summaryrefslogtreecommitdiffstats
path: root/meta-agl-bsp
diff options
context:
space:
mode:
authorJan-Simon Möller <jsmoeller@linuxfoundation.org>2017-06-28 22:01:03 +0200
committerJan-Simon Moeller <jsmoeller@linuxfoundation.org>2017-06-29 09:13:49 +0000
commit462bcd1521094624bddad4c336007b18117a3846 (patch)
tree26449f1028020a5e946ed4589678d1e91464c33e /meta-agl-bsp
parentf6f141ba96c41c613d51a650da6562793d4baced (diff)
Fix for CVE-2017-1000366 in glibc
This changeset fixes CVE-2017-1000366 by using the latest glibc revision from the 2.24 glibc git as of today which includes the needed fix. One patch file had to be adapted and overlayed with our custom version (patching just conflicts on ChangeLog, no source file affected). Bug-AGL: SPEC-705 Change-Id: Ia38344537523808f533c2732deb5ad2ca1202a89 Signed-off-by: Jan-Simon Möller <jsmoeller@linuxfoundation.org> Reviewed-on: https://gerrit.automotivelinux.org/gerrit/9931 Tested-by: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org> ci-image-build: Jenkins Job builder account <agl-jobbuilder@automotivelinux.org>
Diffstat (limited to 'meta-agl-bsp')
-rw-r--r--meta-agl-bsp/meta-core/recipes-core/glibc/files/0028-Bug-20116-Fix-use-after-free-in-pthread_create.patch637
-rw-r--r--meta-agl-bsp/meta-core/recipes-core/glibc/glibc_2.24.bbappend4
2 files changed, 641 insertions, 0 deletions
diff --git a/meta-agl-bsp/meta-core/recipes-core/glibc/files/0028-Bug-20116-Fix-use-after-free-in-pthread_create.patch b/meta-agl-bsp/meta-core/recipes-core/glibc/files/0028-Bug-20116-Fix-use-after-free-in-pthread_create.patch
new file mode 100644
index 000000000..f06f85c62
--- /dev/null
+++ b/meta-agl-bsp/meta-core/recipes-core/glibc/files/0028-Bug-20116-Fix-use-after-free-in-pthread_create.patch
@@ -0,0 +1,637 @@
+From e7ba24f05d86acf7072e066ea6d7b235a106688c Mon Sep 17 00:00:00 2001
+From: Carlos O'Donell <carlos@redhat.com>
+Date: Sat, 28 Jan 2017 19:13:34 -0500
+Subject: [PATCH] Bug 20116: Fix use after free in pthread_create()
+
+The commit documents the ownership rules around 'struct pthread' and
+when a thread can read or write to the descriptor. With those ownership
+rules in place it becomes obvious that pd->stopped_start should not be
+touched in several of the paths during thread startup, particularly so
+for detached threads. In the case of detached threads, between the time
+the thread is created by the OS kernel and the creating thread checks
+pd->stopped_start, the detached thread might have already exited and the
+memory for pd unmapped. As a regression test we add a simple test which
+exercises this exact case by quickly creating detached threads with
+large enough stacks to ensure the thread stack cache is bypassed and the
+stacks are unmapped. Before the fix the testcase segfaults, after the
+fix it works correctly and completes without issue.
