The previous code only made sense if one considers buttons and keys to
be mutually exclusive on a device. That is not necessarily true, causing
a number of issues.
This function allocates and fills in the number of xEvents we need to
send the device state down the wire. This is split across multiple
32-byte devices including one deviceStateNotify event and optional
deviceKeyStateNotify, deviceButtonStateNotify and (possibly multiple)
deviceValuator events.
The previous behavior would instead compose a sequence
of [state, buttonstate, state, keystate, valuator...]. This is not
protocol correct, and on top of that made the code extremely convoluted.
Fix this by streamlining: add both button and key into the deviceStateNotify
and then append the key state and button state, followed by the
valuators. Finally, the deviceValuator events contain up to 6 valuators
per event but we only ever sent through 3 at a time. Let's double that
troughput.
CVE-2024-0229, ZDI-CAN-22678
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
(cherry picked from commit 219c54b8a3)
If a device has both a button class and a key class and numButtons is
zero, we can get an OOB write due to event under-allocation.
This function seems to assume a device has either keys or buttons, not
both. It has two virtually identical code paths, both of which assume
they're applying to the first event in the sequence.
A device with both a key and button class triggered a logic bug - only
one xEvent was allocated but the deviceStateNotify pointer was pushed on
once per type. So effectively this logic code:
int count = 1;
if (button && nbuttons > 32) count++;
if (key && nbuttons > 0) count++;
if (key && nkeys > 32) count++; // this is basically always true
// count is at 2 for our keys + zero button device
ev = alloc(count * sizeof(xEvent));
FixDeviceStateNotify(ev);
if (button)
FixDeviceStateNotify(ev++);
if (key)
FixDeviceStateNotify(ev++); // santa drops into the wrong chimney here
If the device has more than 3 valuators, the OOB is pushed back - we're
off by one so it will happen when the last deviceValuator event is
written instead.
Fix this by allocating the maximum number of events we may allocate.
Note that the current behavior is not protocol-correct anyway, this
patch fixes only the allocation issue.
Note that this issue does not trigger if the device has at least one
button. While the server does not prevent a button class with zero
buttons, it is very unlikely.
CVE-2024-0229, ZDI-CAN-22678
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
(cherry picked from commit ece23be888)
Both DeviceFocusEvent and the XIQueryPointer reply contain a bit for
each logical button currently down. Since buttons can be arbitrarily mapped
to anything up to 255 make sure we have enough bits for the maximum mapping.
CVE-2023-6816, ZDI-CAN-22664, ZDI-CAN-22665
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
(cherry picked from commit 9e2ecb2af8)
In commit 7e1f86d4 monitor support was added to randr. At this time it seemed to be reasonable not to have
more than one (virtual) monitor on a particular physical display. The code was never changed since.
Nowadays, extremely large displays exists (4k displays, ultra-wide displays). In some use cases it makes sense to
split these large physical displays into multiple virtual monitors. An example are ultra-wide screens that can be
split into 2 monitors. The change in this commit makes this work.
Besides that, removing a monitor in a function that is called "RRMonitorAdd" is bad practice and causes
unexpected behaviour.
button->xkb_acts is supposed to be an array sufficiently large for all
our buttons, not just a single XkbActions struct. Allocating
insufficient memory here means when we memcpy() later in
XkbSetDeviceInfo we write into memory that wasn't ours to begin with,
leading to the usual security ooopsiedaisies.
CVE-2023-6377, ZDI-CAN-22412, ZDI-CAN-22413
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
(cherry picked from commit 0c1a93d319)
Affected are ProcRRChangeProviderProperty and ProcRRChangeOutputProperty.
See also xserver@8f454b79 where this same bug was fixed for the core
protocol and XI.
This fixes an OOB read and the resulting information disclosure.
Length calculation for the request was clipped to a 32-bit integer. With
the correct stuff->nUnits value the expected request size was
truncated, passing the REQUEST_FIXED_SIZE check.
