Revision 472e5b056f000a778abb41f1e443de58eb259783 authored by Linus Torvalds on 02 October 2020, 02:14:36 UTC, committed by Linus Torvalds on 02 October 2020, 02:14:36 UTC
The pipe splice code still used the old model of waiting for pipe IO by using a non-specific "pipe_wait()" that waited for any pipe event to happen, which depended on all pipe IO being entirely serialized by the pipe lock. So by checking the state you were waiting for, and then adding yourself to the wait queue before dropping the lock, you were guaranteed to see all the wakeups. Strictly speaking, the actual wakeups were not done under the lock, but the pipe_wait() model still worked, because since the waiter held the lock when checking whether it should sleep, it would always see the current state, and the wakeup was always done after updating the state. However, commit 0ddad21d3e99 ("pipe: use exclusive waits when reading or writing") split the single wait-queue into two, and in the process also made the "wait for event" code wait for _two_ wait queues, and that then showed a race with the wakers that were not serialized by the pipe lock. It's only splice that used that "pipe_wait()" model, so the problem wasn't obvious, but Josef Bacik reports: "I hit a hang with fstest btrfs/187, which does a btrfs send into /dev/null. This works by creating a pipe, the write side is given to the kernel to write into, and the read side is handed to a thread that splices into a file, in this case /dev/null. The box that was hung had the write side stuck here [pipe_write] and the read side stuck here [splice_from_pipe_next -> pipe_wait]. [ more details about pipe_wait() scenario ] The problem is we're doing the prepare_to_wait, which sets our state each time, however we can be woken up either with reads or writes. In the case above we race with the WRITER waking us up, and re-set our state to INTERRUPTIBLE, and thus never break out of schedule" Josef had a patch that avoided the issue in pipe_wait() by just making it set the state only once, but the deeper problem is that pipe_wait() depends on a level of synchonization by the pipe mutex that it really shouldn't. And the whole "wait for any pipe state change" model really isn't very good to begin with. So rather than trying to work around things in pipe_wait(), remove that legacy model of "wait for arbitrary pipe event" entirely, and actually create functions that wait for the pipe actually being readable or writable, and can do so without depending on the pipe lock serializing everything. Fixes: 0ddad21d3e99 ("pipe: use exclusive waits when reading or writing") Link: https://lore.kernel.org/linux-fsdevel/bfa88b5ad6f069b2b679316b9e495a970130416c.1601567868.git.josef@toxicpanda.com/ Reported-by: Josef Bacik <josef@toxicpanda.com> Reviewed-and-tested-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1 parent 44b6e23
init.c
// SPDX-License-Identifier: GPL-2.0
/*
* Routines that mimic syscalls, but don't use the user address space or file
* descriptors. Only for init/ and related early init code.
*/
#include <linux/init.h>
#include <linux/mount.h>
#include <linux/namei.h>
#include <linux/fs.h>
#include <linux/fs_struct.h>
#include <linux/file.h>
#include <linux/init_syscalls.h>
#include <linux/security.h>
#include "internal.h"
int __init init_mount(const char *dev_name, const char *dir_name,
const char *type_page, unsigned long flags, void *data_page)
{
struct path path;
int ret;
ret = kern_path(dir_name, LOOKUP_FOLLOW, &path);
if (ret)
return ret;
ret = path_mount(dev_name, &path, type_page, flags, data_page);
path_put(&path);
return ret;
}
int __init init_umount(const char *name, int flags)
{
int lookup_flags = LOOKUP_MOUNTPOINT;
struct path path;
int ret;
if (!(flags & UMOUNT_NOFOLLOW))
lookup_flags |= LOOKUP_FOLLOW;
ret = kern_path(name, lookup_flags, &path);
if (ret)
return ret;
return path_umount(&path, flags);
}
int __init init_chdir(const char *filename)
{
struct path path;
int error;
error = kern_path(filename, LOOKUP_FOLLOW | LOOKUP_DIRECTORY, &path);
if (error)
return error;
error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
if (!error)
set_fs_pwd(current->fs, &path);
path_put(&path);
return error;
}
int __init init_chroot(const char *filename)
{
struct path path;
int error;
error = kern_path(filename, LOOKUP_FOLLOW | LOOKUP_DIRECTORY, &path);
if (error)
return error;