+
+For a detailed discussion see:
+https://www.sourceware.org/ml/libc-alpha/2017-01/msg00505.html
+
+(cherry-picked from commit f8bf15febcaf137bbec5a61101e88cd5a9d56ca8)
+
+Upstream-Status: Backport [master]
+Signed-off-by: Yuanjie Huang <yuanjie.huang@windriver.com>
+---
+ ChangeLog | 23 ++++
+ nptl/Makefile | 3 +-
+ nptl/createthread.c | 10 +-
+ nptl/pthread_create.c | 207 +++++++++++++++++++++++++++------
+ nptl/pthread_getschedparam.c | 1 +
+ nptl/pthread_setschedparam.c | 1 +
+ nptl/pthread_setschedprio.c | 1 +
+ nptl/tpp.c | 2 +
+ nptl/tst-create-detached.c | 137 ++++++++++++++++++++++
+ sysdeps/nacl/createthread.c | 10 +-
+ sysdeps/unix/sysv/linux/createthread.c | 16 ++-
+ 11 files changed, 356 insertions(+), 55 deletions(-)
+ create mode 100644 nptl/tst-create-detached.c
+
+diff --git a/nptl/Makefile b/nptl/Makefile
+index 0d8aadebed..7dec4edb53 100644
+--- a/nptl/Makefile
++++ b/nptl/Makefile
+@@ -290,7 +290,8 @@ tests = tst-typesizes \
+ tst-initializers1 $(addprefix tst-initializers1-,\
+ c89 gnu89 c99 gnu99 c11 gnu11) \
+ tst-bad-schedattr \
+- tst-thread_local1 tst-mutex-errorcheck tst-robust10
++ tst-thread_local1 tst-mutex-errorcheck tst-robust10 \
++ tst-create-detached \
+ xtests = tst-setuid1 tst-setuid1-static tst-setuid2 \
+ tst-mutexpp1 tst-mutexpp6 tst-mutexpp10
+ test-srcs = tst-oddstacklimit
+diff --git a/nptl/createthread.c b/nptl/createthread.c
+index ba2f9c7167..328f85865d 100644
+--- a/nptl/createthread.c
++++ b/nptl/createthread.c
+@@ -25,16 +25,14 @@
+
+ static int
+ create_thread (struct pthread *pd, const struct pthread_attr *attr,
+- bool stopped_start, STACK_VARIABLES_PARMS, bool *thread_ran)
++ bool *stopped_start, STACK_VARIABLES_PARMS, bool *thread_ran)
+ {
+ /* If the implementation needs to do some tweaks to the thread after
+ it has been created at the OS level, it can set STOPPED_START here. */
+
+- pd->stopped_start = stopped_start;
+- if (__glibc_unlikely (stopped_start))
+- /* We make sure the thread does not run far by forcing it to get a
+- lock. We lock it here too so that the new thread cannot continue
+- until we tell it to. */
++ pd->stopped_start = *stopped_start;
++ if (__glibc_unlikely (*stopped_start))
++ /* See CONCURRENCY NOTES in nptl/pthread_create.c. */
+ lll_lock (pd->lock, LLL_PRIVATE);
+
+ return ENOSYS;
+diff --git a/nptl/pthread_create.c b/nptl/pthread_create.c
+index a834063ad5..44b17bec86 100644
+--- a/nptl/pthread_create.c
++++ b/nptl/pthread_create.c
+@@ -54,25 +54,141 @@ unsigned int __nptl_nthreads = 1;
+ /* Code to allocate and deallocate a stack. */
+ #include "allocatestack.c"
+
+-/* createthread.c defines this function, and two macros:
++/* CONCURRENCY NOTES:
++
++ Understanding who is the owner of the 'struct pthread' or 'PD'
++ (refers to the value of the 'struct pthread *pd' function argument)
++ is critically important in determining exactly which operations are
++ allowed and which are not and when, particularly when it comes to the
++ implementation of pthread_create, pthread_join, pthread_detach, and
++ other functions which all operate on PD.
++
++ The owner of PD is responsible for freeing the final resources
++ associated with PD, and may examine the memory underlying PD at any
++ point in time until it frees it back to the OS or to reuse by the
++ runtime.
++
++ The thread which calls pthread_create is called the creating thread.
++ The creating thread begins as the owner of PD.
++
++ During startup the new thread may examine PD in coordination with the
++ owner thread (which may be itself).
++
++ The four cases of ownership transfer are:
++
++ (1) Ownership of PD is released to the process (all threads may use it)
++ after the new thread starts in a joinable state
++ i.e. pthread_create returns a usable pthread_t.
