bytes,
 			 struct iov_iter *i);
+size_t copy_to_iter(void *addr, size_t bytes, struct iov_iter *i);
+size_t copy_from_iter(void *addr, size_t bytes, struct iov_iter *i);
+size_t iov_iter_zero(size_t bytes, struct iov_iter *);
 unsigned long iov_iter_alignment(const struct iov_iter *i);
 void iov_iter_init(struct iov_iter *i, int direction, const struct iovec *iov,
 			unsigned long nr_segs, size_t count);
diff --git a/mm/iov_iter.c b/mm/iov_iter.c
index 7b5dbd1..e95ed80 100644
--- a/mm/iov_iter.c
+++ b/mm/iov_iter.c
@@ -4,6 +4,96 @@
 #include <linux/slab.h>
 #include <linux/vmalloc.h>
 
+static size_t copy_to_iter_iovec(void *from, size_t bytes, struct iov_iter *i)
+{
+	size_t skip, copy, left, wanted;
+	const struct iovec *iov;
+	char __user *buf;
+
+	if (unlikely(bytes > i->count))
+		bytes = i->count;
+
+	if (unlikely(!bytes))
+		return 0;
+
+	wanted = bytes;
+	iov = i->iov;
+	skip = i->iov_offset;
+	buf = iov->iov_base + skip;
+	copy = min(bytes, iov->iov_len - skip);
+
+	left = __copy_to_user(buf, from, copy);
+	copy -= left;
+	skip += copy;
+	from += copy;
+	bytes -= copy;
+	while (unlikely(!left && bytes)) {
+		iov++;
+		buf = iov->iov_base;
+		copy = min(bytes, iov->iov_len);
+		left = __copy_to_user(buf, from, copy);
+		copy -= left;
+		skip = copy;
+		from += copy;
+		bytes -= copy;
+	}
+
+	if (skip == iov->iov_len) {
+		iov++;
+		skip = 0;
+	}
+	i->count -= wanted - bytes;
+	i->nr_segs -= iov - i->iov;
+	i->iov = iov;
+	i->iov_offset = skip;
+	return wanted - bytes;
+}
+
+static size_t copy_from_iter_iovec(void *to, size_t bytes, struct iov_iter *i)
+{
+	size_t skip, copy, left, wanted;
+	const struct iovec *iov;
+	char __user *buf;
+
+	if (unlikely(bytes > i->count))
+		bytes = i->count;
+
+	if (unlikely(!bytes))
+		return 0;
+
+	wanted = bytes;
+	iov = i->iov;
+	skip = i->iov_offset;
+	buf = iov->iov_base + skip;
+	copy = min(bytes, iov->iov_len - skip);
+
+	left = __copy_from_user(to, buf, copy);
+	copy -= left;
+	skip += copy;
+	to += copy;
+	bytes -= copy;
+	while (unlikely(!left && bytes)) {
+		iov++;
+		buf = iov->iov_base;
+		copy = min(bytes, iov->iov_len);
+		left = __copy_from_user(to, buf, copy);
+		copy -= left;
+		skip = copy;
+		to += copy;
+		bytes -= copy;
+	}
+
+	if (skip == iov->iov_len) {
+		iov++;
+		skip = 0;
+	}
+	i->count -= wanted - bytes;
+	i->nr_segs -= iov - i->iov;
+	i->iov = iov;
+	i->iov_offset = skip;
+	return wanted - bytes;
+}
+
 static size_t copy_page_to_iter_iovec(struct page *page, size_t offset, size_t bytes,
 			 struct iov_iter *i)
 {
@@ -166,6 +256,50 @@ done:
 	return wanted - bytes;
 }
 
