rpms/libX11
6
0
Fork 0
Code Issues Pull Requests Releases Activity

Backport fix for Xlib lockups due to recursive XError

Browse Source 
Resolves: https://issues.redhat.com/browse/RHEL-23452
This commit is contained in:
Olivier Fourdan 2024-02-08 14:35:00 +01:00
parent 78f64db9c8
commit 39ea10b841
2 changed files with 174 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

166
0001-Avoid-recursing-through-_XError-due-to-sequence-adju.patch Normal file
View File

@ -0,0 +1,166 @@
From 8c92ef59890c6d6e2be456658d3b9c145eda8629 Mon Sep 17 00:00:00 2001
From: Keith Packard <keithp@keithp.com>
Date: Sat, 7 Nov 2020 22:22:47 -0800
Subject: [PATCH libX11] Avoid recursing through _XError due to sequence
adjustment
This patch is based on research done by Dmitry Osipenko to uncover the
cause of a large class of Xlib lockups.
_XError must unlock and re-lock the display around the call to the
user error handler function. When re-locking the display, two
functions are called to ensure that the display is ready to generate a request:
_XIDHandler(dpy);
_XSeqSyncFunction(dpy);
The first ensures that there is at least one XID available to use
(possibly calling _xcb_generate_id to do so). The second makes sure a
reply is received at least every 65535 requests to keep sequence
numbers in sync (possibly generating a GetInputFocus request and
synchronously awaiting the reply).
If the second of these does generate a GetInputFocus request and wait
for the reply, then a pending error will cause recursion into _XError,
which deadlocks the display.
One seemingly easy fix is to have _XError avoid those calls by
invoking InternalLockDisplay instead of LockDisplay. That function
does everything that LockDisplay does *except* call those final two
functions which may end up receiving an error.
However, that doesn't protect the system from applications which call
some legal Xlib function from within their error handler. Any Xlib
function which cannot generate protocol or wait for events is valid,
including many which invoke LockDisplay.
What we need to do is make LockDisplay skip these two function calls
precisely when it is called from within the _XError context for the
same display.
This patch accomplishes this by creating a list of threads in the
display which are in _XError, and then having LockDisplay check the
current thread against those list elements.
Inspired-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
(cherry picked from commit 30ccef3a48029bf4fc31d4abda2d2778d0ad6277)
---
include/X11/Xlibint.h | 2 ++
src/OpenDis.c | 1 +
src/XlibInt.c | 10 ++++++++++
src/locking.c | 12 ++++++++++++
src/locking.h | 12 ++++++++++++
5 files changed, 37 insertions(+)
diff --git a/include/X11/Xlibint.h b/include/X11/Xlibint.h
index 6b95bcf7..09078e3f 100644
--- a/include/X11/Xlibint.h
+++ b/include/X11/Xlibint.h
@@ -202,6 +202,8 @@ struct _XDisplay
unsigned long last_request_read_upper32bit;
unsigned long request_upper32bit;
#endif
+
+ struct _XErrorThreadInfo *error_threads;
};
#define XAllocIDs(dpy,ids,n) (*(dpy)->idlist_alloc)(dpy,ids,n)
diff --git a/src/OpenDis.c b/src/OpenDis.c
index 82723578..85901168 100644
--- a/src/OpenDis.c
+++ b/src/OpenDis.c
@@ -201,6 +201,7 @@ XOpenDisplay (
X_DPY_SET_LAST_REQUEST_READ(dpy, 0);
dpy->default_screen = iscreen; /* Value returned by ConnectDisplay */
dpy->last_req = (char *)&_dummy_request;
+ dpy->error_threads = NULL;
/* Initialize the display lock */
if (InitDisplayLock(dpy) != 0) {
diff --git a/src/XlibInt.c b/src/XlibInt.c
index 4e45e62b..8771b791 100644
--- a/src/XlibInt.c
+++ b/src/XlibInt.c
@@ -1482,6 +1482,11 @@ int _XError (
if (_XErrorFunction != NULL) {
int rtn_val;
#ifdef XTHREADS
+ struct _XErrorThreadInfo thread_info = {
+ .error_thread = xthread_self(),
+ .next = dpy->error_threads
+ }, **prev;
+ dpy->error_threads = &thread_info;
if (dpy->lock)
(*dpy->lock->user_lock_display)(dpy);
UnlockDisplay(dpy);
@@ -1491,6 +1496,11 @@ int _XError (
LockDisplay(dpy);
if (dpy->lock)
(*dpy->lock->user_unlock_display)(dpy);
+
+ /* unlink thread_info from the list */
+ for (prev = &dpy->error_threads; *prev != &thread_info; prev = &(*prev)->next)
+ ;
+ *prev = thread_info.next;
#endif
return rtn_val;
} else {
diff --git a/src/locking.c b/src/locking.c
index 9f4fe067..bcadc857 100644
--- a/src/locking.c
+++ b/src/locking.c
@@ -453,6 +453,9 @@ static void _XLockDisplay(
XTHREADS_FILE_LINE_ARGS
)
{
+#ifdef XTHREADS
+ struct _XErrorThreadInfo *ti;
+#endif
#ifdef XTHREADS_WARN
_XLockDisplayWarn(dpy, file, line);
#else
@@ -460,6 +463,15 @@ static void _XLockDisplay(
#endif
if (dpy->lock->locking_level > 0)
_XDisplayLockWait(dpy);
+#ifdef XTHREADS
+ /*
+ * Skip the two function calls below which may generate requests
+ * when LockDisplay is called from within _XError.
+ */
+ for (ti = dpy->error_threads; ti; ti = ti->next)
+ if (ti->error_thread == xthread_self())
+ return;
+#endif
_XIDHandler(dpy);
_XSeqSyncFunction(dpy);
}
diff --git a/src/locking.h b/src/locking.h
index 5251a60c..59fc866e 100644
--- a/src/locking.h
+++ b/src/locking.h
@@ -149,6 +149,18 @@ typedef struct _LockInfoRec {
xmutex_t lock;
} LockInfoRec;
+/* A list of threads currently invoking error handlers on this display
+ * LockDisplay operates differently for these threads, avoiding
+ * generating any requests or reading any events as that can cause
+ * recursion into the error handling code, which will deadlock the
+ * thread.
+ */
+struct _XErrorThreadInfo
+{
+ struct _XErrorThreadInfo *next;
+ xthread_t error_thread;
+};
+
/* XOpenDis.c */
extern int (*_XInitDisplayLock_fn)(Display *dpy);
extern void (*_XFreeDisplayLock_fn)(Display *dpy);
--
2.43.0