++
++ (2) Ownership of PD is released to the new thread starting in a detached
++ state.
++
++ (3) Ownership of PD is dynamically released to a running thread via
++ pthread_detach.
++
++ (4) Ownership of PD is acquired by the thread which calls pthread_join.
++
++ Implementation notes:
++
++ The PD->stopped_start and thread_ran variables are used to determine
++ exactly which of the four ownership states we are in and therefore
++ what actions can be taken. For example after (2) we cannot read or
++ write from PD anymore since the thread may no longer exist and the
++ memory may be unmapped. The most complicated cases happen during
++ thread startup:
++
++ (a) If the created thread is in a detached (PTHREAD_CREATE_DETACHED),
++ or joinable (default PTHREAD_CREATE_JOINABLE) state and
++ STOPPED_START is true, then the creating thread has ownership of
++ PD until the PD->lock is released by pthread_create. If any
++ errors occur we are in states (c), (d), or (e) below.
++
++ (b) If the created thread is in a detached state
++ (PTHREAD_CREATED_DETACHED), and STOPPED_START is false, then the
++ creating thread has ownership of PD until it invokes the OS
++ kernel's thread creation routine. If this routine returns
++ without error, then the created thread owns PD; otherwise, see
++ (c) and (e) below.
++
++ (c) If the detached thread setup failed and THREAD_RAN is true, then
++ the creating thread releases ownership to the new thread by
++ sending a cancellation signal. All threads set THREAD_RAN to
++ true as quickly as possible after returning from the OS kernel's
++ thread creation routine.
++
++ (d) If the joinable thread setup failed and THREAD_RAN is true, then
++ then the creating thread retains ownership of PD and must cleanup
++ state. Ownership cannot be released to the process via the
++ return of pthread_create since a non-zero result entails PD is
++ undefined and therefore cannot be joined to free the resources.
++ We privately call pthread_join on the thread to finish handling
++ the resource shutdown (Or at least we should, see bug 19511).
++
++ (e) If the thread creation failed and THREAD_RAN is false, then the
++ creating thread retains ownership of PD and must cleanup state.
++ No waiting for the new thread is required because it never
++ started.
++
++ The nptl_db interface:
++
++ The interface with nptl_db requires that we enqueue PD into a linked
++ list and then call a function which the debugger will trap. The PD
++ will then be dequeued and control returned to the thread. The caller
++ at the time must have ownership of PD and such ownership remains
++ after control returns to thread. The enqueued PD is removed from the
++ linked list by the nptl_db callback td_thr_event_getmsg. The debugger
++ must ensure that the thread does not resume execution, otherwise
++ ownership of PD may be lost and examining PD will not be possible.
++
++ Note that the GNU Debugger as of (December 10th 2015) commit
++ c2c2a31fdb228d41ce3db62b268efea04bd39c18 no longer uses
++ td_thr_event_getmsg and several other related nptl_db interfaces. The
++ principal reason for this is that nptl_db does not support non-stop
++ mode where other threads can run concurrently and modify runtime
++ structures currently in use by the debugger and the nptl_db
++ interface.
++
++ Axioms:
++
++ * The create_thread function can never set stopped_start to false.
++ * The created thread can read stopped_start but never write to it.
++ * The variable thread_ran is set some time after the OS thread
++ creation routine returns, how much time after the thread is created
++ is unspecified, but it should be as quickly as possible.
++
++*/
++
++/* CREATE THREAD NOTES:
++
++ createthread.c defines the create_thread function, and two macros:
+ START_THREAD_DEFN and START_THREAD_SELF (see below).
+
+- create_thread is obliged to initialize PD->stopped_start. It
+- should be true if the STOPPED_START parameter is true, or if
+- create_thread needs the new thread to synchronize at startup for
+- some other implementation reason. If PD->stopped_start will be
+- true, then create_thread is obliged to perform the operation
+- "lll_lock (PD->lock, LLL_PRIVATE)" before starting the thread.