+static size_t zero_iovec(size_t bytes, struct iov_iter *i)
+{
+	size_t skip, copy, left, wanted;
+	const struct iovec *iov;
+	char __user *buf;
+
+	if (unlikely(bytes > i->count))
+		bytes = i->count;
+
+	if (unlikely(!bytes))
+		return 0;
+
+	wanted = bytes;
+	iov = i->iov;
+	skip = i->iov_offset;
+	buf = iov->iov_base + skip;
+	copy = min(bytes, iov->iov_len - skip);
+
+	left = __clear_user(buf, copy);
+	copy -= left;
+	skip += copy;
+	bytes -= copy;
+
+	while (unlikely(!left && bytes)) {
+		iov++;
+		buf = iov->iov_base;
+		copy = min(bytes, iov->iov_len);
+		left = __clear_user(buf, copy);
+		copy -= left;
+		skip = copy;
+		bytes -= copy;
+	}
+
+	if (skip == iov->iov_len) {
+		iov++;
+		skip = 0;
+	}
+	i->count -= wanted - bytes;
+	i->nr_segs -= iov - i->iov;
+	i->iov = iov;
+	i->iov_offset = skip;
+	return wanted - bytes;
+}
+
 static size_t __iovec_copy_from_user_inatomic(char *vaddr,
 			const struct iovec *iov, size_t base, size_t bytes)
 {
@@ -412,12 +546,17 @@ static void memcpy_to_page(struct page *page, size_t offset, char *from, size_t
 	kunmap_atomic(to);
 }
 
-static size_t copy_page_to_iter_bvec(struct page *page, size_t offset, size_t bytes,
-			 struct iov_iter *i)
+static void memzero_page(struct page *page, size_t offset, size_t len)
+{
+	char *addr = kmap_atomic(page);
+	memset(addr + offset, 0, len);
+	kunmap_atomic(addr);
+}
+
+static size_t copy_to_iter_bvec(void *from, size_t bytes, struct iov_iter *i)
 {
 	size_t skip, copy, wanted;
 	const struct bio_vec *bvec;
-	void *kaddr, *from;
 
 	if (unlikely(bytes > i->count))
 		bytes = i->count;
@@ -430,8 +569,6 @@ static size_t copy_page_to_iter_bvec(struct page *page, size_t offset, size_t by
 	skip = i->iov_offset;
 	copy = min_t(size_t, bytes, bvec->bv_len - skip);
 
-	kaddr = kmap_atomic(page);
-	from = kaddr + offset;
 	memcpy_to_page(bvec->bv_page, skip + bvec->bv_offset, from, copy);
 	skip += copy;
 	from += copy;
@@ -444,7 +581,6 @@ static size_t copy_page_to_iter_bvec(struct page *page, size_t offset, size_t by
 		from += copy;
 		bytes -= copy;
 	}
-	kunmap_atomic(kaddr);
 	if (skip == bvec->bv_len) {
 		bvec++;
 		skip = 0;
@@ -456,12 +592,10 @@ static size_t copy_page_to_iter_bvec(struct page *page, size_t offset, size_t by
 	return wanted - bytes;
 }
 
-static size_t copy_page_from_iter_bvec(struct page *page, size_t offset, size_t bytes,
-			 struct iov_iter *i)
+static size_t copy_from_iter_bvec(void *to, size_t bytes, struct iov_iter *i)
 {
 	size_t skip, copy, wanted;
 	const struct bio_vec *bvec;
-	void *kaddr, *to;
 
 	if (unlikely(bytes > i->count))
 		bytes = i->count;
@@ -473,10 +607,6 @@ static size_t copy_page_from_iter_bvec(struct page *page, size_t offset, size_t
 	bvec = i->bvec;
 	skip = i->iov_offset;
 
-	kaddr = kmap_atomic(page);
-
-	to = kaddr + offset;
-
 	copy = min(bytes, bvec->bv_len - skip);
 
 	memcpy_from_page(to, bvec->bv_page, bvec->bv_offset + skip, copy);
@@ -493,7 +623,6 @@ static size_t copy_page_from_iter_bvec(struct page *page, size_t offset, size_t
 		to += copy;
 		bytes -= copy;
 	}
-	kunmap_atomic(kaddr);
 	if (skip == bvec->bv_len) {
 		bvec++;
 		skip = 0;
@@ -505,6 +634,61 @@ static size_t copy_page_from_iter_bvec(struct page *page, size_t offset, size_t
 	return wanted;
 }
 
