[PATCH v5 0/5] io_uring: add xattr support

[PATCH v5 0/5] io_uring: add xattr support

[Stefan Roesch]

This adds the xattr support to io_uring. The intent is to have a more
complete support for file operations in io_uring.

This change adds support for the following functions to io_uring:
- fgetxattr
- fsetxattr
- getxattr
- setxattr

Patch 1: fs: split off do_user_path_at_empty from user_path_at_empty()
  This splits off a new function do_user_path_at_empty from
  user_path_at_empty that is based on filename and not on a
  user-specified string.

Patch 2: fs: split off setxattr_setup function from setxattr
  Split off the setup part of the setxattr function.

Patch 3: fs: split off do_getxattr from getxattr
  Split of the do_getxattr part from getxattr. This will
  allow it to be invoked it from io_uring.

Patch 4: io_uring: add fsetxattr and setxattr support
  This adds new functions to support the fsetxattr and setxattr
  functions.

Patch 5: io_uring: add fgetxattr and getxattr support
  This adds new functions to support the fgetxattr and getxattr
  functions.


There are two additional patches:
  liburing: Add support for xattr api's.
            This also includes the tests for the new code.
  xfstests: Add support for io_uring xattr support.

V5: - add kname array to xattr_ctx structure
V4: - rebased patch series
V3: - remove req->file checks in prep functions
    - change size parameter in do_xattr
V2: - split off function do_user_path_empty instead of changing
      the function signature of user_path_at
    - Fix datatype size problem in do_getxattr

Stefan Roesch (5):
  fs: split off do_user_path_at_empty from user_path_at_empty()
  fs: split off setxattr_setup function from setxattr
  fs: split off do_getxattr from getxattr
  io_uring: add fsetxattr and setxattr support
  io_uring: add fgetxattr and getxattr support

 fs/internal.h                 |  22 +++
 fs/io_uring.c                 | 301 ++++++++++++++++++++++++++++++++++
 fs/namei.c                    |  10 +-
 fs/xattr.c                    | 105 +++++++-----
 include/linux/namei.h         |   2 +
 include/uapi/linux/io_uring.h |   8 +-
 6 files changed, 408 insertions(+), 40 deletions(-)


base-commit: d09358c3d161dcea8f02eae1281bc996819cc769
-- 
2.30.2

[PATCH v5 1/5] fs: split off do_user_path_at_empty from user_path_at_empty()

[Stefan Roesch]

This splits off a do_user_path_at_empty function from the
user_path_at_empty_function. This is required so it can be
called from io_uring.

Signed-off-by: Stefan Roesch <shr@fb.com>
---
 fs/namei.c            | 10 ++++++++--
 include/linux/namei.h |  2 ++
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/fs/namei.c b/fs/namei.c
index 1f9d2187c765..d988e241b32c 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2794,12 +2794,18 @@ int path_pts(struct path *path)
 }
 #endif
 
+int do_user_path_at_empty(int dfd, struct filename *filename, unsigned int flags,
+		       struct path *path)
+{
+	return filename_lookup(dfd, filename, flags, path, NULL);
+}
+
 int user_path_at_empty(int dfd, const char __user *name, unsigned flags,
-		 struct path *path, int *empty)
+		struct path *path, int *empty)
 {
 	struct filename *filename = getname_flags(name, flags, empty);
-	int ret = filename_lookup(dfd, filename, flags, path, NULL);
 
+	int ret = do_user_path_at_empty(dfd, filename, flags, path);
 	putname(filename);
 	return ret;
 }
diff --git a/include/linux/namei.h b/include/linux/namei.h
index e89329bb3134..8f3ef38c057b 100644
--- a/include/linux/namei.h
+++ b/include/linux/namei.h
@@ -49,6 +49,8 @@ enum {LAST_NORM, LAST_ROOT, LAST_DOT, LAST_DOTDOT};
 
 extern int path_pts(struct path *path);
 
+extern int do_user_path_at_empty(int dfd, struct filename *filename,
+				unsigned int flags, struct path *path);
 extern int user_path_at_empty(int, const char __user *, unsigned, struct path *, int *empty);
 
 static inline int user_path_at(int dfd, const char __user *name, unsigned flags,
-- 
2.30.2

[PATCH v5 2/5] fs: split off setxattr_setup function from setxattr

[Stefan Roesch]

This splits of the setup part of the function
setxattr in its own dedicated function called
setxattr_setup.

