Delay removing the client from these two queues until all potential
I/O has completed in case we mark the client as ready for reading or
with pending output during the close operation.
Bugzilla: https://bugs.freedesktop.org/100957
Signed-off-by: Keith Packard <keithp@keithp.com>
Tested-by: Nick Sarnie <commendsarnex@gmail.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Found when the meson conversion set the symbol to defined, instead of
defined to 1.
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
A client which is attended while a grab is blocking execution of its
requests needs to be placed in the saved_ready_clients list so that it
will get scheduled once the grab terminates. Otherwise, if the client
never sends another request, there is no way for it to be placed in
the ready_clients list.
v2: Wrap comment above mark_client_saved_ready.
Remove test for OS_COMM_IGNORED which will always be true.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99333
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
I think it is possible that output could get queued to a client during
CloseDownClient. After it is removed from the pending queue, active
grabs are released, the client is awoken if sleeping and any work
queue entries related to the client are processed.
To fix this, move the call removing it from the output_pending chain
until after clientGone has been set and then check clientGone in
output_pending_mark.
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1382444
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
GetImage is allowed to return window border contents, so don't remove
that from the returned image.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
pClient does not contain a live value after the transition to lists
https://bugs.freedesktop.org/show_bug.cgi?id=97765
Application Specific Information:
X.Org X Server 1.18.99.1 Build Date: 20160910
=================================================================
==16921==ERROR: AddressSanitizer: global-buffer-overflow on address 0x000108ce3834 at pc 0x000108880766 bp 0x7000045f76c0 sp 0x7000045f76b8
READ of size 4 at 0x000108ce3834 thread T6
#0 0x108880765 in SmartScheduleClient dispatch.c:365
#1 0x10887ecc5 in Dispatch dispatch.c:422
#2 0x1088c05f1 in dix_main main.c:301
#3 0x1082aabba in server_thread quartzStartup.c:66
#4 0x7fffc5f16aaa in _pthread_body (libsystem_pthread.dylib+0x3aaa)
#5 0x7fffc5f169f6 in _pthread_start (libsystem_pthread.dylib+0x39f6)
#6 0x7fffc5f161fc in thread_start (libsystem_pthread.dylib+0x31fc)
Regressed-in: 8f1edf4bd3
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Currently with PRIME if we detect a secondary GPU,
we switch to using SW cursors, this isn't optimal,
esp for the intel/nvidia combinations, we have
no choice for the USB offload devices.
This patch checks on each slave screen if hw
cursors are enabled, and also calls set cursor
and move cursor on all screens.
Cc: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Eliminates all of the fd_set mangling in the server main thread
v2: Listen for POLLOUT while writes are blocked.
v3: Only mark client not ready on EAGAIN return from read
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
This converts the dispatch loop into using a list of ready clients
instead of an array. This changes the WaitForSomething API so that it
notifies DIX when a client becomes ready to read, instead of returning
the set of ready clients.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
The intermediate grabState, "GrabKickout", was used to trigger
dispatch into going back to WaitForSomething after doing a GrabServer
so that the set of ready clients would be recomputed to match what the
server should be processing. As we only process one client per
WaitForSomething call, we will always hit WaitForSomething after
finishing the current client, and so don't need any special case here.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Instead of having scheduling done in two places (one in
WaitForSomething, and the other in SmartScheduleClient), just stick
all of the scheduling in SmartScheduleClient.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
This new libXfont API eliminates exposing internal X server symbols to
the font library, replacing those with a struct full of the entire API
needed to use that library.
v2: Use libXfont2 instead of libXfont_2
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
A single provider can be both a offload and source slave at the same time,
the use of seperate lists breaks in this case e.g. :
xrandr --listproviders
Providers: number : 2
Provider 0: id: 0x7b cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 3 outputs: 2 associated providers: 0 name:modesetting
Provider 1: id: 0x46 cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 2 outputs: 5 associated providers: 0 name:modesetting
xrandr --setprovideroutputsource 1 0x7b
xrandr --listproviders
Providers: number : 2
Provider 0: id: 0x7b cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 3 outputs: 2 associated providers: 1 name:modesetting
Provider 1: id: 0x46 cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 2 outputs: 5 associated providers: 1 name:modesetting
xrandr --setprovideroffloadsink 1 0x7b
xrandr --listproviders
Providers: number : 3
Provider 0: id: 0x7b cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 3 outputs: 2 associated providers: 2 name:modesetting
Provider 1: id: 0x46 cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 2 outputs: 5 associated providers: 2 name:modesetting
Provider 2: id: 0x46 cap: 0xf, Source Output, Sink Output, Source Offload, Sink Offload crtcs: 2 outputs: 5 associated providers: 2 name:modesetting
Not good. The problem is that the provider with id 0x46 now is on both
the output_slave_list and the offload_slave_list of the master screen.
