From: Greg KH <gregkh@suse.de>
To: linux-kernel@vger.kernel.org, stable@kernel.org
Cc: stable-review@kernel.org, torvalds@linux-foundation.org,
akpm@linux-foundation.org, alan@lxorguk.ukuu.org.uk,
Shawn Bohrer <sbohrer@rgmadvisors.com>,
Peter Zijlstra <a.p.zijlstra@chello.nl>,
Darren Hart <dvhart@linux.intel.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
peterz@infradead.org, eric.dumazet@gmail.com,
zvonler@rgmadvisors.com, hughd@google.com,
Thomas Gleixner <tglx@linutronix.de>
Subject: [05/90] futex: Fix regression with read only mappings
Date: Fri, 12 Aug 2011 14:03:50 -0700 [thread overview]
Message-ID: <20110812210552.559922734@clark.kroah.org> (raw)
In-Reply-To: <20110812210626.GA4725@kroah.com>
3.0-stable review patch. If anyone has any objections, please let us know.
------------------
From: Shawn Bohrer <sbohrer@rgmadvisors.com>
commit 9ea71503a8ed9184d2d0b8ccc4d269d05f7940ae upstream.
commit 7485d0d3758e8e6491a5c9468114e74dc050785d (futexes: Remove rw
parameter from get_futex_key()) in 2.6.33 fixed two problems: First, It
prevented a loop when encountering a ZERO_PAGE. Second, it fixed RW
MAP_PRIVATE futex operations by forcing the COW to occur by
unconditionally performing a write access get_user_pages_fast() to get
the page. The commit also introduced a user-mode regression in that it
broke futex operations on read-only memory maps. For example, this
breaks workloads that have one or more reader processes doing a
FUTEX_WAIT on a futex within a read only shared file mapping, and a
writer processes that has a writable mapping issuing the FUTEX_WAKE.
This fixes the regression for valid futex operations on RO mappings by
trying a RO get_user_pages_fast() when the RW get_user_pages_fast()
fails. This change makes it necessary to also check for invalid use
cases, such as anonymous RO mappings (which can never change) and the
ZERO_PAGE which the commit referenced above was written to address.
This patch does restore the original behavior with RO MAP_PRIVATE
mappings, which have inherent user-mode usage problems and don't really
make sense. With this patch performing a FUTEX_WAIT within a RO
MAP_PRIVATE mapping will be successfully woken provided another process
updates the region of the underlying mapped file. However, the mmap()
man page states that for a MAP_PRIVATE mapping:
It is unspecified whether changes made to the file after
the mmap() call are visible in the mapped region.
So user-mode users attempting to use futex operations on RO MAP_PRIVATE
mappings are depending on unspecified behavior. Additionally a
RO MAP_PRIVATE mapping could fail to wake up in the following case.
Thread-A: call futex(FUTEX_WAIT, memory-region-A).
get_futex_key() return inode based key.
sleep on the key
Thread-B: call mprotect(PROT_READ|PROT_WRITE, memory-region-A)
Thread-B: write memory-region-A.
COW happen. This process's memory-region-A become related
to new COWed private (ie PageAnon=1) page.
Thread-B: call futex(FUETX_WAKE, memory-region-A).
get_futex_key() return mm based key.
IOW, we fail to wake up Thread-A.
Once again doing something like this is just silly and users who do
something like this get what they deserve.
While RO MAP_PRIVATE mappings are nonsensical, checking for a private
mapping requires walking the vmas and was deemed too costly to avoid a
userspace hang.
This Patch is based on Peter Zijlstra's initial patch with modifications to
only allow RO mappings for futex operations that need VERIFY_READ access.
Reported-by: David Oliver <david@rgmadvisors.com>
Signed-off-by: Shawn Bohrer <sbohrer@rgmadvisors.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: peterz@infradead.org
Cc: eric.dumazet@gmail.com
Cc: zvonler@rgmadvisors.com
Cc: hughd@google.com
Link: http://lkml.kernel.org/r/1309450892-30676-1-git-send-email-sbohrer@rgmadvisors.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
---
kernel/futex.c | 54 ++++++++++++++++++++++++++++++++++++++++++------------
1 file changed, 42 insertions(+), 12 deletions(-)
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -218,6 +218,8 @@ static void drop_futex_key_refs(union fu
* @uaddr: virtual address of the futex
* @fshared: 0 for a PROCESS_PRIVATE futex, 1 for PROCESS_SHARED
* @key: address where result is stored.