This makes it possible to call this function
from io_uring in the pre-processing of an
xattr request.

Signed-off-by: Stefan Roesch <shr@fb.com>
---
 fs/internal.h | 16 +++++++++++
 fs/xattr.c    | 74 +++++++++++++++++++++++++++++++++------------------
 2 files changed, 64 insertions(+), 26 deletions(-)

diff --git a/fs/internal.h b/fs/internal.h
index 7979ff8d168c..d13b9f13df09 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -194,3 +194,19 @@ long splice_file_to_pipe(struct file *in,
 			 struct pipe_inode_info *opipe,
 			 loff_t *offset,
 			 size_t len, unsigned int flags);
+
+ /*
+  * fs/xattr.c:
+  */
+struct xattr_ctx {
+	/* Value of attribute */
+	const void __user *value;
+	size_t size;
+	/* Attribute name */
+	char kname[XATTR_NAME_MAX + 1];
+	unsigned int flags;
+};
+
+void *setxattr_setup(struct user_namespace *mnt_userns,
+		     const char __user *name,
+		     struct xattr_ctx *data);
diff --git a/fs/xattr.c b/fs/xattr.c
index 5c8c5175b385..a4b59523bd0e 100644
--- a/fs/xattr.c
+++ b/fs/xattr.c
@@ -25,6 +25,8 @@
 
 #include <linux/uaccess.h>
 
+#include "internal.h"
+
 static const char *
 strcmp_prefix(const char *a, const char *a_prefix)
 {
@@ -539,43 +541,63 @@ EXPORT_SYMBOL_GPL(vfs_removexattr);
 /*
  * Extended attribute SET operations
  */
-static long
-setxattr(struct user_namespace *mnt_userns, struct dentry *d,
-	 const char __user *name, const void __user *value, size_t size,
-	 int flags)
+
+void *setxattr_setup(struct user_namespace *mnt_userns, const char __user *name,
+		struct xattr_ctx *ctx)
 {
-	int error;
 	void *kvalue = NULL;
-	char kname[XATTR_NAME_MAX + 1];
+	int error;
 
-	if (flags & ~(XATTR_CREATE|XATTR_REPLACE))
-		return -EINVAL;
+	if (ctx->flags & ~(XATTR_CREATE|XATTR_REPLACE))
+		return ERR_PTR(-EINVAL);
 
-	error = strncpy_from_user(kname, name, sizeof(kname));
-	if (error == 0 || error == sizeof(kname))
-		error = -ERANGE;
+	error = strncpy_from_user(ctx->kname, name, sizeof(ctx->kname));
+	if (error == 0 || error == sizeof(ctx->kname))
+		return  ERR_PTR(-ERANGE);
 	if (error < 0)
-		return error;
+		return ERR_PTR(error);
 
-	if (size) {
-		if (size > XATTR_SIZE_MAX)
-			return -E2BIG;
-		kvalue = kvmalloc(size, GFP_KERNEL);
+	if (ctx->size) {
+		if (ctx->size > XATTR_SIZE_MAX)
+			return ERR_PTR(-E2BIG);
+
+		kvalue = kvmalloc(ctx->size, GFP_KERNEL);
 		if (!kvalue)
-			return -ENOMEM;
-		if (copy_from_user(kvalue, value, size)) {
-			error = -EFAULT;
-			goto out;
+			return ERR_PTR(-ENOMEM);
+
+		if (copy_from_user(kvalue, ctx->value, ctx->size)) {
+			kvfree(kvalue);
+			return ERR_PTR(-EFAULT);
 		}
-		if ((strcmp(kname, XATTR_NAME_POSIX_ACL_ACCESS) == 0) ||
-		    (strcmp(kname, XATTR_NAME_POSIX_ACL_DEFAULT) == 0))
-			posix_acl_fix_xattr_from_user(mnt_userns, kvalue, size);
+
+		if ((strcmp(ctx->kname, XATTR_NAME_POSIX_ACL_ACCESS) == 0) ||
+		    (strcmp(ctx->kname, XATTR_NAME_POSIX_ACL_DEFAULT) == 0))
+			posix_acl_fix_xattr_from_user(mnt_userns, kvalue, ctx->size);
 	}
 
