This is a note to let you know that I've just added the patch titled
media: v4l2-compat-ioctl32.c: don't copy back the result for certain errors
to the 4.15-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
The filename of the patch is:
media-v4l2-compat-ioctl32.c-don-t-copy-back-the-result-for-certain-errors.patch
and it can be found in the queue-4.15 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable(a)vger.kernel.org> know about it.
>From d83a8243aaefe62ace433e4384a4f077bed86acb Mon Sep 17 00:00:00 2001
From: Hans Verkuil <hans.verkuil(a)cisco.com>
Date: Fri, 26 Jan 2018 03:24:53 -0500
Subject: media: v4l2-compat-ioctl32.c: don't copy back the result for certain errors
From: Hans Verkuil <hans.verkuil(a)cisco.com>
commit d83a8243aaefe62ace433e4384a4f077bed86acb upstream.
Some ioctls need to copy back the result even if the ioctl returned
an error. However, don't do this for the error code -ENOTTY.
It makes no sense in that cases.
Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com>
Acked-by: Sakari Ailus <sakari.ailus(a)linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab(a)s-opensource.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
---
drivers/media/v4l2-core/v4l2-compat-ioctl32.c | 3 +++
1 file changed, 3 insertions(+)
--- a/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
+++ b/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
@@ -968,6 +968,9 @@ static long do_video_ioctl(struct file *
set_fs(old_fs);
}
+ if (err == -ENOTTY)
+ return err;
+
/* Special case: even after an error we need to put the
results back for these ioctls since the error_idx will
contain information on which control failed. */
Patches currently in stable-queue which might be from hans.verkuil(a)cisco.com are
queue-4.15/media-hdpvr-fix-an-error-handling-path-in-hdpvr_probe.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-m.userptr-in-put_v4l2_plane32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-avoid-sizeof-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-drop-pr_info-for-unknown-buffer-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-add-missing-vidioc_prepare_buf.patch
queue-4.15/media-v4l2-compat-ioctl32.c-refactor-compat-ioctl32-logic.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-ctrl_is_pointer.patch
queue-4.15/media-v4l2-compat-ioctl32.c-move-helper-functions-to-__get-put_v4l2_format32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-don-t-copy-back-the-result-for-certain-errors.patch
queue-4.15/media-v4l2-compat-ioctl32.c-make-ctrl_is_pointer-work-for-subdevs.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-the-indentation.patch
queue-4.15/media-v4l2-ioctl.c-don-t-copy-back-the-result-for-enotty.patch
queue-4.15/media-v4l2-ioctl.c-use-check_fmt-for-enum-g-s-try_fmt.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-clip-list-in-put_v4l2_window32.patch
This is a note to let you know that I've just added the patch titled
media: v4l2-compat-ioctl32.c: drop pr_info for unknown buffer type
to the 4.15-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
The filename of the patch is:
media-v4l2-compat-ioctl32.c-drop-pr_info-for-unknown-buffer-type.patch
and it can be found in the queue-4.15 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable(a)vger.kernel.org> know about it.
>From 169f24ca68bf0f247d111aef07af00dd3a02ae88 Mon Sep 17 00:00:00 2001
From: Hans Verkuil <hans.verkuil(a)cisco.com>
Date: Wed, 24 Jan 2018 09:33:57 -0500
Subject: media: v4l2-compat-ioctl32.c: drop pr_info for unknown buffer type
From: Hans Verkuil <hans.verkuil(a)cisco.com>
commit 169f24ca68bf0f247d111aef07af00dd3a02ae88 upstream.
There is nothing wrong with using an unknown buffer type. So
stop spamming the kernel log whenever this happens. The kernel
will just return -EINVAL to signal this.
Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com>
Acked-by: Sakari Ailus <sakari.ailus(a)linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab(a)s-opensource.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
---
drivers/media/v4l2-core/v4l2-compat-ioctl32.c | 4 ----
1 file changed, 4 deletions(-)
--- a/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
+++ b/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
@@ -179,8 +179,6 @@ static int __get_v4l2_format32(struct v4
return copy_from_user(&kp->fmt.meta, &up->fmt.meta,
sizeof(kp->fmt.meta)) ? -EFAULT : 0;
default:
- pr_info("compat_ioctl32: unexpected VIDIOC_FMT type %d\n",
- kp->type);
return -EINVAL;
}
}
@@ -233,8 +231,6 @@ static int __put_v4l2_format32(struct v4
return copy_to_user(&up->fmt.meta, &kp->fmt.meta,
sizeof(kp->fmt.meta)) ? -EFAULT : 0;
default:
- pr_info("compat_ioctl32: unexpected VIDIOC_FMT type %d\n",
- kp->type);
return -EINVAL;
}
}
Patches currently in stable-queue which might be from hans.verkuil(a)cisco.com are
queue-4.15/media-hdpvr-fix-an-error-handling-path-in-hdpvr_probe.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-m.userptr-in-put_v4l2_plane32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-avoid-sizeof-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-drop-pr_info-for-unknown-buffer-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-add-missing-vidioc_prepare_buf.patch
queue-4.15/media-v4l2-compat-ioctl32.c-refactor-compat-ioctl32-logic.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-ctrl_is_pointer.patch
queue-4.15/media-v4l2-compat-ioctl32.c-move-helper-functions-to-__get-put_v4l2_format32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-don-t-copy-back-the-result-for-certain-errors.patch
queue-4.15/media-v4l2-compat-ioctl32.c-make-ctrl_is_pointer-work-for-subdevs.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-the-indentation.patch
queue-4.15/media-v4l2-ioctl.c-don-t-copy-back-the-result-for-enotty.patch
queue-4.15/media-v4l2-ioctl.c-use-check_fmt-for-enum-g-s-try_fmt.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-clip-list-in-put_v4l2_window32.patch
This is a note to let you know that I've just added the patch titled
media: v4l2-compat-ioctl32.c: copy m.userptr in put_v4l2_plane32
to the 4.15-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
The filename of the patch is:
media-v4l2-compat-ioctl32.c-copy-m.userptr-in-put_v4l2_plane32.patch
and it can be found in the queue-4.15 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable(a)vger.kernel.org> know about it.
>From 8ed5a59dcb47a6f76034ee760b36e089f3e82529 Mon Sep 17 00:00:00 2001
From: Hans Verkuil <hans.verkuil(a)cisco.com>
Date: Wed, 24 Jan 2018 05:19:59 -0500
Subject: media: v4l2-compat-ioctl32.c: copy m.userptr in put_v4l2_plane32
From: Hans Verkuil <hans.verkuil(a)cisco.com>
commit 8ed5a59dcb47a6f76034ee760b36e089f3e82529 upstream.
The struct v4l2_plane32 should set m.userptr as well. The same
happens in v4l2_buffer32 and v4l2-compliance tests for this.
Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com>
Acked-by: Sakari Ailus <sakari.ailus(a)linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab(a)s-opensource.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
---
drivers/media/v4l2-core/v4l2-compat-ioctl32.c | 47 +++++++++++++++-----------
1 file changed, 28 insertions(+), 19 deletions(-)
--- a/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
+++ b/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
@@ -310,19 +310,24 @@ static int get_v4l2_plane32(struct v4l2_
sizeof(up->data_offset)))
return -EFAULT;
- if (memory == V4L2_MEMORY_USERPTR) {
+ switch (memory) {
+ case V4L2_MEMORY_MMAP:
+ case V4L2_MEMORY_OVERLAY:
+ if (copy_in_user(&up->m.mem_offset, &up32->m.mem_offset,
+ sizeof(up32->m.mem_offset)))
+ return -EFAULT;
+ break;
+ case V4L2_MEMORY_USERPTR:
if (get_user(p, &up32->m.userptr))
return -EFAULT;
up_pln = compat_ptr(p);
if (put_user((unsigned long)up_pln, &up->m.userptr))
return -EFAULT;
- } else if (memory == V4L2_MEMORY_DMABUF) {
+ break;
+ case V4L2_MEMORY_DMABUF:
if (copy_in_user(&up->m.fd, &up32->m.fd, sizeof(up32->m.fd)))
return -EFAULT;
- } else {
- if (copy_in_user(&up->m.mem_offset, &up32->m.mem_offset,
- sizeof(up32->m.mem_offset)))
- return -EFAULT;
+ break;
}
return 0;
@@ -331,22 +336,32 @@ static int get_v4l2_plane32(struct v4l2_
static int put_v4l2_plane32(struct v4l2_plane __user *up, struct v4l2_plane32 __user *up32,
enum v4l2_memory memory)
{
+ unsigned long p;
+
if (copy_in_user(up32, up, 2 * sizeof(__u32)) ||
copy_in_user(&up32->data_offset, &up->data_offset,
sizeof(up->data_offset)))
return -EFAULT;
- /* For MMAP, driver might've set up the offset, so copy it back.