++ create_thread must initialize PD->stopped_start. It should be true
++ if the STOPPED_START parameter is true, or if create_thread needs the
++ new thread to synchronize at startup for some other implementation
++ reason. If STOPPED_START will be true, then create_thread is obliged
++ to lock PD->lock before starting the thread. Then pthread_create
++ unlocks PD->lock which synchronizes-with START_THREAD_DEFN in the
++ child thread which does an acquire/release of PD->lock as the last
++ action before calling the user entry point. The goal of all of this
++ is to ensure that the required initial thread attributes are applied
++ (by the creating thread) before the new thread runs user code. Note
++ that the the functions pthread_getschedparam, pthread_setschedparam,
++ pthread_setschedprio, __pthread_tpp_change_priority, and
++ __pthread_current_priority reuse the same lock, PD->lock, for a
++ similar purpose e.g. synchronizing the setting of similar thread
++ attributes. These functions are never called before the thread is
++ created, so don't participate in startup syncronization, but given
++ that the lock is present already and in the unlocked state, reusing
++ it saves space.
+
+ The return value is zero for success or an errno code for failure.
+ If the return value is ENOMEM, that will be translated to EAGAIN,
+ so create_thread need not do that. On failure, *THREAD_RAN should
+ be set to true iff the thread actually started up and then got
+- cancelled before calling user code (*PD->start_routine), in which
+- case it is responsible for doing its own cleanup. */
+-
++ canceled before calling user code (*PD->start_routine). */
+ static int create_thread (struct pthread *pd, const struct pthread_attr *attr,
+- bool stopped_start, STACK_VARIABLES_PARMS,
++ bool *stopped_start, STACK_VARIABLES_PARMS,
+ bool *thread_ran);
+
+ #include <createthread.c>
+@@ -314,12 +430,19 @@ START_THREAD_DEFN
+ /* Store the new cleanup handler info. */
+ THREAD_SETMEM (pd, cleanup_jmp_buf, &unwind_buf);
+
++ /* We are either in (a) or (b), and in either case we either own
++ PD already (2) or are about to own PD (1), and so our only
++ restriction would be that we can't free PD until we know we
++ have ownership (see CONCURRENCY NOTES above). */
+ if (__glibc_unlikely (pd->stopped_start))
+ {
+ int oldtype = CANCEL_ASYNC ();
+
+ /* Get the lock the parent locked to force synchronization. */
+ lll_lock (pd->lock, LLL_PRIVATE);
++
++ /* We have ownership of PD now. */
++
+ /* And give it up right away. */
+ lll_unlock (pd->lock, LLL_PRIVATE);
+
+@@ -378,7 +501,8 @@ START_THREAD_DEFN
+ pd, pd->nextevent));
+ }
+
+- /* Now call the function to signal the event. */
++ /* Now call the function which signals the event. See
++ CONCURRENCY NOTES for the nptl_db interface comments. */
+ __nptl_death_event ();
+ }
+ }
+@@ -642,19 +766,28 @@ __pthread_create_2_1 (pthread_t *newthread, const pthread_attr_t *attr,
+ that cares whether the thread count is correct. */
+ atomic_increment (&__nptl_nthreads);
+
+- bool thread_ran = false;