-	error = vfs_setxattr(mnt_userns, d, kname, kvalue, size, flags);
-out:
-	kvfree(kvalue);
+	return kvalue;
+}
 
+static long
+setxattr(struct user_namespace *mnt_userns, struct dentry *d,
+	const char __user *name, const void __user *value, size_t size,
+	int flags)
+{
+	struct xattr_ctx ctx = {
+		.value    = value,
+		.size     = size,
+		.flags    = flags,
+	};
+	void *kvalue;
+	int error;
+
+	kvalue = setxattr_setup(mnt_userns, name, &ctx);
+	if (IS_ERR(kvalue))
+		return PTR_ERR(kvalue);
+
+	error = vfs_setxattr(mnt_userns, d, ctx.kname, kvalue, size, flags);
+
+	kvfree(kvalue);
 	return error;
 }
 
-- 
2.30.2

[PATCH v5 3/5] fs: split off do_getxattr from getxattr

[Stefan Roesch]

This splits off do_getxattr function from the getxattr
function. This will allow io_uring to call it from its
io worker.

Signed-off-by: Stefan Roesch <shr@fb.com>
---
 fs/internal.h |  6 ++++++
 fs/xattr.c    | 31 ++++++++++++++++++++-----------
 2 files changed, 26 insertions(+), 11 deletions(-)

diff --git a/fs/internal.h b/fs/internal.h
index d13b9f13df09..efd2c4f7a536 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -207,6 +207,12 @@ struct xattr_ctx {
 	unsigned int flags;
 };
 
+ssize_t do_getxattr(struct user_namespace *mnt_userns,
+		    struct dentry *d,
+		    const char *kname,
+		    void __user *value,
+		    size_t size);
+
 void *setxattr_setup(struct user_namespace *mnt_userns,
 		     const char __user *name,
 		     struct xattr_ctx *data);
diff --git a/fs/xattr.c b/fs/xattr.c
index a4b59523bd0e..1d9795bc8be6 100644
--- a/fs/xattr.c
+++ b/fs/xattr.c
@@ -663,19 +663,12 @@ SYSCALL_DEFINE5(fsetxattr, int, fd, const char __user *, name,
 /*
  * Extended attribute GET operations
  */
-static ssize_t
-getxattr(struct user_namespace *mnt_userns, struct dentry *d,
-	 const char __user *name, void __user *value, size_t size)
+ssize_t
+do_getxattr(struct user_namespace *mnt_userns, struct dentry *d,
+	 const char *kname, void __user *value, size_t size)
 {
-	ssize_t error;
 	void *kvalue = NULL;
-	char kname[XATTR_NAME_MAX + 1];
-
-	error = strncpy_from_user(kname, name, sizeof(kname));
-	if (error == 0 || error == sizeof(kname))
-		error = -ERANGE;
-	if (error < 0)
-		return error;
+	ssize_t error;
 
 	if (size) {
 		if (size > XATTR_SIZE_MAX)
@@ -703,6 +696,22 @@ getxattr(struct user_namespace *mnt_userns, struct dentry *d,
 	return error;
 }
 
+static ssize_t
+getxattr(struct user_namespace *mnt_userns, struct dentry *d,
+	 const char __user *name, void __user *value, size_t size)
+{
+	ssize_t error;
+	char kname[XATTR_NAME_MAX + 1];
+
+	error = strncpy_from_user(kname, name, sizeof(kname));
+	if (error == 0 || error == sizeof(kname))
+		error = -ERANGE;
+	if (error < 0)
+		return error;
+
+	return do_getxattr(mnt_userns, d, kname, value, size);
+}
+
 static ssize_t path_getxattr(const char __user *pathname,
 			     const char __user *name, void __user *value,
 			     size_t size, unsigned int lookup_flags)
-- 
2.30.2

[PATCH v5 4/5] io_uring: add fsetxattr and setxattr support

[Stefan Roesch]

This adds support to io_uring for the fsetxattr and setxattr API.