+ * @rw: mapping needs to be read/write (values: VERIFY_READ,
+ * VERIFY_WRITE)
*
* Returns a negative error code or 0
* The key words are stored in *key on success.
@@ -229,12 +231,12 @@ static void drop_futex_key_refs(union fu
* lock_page() might sleep, the caller should not hold a spinlock.
*/
static int
-get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key)
+get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw)
{
unsigned long address = (unsigned long)uaddr;
struct mm_struct *mm = current->mm;
struct page *page, *page_head;
- int err;
+ int err, ro = 0;
/*
* The futex address must be "naturally" aligned.
@@ -262,8 +264,18 @@ get_futex_key(u32 __user *uaddr, int fsh
again:
err = get_user_pages_fast(address, 1, 1, &page);
+ /*
+ * If write access is not required (eg. FUTEX_WAIT), try
+ * and get read-only access.
+ */
+ if (err == -EFAULT && rw == VERIFY_READ) {
+ err = get_user_pages_fast(address, 1, 0, &page);
+ ro = 1;
+ }
if (err < 0)
return err;
+ else
+ err = 0;
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
page_head = page;
@@ -305,6 +317,13 @@ again:
if (!page_head->mapping) {
unlock_page(page_head);
put_page(page_head);
+ /*
+ * ZERO_PAGE pages don't have a mapping. Avoid a busy loop
+ * trying to find one. RW mapping would have COW'd (and thus
+ * have a mapping) so this page is RO and won't ever change.
+ */
+ if ((page_head == ZERO_PAGE(address)))
+ return -EFAULT;
goto again;
}
@@ -316,6 +335,15 @@ again:
* the object not the particular process.
*/
if (PageAnon(page_head)) {
+ /*
+ * A RO anonymous page will never change and thus doesn't make
+ * sense for futex operations.
+ */
+ if (ro) {
+ err = -EFAULT;
+ goto out;
+ }
+
key->both.offset |= FUT_OFF_MMSHARED; /* ref taken on mm */
key->private.mm = mm;
key->private.address = address;
@@ -327,9 +355,10 @@ again:
get_futex_key_refs(key);
+out:
unlock_page(page_head);
put_page(page_head);
- return 0;
+ return err;
}
static inline void put_futex_key(union futex_key *key)
@@ -940,7 +969,7 @@ futex_wake(u32 __user *uaddr, unsigned i
if (!bitset)
return -EINVAL;
- ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &key);
+ ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &key, VERIFY_READ);
if (unlikely(ret != 0))
goto out;
@@ -986,10 +1015,10 @@ futex_wake_op(u32 __user *uaddr1, unsign
int ret, op_ret;
retry:
- ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1);
+ ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1, VERIFY_READ);
if (unlikely(ret != 0))
goto out;
- ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2);
+ ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2, VERIFY_WRITE);
if (unlikely(ret != 0))
goto out_put_key1;
@@ -1243,10 +1272,11 @@ retry:
pi_state = NULL;
}
- ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1);
+ ret = get_futex_key(uaddr1, flags & FLAGS_SHARED, &key1, VERIFY_READ);
if (unlikely(ret != 0))
goto out;
- ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2);
+ ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2,
+ requeue_pi ? VERIFY_WRITE : VERIFY_READ);
if (unlikely(ret != 0))
goto out_put_key1;
@@ -1790,7 +1820,7 @@ static int futex_wait_setup(u32 __user *
* while the syscall executes.