The server then proceeded with reading at least stuff->num_items bytes
(depending on stuff->format) from the request and stuffing whatever it
finds into the property. In the process it would also allocate at least
stuff->nUnits bytes, i.e. 4GB.
CVE-2023-6478, ZDI-CAN-22561
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
(cherry picked from commit 14f480010a)
After a change for the xserver to automatically determine the seat
based on the XDG_SEAT variable, xephyr stopped working. This was
because of an old feature where xephyr used to handle evdev
directly. This was dropped some time ago, and now this check is
not needed
(cherry picked from commit 4c03b67d33)
PointerWindows[] keeps a reference to the last window our sprite
entered - changes are usually handled by CheckMotion().
If we switch between screens via XWarpPointer our
dev->spriteInfo->sprite->win is set to the new screen's root window.
If there's another window at the cursor location CheckMotion() will
trigger the right enter/leave events later. If there is not, it skips
that process and we never trigger LeaveWindow() - PointerWindows[] for
the device still refers to the previous window.
If that window is destroyed we have a dangling reference that will
eventually cause a use-after-free bug when checking the window hierarchy
later.
To trigger this, we require:
- two protocol screens
- XWarpPointer to the other screen's root window
- XDestroyWindow before entering any other window
This is a niche bug so we hack around it by making sure we reset the
PointerWindows[] entry so we cannot have a dangling pointer. This
doesn't handle Enter/Leave events correctly but the previous code didn't
either.
CVE-2023-5380, ZDI-CAN-21608
This vulnerability was discovered by:
Sri working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit 564ccf2ce9)
The handling of appending/prepending properties was incorrect, with at
least two bugs: the property length was set to the length of the new
part only, i.e. appending or prepending N elements to a property with P
existing elements always resulted in the property having N elements
instead of N + P.
Second, when pre-pending a value to a property, the offset for the old
values was incorrect, leaving the new property with potentially
uninitalized values and/or resulting in OOB memory writes.
For example, prepending a 3 element value to a 5 element property would
result in this 8 value array:
[N, N, N, ?, ?, P, P, P ] P, P
^OOB write
The XI2 code is a copy/paste of the RandR code, so the bug exists in
both.
CVE-2023-5367, ZDI-CAN-22153
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 541ab2ecd4)
This is more portable than libbsd as everything Just Works, even on BSD systems,
and is the recommended method of consuming libbsd nowadays.
It also helpfully lets things work with glibc-provided functions for new
enough glibc.
[For the 21.1.x backport, take inspiration from @alanc's commit to libxdmcp
at c01da8ebd0.]
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/973
Co-authored-by: Guillem Jover <guillem@hadrons.org>
(cherry picked from commit 94945a5274)
Signed-off-by: Sam James <sam@gentoo.org>
This enables fixing a deadlock case on the client side, where the client
ends up blocked waiting for a Present event that will never come because
the window was destroyed. The new PresentWindowDestroyed flag allows the
client to avoid blocking indefinitely.
Signed-off-by: Adam Jackson <ajax@redhat.com>
See-also: https://gitlab.freedesktop.org/mesa/mesa/-/issues/116
See-also: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6685
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
(cherry picked from commit 462b06033e)
ZDI-CAN-19866/CVE-2023-1393
If a client explicitly destroys the compositor overlay window (aka COW),
we would leave a dangling pointer to that window in the CompScreen
structure, which will trigger a use-after-free later.
Make sure to clear the CompScreen pointer to the COW when the latter gets
destroyed explicitly by the client.
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit 26ef545b35)
CVE-2023-0494, ZDI-CAN-19596
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 0ba6d8c370)
The event might be a DeviceEvent allocated on the stack, in
AccessXKeyboardEvent for instance. Fixes out-of-bounds read.
Signed-off-by: Mike Gorse <mgorse@suse.com>
(cherry picked from commit 2ef5ef57bd)
Withoug a proper implementation of DetermineClientCmd, clients that
connect via an ssh tunnel are miscategorized as local. This results
in failures when we try to use SCM_RIGHTS (eg: in MIT-SHM).