++ /* Our local value of stopped_start and thread_ran can be accessed at
++ any time. The PD->stopped_start may only be accessed if we have
++ ownership of PD (see CONCURRENCY NOTES above). */
++ bool stopped_start = false; bool thread_ran = false;
+
+ /* Start the thread. */
+ if (__glibc_unlikely (report_thread_creation (pd)))
+ {
+- /* Create the thread. We always create the thread stopped
+- so that it does not get far before we tell the debugger. */
+- retval = create_thread (pd, iattr, true, STACK_VARIABLES_ARGS,
+- &thread_ran);
++ stopped_start = true;
++
++ /* We always create the thread stopped at startup so we can
++ notify the debugger. */
++ retval = create_thread (pd, iattr, &stopped_start,
++ STACK_VARIABLES_ARGS, &thread_ran);
+ if (retval == 0)
+ {
+- /* create_thread should have set this so that the logic below can
+- test it. */
++ /* We retain ownership of PD until (a) (see CONCURRENCY NOTES
++ above). */
++
++ /* Assert stopped_start is true in both our local copy and the
++ PD copy. */
++ assert (stopped_start);
+ assert (pd->stopped_start);
+
+ /* Now fill in the information about the new thread in
+@@ -671,26 +804,30 @@ __pthread_create_2_1 (pthread_t *newthread, const pthread_attr_t *attr,
+ pd, pd->nextevent)
+ != 0);
+
+- /* Now call the function which signals the event. */
++ /* Now call the function which signals the event. See
++ CONCURRENCY NOTES for the nptl_db interface comments. */
+ __nptl_create_event ();
+ }
+ }
+ else
+- retval = create_thread (pd, iattr, false, STACK_VARIABLES_ARGS,
+- &thread_ran);
++ retval = create_thread (pd, iattr, &stopped_start,
++ STACK_VARIABLES_ARGS, &thread_ran);
+
+ if (__glibc_unlikely (retval != 0))
+ {
+- /* If thread creation "failed", that might mean that the thread got
+- created and ran a little--short of running user code--but then
+- create_thread cancelled it. In that case, the thread will do all
+- its own cleanup just like a normal thread exit after a successful
+- creation would do. */
+-
+ if (thread_ran)
+- assert (pd->stopped_start);
++ /* State (c) or (d) and we may not have PD ownership (see
++ CONCURRENCY NOTES above). We can assert that STOPPED_START
++ must have been true because thread creation didn't fail, but
++ thread attribute setting did. */
++ /* See bug 19511 which explains why doing nothing here is a
++ resource leak for a joinable thread. */
++ assert (stopped_start);
+ else
+ {
++ /* State (e) and we have ownership of PD (see CONCURRENCY
++ NOTES above). */
++
+ /* Oops, we lied for a second. */
+ atomic_decrement (&__nptl_nthreads);
+
+@@ -710,10 +847,14 @@ __pthread_create_2_1 (pthread_t *newthread, const pthread_attr_t *attr,
+ }
+ else
+ {
+- if (pd->stopped_start)
+- /* The thread blocked on this lock either because we're doing TD_CREATE
+- event reporting, or for some other reason that create_thread chose.
+- Now let it run free. */
++ /* We don't know if we have PD ownership. Once we check the local
++ stopped_start we'll know if we're in state (a) or (b) (see
++ CONCURRENCY NOTES above). */
++ if (stopped_start)