+static size_t copy_page_to_iter_bvec(struct page *page, size_t offset,
+					size_t bytes, struct iov_iter *i)
+{
+	void *kaddr = kmap_atomic(page);
+	size_t wanted = copy_to_iter_bvec(kaddr + offset, bytes, i);
+	kunmap_atomic(kaddr);
+	return wanted;
+}
+
+static size_t copy_page_from_iter_bvec(struct page *page, size_t offset,
+					size_t bytes, struct iov_iter *i)
+{
+	void *kaddr = kmap_atomic(page);
+	size_t wanted = copy_from_iter_bvec(kaddr + offset, bytes, i);
+	kunmap_atomic(kaddr);
+	return wanted;
+}
+
+static size_t zero_bvec(size_t bytes, struct iov_iter *i)
+{
+	size_t skip, copy, wanted;
+	const struct bio_vec *bvec;
+
+	if (unlikely(bytes > i->count))
+		bytes = i->count;
+
+	if (unlikely(!bytes))
+		return 0;
+
+	wanted = bytes;
+	bvec = i->bvec;
+	skip = i->iov_offset;
+	copy = min_t(size_t, bytes, bvec->bv_len - skip);
+
+	memzero_page(bvec->bv_page, skip + bvec->bv_offset, copy);
+	skip += copy;
+	bytes -= copy;
+	while (bytes) {
+		bvec++;
+		copy = min(bytes, (size_t)bvec->bv_len);
+		memzero_page(bvec->bv_page, bvec->bv_offset, copy);
+		skip = copy;
+		bytes -= copy;
+	}
+	if (skip == bvec->bv_len) {
+		bvec++;
+		skip = 0;
+	}
+	i->count -= wanted - bytes;
+	i->nr_segs -= bvec - i->bvec;
+	i->bvec = bvec;
+	i->iov_offset = skip;
+	return wanted - bytes;
+}
+
 static size_t copy_from_user_bvec(struct page *page,
 		struct iov_iter *i, unsigned long offset, size_t bytes)
 {
@@ -669,6 +853,31 @@ size_t copy_page_from_iter(struct page *page, size_t offset, size_t bytes,
 }
 EXPORT_SYMBOL(copy_page_from_iter);
 
+size_t copy_to_iter(void *addr, size_t bytes, struct iov_iter *i)
+{
+	if (i->type & ITER_BVEC)
+		return copy_to_iter_bvec(addr, bytes, i);
+	else
+		return copy_to_iter_iovec(addr, bytes, i);
+}
+
+size_t copy_from_iter(void *addr, size_t bytes, struct iov_iter *i)
+{
+	if (i->type & ITER_BVEC)
+		return copy_from_iter_bvec(addr, bytes, i);
+	else
+		return copy_from_iter_iovec(addr, bytes, i);
+}
+
+size_t iov_iter_zero(size_t bytes, struct iov_iter *i)
+{
+	if (i->type & ITER_BVEC) {
+		return zero_bvec(bytes, i);
+	} else {
+		return zero_iovec(bytes, i);
+	}
+}
+
 size_t iov_iter_copy_from_user_atomic(struct page *page,
 		struct iov_iter *i, unsigned long offset, size_t bytes)
 {
-- 
2.0.0


  parent reply	other threads:[~2014-07-22 19:50 UTC|newest]