*/
retry:
- ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q->key);
+ ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q->key, VERIFY_READ);
if (unlikely(ret != 0))
return ret;
@@ -1941,7 +1971,7 @@ static int futex_lock_pi(u32 __user *uad
}
retry:
- ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q.key);
+ ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &q.key, VERIFY_WRITE);
if (unlikely(ret != 0))
goto out;
@@ -2060,7 +2090,7 @@ retry:
if ((uval & FUTEX_TID_MASK) != vpid)
return -EPERM;
- ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &key);
+ ret = get_futex_key(uaddr, flags & FLAGS_SHARED, &key, VERIFY_WRITE);
if (unlikely(ret != 0))
goto out;
@@ -2249,7 +2279,7 @@ static int futex_wait_requeue_pi(u32 __u
debug_rt_mutex_init_waiter(&rt_waiter);
rt_waiter.task = NULL;
- ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2);
+ ret = get_futex_key(uaddr2, flags & FLAGS_SHARED, &key2, VERIFY_WRITE);
if (unlikely(ret != 0))
goto out;
next prev parent reply other threads:[~2011-08-14 16:05 UTC|newest]
Thread overview: 99+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-08-12 21:06 [00/90] 3.0.2-stable review Greg KH
2011-08-12 21:03 ` [01/90] cris: fix a build error in kernel/fork.c Greg KH
2011-08-12 21:03 ` [02/90] cris: fix a build error in sync_serial_open() Greg KH
2011-08-12 21:03 ` [03/90] cris: fix the prototype of sync_serial_ioctl() Greg KH
2011-08-12 21:03 ` [04/90] cris: add missing declaration of kgdb_init() and breakpoint() Greg KH
2011-08-12 21:03 ` Greg KH [this message]
2011-08-12 21:03 ` [06/90] [PARISC] wire up sendmmsg syscall Greg KH
2011-08-12 21:03 ` [07/90] [PARISC] Fix futex support Greg KH
2011-08-12 21:03 ` [08/90] [PARISC] fix return type of __atomic64_add_return Greg KH
2011-08-12 21:03 ` [09/90] rt2x00: rt2800: fix zeroing skb structure Greg KH
2011-08-12 21:03 ` [10/90] rt2x00: fix usage of NULL queue Greg KH
2011-08-12 21:03 ` [11/90] rtlwifi: Fix kernel oops on ARM SOC Greg KH
2011-08-12 21:03 ` [12/90] iwlegacy: set tx power after rxon_assoc Greg KH
2011-08-12 21:03 ` [13/90] ath9k: initialize tx chainmask before testing channel tx power values Greg KH
2011-08-12 21:03 ` [14/90] ath9k: skip ->config_pci_powersave() if PCIe port has ASPM disabled Greg KH
2011-08-12 21:04 ` [15/90] net: sendmmsg should only return an error if no messages were sent Greg KH
2011-08-12 21:04 ` [16/90] net: Cap number of elements for sendmmsg Greg KH
2011-08-12 21:04 ` [17/90] net: Fix security_socket_sendmsg() bypass problem Greg KH
2011-08-12 21:04 ` [18/90] xen: allow enable use of VGA console on dom0 Greg KH
2011-08-12 21:04 ` [19/90] drm: Separate EDID Header Check from EDID Block Check Greg KH
2011-08-12 21:04 ` [20/90] drm/radeon: Extended DDC Probing for Connectors with Improperly Wired DDC Lines (here: Asus M2A-VM HDMI) Greg KH
2011-08-12 21:04 ` [21/90] drm/radeon: Extended DDC Probing for ECS A740GM-M DVI-D Connector Greg KH
2011-08-12 21:04 ` [22/90] drm/radeon: Log Subsystem Vendor and Device Information Greg KH
2011-08-12 21:04 ` [23/90] drm/i915/pch: Fix integer math bugs in panel fitting Greg KH
2011-08-12 21:04 ` [24/90] drm/i915: load the LUT before pipe enable on ILK+ Greg KH
