linux-erofs.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
From: Gao Xiang <hsiangkao@linux.alibaba.com>
To: linux-erofs@lists.ozlabs.org
Cc: nvdimm@lists.linux.dev, "Darrick J. Wong" <djwong@kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Joseqh Qi <joseph.qi@linux.alibaba.com>,
	Liu Bo <bo.liu@linux.alibaba.com>,
	linux-fsdevel@vger.kernel.org,
	Gao Xiang <hsiangkao@linux.alibaba.com>,
	Liu Jiang <gerry@linux.alibaba.com>
Subject: [RFC PATCH v1.1 2/2] erofs: dax support for non-tailpacking regular file
Date: Mon,  5 Jul 2021 21:21:53 +0800	[thread overview]
Message-ID: <20210705132153.223839-1-hsiangkao@linux.alibaba.com> (raw)
In-Reply-To: <20210704135056.42723-3-hsiangkao@linux.alibaba.com>

DAX is quite useful for some VM use cases in order to save guest
memory extremely with minimal lightweight EROFS.

In order to prepare for such use cases, add preliminary dax support
for non-tailpacking regular files for now.

Tested with the DRAM-emulated PMEM and the EROFS image generated by
"mkfs.erofs -Enoinline_data enwik9.fsdax.img enwik9"

Cc: nvdimm@lists.linux.dev
Cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
---
change since v1:
 - update missing hunks due to patch spliting...
    bdev_dax_supported(...)
    erofs_file_mmap(...)   

 fs/erofs/data.c     | 43 +++++++++++++++++++++++++++++++++++++++++--
 fs/erofs/inode.c    |  5 +++++
 fs/erofs/internal.h |  2 ++
 fs/erofs/super.c    | 26 ++++++++++++++++++++++++--
 4 files changed, 72 insertions(+), 4 deletions(-)

diff --git a/fs/erofs/data.c b/fs/erofs/data.c
index 0f82b4cb474c..c188c629be45 100644
--- a/fs/erofs/data.c
+++ b/fs/erofs/data.c
@@ -6,7 +6,7 @@
 #include "internal.h"
 #include <linux/prefetch.h>
 #include <linux/iomap.h>
-
+#include <linux/dax.h>
 #include <trace/events/erofs.h>
 
 static void erofs_readendio(struct bio *bio)
@@ -323,6 +323,7 @@ static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
 		return ret;
 
 	iomap->bdev = inode->i_sb->s_bdev;
+	iomap->dax_dev = EROFS_I_SB(inode)->dax_dev;
 	iomap->offset = map.m_la;
 	iomap->length = map.m_llen;
 
@@ -382,6 +383,11 @@ static ssize_t erofs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
 	if (!iov_iter_count(to))
 		return 0;
 
+#ifdef CONFIG_FS_DAX
+	if (IS_DAX(iocb->ki_filp->f_mapping->host))
+		return dax_iomap_rw(iocb, to, &erofs_iomap_ops);
+#endif
+
 	if (iocb->ki_flags & IOCB_DIRECT) {
 		int err = erofs_prepare_dio(iocb, to);
 
@@ -410,9 +416,42 @@ const struct address_space_operations erofs_raw_access_aops = {
 	.direct_IO = noop_direct_IO,
 };
 