- * USERPTR stays the same (was userspace-provided), so no copying. */
- if (memory == V4L2_MEMORY_MMAP)
+ switch (memory) {
+ case V4L2_MEMORY_MMAP:
+ case V4L2_MEMORY_OVERLAY:
if (copy_in_user(&up32->m.mem_offset, &up->m.mem_offset,
sizeof(up->m.mem_offset)))
return -EFAULT;
- /* For DMABUF, driver might've set up the fd, so copy it back. */
- if (memory == V4L2_MEMORY_DMABUF)
+ break;
+ case V4L2_MEMORY_USERPTR:
+ if (get_user(p, &up->m.userptr) ||
+ put_user((compat_ulong_t)ptr_to_compat((__force void *)p),
+ &up32->m.userptr))
+ return -EFAULT;
+ break;
+ case V4L2_MEMORY_DMABUF:
if (copy_in_user(&up32->m.fd, &up->m.fd,
sizeof(up->m.fd)))
return -EFAULT;
+ break;
+ }
return 0;
}
@@ -408,6 +423,7 @@ static int get_v4l2_buffer32(struct v4l2
} else {
switch (kp->memory) {
case V4L2_MEMORY_MMAP:
+ case V4L2_MEMORY_OVERLAY:
if (get_user(kp->m.offset, &up->m.offset))
return -EFAULT;
break;
@@ -421,10 +437,6 @@ static int get_v4l2_buffer32(struct v4l2
kp->m.userptr = (unsigned long)compat_ptr(tmp);
}
break;
- case V4L2_MEMORY_OVERLAY:
- if (get_user(kp->m.offset, &up->m.offset))
- return -EFAULT;
- break;
case V4L2_MEMORY_DMABUF:
if (get_user(kp->m.fd, &up->m.fd))
return -EFAULT;
@@ -481,6 +493,7 @@ static int put_v4l2_buffer32(struct v4l2
} else {
switch (kp->memory) {
case V4L2_MEMORY_MMAP:
+ case V4L2_MEMORY_OVERLAY:
if (put_user(kp->m.offset, &up->m.offset))
return -EFAULT;
break;
@@ -488,10 +501,6 @@ static int put_v4l2_buffer32(struct v4l2
if (put_user(kp->m.userptr, &up->m.userptr))
return -EFAULT;
break;
- case V4L2_MEMORY_OVERLAY:
- if (put_user(kp->m.offset, &up->m.offset))
- return -EFAULT;
- break;
case V4L2_MEMORY_DMABUF:
if (put_user(kp->m.fd, &up->m.fd))
return -EFAULT;
Patches currently in stable-queue which might be from hans.verkuil(a)cisco.com are
queue-4.15/media-hdpvr-fix-an-error-handling-path-in-hdpvr_probe.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-m.userptr-in-put_v4l2_plane32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-avoid-sizeof-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-drop-pr_info-for-unknown-buffer-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-add-missing-vidioc_prepare_buf.patch
queue-4.15/media-v4l2-compat-ioctl32.c-refactor-compat-ioctl32-logic.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-ctrl_is_pointer.patch
queue-4.15/media-v4l2-compat-ioctl32.c-move-helper-functions-to-__get-put_v4l2_format32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-don-t-copy-back-the-result-for-certain-errors.patch
queue-4.15/media-v4l2-compat-ioctl32.c-make-ctrl_is_pointer-work-for-subdevs.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-the-indentation.patch
queue-4.15/media-v4l2-ioctl.c-don-t-copy-back-the-result-for-enotty.patch
queue-4.15/media-v4l2-ioctl.c-use-check_fmt-for-enum-g-s-try_fmt.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-clip-list-in-put_v4l2_window32.patch
This is a note to let you know that I've just added the patch titled
media: v4l2-compat-ioctl32.c: add missing VIDIOC_PREPARE_BUF
to the 4.15-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
The filename of the patch is:
media-v4l2-compat-ioctl32.c-add-missing-vidioc_prepare_buf.patch
and it can be found in the queue-4.15 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable(a)vger.kernel.org> know about it.
>From 3ee6d040719ae09110e5cdf24d5386abe5d1b776 Mon Sep 17 00:00:00 2001
From: Hans Verkuil <hans.verkuil(a)cisco.com>
Date: Wed, 24 Jan 2018 08:37:04 -0500
Subject: media: v4l2-compat-ioctl32.c: add missing VIDIOC_PREPARE_BUF
From: Hans Verkuil <hans.verkuil(a)cisco.com>
commit 3ee6d040719ae09110e5cdf24d5386abe5d1b776 upstream.
The result of the VIDIOC_PREPARE_BUF ioctl was never copied back
to userspace since it was missing in the switch.
Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com>
Acked-by: Sakari Ailus <sakari.ailus(a)linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab(a)s-opensource.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
---
drivers/media/v4l2-core/v4l2-compat-ioctl32.c | 1 +
1 file changed, 1 insertion(+)
--- a/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
+++ b/drivers/media/v4l2-core/v4l2-compat-ioctl32.c
@@ -1052,6 +1052,7 @@ static long do_video_ioctl(struct file *
err = put_v4l2_create32(&karg.v2crt, up);
break;
+ case VIDIOC_PREPARE_BUF:
case VIDIOC_QUERYBUF:
case VIDIOC_QBUF:
case VIDIOC_DQBUF:
Patches currently in stable-queue which might be from hans.verkuil(a)cisco.com are
queue-4.15/media-hdpvr-fix-an-error-handling-path-in-hdpvr_probe.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-m.userptr-in-put_v4l2_plane32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-avoid-sizeof-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-drop-pr_info-for-unknown-buffer-type.patch
queue-4.15/media-v4l2-compat-ioctl32.c-add-missing-vidioc_prepare_buf.patch
queue-4.15/media-v4l2-compat-ioctl32.c-refactor-compat-ioctl32-logic.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-ctrl_is_pointer.patch
queue-4.15/media-v4l2-compat-ioctl32.c-move-helper-functions-to-__get-put_v4l2_format32.patch
queue-4.15/media-v4l2-compat-ioctl32.c-don-t-copy-back-the-result-for-certain-errors.patch
queue-4.15/media-v4l2-compat-ioctl32.c-make-ctrl_is_pointer-work-for-subdevs.patch
queue-4.15/media-v4l2-compat-ioctl32.c-fix-the-indentation.patch
queue-4.15/media-v4l2-ioctl.c-don-t-copy-back-the-result-for-enotty.patch
queue-4.15/media-v4l2-ioctl.c-use-check_fmt-for-enum-g-s-try_fmt.patch
queue-4.15/media-v4l2-compat-ioctl32.c-copy-clip-list-in-put_v4l2_window32.patch
Hi Greg, can you please apply commit 073c516ff735 ("nsfs: mark dentry with
DCACHE_RCUACCESS") to 4.9-stable and 4.4-stable? I can reproduce the
use-after-free on those kernel versions (it was introduced in v3.19).
Thanks,
- Eric
This is a note to let you know that I've just added the patch titled
nsfs: mark dentry with DCACHE_RCUACCESS
to the 4.9-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
The filename of the patch is:
nsfs-mark-dentry-with-dcache_rcuaccess.patch
and it can be found in the queue-4.9 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable(a)vger.kernel.org> know about it.
>From 073c516ff73557a8f7315066856c04b50383ac34 Mon Sep 17 00:00:00 2001
From: Cong Wang <xiyou.wangcong(a)gmail.com>
Date: Wed, 19 Apr 2017 15:11:00 -0700
Subject: nsfs: mark dentry with DCACHE_RCUACCESS
From: Cong Wang <xiyou.wangcong(a)gmail.com>
commit 073c516ff73557a8f7315066856c04b50383ac34 upstream.
Andrey reported a use-after-free in __ns_get_path():
spin_lock include/linux/spinlock.h:299 [inline]
lockref_get_not_dead+0x19/0x80 lib/lockref.c:179
__ns_get_path+0x197/0x860 fs/nsfs.c:66
open_related_ns+0xda/0x200 fs/nsfs.c:143
sock_ioctl+0x39d/0x440 net/socket.c:1001
vfs_ioctl fs/ioctl.c:45 [inline]
do_vfs_ioctl+0x1bf/0x1780 fs/ioctl.c:685
SYSC_ioctl fs/ioctl.c:700 [inline]
SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691
We are under rcu read lock protection at that point:
rcu_read_lock();
d = atomic_long_read(&ns->stashed);
if (!d)
goto slow;
dentry = (struct dentry *)d;
if (!lockref_get_not_dead(&dentry->d_lockref))
goto slow;
rcu_read_unlock();
but don't use a proper RCU API on the free path, therefore a parallel
__d_free() could free it at the same time. We need to mark the stashed
dentry with DCACHE_RCUACCESS so that __d_free() will be called after all
readers leave RCU.