Thread overview: 61+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-07-22 19:47 [PATCH v8 00/22] Support ext4 on NV-DIMMs Matthew Wilcox
2014-07-22 19:47 ` [PATCH v8 01/22] Fix XIP fault vs truncate race Matthew Wilcox
2014-07-23 11:21   ` Kirill A. Shutemov
2014-07-22 19:47 ` [PATCH v8 02/22] Allow page fault handlers to perform the COW Matthew Wilcox
2014-07-23 11:23   ` Kirill A. Shutemov
2014-07-22 19:47 ` [PATCH v8 03/22] axonram: Fix bug in direct_access Matthew Wilcox
2014-07-23 11:24   ` Kirill A. Shutemov
2014-07-22 19:47 ` [PATCH v8 04/22] Change direct_access calling convention Matthew Wilcox
2014-07-30 16:03   ` Boaz Harrosh
2014-07-30 16:12     ` Boaz Harrosh
2014-07-30 20:34       ` Matthew Wilcox
2014-07-31 10:16         ` Boaz Harrosh
2014-07-30 19:45     ` Matthew Wilcox
2014-07-31 10:11       ` Boaz Harrosh
2014-07-31 14:13         ` Matthew Wilcox
2014-07-31 15:28           ` Boaz Harrosh
2014-07-31 17:19             ` Matthew Wilcox
2014-07-31 18:04               ` Boaz Harrosh
2014-07-31 20:30                 ` Zwisler, Ross
2014-08-01 18:45                   ` Zwisler, Ross
2014-07-22 19:47 ` [PATCH v8 05/22] Add vm_replace_mixed() Matthew Wilcox
2014-07-23  9:10   ` Jan Kara
2014-07-23 11:45   ` Kirill A. Shutemov
2014-07-23 13:52     ` Matthew Wilcox
2014-07-23 14:20       ` Kirill A. Shutemov
2014-07-23 14:27         ` Matthew Wilcox
2014-07-23 15:55           ` Kirill A. Shutemov
2014-07-24  1:36             ` Zhang, Tianfei
2014-07-25 19:44             ` Matthew Wilcox
2014-07-28 13:25               ` Kirill A. Shutemov
2014-07-29  1:55                 ` Zhang, Tianfei
2014-07-22 19:47 ` [PATCH v8 06/22] Introduce IS_DAX(inode) Matthew Wilcox
2014-07-22 19:47 ` Matthew Wilcox [this message]
2014-07-22 19:47 ` [PATCH v8 08/22] Replace XIP read and write with DAX I/O Matthew Wilcox
2014-07-22 19:47 ` [PATCH v8 09/22] Replace ext2_clear_xip_target with dax_clear_blocks Matthew Wilcox
2014-07-22 19:47 ` [PATCH v8 10/22] Replace the XIP page fault handler with the DAX page fault handler Matthew Wilcox
2014-07-23 12:10   ` Kirill A. Shutemov
2014-07-23 13:55     ` Matthew Wilcox
2014-07-23 14:10       ` [PATCH v8 00/22] Support ext4 on NV-DIMMs Howard Chu
2014-07-23 14:34         ` Matthew Wilcox
2014-07-23 15:28           ` Howard Chu
2014-07-23 20:55             ` Theodore Ts'o
2014-07-23 16:57   ` [PATCH v8 10/22] Replace the XIP page fault handler with the DAX page fault handler Boaz Harrosh
2014-07-23 19:57     ` Matthew Wilcox
2014-07-22 19:47 ` [PATCH v8 11/22] Replace xip_truncate_page with dax_truncate_page Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 12/22] Replace XIP documentation with DAX documentation Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 13/22] Remove get_xip_mem Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 14/22] ext2: Remove ext2_xip_verify_sb() Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 15/22] ext2: Remove ext2_use_xip Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 16/22] ext2: Remove xip.c and xip.h Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 17/22] Remove CONFIG_EXT2_FS_XIP and rename CONFIG_FS_XIP to CONFIG_FS_DAX Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 18/22] ext2: Remove ext2_aops_xip Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 19/22] Get rid of most mentions of XIP in ext2 Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 20/22] xip: Add xip_zero_page_range Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 21/22] ext4: Add DAX functionality Matthew Wilcox
2014-07-22 19:48 ` [PATCH v8 22/22] brd: Rename XIP to DAX Matthew Wilcox
2014-07-23 12:30 ` [PATCH v8 00/22] Support ext4 on NV-DIMMs Kirill A. Shutemov
2014-07-23 13:59   ` Matthew Wilcox
2014-07-23 15:58 ` Boaz Harrosh
2014-07-23 19:50   ` Matthew Wilcox
2014-07-24 18:51     ` Ross Zwisler

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=02a76f478c739ad87d7d5354b0e63818ebea8541.1406058387.git.matthew.r.wilcox@intel.com \
    --to=matthew.r.wilcox@intel.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=willy@linux.intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).