9
libX11.spec
View File

@ -5,7 +5,7 @@
Summary: Core X11 protocol client library
Name: libX11
Version: 1.6.8
Release: 7%{?gitdate:.%{gitdate}git%{gitversion}}%{?dist}
Release: 8%{?gitdate:.%{gitdate}git%{gitversion}}%{?dist}
License: MIT
Group: System Environment/Libraries
URL: http://www.x.org
@ -39,6 +39,9 @@ Patch11: 0003-XCreatePixmap-trigger-BadValue-error-for-out-of-rang.patch
# CVE-2023-43787
Patch12: 0001-CVE-2023-43787-Integer-overflow-in-XCreateImage-lead.patch
# RHEL-23452
Patch13: 0001-Avoid-recursing-through-_XError-due-to-sequence-adju.patch
BuildRequires: xorg-x11-util-macros >= 1.11
BuildRequires: pkgconfig(xproto) >= 7.0.15
BuildRequires: xorg-x11-xtrans-devel >= 1.0.3-4
@ -89,6 +92,7 @@ libX11/libxcb interoperability library
%patch10 -p1 -b .xputimage-clip-images-to-maximum-height-width-allowe
%patch11 -p1 -b .xcreatepixmap-trigger-badvalue-error-for-out-of-rang
%patch12 -p1 -b .cve-2023-43787
%patch13 -p1 -b .rhel-23452
%build
autoreconf -v --install --force
@ -153,6 +157,9 @@ make %{?_smp_mflags} check
%{_mandir}/man5/*.5*
%changelog
* Tue Jan 30 2024 Olivier Fourdan <ofourdan@redhat.com> - 1.6.8-8
- Backport fix for Xlib lockups due to recursive XError (RHEL-23452)
* Wed Oct 11 2023 José Expósito <jexposit@redhat.com> - 1.6.8-7
- Fix CVE-2023-43785: out-of-bounds memory access in _XkbReadKeySyms()
- Fix CVE-2023-43786: stack exhaustion from infinite recursion in