lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180917201054.3530-15-martin@omnibond.com>
Date:   Mon, 17 Sep 2018 20:10:51 +0000
From:   Martin Brandenburg <martin@...ibond.com>
To:     linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
        devel@...ts.orangefs.org
Cc:     Martin Brandenburg <martin@...ibond.com>
Subject: [PATCH 14/17] orangefs: write range tracking

This is necessary to ensure the uid/gid responsible for the write is
communicated with the server.  Only one uid/gid may have outstanding
changes at a time.  If another uid/gid writes while there are
outstanding changes, the changes must be written out before the new
data is put into the page.

Signed-off-by: Martin Brandenburg <martin@...ibond.com>
---
 fs/orangefs/file.c            |  12 +-
 fs/orangefs/inode.c           | 243 ++++++++++++++++++++++++++++++----
 fs/orangefs/orangefs-kernel.h |  12 +-
 3 files changed, 237 insertions(+), 30 deletions(-)

diff --git a/fs/orangefs/file.c b/fs/orangefs/file.c
index ba580a5c6fd2..5eda483263ae 100644
--- a/fs/orangefs/file.c
+++ b/fs/orangefs/file.c
@@ -46,8 +46,8 @@ static int flush_racache(struct inode *inode)
  * Post and wait for the I/O upcall to finish
  */
 ssize_t wait_for_direct_io(enum ORANGEFS_io_type type, struct inode *inode,
-		loff_t *offset, struct iov_iter *iter,
-		size_t total_size, loff_t readahead_size)
+    loff_t *offset, struct iov_iter *iter, size_t total_size,
+    loff_t readahead_size, struct orangefs_write_request *wr)
 {
 	struct orangefs_inode_s *orangefs_inode = ORANGEFS_I(inode);
 	struct orangefs_khandle *handle = &orangefs_inode->refn.khandle;
@@ -103,6 +103,10 @@ ssize_t wait_for_direct_io(enum ORANGEFS_io_type type, struct inode *inode,
 			    __func__, (long)ret);
 			goto out;
 		}
+		if (wr) {
+			new_op->upcall.uid = from_kuid(&init_user_ns, wr->uid);
+			new_op->upcall.gid = from_kgid(&init_user_ns, wr->gid);
+		}
 	}
 
 	gossip_debug(GOSSIP_FILE_DEBUG,
@@ -292,7 +296,7 @@ ssize_t do_readv_writev(enum ORANGEFS_io_type type, struct file *file,
 			     (int)*offset);
 
 		ret = wait_for_direct_io(type, inode, offset, iter,
-				each_count, 0);
+				each_count, 0, NULL);
 		gossip_debug(GOSSIP_FILE_DEBUG,
 			     "%s(%pU): return from wait_for_io:%d\n",
 			     __func__,
@@ -434,7 +438,7 @@ static vm_fault_t orangefs_fault(struct vm_fault *vmf)
 static const struct vm_operations_struct orangefs_file_vm_ops = {
 	.fault = orangefs_fault,
 	.map_pages = filemap_map_pages,
-	.page_mkwrite = filemap_page_mkwrite,
+	.page_mkwrite = orangefs_page_mkwrite,
 };
 
 /*
diff --git a/fs/orangefs/inode.c b/fs/orangefs/inode.c
index 5832104cab6a..efb00cd50b61 100644
--- a/fs/orangefs/inode.c
+++ b/fs/orangefs/inode.c
@@ -15,9 +15,11 @@
 #include "orangefs-kernel.h"
 #include "orangefs-bufmap.h"
 
-static int orangefs_writepage(struct page *page, struct writeback_control *wbc)
+static int orangefs_writepage_locked(struct page *page,
+    struct writeback_control *wbc)
 {
 	struct inode *inode = page->mapping->host;
+	struct orangefs_write_request *wr;
 	struct iov_iter iter;
 	struct bio_vec bv;
 	size_t len, wlen;
@@ -26,33 +28,175 @@ static int orangefs_writepage(struct page *page, struct writeback_control *wbc)
 
 	set_page_writeback(page);
 