Fixes: https://github.com/XQuartz/XQuartz/issues/314
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
(cherry picked from commit 0ea9b59589)
This provides a way to determine the pid of a peer connection on
systems like darwin that do not support getpeerucred() nor
SO_PEERCRED.
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
(cherry picked from commit 8a4ab22873)
GetLocalClientCreds() was preferring getpeereid() above other implementations.
getpeereid(), however, only returns the effective uid and gid of the peer,
leaving the pid unset. When this happens, we are unable to use the pid to
determine the peer's command line arguments and incorrectly treat ssh-tunneled
traffic as local.
To address this, we now prioritize getpeerucred() or SO_PEERCRED as those two
implementations will return the pid in addition to uid and gid.
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
(cherry picked from commit 165d5c1260)
This will allow us to remove build-time conditionalization on COMPOSITE
while still allowing XQuartz to disable it and use ROOTLESS.
This reverts commit 5f2d652377
(cherry picked from commit 66e7b7349d)
Commit c7311654 cached the value of ResourceClientBits(), but that value
depends on the `MaxClients` value set either from the command line or
from the configuration file.
For the latter, a call to ResourceClientBits() is issued before the
configuration file is read, meaning that the cached value is from the
default, not from the maximum number of clients set in the configuration
file.
That obviously causes all sort of issues, including memory corruption
and crashes of the Xserver when reaching the default limit value.
To avoid that issue, also keep the LimitClient value, and recompute the
ilog2() value if that changes, as on startup when the value is set from
the the xorg.conf ServerFlags section.
v2: Drop the `cache == 0` test
Rename cache vars
Fixes: c7311654 - dix: cache ResourceClientBits() value
Closes: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1310
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
(cherry picked from commit 2efa6d6595)
In commit b320ca0 the mask was inadvertently changed from octal 0177 to
hexadecimal 0x177.
Fixes commit b320ca0ffe
Xtest: disallow GenericEvents in XTestSwapFakeInput
Found by Stuart Cassoff
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit bb1711b7fb)
GetComponentByName returns an allocated string, so let's free that if we
fail somewhere.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 18f91b950e)
GetCountedString did a check for the whole string to be within the
request buffer but not for the initial 2 bytes that contain the length
field. A swapped client could send a malformed request to trigger a
swaps() on those bytes, writing into random memory.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 11beef0b7f)
Unlike other elements of the keymap, this pointer was freed but not
reset. On a subsequent XkbGetKbdByName request, the server may access
already freed memory.
CVE-2022-4283, ZDI-CAN-19530
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit ccdd431cd8)
This fixes an OOB read and the resulting information disclosure.
Length calculation for the request was clipped to a 32-bit integer. With
the correct stuff->num_items value the expected request size was
truncated, passing the REQUEST_FIXED_SIZE check.
The server then proceeded with reading at least stuff->num_items bytes
(depending on stuff->format) from the request and stuffing whatever it
finds into the property. In the process it would also allocate at least
stuff->num_items bytes, i.e. 4GB.
The same bug exists in ProcChangeProperty and ProcXChangeDeviceProperty,
so let's fix that too.
CVE-2022-46344, ZDI-CAN 19405
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit 8f454b793e)
Both ProcXChangeDeviceProperty and ProcXIChangeProperty checked the
property for validity but didn't actually return the potential error.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit b8a84cb0f2)
This fixes a use-after-free bug:
When a client first calls ScreenSaverSetAttributes(), a struct
ScreenSaverAttrRec is allocated and added to the client's
resources.
When the same client calls ScreenSaverSetAttributes() again, a new
struct ScreenSaverAttrRec is allocated, replacing the old struct. The
old struct was freed but not removed from the clients resources.
Later, when the client is destroyed the resource system invokes
ScreenSaverFreeAttr and attempts to clean up the already freed struct.
Fix this by letting the resource system free the old attrs instead.