This commit fixes this by unifying all 3 lists into a single slaves list.
Note that this does change the struct _Screen definition, so this is an ABI
break. I do not expect any of the drivers to actually use the removed / changed
fields so a recompile should suffice.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
There are no in-tree consumers of the audit hooks, and they are in any
case redundant with the dtrace dispatch hooks. Neither is there any
in-tree user of the core request dispatch hook. The extension hook is
only used for non-default security cases, but in the absence of LTO we
always have to take the function call into XaceHookDispatch to find out
that there's no callback registered.
Cc: Eamon Walsh <ewalsh@tycho.nsa.gov>
Signed-off-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
This was added in:
commit 312910b4e3
Author: Chase Douglas <chase.douglas@canonical.com>
Date: Wed Apr 18 11:15:40 2012 -0700
Update currentTime in dispatch loop
Unfortunately this is equivalent to calling GetTimeInMillis() once per
request. In the absolute best case (as on Linux) you're only hitting the
vDSO; on other platforms that's a syscall. Either way it puts a pretty
hard ceiling on request throughput.
Instead, push the call down to the requests that need it; basically,
grab processing and event generation.
Cc: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Adam Jackson <ajax@redhat.com>
This allows the server to call GetTimeInMillis() after each request is
processed to avoid needing setitimer. -dumbSched now turns off the
setitimer.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Make the maximum number of clients user configurable, either from the command
line or from xorg.conf
This patch works by using the MAXCLIENTS (raised to 512) as the maximum
allowed number of clients, but allowing the actual limit to be set by the
user to a lower value (keeping the default of 256).
There is a limit size of 29 bits to be used to store both the client ID and
the X resources ID, so by reducing the number of clients allowed to connect to
the X server, the user can increase the number of X resources per client or
vice-versa.
Parts of this patch are based on a similar patch from Adam Jackson
<ajax@redhat.com>
This now requires at least xproto 7.0.28
Signed-off-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Olivier Fourdan <ofourdan@redhat.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
It's going to multiply anyway, so if we have non-constant values, might
as well let it do the multiplication instead of adding another multiply,
and good versions of calloc will check for & avoid overflow in the process.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Use typedefs to work around dtrace dropping const qualifiers from probe
arguments when generating Xserver-dtrace.h. Add new probes.h header to
avoid having to replicate these typedefs in every file with dtrace probes.
Gets rid of these warnings from gcc 4.8:
getevents.c:1096:9:
warning: passing argument 6 of '__dtrace_Xserver___input__event' discards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1096:9:
warning: passing argument 7 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1651:9:
warning: passing argument 6 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1651:9:
warning: passing argument 7 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1791:9:
warning: passing argument 6 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1791:9:
warning: passing argument 7 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1921:9:
warning: passing argument 6 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
getevents.c:1921:9:
warning: passing argument 7 of '__dtrace_Xserver___input__event' disards
'const' qualifier from pointer target type [enabled by default]
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
The length checking code validates PutImage height and byte width by
making sure that byte-width >= INT32_MAX / height. If height is zero,
this generates a divide by zero exception. Allow zero height requests
explicitly, bypassing the INT32_MAX check.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
ProcPutImage() calculates a length field from a width, left pad and depth
specified by the client (if the specified format is XYPixmap).
The calculations for the total amount of memory the server needs for the
pixmap can overflow a 32-bit number, causing out-of-bounds memory writes
on 32-bit systems (since the length is stored in a long int variable).
Reported-by: Ilja Van Sprundel <ivansprundel@ioactive.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
No DDX is overriding this and it's fairly absurd to expose it as a
screen operation anyway.
Reviewed-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Adam Jackson <ajax@redhat.com>
isItTimeToYield in the conditional effectively didn't do anything here.
Take it out, and remove the comment since LBX proxies aren't a thing for
us anymore.
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Adam Jackson <ajax@redhat.com>
dispatch.c: In function 'SetVendorString':
dispatch.c:481:29: warning: declaration of 'string' shadows a global declaration [-Wshadow]
SetVendorString(const char *string)
^
dispatch.c:135:21: warning: shadowed declaration is here [-Wshadow]
typedef const char *string;
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
A client which is ready, but hasn't run for a while, should receive
the same benefit as one which has simply been idle for a while. Use
the smart_stop_tick to see how long it has been since a client has
run instead of smart_check_tick, which got reset each time a client
was ready, even if it didn't get to run.
Reported-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Keith Packard <keithp@keithp.com>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
This lets us stop using the 'pointer' typedef in Xdefs.h as 'pointer'
is used throughout the X server for other things, and having duplicate
names generates compiler warnings.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
As usual, mostly const char changes. However, filter_device_events had
a potentially uninitialized value, 'raw', which I added a bunch of
checks for. I suspect most of those are 'can't happen', but it's hard
to see that inside the function.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
A default timeslice of 20ms means a pathological client can ruin up to
two frames per scheduler tick. And a fifth of a second is just insane.