Signed-off-by: Stefan Roesch <shr@fb.com>
---
 fs/io_uring.c                 | 159 ++++++++++++++++++++++++++++++++++
 include/uapi/linux/io_uring.h |   6 +-
 2 files changed, 164 insertions(+), 1 deletion(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 5092dfe56da6..9f56a2e3bc32 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -82,6 +82,7 @@
 #include <linux/audit.h>
 #include <linux/security.h>
 #include <linux/atomic-ref.h>
+#include <linux/xattr.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/io_uring.h>
@@ -719,6 +720,13 @@ struct io_async_rw {
 	struct wait_page_queue		wpq;
 };
 
+struct io_xattr {
+	struct file			*file;
+	struct xattr_ctx		ctx;
+	void				*value;
+	struct filename			*filename;
+};
+
 enum {
 	REQ_F_FIXED_FILE_BIT	= IOSQE_FIXED_FILE_BIT,
 	REQ_F_IO_DRAIN_BIT	= IOSQE_IO_DRAIN_BIT,
@@ -858,6 +866,7 @@ struct io_kiocb {
 		struct io_mkdir		mkdir;
 		struct io_symlink	symlink;
 		struct io_hardlink	hardlink;
+		struct io_xattr		xattr;
 	};
 
 	u8				opcode;
@@ -1107,6 +1116,10 @@ static const struct io_op_def io_op_defs[] = {
 	[IORING_OP_MKDIRAT] = {},
 	[IORING_OP_SYMLINKAT] = {},
 	[IORING_OP_LINKAT] = {},
+	[IORING_OP_FSETXATTR] = {
+		.needs_file = 1
+	},
+	[IORING_OP_SETXATTR] = {},
 };
 
 /* requests with any of those set should undergo io_disarm_next() */
@@ -3876,6 +3889,133 @@ static int io_renameat(struct io_kiocb *req, unsigned int issue_flags)
 	return 0;
 }
 
+static int __io_setxattr_prep(struct io_kiocb *req,
+			const struct io_uring_sqe *sqe,
+			struct user_namespace *user_ns)
+{
+	struct io_xattr *ix = &req->xattr;
+	const char __user *name;
+	void *ret;
+
+	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
+		return -EINVAL;
+	if (unlikely(sqe->ioprio))
+		return -EINVAL;
+	if (unlikely(req->flags & REQ_F_FIXED_FILE))
+		return -EBADF;
+
+	ix->filename = NULL;
+	name = u64_to_user_ptr(READ_ONCE(sqe->addr));
+	ix->ctx.value = u64_to_user_ptr(READ_ONCE(sqe->addr2));
+	ix->ctx.size = READ_ONCE(sqe->len);
+	ix->ctx.flags = READ_ONCE(sqe->xattr_flags);
+
+	ret = setxattr_setup(user_ns, name, &ix->ctx);
+	if (IS_ERR(ret))
+		return PTR_ERR(ret);
+
+	ix->value = ret;
+	req->flags |= REQ_F_NEED_CLEANUP;
+	return 0;
+}
+
+static int io_setxattr_prep(struct io_kiocb *req,
+			const struct io_uring_sqe *sqe)
+{
+	struct io_xattr *ix = &req->xattr;
+	const char __user *path;
+	int ret;
+
+	ret = __io_setxattr_prep(req, sqe, current_user_ns());
+	if (ret)
+		return ret;
+
+	path = u64_to_user_ptr(READ_ONCE(sqe->addr3));
+
+	ix->filename = getname_flags(path, LOOKUP_FOLLOW, NULL);
+	if (IS_ERR(ix->filename)) {
+		ret = PTR_ERR(ix->filename);
+		ix->filename = NULL;
+	}
+
+	return ret;
+}
+
+static int io_fsetxattr_prep(struct io_kiocb *req,
+			const struct io_uring_sqe *sqe)
+{
+	return __io_setxattr_prep(req, sqe, file_mnt_user_ns(req->file));
+}
+
+static int __io_setxattr(struct io_kiocb *req, unsigned int issue_flags,
+			struct path *path)
+{
+	struct io_xattr *ix = &req->xattr;
+	int ret;
+
+	ret = mnt_want_write(path->mnt);
+	if (!ret) {
+		ret = vfs_setxattr(mnt_user_ns(path->mnt), path->dentry,
+				ix->ctx.kname, ix->value, ix->ctx.size,
+				ix->ctx.flags);
+		mnt_drop_write(path->mnt);
+	}
+
+	return ret;
+}
+
+static int io_fsetxattr(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_xattr *ix = &req->xattr;
+	int ret;
+
+	if (issue_flags & IO_URING_F_NONBLOCK)
+		return -EAGAIN;
+
+	ret = __io_setxattr(req, issue_flags, &req->file->f_path);
+
+	req->flags &= ~REQ_F_NEED_CLEANUP;
+	if (ix->value)
+		kvfree(ix->value);
+	if (ret < 0)
+		req_set_fail(req);
+
+	io_req_complete(req, ret);
+	return 0;
+}
+
+static int io_setxattr(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_xattr *ix = &req->xattr;
+	unsigned int lookup_flags = LOOKUP_FOLLOW;
+	struct path path;
+	int ret;
+
+	if (issue_flags & IO_URING_F_NONBLOCK)
+		return -EAGAIN;
+
+retry:
+	ret = do_user_path_at_empty(AT_FDCWD, ix->filename, lookup_flags, &path);
+	putname(ix->filename);
+	if (!ret) {
+		ret = __io_setxattr(req, issue_flags, &path);
+		path_put(&path);
+		if (retry_estale(ret, lookup_flags)) {
+			lookup_flags |= LOOKUP_REVAL;
+			goto retry;
+		}
+	}
+
+	req->flags &= ~REQ_F_NEED_CLEANUP;
+	if (ix->value)
+		kvfree(ix->value);
+	if (ret < 0)
+		req_set_fail(req);
+
+	io_req_complete(req, ret);
+	return 0;
+}
+
 static int io_unlinkat_prep(struct io_kiocb *req,
 			    const struct io_uring_sqe *sqe)
 {
@@ -6574,6 +6714,10 @@ static int io_req_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 		return io_symlinkat_prep(req, sqe);
 	case IORING_OP_LINKAT:
 		return io_linkat_prep(req, sqe);
+	case IORING_OP_FSETXATTR:
+		return io_fsetxattr_prep(req, sqe);
+	case IORING_OP_SETXATTR:
+		return io_setxattr_prep(req, sqe);
 	}
 