+#ifdef CONFIG_FS_DAX
+static vm_fault_t erofs_dax_huge_fault(struct vm_fault *vmf,
+		enum page_entry_size pe_size)
+{
+	return dax_iomap_fault(vmf, pe_size, NULL, NULL, &erofs_iomap_ops);
+}
+
+static vm_fault_t erofs_dax_fault(struct vm_fault *vmf)
+{
+	return erofs_dax_huge_fault(vmf, PE_SIZE_PTE);
+}
+
+static const struct vm_operations_struct erofs_dax_vm_ops = {
+	.fault		= erofs_dax_fault,
+	.huge_fault	= erofs_dax_huge_fault,
+};
+
+static int erofs_file_mmap(struct file *file, struct vm_area_struct *vma)
+{
+	if (!IS_DAX(file_inode(file)))
+		return generic_file_readonly_mmap(file, vma);
+
+	if ((vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE))
+		return -EINVAL;
+
+	vma->vm_ops = &erofs_dax_vm_ops;
+	vma->vm_flags |= VM_HUGEPAGE;
+	return 0;
+}
+#else
+#define erofs_file_mmap	generic_file_readonly_mmap
+#endif
+
 const struct file_operations erofs_file_fops = {
 	.llseek		= generic_file_llseek,
 	.read_iter	= erofs_file_read_iter,
-	.mmap		= generic_file_readonly_mmap,
+	.mmap		= erofs_file_mmap,
 	.splice_read	= generic_file_splice_read,
 };
diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
index 00edb7562fea..695b97acb9a6 100644
--- a/fs/erofs/inode.c
+++ b/fs/erofs/inode.c
@@ -174,6 +174,11 @@ static struct page *erofs_read_inode(struct inode *inode,
 	inode->i_mtime.tv_nsec = inode->i_ctime.tv_nsec;
 	inode->i_atime.tv_nsec = inode->i_ctime.tv_nsec;
 
+	inode->i_flags &= ~S_DAX;
+	if (test_opt(&sbi->ctx, DAX) && S_ISREG(inode->i_mode) &&
+	    vi->datalayout == EROFS_INODE_FLAT_PLAIN)
+		inode->i_flags |= S_DAX;
+
 	if (!nblks)
 		/* measure inode.i_blocks as generic filesystems */
 		inode->i_blocks = roundup(inode->i_size, EROFS_BLKSIZ) >> 9;
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index 2669c785d548..8b0542d35148 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -83,6 +83,7 @@ struct erofs_sb_info {
 
 	struct erofs_sb_lz4_info lz4;
 #endif	/* CONFIG_EROFS_FS_ZIP */
+	struct dax_device *dax_dev;
 	u32 blocks;
 	u32 meta_blkaddr;
 #ifdef CONFIG_EROFS_FS_XATTR
@@ -115,6 +116,7 @@ struct erofs_sb_info {
 /* Mount flags set via mount options or defaults */
 #define EROFS_MOUNT_XATTR_USER		0x00000010
 #define EROFS_MOUNT_POSIX_ACL		0x00000020
+#define EROFS_MOUNT_DAX			0x00000040
 
 #define clear_opt(ctx, option)	((ctx)->mount_opt &= ~EROFS_MOUNT_##option)
 #define set_opt(ctx, option)	((ctx)->mount_opt |= EROFS_MOUNT_##option)
diff --git a/fs/erofs/super.c b/fs/erofs/super.c
index 8fc6c04b54f4..b44a964ab24f 100644
--- a/fs/erofs/super.c
+++ b/fs/erofs/super.c
@@ -11,6 +11,7 @@
 #include <linux/crc32c.h>
 #include <linux/fs_context.h>
 #include <linux/fs_parser.h>
+#include <linux/dax.h>
 #include "xattr.h"
 
 #define CREATE_TRACE_POINTS
@@ -355,6 +356,7 @@ enum {
 	Opt_user_xattr,
 	Opt_acl,
 	Opt_cache_strategy,
+	Opt_dax,
 	Opt_err
 };
 
@@ -370,6 +372,7 @@ static const struct fs_parameter_spec erofs_fs_parameters[] = {
 	fsparam_flag_no("acl",		Opt_acl),
 	fsparam_enum("cache_strategy",	Opt_cache_strategy,
 		     erofs_param_cache_strategy),
+	fsparam_flag("dax",             Opt_dax),
 	{}
 };
 