++ /* State (a), we own PD. The thread blocked on this lock either
++ because we're doing TD_CREATE event reporting, or for some
++ other reason that create_thread chose. Now let it run
++ free. */
+ lll_unlock (pd->lock, LLL_PRIVATE);
+
+ /* We now have for sure more than one thread. The main thread might
+diff --git a/nptl/pthread_getschedparam.c b/nptl/pthread_getschedparam.c
+index b887881baf..de71171a08 100644
+--- a/nptl/pthread_getschedparam.c
++++ b/nptl/pthread_getschedparam.c
+@@ -35,6 +35,7 @@ __pthread_getschedparam (pthread_t threadid, int *policy,
+
+ int result = 0;
+
++ /* See CREATE THREAD NOTES in nptl/pthread_create.c. */
+ lll_lock (pd->lock, LLL_PRIVATE);
+
+ /* The library is responsible for maintaining the values at all
+diff --git a/nptl/pthread_setschedparam.c b/nptl/pthread_setschedparam.c
+index dfb52b9dbf..dcb520f1c8 100644
+--- a/nptl/pthread_setschedparam.c
++++ b/nptl/pthread_setschedparam.c
+@@ -36,6 +36,7 @@ __pthread_setschedparam (pthread_t threadid, int policy,
+
+ int result = 0;
+
++ /* See CREATE THREAD NOTES in nptl/pthread_create.c. */
+ lll_lock (pd->lock, LLL_PRIVATE);
+
+ struct sched_param p;
+diff --git a/nptl/pthread_setschedprio.c b/nptl/pthread_setschedprio.c
+index cefc6481d6..8134b50560 100644
+--- a/nptl/pthread_setschedprio.c
++++ b/nptl/pthread_setschedprio.c
+@@ -38,6 +38,7 @@ pthread_setschedprio (pthread_t threadid, int prio)
+ struct sched_param param;
+ param.sched_priority = prio;
+
++ /* See CREATE THREAD NOTES in nptl/pthread_create.c. */
+ lll_lock (pd->lock, LLL_PRIVATE);
+
+ /* If the thread should have higher priority because of some
+diff --git a/nptl/tpp.c b/nptl/tpp.c
+index e175bf4d53..223bd6bbee 100644
+--- a/nptl/tpp.c
++++ b/nptl/tpp.c
+@@ -114,6 +114,7 @@ __pthread_tpp_change_priority (int previous_prio, int new_prio)
+ if (priomax == newpriomax)
+ return 0;
+
++ /* See CREATE THREAD NOTES in nptl/pthread_create.c. */
+ lll_lock (self->lock, LLL_PRIVATE);
+
+ tpp->priomax = newpriomax;
+@@ -165,6 +166,7 @@ __pthread_current_priority (void)
+
+ int result = 0;
+
++ /* See CREATE THREAD NOTES in nptl/pthread_create.c. */
+ lll_lock (self->lock, LLL_PRIVATE);
+
+ if ((self->flags & ATTR_FLAG_SCHED_SET) == 0)
+diff --git a/nptl/tst-create-detached.c b/nptl/tst-create-detached.c
+new file mode 100644
+index 0000000000..ea93e441c7
+--- /dev/null
++++ b/nptl/tst-create-detached.c
+@@ -0,0 +1,137 @@
++/* Bug 20116: Test rapid creation of detached threads.
++ Copyright (C) 2017 Free Software Foundation, Inc.
++ This file is part of the GNU C Library.
++
++ The GNU C Library is free software; you can redistribute it and/or
++ modify it under the terms of the GNU Lesser General Public
++ License as published by the Free Software Foundation; either
++ version 2.1 of the License, or (at your option) any later version.
++
++ The GNU C Library is distributed in the hope that it will be useful,
++ but WITHOUT ANY WARRANTY; without even the implied warranty of
++ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
++ Lesser General Public License for more details.