error = inode_permission(path.dentry->d_inode, MAY_EXEC | MAY_CHDIR);
if (error)
goto dput_and_out;
error = -EPERM;
if (!ns_capable(current_user_ns(), CAP_SYS_CHROOT))
goto dput_and_out;
error = security_path_chroot(&path);
if (error)
goto dput_and_out;
set_fs_root(current->fs, &path);
dput_and_out:
path_put(&path);
return error;
}
int __init init_chown(const char *filename, uid_t user, gid_t group, int flags)
{
int lookup_flags = (flags & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
struct path path;
int error;
error = kern_path(filename, lookup_flags, &path);
if (error)
return error;
error = mnt_want_write(path.mnt);
if (!error) {
error = chown_common(&path, user, group);
mnt_drop_write(path.mnt);
}
path_put(&path);
return error;
}
int __init init_chmod(const char *filename, umode_t mode)
{
struct path path;
int error;
error = kern_path(filename, LOOKUP_FOLLOW, &path);
if (error)
return error;
error = chmod_common(&path, mode);
path_put(&path);
return error;
}
int __init init_eaccess(const char *filename)
{
struct path path;
int error;
error = kern_path(filename, LOOKUP_FOLLOW, &path);
if (error)
return error;
error = inode_permission(d_inode(path.dentry), MAY_ACCESS);
path_put(&path);
return error;
}
int __init init_stat(const char *filename, struct kstat *stat, int flags)
{
int lookup_flags = (flags & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
struct path path;
int error;
error = kern_path(filename, lookup_flags, &path);
if (error)
return error;
error = vfs_getattr(&path, stat, STATX_BASIC_STATS,
flags | AT_NO_AUTOMOUNT);
path_put(&path);
return error;
}
int __init init_mknod(const char *filename, umode_t mode, unsigned int dev)
{
struct dentry *dentry;
struct path path;
int error;
if (S_ISFIFO(mode) || S_ISSOCK(mode))
dev = 0;
else if (!(S_ISBLK(mode) || S_ISCHR(mode)))
return -EINVAL;
dentry = kern_path_create(AT_FDCWD, filename, &path, 0);
if (IS_ERR(dentry))
return PTR_ERR(dentry);
if (!IS_POSIXACL(path.dentry->d_inode))
mode &= ~current_umask();
error = security_path_mknod(&path, dentry, mode, dev);
if (!error)
error = vfs_mknod(path.dentry->d_inode, dentry, mode,
new_decode_dev(dev));
done_path_create(&path, dentry);
return error;
}
int __init init_link(const char *oldname, const char *newname)
{
struct dentry *new_dentry;
struct path old_path, new_path;
int error;
error = kern_path(oldname, 0, &old_path);
if (error)
return error;
new_dentry = kern_path_create(AT_FDCWD, newname, &new_path, 0);
error = PTR_ERR(new_dentry);
if (IS_ERR(new_dentry))
goto out;
error = -EXDEV;
if (old_path.mnt != new_path.mnt)
goto out_dput;
error = may_linkat(&old_path);
if (unlikely(error))
goto out_dput;
error = security_path_link(old_path.dentry, &new_path, new_dentry);
if (error)
goto out_dput;
error = vfs_link(old_path.dentry, new_path.dentry->d_inode, new_dentry,
NULL);
out_dput:
done_path_create(&new_path, new_dentry);
out:
path_put(&old_path);
return error;
}
int __init init_symlink(const char *oldname, const char *newname)
{
struct dentry *dentry;
struct path path;
int error;
dentry = kern_path_create(AT_FDCWD, newname, &path, 0);
if (IS_ERR(dentry))
return PTR_ERR(dentry);
error = security_path_symlink(&path, dentry, oldname);
if (!error)
error = vfs_symlink(path.dentry->d_inode, dentry, oldname);
done_path_create(&path, dentry);
return error;
}
int __init init_unlink(const char *pathname)
{
return do_unlinkat(AT_FDCWD, getname_kernel(pathname));
}
int __init init_mkdir(const char *pathname, umode_t mode)
{
struct dentry *dentry;
struct path path;
int error;
dentry = kern_path_create(AT_FDCWD, pathname, &path, LOOKUP_DIRECTORY);
if (IS_ERR(dentry))
return PTR_ERR(dentry);
if (!IS_POSIXACL(path.dentry->d_inode))
mode &= ~current_umask();
error = security_path_mkdir(&path, dentry, mode);
if (!error)
error = vfs_mkdir(path.dentry->d_inode, dentry, mode);
done_path_create(&path, dentry);
return error;
}
int __init init_rmdir(const char *pathname)
{
return do_rmdir(AT_FDCWD, getname_kernel(pathname));
}
int __init init_utimes(char *filename, struct timespec64 *ts)
{
struct path path;
int error;
error = kern_path(filename, 0, &path);
if (error)
return error;
error = vfs_utimes(&path, ts);
path_put(&path);
return error;
}
int __init init_dup(struct file *file)
{
int fd;
fd = get_unused_fd_flags(0);
if (fd < 0)
return fd;
fd_install(fd, get_file(file));
return 0;
}
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...