@@ -410,6 +413,14 @@ static int erofs_fc_parse_param(struct fs_context *fc,
 		ctx->cache_strategy = result.uint_32;
 #else
 		errorfc(fc, "compression not supported, cache_strategy ignored");
+#endif
+		break;
+	case Opt_dax:
+#ifdef CONFIG_FS_DAX
+		warnfc(fc, "DAX enabled. Warning: EXPERIMENTAL, use at your own risk");
+		set_opt(ctx, DAX);
+#else
+		errorfc(fc, "dax options not supported");
 #endif
 		break;
 	default:
@@ -496,10 +507,17 @@ static int erofs_fc_fill_super(struct super_block *sb, struct fs_context *fc)
 		return -ENOMEM;
 
 	sb->s_fs_info = sbi;
+	sbi->dax_dev = fs_dax_get_by_bdev(sb->s_bdev);
 	err = erofs_read_superblock(sb);
 	if (err)
 		return err;
 
+	if (test_opt(ctx, DAX) &&
+	    !bdev_dax_supported(sb->s_bdev, EROFS_BLKSIZ)) {
+		errorfc(fc, "DAX unsupported by block device. Turning off DAX.");
+		clear_opt(ctx, DAX);
+	}
+
 	sb->s_flags |= SB_RDONLY | SB_NOATIME;
 	sb->s_maxbytes = MAX_LFS_FILESIZE;
 	sb->s_time_gran = 1;
@@ -609,6 +627,8 @@ static void erofs_kill_sb(struct super_block *sb)
 	sbi = EROFS_SB(sb);
 	if (!sbi)
 		return;
+	if (sbi->dax_dev)
+		fs_put_dax(sbi->dax_dev);
 	kfree(sbi);
 	sb->s_fs_info = NULL;
 }
@@ -711,8 +731,8 @@ static int erofs_statfs(struct dentry *dentry, struct kstatfs *buf)
 
 static int erofs_show_options(struct seq_file *seq, struct dentry *root)
 {
-	struct erofs_sb_info *sbi __maybe_unused = EROFS_SB(root->d_sb);
-	struct erofs_fs_context *ctx __maybe_unused = &sbi->ctx;
+	struct erofs_sb_info *sbi = EROFS_SB(root->d_sb);
+	struct erofs_fs_context *ctx = &sbi->ctx;
 
 #ifdef CONFIG_EROFS_FS_XATTR
 	if (test_opt(ctx, XATTR_USER))
@@ -734,6 +754,8 @@ static int erofs_show_options(struct seq_file *seq, struct dentry *root)
 	else if (ctx->cache_strategy == EROFS_ZIP_CACHE_READAROUND)
 		seq_puts(seq, ",cache_strategy=readaround");
 #endif
+	if (test_opt(ctx, DAX))
+		seq_puts(seq, ",dax");
 	return 0;
 }
 
-- 
2.24.4


  reply	other threads:[~2021-07-05 13:22 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-04 13:50 [RFC PATCH 0/2] erofs: dio/dax support for non-tailpacking cases Gao Xiang
2021-07-04 13:50 ` [RFC PATCH 1/2] erofs: iomap support for non-tailpacking DIO Gao Xiang
2021-07-04 13:50 ` [RFC PATCH 2/2] erofs: dax support for non-tailpacking regular file Gao Xiang
2021-07-05 13:21   ` Gao Xiang [this message]
2021-07-09  1:47     ` [RFC PATCH v1.1 " Darrick J. Wong
2021-07-09  2:28       ` Gao Xiang

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=20210705132153.223839-1-hsiangkao@linux.alibaba.com \
    --to=hsiangkao@linux.alibaba.com \
    --cc=bo.liu@linux.alibaba.com \
    --cc=djwong@kernel.org \
    --cc=gerry@linux.alibaba.com \
    --cc=joseph.qi@linux.alibaba.com \
    --cc=linux-erofs@lists.ozlabs.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=nvdimm@lists.linux.dev \
    /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).