 	printk_once(KERN_WARNING "io_uring: unhandled opcode %d\n",
@@ -6715,6 +6859,14 @@ static void io_clean_op(struct io_kiocb *req)
 			putname(req->hardlink.oldpath);
 			putname(req->hardlink.newpath);
 			break;
+		case IORING_OP_SETXATTR:
+			if (req->xattr.filename)
+				putname(req->xattr.filename);
+			fallthrough;
+		case IORING_OP_FSETXATTR:
+			if (req->xattr.value)
+				kvfree(req->xattr.value);
+			break;
 		}
 	}
 	if ((req->flags & REQ_F_POLLED) && req->apoll) {
@@ -6857,6 +7009,12 @@ static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
 	case IORING_OP_LINKAT:
 		ret = io_linkat(req, issue_flags);
 		break;
+	case IORING_OP_FSETXATTR:
+		ret = io_fsetxattr(req, issue_flags);
+		break;
+	case IORING_OP_SETXATTR:
+		ret = io_setxattr(req, issue_flags);
+		break;
 	default:
 		ret = -EINVAL;
 		break;
@@ -11225,6 +11383,7 @@ static int __init io_uring_init(void)
 	BUILD_BUG_SQE_ELEM(42, __u16,  personality);
 	BUILD_BUG_SQE_ELEM(44, __s32,  splice_fd_in);
 	BUILD_BUG_SQE_ELEM(44, __u32,  file_index);
+	BUILD_BUG_SQE_ELEM(48, __u64,  addr3);
 
 	BUILD_BUG_ON(sizeof(struct io_uring_files_update) !=
 		     sizeof(struct io_uring_rsrc_update));
diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
index 787f491f0d2a..dbf473900da2 100644
--- a/include/uapi/linux/io_uring.h
+++ b/include/uapi/linux/io_uring.h
@@ -45,6 +45,7 @@ struct io_uring_sqe {
 		__u32		rename_flags;
 		__u32		unlink_flags;
 		__u32		hardlink_flags;
+		__u32		xattr_flags;
 	};
 	__u64	user_data;	/* data to be passed back at completion time */
 	/* pack this to avoid bogus arm OABI complaints */
@@ -60,7 +61,8 @@ struct io_uring_sqe {
 		__s32	splice_fd_in;
 		__u32	file_index;
 	};
-	__u64	__pad2[2];
+	__u64	addr3;
+	__u64	__pad2[1];
 };
 
 enum {
@@ -143,6 +145,8 @@ enum {
 	IORING_OP_MKDIRAT,
 	IORING_OP_SYMLINKAT,
 	IORING_OP_LINKAT,
+	IORING_OP_FSETXATTR,
+	IORING_OP_SETXATTR,
 
 	/* this goes last, obviously */
 	IORING_OP_LAST,
-- 
2.30.2

[PATCH v5 5/5] io_uring: add fgetxattr and getxattr support

[Stefan Roesch]

This adds support to io_uring for the fgetxattr and getxattr API.