Fixes: e149ed2b805f ("take the targets of /proc/*/ns/* symlinks to separate fs")
Cc: Alexander Viro <viro(a)zeniv.linux.org.uk>
Cc: Andrew Morton <akpm(a)linux-foundation.org>
Reported-by: Andrey Konovalov <andreyknvl(a)google.com>
Signed-off-by: Cong Wang <xiyou.wangcong(a)gmail.com>
Signed-off-by: Linus Torvalds <torvalds(a)linux-foundation.org>
Cc: Eric Biggers <ebiggers3(a)gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
---
fs/nsfs.c | 1 +
1 file changed, 1 insertion(+)
--- a/fs/nsfs.c
+++ b/fs/nsfs.c
@@ -90,6 +90,7 @@ slow:
return ERR_PTR(-ENOMEM);
}
d_instantiate(dentry, inode);
+ dentry->d_flags |= DCACHE_RCUACCESS;
dentry->d_fsdata = (void *)ns->ops;
d = atomic_long_cmpxchg(&ns->stashed, 0, (unsigned long)dentry);
if (d) {
Patches currently in stable-queue which might be from xiyou.wangcong(a)gmail.com are
queue-4.9/nsfs-mark-dentry-with-dcache_rcuaccess.patch
This is a note to let you know that I've just added the patch titled
nsfs: mark dentry with DCACHE_RCUACCESS
to the 4.4-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
The filename of the patch is:
nsfs-mark-dentry-with-dcache_rcuaccess.patch
and it can be found in the queue-4.4 subdirectory.
If you, or anyone else, feels it should not be added to the stable tree,
please let <stable(a)vger.kernel.org> know about it.
>From 073c516ff73557a8f7315066856c04b50383ac34 Mon Sep 17 00:00:00 2001
From: Cong Wang <xiyou.wangcong(a)gmail.com>
Date: Wed, 19 Apr 2017 15:11:00 -0700
Subject: nsfs: mark dentry with DCACHE_RCUACCESS
From: Cong Wang <xiyou.wangcong(a)gmail.com>
commit 073c516ff73557a8f7315066856c04b50383ac34 upstream.
Andrey reported a use-after-free in __ns_get_path():
spin_lock include/linux/spinlock.h:299 [inline]
lockref_get_not_dead+0x19/0x80 lib/lockref.c:179
__ns_get_path+0x197/0x860 fs/nsfs.c:66
open_related_ns+0xda/0x200 fs/nsfs.c:143
sock_ioctl+0x39d/0x440 net/socket.c:1001
vfs_ioctl fs/ioctl.c:45 [inline]
do_vfs_ioctl+0x1bf/0x1780 fs/ioctl.c:685
SYSC_ioctl fs/ioctl.c:700 [inline]
SyS_ioctl+0x8f/0xc0 fs/ioctl.c:691
We are under rcu read lock protection at that point:
rcu_read_lock();
d = atomic_long_read(&ns->stashed);
if (!d)
goto slow;
dentry = (struct dentry *)d;
if (!lockref_get_not_dead(&dentry->d_lockref))
goto slow;
rcu_read_unlock();
but don't use a proper RCU API on the free path, therefore a parallel
__d_free() could free it at the same time. We need to mark the stashed
dentry with DCACHE_RCUACCESS so that __d_free() will be called after all
readers leave RCU.
Fixes: e149ed2b805f ("take the targets of /proc/*/ns/* symlinks to separate fs")
Cc: Alexander Viro <viro(a)zeniv.linux.org.uk>
Cc: Andrew Morton <akpm(a)linux-foundation.org>
Reported-by: Andrey Konovalov <andreyknvl(a)google.com>
Signed-off-by: Cong Wang <xiyou.wangcong(a)gmail.com>
Signed-off-by: Linus Torvalds <torvalds(a)linux-foundation.org>
Cc: Eric Biggers <ebiggers3(a)gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
---
fs/nsfs.c | 1 +
1 file changed, 1 insertion(+)
--- a/fs/nsfs.c
+++ b/fs/nsfs.c
@@ -95,6 +95,7 @@ slow:
return ERR_PTR(-ENOMEM);
}
d_instantiate(dentry, inode);
+ dentry->d_flags |= DCACHE_RCUACCESS;
dentry->d_fsdata = (void *)ns_ops;
d = atomic_long_cmpxchg(&ns->stashed, 0, (unsigned long)dentry);
if (d) {
Patches currently in stable-queue which might be from xiyou.wangcong(a)gmail.com are
queue-4.4/nsfs-mark-dentry-with-dcache_rcuaccess.patch
Nikola Ciprich wrote...
> it's been introduced by 4.14 and seems to be fixed by
> 2a266f23550be997d783f27e704b9b40c4010292 (no crash with it applied so
> far for me).
Thanks for sharing, running 4.14.19 with this additional one made all my
problems go away, although they were a bit different: Using the "Resume"
button in virt-manager after an earlier "Pause" caused the guest to
freeze and using 100% CPU - especially when the guest is the grml rescue
system, this happend almost always.
Christoph