-	off = page_offset(page);
-	len = i_size_read(inode);
-	if (off + PAGE_SIZE > len)
-		wlen = len - off;
-	else
-		wlen = PAGE_SIZE;
+	if (PagePrivate(page)) {
+		wr = (struct orangefs_write_request *)page_private(page);
+		BUG_ON(!wr);
+		if (wr->mwrite) {
+			off = page_offset(page);
+			len = i_size_read(inode);
+			if (off + PAGE_SIZE > len)
+				wlen = len - off;
+			else
+				wlen = PAGE_SIZE;
+		} else {
+			off = wr->pos;
+			wlen = wr->len;
+			len = i_size_read(inode);
+		}
+	} else {
+/*		BUG();*/
+		/* It's not private so there's nothing to write, right? */
+		printk("writepage not private!\n");
+		end_page_writeback(page);
+		return 0;
+
+	}
 
 	bv.bv_page = page;
 	bv.bv_len = wlen;
 	bv.bv_offset = 0;
-	if (wlen == 0)
-		dump_stack();
 	iov_iter_bvec(&iter, ITER_BVEC | WRITE, &bv, 1, wlen);
 
 	ret = wait_for_direct_io(ORANGEFS_IO_WRITE, inode, &off, &iter, wlen,
-	    len);
+	    len, wr);
 	if (ret < 0) {
 		SetPageError(page);
 		mapping_set_error(page->mapping, ret);
 	} else {
 		ret = 0;
+		if (wr) {
+			ClearPagePrivate(page);
+			kfree(wr);
+		}
 	}
 	end_page_writeback(page);
-	unlock_page(page);
 	return ret;
 }
 
+static int do_writepage_if_necessary(struct page *page, loff_t pos,
+    unsigned len)
+{
+	struct orangefs_write_request *wr;
+	struct writeback_control wbc = {
+		.sync_mode = WB_SYNC_ALL,
+		.nr_to_write = 0,
+	};
+	int r;
+	if (PagePrivate(page)) {
+		wr = (struct orangefs_write_request *)page_private(page);
+		BUG_ON(!wr);
+		/*
+		 * If the new request is not contiguous with the last one or if
+		 * the uid or gid is different, the page must be written out
+		 * before continuing.
+		 */
+		if (pos + len < wr->pos || wr->pos + wr->len < pos ||
+		    !uid_eq(current_fsuid(), wr->uid) ||
+		    !gid_eq(current_fsgid(), wr->gid)) {
+			wbc.range_start = page_file_offset(page);
+			wbc.range_end = wbc.range_start + PAGE_SIZE - 1;
+			wait_on_page_writeback(page);
+			if (clear_page_dirty_for_io(page)) {
+				r = orangefs_writepage_locked(page, &wbc);
+				if (r)
+					return r;
+			}
+			BUG_ON(PagePrivate(page));
+		}
+	}
+	return 0;
+}
+
+static int update_wr(struct page *page, loff_t pos, unsigned len, int mwrite)
+{
+	struct orangefs_write_request *wr;
+	if (PagePrivate(page)) {
