linux-kernel - [PATCH 3.16 229/346] drm/msm: protect against faults from copy_from

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <lsq.1479082460.697775777@decadent.org.uk>
Date:   Mon, 14 Nov 2016 00:14:20 +0000
From:   Ben Hutchings <ben@...adent.org.uk>
To:     linux-kernel@...r.kernel.org, stable@...r.kernel.org
CC:     akpm@...ux-foundation.org, "Rob Clark" <robdclark@...il.com>
Subject: [PATCH 3.16 229/346] drm/msm: protect against faults from
 copy_from_user() in submit ioctl

3.16.39-rc1 review patch.  If anyone has any objections, please let me know.

------------------

From: Rob Clark <robdclark@...il.com>

commit d78d383ab354b0b9e1d23404ae0d9fbdeb9aa035 upstream.

An evil userspace could try to cause deadlock by passing an unfaulted-in
GEM bo as submit->bos (or submit->cmds) table.  Which will trigger
msm_gem_fault() while we already hold struct_mutex.  See:

https://github.com/freedreno/msmtest/blob/master/evilsubmittest.c

Signed-off-by: Rob Clark <robdclark@...il.com>
[bwh: Backported to 3.16: adjust context]
Signed-off-by: Ben Hutchings <ben@...adent.org.uk>
---
 drivers/gpu/drm/msm/msm_drv.h        | 6 ++++++
 drivers/gpu/drm/msm/msm_gem.c        | 9 +++++++++
 drivers/gpu/drm/msm/msm_gem_submit.c | 3 +++
 3 files changed, 18 insertions(+)

--- a/drivers/gpu/drm/msm/msm_drv.h
+++ b/drivers/gpu/drm/msm/msm_drv.h
@@ -124,6 +124,12 @@ struct msm_drm_private {
 		 */
 		struct drm_mm mm;
 	} vram;
+
+	/* task holding struct_mutex.. currently only used in submit path
+	 * to detect and reject faults from copy_from_user() for submit
+	 * ioctl.
+	 */
+	struct task_struct *struct_mutex_task;
 };
 
 struct msm_format {
--- a/drivers/gpu/drm/msm/msm_gem.c
+++ b/drivers/gpu/drm/msm/msm_gem.c
@@ -188,11 +188,20 @@ int msm_gem_fault(struct vm_area_struct
 {
 	struct drm_gem_object *obj = vma->vm_private_data;
 	struct drm_device *dev = obj->dev;
+	struct msm_drm_private *priv = dev->dev_private;
 	struct page **pages;
 	unsigned long pfn;
 	pgoff_t pgoff;
 	int ret;
 
+	/* This should only happen if userspace tries to pass a mmap'd
+	 * but unfaulted gem bo vaddr into submit ioctl, triggering
+	 * a page fault while struct_mutex is already held.  This is
+	 * not a valid use-case so just bail.
+	 */
+	if (priv->struct_mutex_task == current)
+		return VM_FAULT_SIGBUS;
+
 	/* Make sure we don't parallel update on a fault, nor move or remove
 	 * something from beneath our feet
 	 */
--- a/drivers/gpu/drm/msm/msm_gem_submit.c
+++ b/drivers/gpu/drm/msm/msm_gem_submit.c
@@ -360,6 +360,8 @@ int msm_ioctl_gem_submit(struct drm_devi
 	if (ret)
 		return ret;
 
+	priv->struct_mutex_task = current;
+
 	submit = submit_create(dev, gpu, args->nr_bos);
 	if (!submit) {
 		ret = -ENOMEM;
@@ -442,6 +444,7 @@ out:
 	if (submit)
 		submit_cleanup(submit, !!ret);
 out_unlock:
+	priv->struct_mutex_task = NULL;
 	mutex_unlock(&dev->struct_mutex);
 	return ret;
 }