Signed-off-by: Stefan Roesch <shr@fb.com>
---
 fs/io_uring.c                 | 142 ++++++++++++++++++++++++++++++++++
 include/uapi/linux/io_uring.h |   2 +
 2 files changed, 144 insertions(+)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 9f56a2e3bc32..cd9fb60ecf72 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -1116,6 +1116,10 @@ static const struct io_op_def io_op_defs[] = {
 	[IORING_OP_MKDIRAT] = {},
 	[IORING_OP_SYMLINKAT] = {},
 	[IORING_OP_LINKAT] = {},
+	[IORING_OP_FGETXATTR] = {
+		.needs_file = 1
+	},
+	[IORING_OP_GETXATTR] = {},
 	[IORING_OP_FSETXATTR] = {
 		.needs_file = 1
 	},
@@ -3889,6 +3893,125 @@ static int io_renameat(struct io_kiocb *req, unsigned int issue_flags)
 	return 0;
 }
 
+static int __io_getxattr_prep(struct io_kiocb *req,
+			      const struct io_uring_sqe *sqe)
+{
+	struct io_xattr *ix = &req->xattr;
+	const char __user *name;
+	int ret;
+
+	if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL))
+		return -EINVAL;
+	if (unlikely(sqe->ioprio))
+		return -EINVAL;
+	if (unlikely(req->flags & REQ_F_FIXED_FILE))
+		return -EBADF;
+
+	ix->filename = NULL;
+	ix->value = NULL;
+	name = u64_to_user_ptr(READ_ONCE(sqe->addr));
+	ix->ctx.value = u64_to_user_ptr(READ_ONCE(sqe->addr2));
+	ix->ctx.size = READ_ONCE(sqe->len);
+	ix->ctx.flags = READ_ONCE(sqe->xattr_flags);
+
+	if (ix->ctx.flags)
+		return -EINVAL;
+
+	ret = strncpy_from_user(ix->ctx.kname, name, sizeof(ix->ctx.kname));
+	if (!ret || ret == sizeof(ix->ctx.kname))
+		ret = -ERANGE;
+	if (ret < 0)
+		return ret;
+
+	req->flags |= REQ_F_NEED_CLEANUP;
+	return 0;
+}
+
+static int io_fgetxattr_prep(struct io_kiocb *req,
+			     const struct io_uring_sqe *sqe)
+{
+	return __io_getxattr_prep(req, sqe);
+}
+
+static int io_getxattr_prep(struct io_kiocb *req,
+			    const struct io_uring_sqe *sqe)
+{
+	struct io_xattr *ix = &req->xattr;
+	const char __user *path;
+	int ret;
+
+	ret = __io_getxattr_prep(req, sqe);
+	if (ret)
+		return ret;
+
+	path = u64_to_user_ptr(READ_ONCE(sqe->addr3));
+
+	ix->filename = getname_flags(path, LOOKUP_FOLLOW, NULL);
+	if (IS_ERR(ix->filename)) {
+		ret = PTR_ERR(ix->filename);
+		ix->filename = NULL;
+	}
+
+	return ret;
+}
+
+static int io_fgetxattr(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_xattr *ix = &req->xattr;
+	int ret;
+
+	if (issue_flags & IO_URING_F_NONBLOCK)
+		return -EAGAIN;
+
+	ret = do_getxattr(mnt_user_ns(req->file->f_path.mnt),
+			req->file->f_path.dentry,
+			ix->ctx.kname,
+			(void __user *)ix->ctx.value,
+			ix->ctx.size);
+
+	req->flags &= ~REQ_F_NEED_CLEANUP;
+	if (ret < 0)
+		req_set_fail(req);
+
+	io_req_complete(req, ret);
+	return 0;
+}
+
+static int io_getxattr(struct io_kiocb *req, unsigned int issue_flags)
+{
+	struct io_xattr *ix = &req->xattr;
+	unsigned int lookup_flags = LOOKUP_FOLLOW;
+	struct path path;
+	int ret;
+
+	if (issue_flags & IO_URING_F_NONBLOCK)
+		return -EAGAIN;
+
+retry:
+	ret = do_user_path_at_empty(AT_FDCWD, ix->filename, lookup_flags, &path);