+		wr = (struct orangefs_write_request *)page_private(page);
+		BUG_ON(!wr);
+		if (mwrite) {
+			wr->mwrite = 1;
+			return 0;
+		}
+		if (pos < wr->pos) {
+			wr->len += wr->pos - pos;
+			wr->pos = pos;
+		}
+		if (pos + len > wr->pos + wr->len)
+			wr->len = pos + len - wr->pos;
+		else
+			wr->len = wr->pos + wr->len - wr->pos;
+	} else {
+		wr = kmalloc(sizeof *wr, GFP_KERNEL);
+		if (wr) {
+			wr->pos = pos;
+			wr->len = len;
+			wr->uid = current_fsuid();
+			wr->gid = current_fsgid();
+			wr->mwrite = mwrite;
+			SetPagePrivate(page);
+			set_page_private(page, (unsigned long)wr);
+		} else {
+			return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+int orangefs_page_mkwrite(struct vm_fault *vmf)
+{
+	struct page *page = vmf->page;
+	struct inode *inode = file_inode(vmf->vma->vm_file);
+	unsigned len;
+	int r;
+
+	/* Do not write past the file size. */
+	len = i_size_read(inode) - page_file_offset(page);
+	if (len > PAGE_SIZE)
+		len = PAGE_SIZE;
+
+	lock_page(page);
+	r = do_writepage_if_necessary(page, page_file_offset(page),
+	    len);
+	if (r) {
+		r = VM_FAULT_RETRY;
+		unlock_page(vmf->page);
+		return r;
+	}
+	r = update_wr(page, page_file_offset(page), len, 1);
+	if (r) {
+		r = VM_FAULT_RETRY;
+		unlock_page(vmf->page);
+		return r;
+	}
+
+	r = VM_FAULT_LOCKED;
+	sb_start_pagefault(inode->i_sb);
+	file_update_time(vmf->vma->vm_file);
+	if (page->mapping != inode->i_mapping) {
+		unlock_page(page);
+		r = VM_FAULT_NOPAGE;
+		goto out;
+	}
+	/*
+	 * We mark the page dirty already here so that when freeze is in
+	 * progress, we are guaranteed that writeback during freezing will
+	 * see the dirty page and writeprotect it again.
+	 */
+	set_page_dirty(page);
+	wait_for_stable_page(page);
+out:
+	sb_end_pagefault(inode->i_sb);
+	return r;
+}
+
+static int orangefs_writepage(struct page *page, struct writeback_control *wbc)
+{
+	int r;
+	r = orangefs_writepage_locked(page, wbc);
+	unlock_page(page);
+	return r;
+}
+
 static int orangefs_readpage(struct file *file, struct page *page)
 {
 	struct inode *inode = page->mapping->host;
@@ -68,7 +212,7 @@ static int orangefs_readpage(struct file *file, struct page *page)
 	iov_iter_bvec(&iter, ITER_BVEC | READ, &bv, 1, PAGE_SIZE);
 