2011-08-12 21:04 ` [25/90] drm/i915: Fix typo in DRM_I915_OVERLAY_PUT_IMAGE ioctl define Greg KH
2011-08-12 21:04 ` [26/90] drm/i915: Initialize RCS ring status page address in intel_render_ring_init_dri Greg KH
2011-08-12 21:04 ` [27/90] drm/i915: Hold mode_config->mutex during hotplug processing Greg KH
2011-08-12 21:04 ` [28/90] drm/i915: Fixup for Hold mode_config->mutex during hotplug Greg KH
2011-08-12 21:04 ` [29/90] crypto: Move md5_transform to lib/md5.c Greg KH
2011-08-12 21:04 ` [30/90] net: Compute protocol sequence numbers and fragment IDs using MD5 Greg KH
2011-08-12 21:04 ` [31/90] asus-wmi: fix hwmon/pwm1 Greg KH
2011-08-12 21:04 ` [32/90] asus-wmi: return proper value in store_cpufv() Greg KH
2011-08-12 21:04 ` [33/90] CIFS: Fix missing a decrement of inFlight value Greg KH
2011-08-12 21:04 ` [34/90] cifs: cope with negative dentries in cifs_get_root Greg KH
2011-08-12 21:04 ` [35/90] cifs: convert prefixpath delimiters in Greg KH
2011-08-12 21:04 ` [36/90] ASoC: sgtl5000: fix cache handling Greg KH
2011-08-12 21:04 ` [37/90] ALSA: timer - Fix Oops at closing slave timer Greg KH
2011-08-12 21:04 ` [38/90] ALSA: snd-usb-caiaq: Fix keymap for RigKontrol3 Greg KH
2011-08-12 21:04 ` [39/90] ALSA: snd-usb: avoid dividing by zero on invalid input Greg KH
2011-08-12 21:04 ` [40/90] ALSA: snd-usb: operate on given mixer interface only Greg KH
2011-08-12 21:04 ` [41/90] ALSA: snd-usb: Accept UAC2 FORMAT_TYPE descriptors with Greg KH
2011-08-12 21:04 ` [42/90] ipv6: make fragment identifications less predictable Greg KH
2011-08-12 21:04 ` [43/90] icmp: Fix regression in nexthop resolution during replies Greg KH
2011-08-12 21:04 ` [44/90] ipv4: Constrain UFO fragment sizes to multiples of 8 bytes Greg KH
2011-08-12 21:04 ` [45/90] ipv4: fix the reusing of routing cache entries Greg KH
2011-08-12 21:04 ` [46/90] IPVS: Free resources on module removal Greg KH
2011-08-12 21:04 ` [47/90] net: adjust array index Greg KH
2011-08-12 21:04 ` [48/90] drivers/net/niu.c: " Greg KH
2011-08-12 21:04 ` [49/90] sch_sfq: fix sfq_enqueue() Greg KH
2011-08-12 21:04 ` [50/90] IPv4: Send gratuitous ARP for secondary IP addresses also Greg KH
2011-08-12 21:04 ` [51/90] net: add IFF_SKB_TX_SHARED flag to priv_flags Greg KH
2011-08-12 21:04 ` [52/90] net: Audit drivers to identify those needing IFF_TX_SKB_SHARING cleared Greg KH
2011-08-12 21:04 ` [53/90] bonding: fix string comparison errors Greg KH
2011-08-12 21:04 ` [54/90] Fix cdc-phonet build Greg KH
2011-08-12 21:04 ` [55/90] xfrm: Fix key lengths for rfc3686(ctr(aes)) Greg KH
2011-08-12 21:04 ` [56/90] sis190: Rx filter init is needed for MAC address change Greg KH
2011-08-12 21:04 ` [57/90] r8169: Add support for D-Link 530T rev C1 (Kernel Bug 38862) Greg KH
2011-08-12 21:04 ` [58/90] net: allow netif_carrier to be called safely from IRQ Greg KH
2011-08-12 21:04 ` [59/90] ipv4: use RT_TOS after some rt_tos conversions Greg KH
2011-08-12 21:04 ` [60/90] gre: fix improper error handling Greg KH
2011-08-12 21:04 ` [61/90] iwlagn: 5000 do not support idle mode Greg KH
2011-08-12 21:04 ` [62/90] [SCSI] mpt2sas: Fixed Big Indian Issues on 32 bit PPC Greg KH
2011-08-12 21:04 ` [63/90] sparc: Dont leave sparc_pmu_type NULL on sun4v Greg KH