+	putname(ix->filename);
+	if (!ret) {
+		ret = do_getxattr(mnt_user_ns(path.mnt),
+				  path.dentry,
+				  ix->ctx.kname,
+				  (void __user *)ix->ctx.value,
+				  ix->ctx.size);
+
+		path_put(&path);
+		if (retry_estale(ret, lookup_flags)) {
+			lookup_flags |= LOOKUP_REVAL;
+			goto retry;
+		}
+	}
+
+	req->flags &= ~REQ_F_NEED_CLEANUP;
+	if (ret < 0)
+		req_set_fail(req);
+
+	io_req_complete(req, ret);
+	return 0;
+}
+
 static int __io_setxattr_prep(struct io_kiocb *req,
 			const struct io_uring_sqe *sqe,
 			struct user_namespace *user_ns)
@@ -6714,6 +6837,10 @@ static int io_req_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 		return io_symlinkat_prep(req, sqe);
 	case IORING_OP_LINKAT:
 		return io_linkat_prep(req, sqe);
+	case IORING_OP_FGETXATTR:
+		return io_fgetxattr_prep(req, sqe);
+	case IORING_OP_GETXATTR:
+		return io_getxattr_prep(req, sqe);
 	case IORING_OP_FSETXATTR:
 		return io_fsetxattr_prep(req, sqe);
 	case IORING_OP_SETXATTR:
@@ -6859,6 +6986,15 @@ static void io_clean_op(struct io_kiocb *req)
 			putname(req->hardlink.oldpath);
 			putname(req->hardlink.newpath);
 			break;
+
+		case IORING_OP_GETXATTR:
+			if (req->xattr.filename)
+				putname(req->xattr.filename);
+			break;
+
+		case IORING_OP_FGETXATTR:
+			break;
+
 		case IORING_OP_SETXATTR:
 			if (req->xattr.filename)
 				putname(req->xattr.filename);
@@ -7009,6 +7145,12 @@ static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
 	case IORING_OP_LINKAT:
 		ret = io_linkat(req, issue_flags);
 		break;
+	case IORING_OP_FGETXATTR:
+		ret = io_fgetxattr(req, issue_flags);
+		break;
+	case IORING_OP_GETXATTR:
+		ret = io_getxattr(req, issue_flags);
+		break;
 	case IORING_OP_FSETXATTR:
 		ret = io_fsetxattr(req, issue_flags);
 		break;
diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
index dbf473900da2..cd9160272308 100644
--- a/include/uapi/linux/io_uring.h
+++ b/include/uapi/linux/io_uring.h
@@ -145,7 +145,9 @@ enum {
 	IORING_OP_MKDIRAT,
 	IORING_OP_SYMLINKAT,
 	IORING_OP_LINKAT,
+	IORING_OP_FGETXATTR,
 	IORING_OP_FSETXATTR,
+	IORING_OP_GETXATTR,
 	IORING_OP_SETXATTR,
 
 	/* this goes last, obviously */
-- 
2.30.2

Re: [PATCH v5 3/5] fs: split off do_getxattr from getxattr

[Linus Torvalds]

On Tue, Dec 21, 2021 at 8:50 AM Stefan Roesch <shr@fb.com> wrote:
>
> This splits off do_getxattr function from the getxattr
> function. This will allow io_uring to call it from its
> io worker.

Hmm.

My reaction to this one is

 "Why isn't do_getxattr() using 'struct xattr_ctx' for its context?"

As far as I can tell, that's *exactly* what it wants, and it would be
logical to match up with the setxattr side.

Yeah, yeah, setxattr has a 'const void __user *value' while getxattr
obviously has just a 'void __user *value'. But if the cost of having a
unified interface is that you lose the 'const' part for the setxattr,
I think that's still a good thing.

Yes? No? Comments?