 	ret = wait_for_direct_io(ORANGEFS_IO_READ, inode, &off, &iter,
-	    PAGE_SIZE, inode->i_size);
+	    PAGE_SIZE, inode->i_size, NULL);
 	/* this will only zero remaining unread portions of the page data */
 	iov_iter_zero(~0U, &iter);
 	/* takes care of potential aliasing */
@@ -86,10 +230,26 @@ static int orangefs_readpage(struct file *file, struct page *page)
 	return ret;
 }
 
+static int orangefs_write_begin(struct file *file,
+    struct address_space *mapping, loff_t pos, unsigned len, unsigned flags,
+    struct page **pagep, void **fsdata)
+{
+	int r;
+	r = simple_write_begin(file, mapping, pos, len, flags, pagep, fsdata);
+	if (r)
+		return r;
+	r = do_writepage_if_necessary(*pagep, pos, len);
+	if (r)
+		unlock_page(*pagep);
+	return r;
+}
+
 int orangefs_write_end(struct file *file, struct address_space *mapping,
     loff_t pos, unsigned len, unsigned copied, struct page *page, void *fsdata)
 {
 	int r;
+	if (update_wr(page, pos, len, 0))
+		return -ENOMEM;
 	r = simple_write_end(file, mapping, pos, len, copied, page, fsdata);
 	mark_inode_dirty_sync(file_inode(file));
 	return r;
@@ -99,24 +259,57 @@ static void orangefs_invalidatepage(struct page *page,
 				 unsigned int offset,
 				 unsigned int length)
 {
-	gossip_debug(GOSSIP_INODE_DEBUG,
-		     "orangefs_invalidatepage called on page %p "
-		     "(offset is %u)\n",
-		     page,
-		     offset);
-
-	ClearPageUptodate(page);
-	ClearPageMappedToDisk(page);
+	struct orangefs_write_request *wr;
+	/* XXX move to releasepage and call + rebase */
+	struct writeback_control wbc = {
+		.sync_mode = WB_SYNC_ALL,
+		.nr_to_write = 0,
+	};
+	int r;
+	if (PagePrivate(page)) {
+		wr = (struct orangefs_write_request *)page_private(page);
+		BUG_ON(!wr);
+/* XXX prove */
+		if (offset == 0 && length == PAGE_SIZE) {
+			ClearPagePrivate(page);
+			kfree(wr);
+		} else if (wr->pos - page_offset(page) < offset &&
+		    wr->pos - page_offset(page) + wr->len > offset + length) {
+			wbc.range_start = page_file_offset(page);
+			wbc.range_end = wbc.range_start + PAGE_SIZE - 1;
+			wait_on_page_writeback(page);
+			if (clear_page_dirty_for_io(page)) {
+				r = orangefs_writepage_locked(page, &wbc);
+				if (r)
+					return;
+			} else {
+				ClearPagePrivate(page);
+				kfree(wr);
+			}
+		} else if (wr->pos - page_offset(page) < offset &&
+		    wr->pos - page_offset(page) + wr->len <= offset + length) {
+			wr->len = offset;
+		} else if (wr->pos - page_offset(page) >= offset &&
+		    wr->pos - page_offset(page) + wr->len > offset + length) {
+			wr->pos += length - wr->pos + page_offset(page);
+			wr->len -= length - wr->pos + page_offset(page);
+		} else {
+			/*
+			 * Invalidate range is bigger than write range but
+			 * entire write range is to be invalidated.
+			 */
+			ClearPagePrivate(page);
+			kfree(wr);
+		}
+	}
 	return;
 
 }
 
 static int orangefs_releasepage(struct page *page, gfp_t foo)
 {
-	gossip_debug(GOSSIP_INODE_DEBUG,
-		     "orangefs_releasepage called on page %p\n",
-		     page);
-	return 0;
+	BUG();
+	return !PagePrivate(page);
 }
 
 static ssize_t orangefs_direct_IO(struct kiocb *iocb,
@@ -133,7 +326,7 @@ static const struct address_space_operations orangefs_address_operations = {
 	.writepage = orangefs_writepage,
 	.readpage = orangefs_readpage,
 	.set_page_dirty = __set_page_dirty_nobuffers,
-	.write_begin = simple_write_begin,
+	.write_begin = orangefs_write_begin,
 	.write_end = orangefs_write_end,
 	.invalidatepage = orangefs_invalidatepage,
 	.releasepage = orangefs_releasepage,
diff --git a/fs/orangefs/orangefs-kernel.h b/fs/orangefs/orangefs-kernel.h
index e128500e33b4..2e9726d1de7d 100644
--- a/fs/orangefs/orangefs-kernel.h
+++ b/fs/orangefs/orangefs-kernel.h
@@ -178,6 +178,14 @@ static inline void set_op_state_purged(struct orangefs_kernel_op_s *op)
 	}
 }
 
+struct orangefs_write_request {
+	loff_t pos;
+	unsigned len;
+	kuid_t uid;
+	kgid_t gid;
+	int mwrite;
+};
+
 /* per inode private orangefs info */
 struct orangefs_inode_s {
 	struct orangefs_object_kref refn;
@@ -341,6 +349,8 @@ void fsid_key_table_finalize(void);
 /*
  * defined in inode.c
  */
+int orangefs_page_mkwrite(struct vm_fault *);
+
 struct inode *orangefs_new_inode(struct super_block *sb,
 			      struct inode *dir,
 			      int mode,
@@ -382,7 +392,7 @@ bool __is_daemon_in_service(void);
  * defined in file.c
  */
 ssize_t wait_for_direct_io(enum ORANGEFS_io_type, struct inode *, loff_t *,
-    struct iov_iter *, size_t, loff_t);
+    struct iov_iter *, size_t, loff_t, struct orangefs_write_request *);
 ssize_t do_readv_writev(enum ORANGEFS_io_type, struct file *, loff_t *,
     struct iov_iter *);
 
-- 
2.19.0

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