2011-08-12 21:04 ` [64/90] sparc: Add T3 sun4v cpu type and hypervisor group defines Greg KH
2011-08-12 21:04 ` [65/90] sparc: Dont do expensive hypervisor PCR write unless necessary Greg KH
2011-08-12 21:04 ` [66/90] sparc: Detect and handle UltraSPARC-T3 cpu types Greg KH
2011-08-12 21:04 ` [67/90] sparc: Sanitize cpu feature detection and reporting Greg KH
2011-08-12 21:04 ` [68/90] sparc: Minor tweaks to Niagara page copy/clear Greg KH
2011-08-12 21:04 ` [69/90] sparc: Use popc if possible for hweight routines Greg KH
2011-08-12 21:04 ` [70/90] sparc: Use hweight64() in popc emulation Greg KH
2011-08-12 21:04 ` [71/90] sparc: Add some missing hypervisor API groups Greg KH
2011-08-12 21:04 ` [72/90] sparc: Set reboot-cmd using reboot data hypervisor call if available Greg KH
2011-08-12 21:04 ` [73/90] sparc: Use popc when possible for ffs/__ffs/ffz Greg KH
2011-08-12 21:04 ` [74/90] sparc: Access kernel TSB using physical addressing when possible Greg KH
2011-08-12 21:05 ` [75/90] sparc: Size mondo queues more sanely Greg KH
2011-08-12 21:05 ` [76/90] sparc: Fix build with DEBUG_PAGEALLOC enabled Greg KH
2011-08-12 21:05 ` [77/90] Ecryptfs: Add mount option to check uid of device being Greg KH
2011-08-12 21:05 ` [78/90] eCryptfs: Return error when lower file pointer is NULL Greg KH
2011-08-12 21:05 ` [79/90] powerpc/pseries: Fix kexec on recent firmware versions Greg KH
2011-08-12 21:05 ` [80/90] powerpc: Fix device tree claim code Greg KH
2011-08-12 21:05 ` [81/90] powerpc: pseries: Fix kexec on machines with more than 4TB of RAM Greg KH
2011-08-12 21:05 ` [82/90] ext3: Properly count journal credits for long symlinks Greg KH
2011-08-12 21:05 ` [83/90] ext4: " Greg KH
2011-08-12 21:05 ` [84/90] e1000e: alternate MAC address does not work on device id 0x1060 Greg KH
2011-08-12 21:05 ` [85/90] gianfar: fix fiper alignment after resetting the time Greg KH
2011-08-12 21:05 ` [86/90] dp83640: increase receive time stamp buffer size Greg KH
2011-08-12 21:05 ` [87/90] ath9k_hw: Fix incorrect Tx control power in AR9003 template Greg KH
2011-08-12 21:05 ` [88/90] ath9k: fix a misprint which leads to incorrect calibration Greg KH
2011-08-12 21:05 ` [89/90] sparc: Dont do hypervisor calls on non-sun4v in DS driver Greg KH
2011-08-12 21:05 ` [90/90] mm: Fix fixup_user_fault() for MMU=n Greg KH
2011-08-14 10:31 ` [00/90] 3.0.2-stable review Jiri Slaby
2011-08-14 16:18 ` Greg KH
2011-08-15 20:56 ` Dave Jones
2011-08-16 1:24 ` Greg KH
2011-08-16 17:46 ` Arnaud Lacombe
2011-08-16 17:59 ` Greg KH
2011-08-16 18:10 ` Linus Torvalds
2011-08-16 21:11 ` H. Peter Anvin
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=20110812210552.559922734@clark.kroah.org \
--to=gregkh@suse.de \
--cc=a.p.zijlstra@chello.nl \
--cc=akpm@linux-foundation.org \
--cc=alan@lxorguk.ukuu.org.uk \
--cc=dvhart@linux.intel.com \
--cc=eric.dumazet@gmail.com \
--cc=hughd@google.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=sbohrer@rgmadvisors.com \
--cc=stable-review@kernel.org \
--cc=stable@kernel.org \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=zvonler@rgmadvisors.com \
/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
all inboxes | Powered by JetHome®