CVE-2022-46343, ZDI-CAN 19404
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit 842ca3ccef)
This fixes a use-after-free bug:
When a client first calls XvdiSelectVideoNotify() on a drawable with a
TRUE onoff argument, a struct XvVideoNotifyRec is allocated. This struct
is added twice to the resources:
- as the drawable's XvRTVideoNotifyList. This happens only once per
drawable, subsequent calls append to this list.
- as the client's XvRTVideoNotify. This happens for every client.
The struct keeps the ClientPtr around once it has been added for a
client. The idea, presumably, is that if the client disconnects we can remove
all structs from the drawable's list that match the client (by resetting
the ClientPtr to NULL), but if the drawable is destroyed we can remove
and free the whole list.
However, if the same client then calls XvdiSelectVideoNotify() on the
same drawable with a FALSE onoff argument, only the ClientPtr on the
existing struct was set to NULL. The struct itself remained in the
client's resources.
If the drawable is now destroyed, the resource system invokes
XvdiDestroyVideoNotifyList which frees the whole list for this drawable
- including our struct. This function however does not free the resource
for the client since our ClientPtr is NULL.
Later, when the client is destroyed and the resource system invokes
XvdiDestroyVideoNotify, we unconditionally set the ClientPtr to NULL. On
a struct that has been freed previously. This is generally frowned upon.
Fix this by calling FreeResource() on the second call instead of merely
setting the ClientPtr to NULL. This removes the struct from the client
resources (but not from the list), ensuring that it won't be accessed
again when the client quits.
Note that the assignment tpn->client = NULL; is superfluous since the
XvdiDestroyVideoNotify function will do this anyway. But it's left for
clarity and to match a similar invocation in XvdiSelectPortNotify.
CVE-2022-46342, ZDI-CAN 19400
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit b79f32b57c)
The XKB protocol effectively prevents us from ever using keycodes above
255. For buttons it's theoretically possible but realistically too niche
to worry about. For all other passive grabs, the detail must be zero
anyway.
This fixes an OOB write:
ProcXIPassiveUngrabDevice() calls DeletePassiveGrabFromList with a
temporary grab struct which contains tempGrab->detail.exact = stuff->detail.
For matching existing grabs, DeleteDetailFromMask is called with the
stuff->detail value. This function creates a new mask with the one bit
representing stuff->detail cleared.
However, the array size for the new mask is 8 * sizeof(CARD32) bits,
thus any detail above 255 results in an OOB array write.
CVE-2022-46341, ZDI-CAN 19381
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit 51eb63b0ee)
XTestSwapFakeInput assumes all events in this request are
sizeof(xEvent) and iterates through these in 32-byte increments.
However, a GenericEvent may be of arbitrary length longer than 32 bytes,
so any GenericEvent in this list would result in subsequent events to be
misparsed.
Additional, the swapped event is written into a stack-allocated struct
xEvent (size 32 bytes). For any GenericEvent longer than 32 bytes,
swapping the event may thus smash the stack like an avocado on toast.
Catch this case early and return BadValue for any GenericEvent.
Which is what would happen in unswapped setups anyway since XTest
doesn't support GenericEvent.
CVE-2022-46340, ZDI-CAN 19265
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Olivier Fourdan <ofourdan@redhat.com>
(cherry picked from commit b320ca0ffe)
No validation of the various fields on that report were done, so a
malicious client could send a short request that claims it had N
sections, or rows, or keys, and the server would process the request for
N sections, running out of bounds of the actual request data.
Fix this by adding size checks to ensure our data is valid.
ZDI-CAN 16062, CVE-2022-2319.
This vulnerability was discovered by:
Jan-Niklas Sohn working with Trend Micro Zero Day Initiative
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
(cherry picked from commit 6907b6ea2b)
Peter Hutterer
Michael Wyraz
nerdopolis
Sam James
Adam Jackson
Olivier Fourdan
Benno Schulenberg
Mike Gorse
Jeremy Huddleston Sequoia
John D Pell
Povilas Kanapickas