++
++ You should have received a copy of the GNU Lesser General Public
++ License along with the GNU C Library; see the file COPYING.LIB. If
++ not, see <http://www.gnu.org/licenses/>. */
++
++/* The goal of the test is to trigger a failure if the parent touches
++ any part of the thread descriptor after the detached thread has
++ exited. We test this by creating many detached threads with large
++ stacks. The stacks quickly fill the the stack cache and subsequent
++ threads will start to cause the thread stacks to be immediately
++ unmapped to satisfy the stack cache max. With the stacks being
++ unmapped the parent's read of any part of the thread descriptor will
++ trigger a segfault. That segfault is what we are trying to cause,
++ since any segfault is a defect in the implementation. */
++
++#include <pthread.h>
++#include <stdio.h>
++#include <errno.h>
++#include <unistd.h>
++#include <stdbool.h>
++#include <sys/resource.h>
++#include <support/xthread.h>
++
++/* Number of threads to create. */
++enum { threads_to_create = 100000 };
++
++/* Number of threads which should spawn other threads. */
++enum { creator_threads = 2 };
++
++/* Counter of threads created so far. This is incremented by all the
++ running creator threads. */
++static unsigned threads_created;
++
++/* Thread callback which does nothing, so that the thread exits
++ immediatedly. */
++static void *
++do_nothing (void *arg)
++{
++ return NULL;
++}
++
++/* Attribute indicating that the thread should be created in a detached
++ fashion. */
++static pthread_attr_t detached;
++
++/* Barrier to synchronize initialization. */
++static pthread_barrier_t barrier;
++
++static void *
++creator_thread (void *arg)
++{
++ int ret;
++ xpthread_barrier_wait (&barrier);
++
++ while (true)
++ {
++ pthread_t thr;
++ /* Thread creation will fail if the kernel does not free old
++ threads quickly enough, so we do not report errors. */
++ ret = pthread_create (&thr, &detached, do_nothing, NULL);
++ if (ret == 0 && __atomic_add_fetch (&threads_created, 1, __ATOMIC_SEQ_CST)
++ >= threads_to_create)
++ break;
++ }
++
++ return NULL;
++}
++
++static int
++do_test (void)
++{
++ /* Limit the size of the process, so that memory allocation will
++ fail without impacting the entire system. */
++ {
++ struct rlimit limit;
++ if (getrlimit (RLIMIT_AS, &limit) != 0)
++ {
++ printf ("FAIL: getrlimit (RLIMIT_AS) failed: %m\n");
++ return 1;
++ }
++ /* This limit, 800MB, is just a heuristic. Any value can be
++ picked. */
++ long target = 800 * 1024 * 1024;
++ if (limit.rlim_cur == RLIM_INFINITY || limit.rlim_cur > target)
++ {
++ limit.rlim_cur = target;
++ if (setrlimit (RLIMIT_AS, &limit) != 0)
++ {
++ printf ("FAIL: setrlimit (RLIMIT_AS) failed: %m\n");
++ return 1;
++ }
++ }
++ }
++
++ xpthread_attr_init (&detached);
++
++ xpthread_attr_setdetachstate (&detached, PTHREAD_CREATE_DETACHED);
++
++ /* A large thread stack seems beneficial for reproducing a race
++ condition in detached thread creation. The goal is to reach the
++ limit of the runtime thread stack cache such that the detached
++ thread's stack is unmapped after exit and causes a segfault when
++ the parent reads the thread descriptor data stored on the the
++ unmapped stack. */
++ xpthread_attr_setstacksize (&detached, 16 * 1024 * 1024);
++
++ xpthread_barrier_init (&barrier, NULL, creator_threads);
++
++ pthread_t threads[creator_threads];
++
++ for (int i = 0; i < creator_threads; ++i)
++ threads[i] = xpthread_create (NULL, creator_thread, NULL);
++
++ for (int i = 0; i < creator_threads; ++i)
++ xpthread_join (threads[i]);
++
++ xpthread_attr_destroy (&detached);