              Linus

Re: [PATCH v5 3/5] fs: split off do_getxattr from getxattr

[Stefan Roesch]

On 12/21/21 9:22 AM, Linus Torvalds wrote:
> On Tue, Dec 21, 2021 at 8:50 AM Stefan Roesch <shr@fb.com> wrote:
>>
>> This splits off do_getxattr function from the getxattr
>> function. This will allow io_uring to call it from its
>> io worker.
> 
> Hmm.
> 
> My reaction to this one is
> 
>  "Why isn't do_getxattr() using 'struct xattr_ctx' for its context?"
> 
> As far as I can tell, that's *exactly* what it wants, and it would be
> logical to match up with the setxattr side.
> 
> Yeah, yeah, setxattr has a 'const void __user *value' while getxattr
> obviously has just a 'void __user *value'. But if the cost of having a
> unified interface is that you lose the 'const' part for the setxattr,
> I think that's still a good thing.
> 
> Yes? No? Comments?

Linus, if we remove the constness, then we either need to cast away the constness (the system call
is defined as const) or change the definition of the system call.


> 
>               Linus
> 

Re: [PATCH v5 3/5] fs: split off do_getxattr from getxattr

[Linus Torvalds]

On Tue, Dec 21, 2021 at 11:15 AM Stefan Roesch <shr@fb.com> wrote:
>
> Linus, if we remove the constness, then we either need to cast away the constness (the system call
> is defined as const) or change the definition of the system call.

You could also do it as

        union {
                const void __user *setxattr_value;
                void __user *getxattr_value;
        };

if you wanted to..

                 Linus

Re: [PATCH v5 3/5] fs: split off do_getxattr from getxattr

[Stefan Roesch]

On 12/21/21 11:18 AM, Linus Torvalds wrote:
> On Tue, Dec 21, 2021 at 11:15 AM Stefan Roesch <shr@fb.com> wrote:
>>
>> Linus, if we remove the constness, then we either need to cast away the constness (the system call
>> is defined as const) or change the definition of the system call.
> 
> You could also do it as
> 
>         union {
>                 const void __user *setxattr_value;
>                 void __user *getxattr_value;
>         };
>

Pavel brought up a very good point. By adding the kname array into the 
xarray_ctx we increase the size of io_xattr structure too much. In addition
this will also increase the size of the io_kiocb structure. The original
solution did not increase the size.

Per opcode we limit the storage space to 64 bytes. However the array itself
requires 256 bytes.
 
> if you wanted to..
> 
>                  Linus
> 

Re: [PATCH v5 3/5] fs: split off do_getxattr from getxattr

[Jens Axboe]

On 12/21/21 2:59 PM, Stefan Roesch wrote:
> 
> 
> On 12/21/21 11:18 AM, Linus Torvalds wrote:
>> On Tue, Dec 21, 2021 at 11:15 AM Stefan Roesch <shr@fb.com> wrote:
>>>
>>> Linus, if we remove the constness, then we either need to cast away the constness (the system call
>>> is defined as const) or change the definition of the system call.
>>
>> You could also do it as
>>
>>         union {
>>                 const void __user *setxattr_value;
>>                 void __user *getxattr_value;
>>         };
>>
> 
> Pavel brought up a very good point. By adding the kname array into the
> xarray_ctx we increase the size of io_xattr structure too much. In
> addition this will also increase the size of the io_kiocb structure.
> The original solution did not increase the size.
> 
> Per opcode we limit the storage space to 64 bytes. However the array
> itself requires 256 bytes.

Just to expand on that a bit - part of struct io_kiocb is per-command
data, and we try pretty hard to keep that at 64-bytes as that's the
largest one we currently have. If we add the array to the io_xattr
structure, then that will increase the whole io_kiocb from 224 bytes to
more than twice that.

So there are really two options here:

1) The xattr_ctx structure goes into the async data that a command has
   to allocate for deferred execution. This isn't a _huge_ deal as we
   have to defer the xattr commands for now anyway, as the VFS doesn't
   support a nonblocking version of that yet. But it would still be nice
   not to have to do that.

2) We keep the original interface that Stefan proposed, leaving the
   xattr_ctx bits embedded as they fit fine like that.

#2 would be a bit more efficient, but I don't feel that strongly about
it for this particular case.

Comments?

-- 
Jens Axboe