++
++ xpthread_barrier_destroy (&barrier);
++
++ return 0;
++}
++
++#include <support/test-driver.c>
+diff --git a/sysdeps/nacl/createthread.c b/sysdeps/nacl/createthread.c
+index 7b571c34e2..5465558cc1 100644
+--- a/sysdeps/nacl/createthread.c
++++ b/sysdeps/nacl/createthread.c
+@@ -32,15 +32,13 @@ static void start_thread (void) __attribute__ ((noreturn));
+
+ static int
+ create_thread (struct pthread *pd, const struct pthread_attr *attr,
+- bool stopped_start, STACK_VARIABLES_PARMS, bool *thread_ran)
++ bool *stopped_start, STACK_VARIABLES_PARMS, bool *thread_ran)
+ {
+ pd->tid = __nacl_get_tid (pd);
+
+- pd->stopped_start = stopped_start;
+- if (__glibc_unlikely (stopped_start))
+- /* We make sure the thread does not run far by forcing it to get a
+- lock. We lock it here too so that the new thread cannot continue
+- until we tell it to. */
++ pd->stopped_start = *stopped_start;
++ if (__glibc_unlikely (*stopped_start))
++ /* See CONCURRENCY NOTES in nptl/pthread_create.c. */
+ lll_lock (pd->lock, LLL_PRIVATE);
+
+ TLS_DEFINE_INIT_TP (tp, pd);
+diff --git a/sysdeps/unix/sysv/linux/createthread.c b/sysdeps/unix/sysv/linux/createthread.c
+index 6d32cece48..66ddae61d4 100644
+--- a/sysdeps/unix/sysv/linux/createthread.c
++++ b/sysdeps/unix/sysv/linux/createthread.c
+@@ -46,7 +46,7 @@ static int start_thread (void *arg) __attribute__ ((noreturn));
+
+ static int
+ create_thread (struct pthread *pd, const struct pthread_attr *attr,
+- bool stopped_start, STACK_VARIABLES_PARMS, bool *thread_ran)
++ bool *stopped_start, STACK_VARIABLES_PARMS, bool *thread_ran)
+ {
+ /* Determine whether the newly created threads has to be started
+ stopped since we have to set the scheduling parameters or set the
+@@ -54,13 +54,11 @@ create_thread (struct pthread *pd, const struct pthread_attr *attr,
+ if (attr != NULL
+ && (__glibc_unlikely (attr->cpuset != NULL)
+ || __glibc_unlikely ((attr->flags & ATTR_FLAG_NOTINHERITSCHED) != 0)))
+- stopped_start = true;
++ *stopped_start = true;
+
+- pd->stopped_start = stopped_start;
+- if (__glibc_unlikely (stopped_start))
+- /* We make sure the thread does not run far by forcing it to get a
+- lock. We lock it here too so that the new thread cannot continue
+- until we tell it to. */
++ pd->stopped_start = *stopped_start;
++ if (__glibc_unlikely (*stopped_start))
++ /* See CONCURRENCY NOTES in nptl/pthread_creat.c. */
+ lll_lock (pd->lock, LLL_PRIVATE);
+
+ /* We rely heavily on various flags the CLONE function understands:
+@@ -117,7 +115,7 @@ create_thread (struct pthread *pd, const struct pthread_attr *attr,
+ /* Set the affinity mask if necessary. */
+ if (attr->cpuset != NULL)
+ {
+- assert (stopped_start);
++ assert (*stopped_start);
+
+ res = INTERNAL_SYSCALL (sched_setaffinity, err, 3, pd->tid,
+ attr->cpusetsize, attr->cpuset);
+@@ -140,7 +138,7 @@ create_thread (struct pthread *pd, const struct pthread_attr *attr,
+ /* Set the scheduling parameters. */
+ if ((attr->flags & ATTR_FLAG_NOTINHERITSCHED) != 0)
+ {
+- assert (stopped_start);
++ assert (*stopped_start);
+
+ res = INTERNAL_SYSCALL (sched_setscheduler, err, 3, pd->tid,
+ pd->schedpolicy, &pd->schedparam);
+--
+2.11.0
+
diff --git a/meta-agl-bsp/meta-core/recipes-core/glibc/glibc_2.24.bbappend b/meta-agl-bsp/meta-core/recipes-core/glibc/glibc_2.24.bbappend
new file mode 100644
index 000000000..a9f2a238d
--- /dev/null
+++ b/meta-agl-bsp/meta-core/recipes-core/glibc/glibc_2.24.bbappend
@@ -0,0 +1,4 @@
+FILESEXTRAPATHS_prepend := "${THISDIR}/files:"
+
+# include fix for CVE-2017-1000366
+SRCREV = "86ac4a78a9218d1e1dcfbacc6f7d09957c1fe3a4" \ No newline at end of file