Pick two different numbers out of the hat. A 5ms slice means you can
probably keep up with two or three abusive clients, and letting it burst
to 15ms should give you about all the timeslice you need for a
fullscreen game (that's doing server-side rendering for some reason).
If you're running on a system with a 10ms granularity on SIGALRM, then
this effectively changes the intervals to 10ms and 30ms. Which is still
better, just not as better.
I suspect this is about as good as we can do without actually going
preemptive, which is an entire other nightmare.
Reviewed-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Adam Jackson <ajax@redhat.com>
CreateCursor (Xlib call XCreatePixmapCursor) with a non-bitmap
source pixmap and a None mask is supposed to error out with BadMatch,
but didn't.
From der Mouse <mouse@Rodents-Montreal.ORG>, changed following
comments by Alan Coopersmith <alan.coopersmith@oracle.com>.
Signed-off-by: Thomas Klausner <wiz@NetBSD.org>
Reviewed-by: Jasper St. Pierre <jstpierre@mecheye.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
TouchListenerGone cleans up if a client disappears. Having this in
FreeGrab() triggers cyclic removal of grabs, emitting wrong events. In
particular, it would clean up a passive grab record while that grab is
active.
Move it to CloseDownClient() instead, cleaning up before we go.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
==21860== 24 bytes in 1 blocks are still reachable in loss record 85 of 397
==21860== at 0x4C2B3F8: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==21860== by 0x61ED93: AllocateOutputBuffer (io.c:1037)
==21860== by 0x61E15A: WriteToClient (io.c:764)
==21860== by 0x457B30: ProcQueryExtension (extension.c:275)
==21860== by 0x43596B: Dispatch (dispatch.c:432)
==21860== by 0x425DAB: main (main.c:295)
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
We should have no problem allowing output/offload from the same slave,
I asserted here, but in order to implement reverse optimus this makes
perfect sense. (reverse optimus is intel outputting to nvidia).
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The formatter confused address operators preceded by casts with
bitwise-and expressions, placing spaces on either side of both.
That syntax isn't used by ordinary address operators, however,
so fix them for consistency.
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
setitimer() and SIGALRM aren't available on WIN32, so smart scheduler
code cannot be built. Provide only stubs for smart scheduler timer
code, and disable smart scheduler by default.
Signed-off-by: Ryan Pavlik <rpavlik@iastate.edu>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Forwarding proxies like sshd will appear to be local, even though they
aren't really. This leads to weird behaviour for extensions that truly
require running under the same OS services as the client, like MIT-SHM
and DRI2.
Add two new legal values for the initial connection's byteOrder field,
'r' and 'R'. These act like 'l' and 'B' respectively, but have the side
effect of forcing the client to be treated as non-local. Forwarding
proxies should attempt to munge the first packet of the connection
accordingly; older servers will reject connections thusly munged, so the
proxy should fall back to passthrough if the munged connection attempt
fails.
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Adam Jackson <ajax@redhat.com>
If we don't free this here, it gets freed later in the resource
cleanups, however it then looks up up pmap->pScreen, which we
freed already in this function. So free the default colormap
when we should.
This fixes a bug after a couple of hotplug cycles when you try
to exit the X server and it crashes.
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Adds new function padding_for_int32() and uses existing pad_to_int32()
depending on required results.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Tested-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Tested-by: Daniel Stone <daniel@fooishbar.org>
Let the compiler worry about 0-filling the rest of the fields,
instead of memsetting the whole struct and then going back to
overwrite some of the fields.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Tested-by: Daniel Stone <daniel@fooishbar.org>
When passing variable pointers to functions or otherwise doing long
sequences to compute values for replies, create & use some new
temporary variables, to allow for simpler initialization of reply
structures in the following patches.
Move memsets & other initializations to group with the rest of the
filling in of the reply structure, now that they're not needed so
early in the code path.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Tested-by: Daniel Stone <daniel@fooishbar.org>
Casting return to (void) was used to tell lint that you intended
to ignore the return value, so it didn't warn you about it.
Casting the third argument to (char *) was used as the most generic
pointer type in the days before compilers supported C89 (void *)
(except for a couple places it's used for byte-sized pointer math).
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Tested-by: Daniel Stone <daniel@fooishbar.org>
add the linked list and provider hooks.
v1.1: add another assert in the add path.
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This adds two functions for drivers to use directly to keep a
linked list of slave pixmaps to do damage tracking on and keep
updated. It also adds a helper function that drivers may optionally
call to do a simple copy area damage update.
v2: use damage.h not damagestr.h, fixes ephyr build.
v3: address ajax review: use slave_dst, drop unused dst member.
v4: check DamageCreate return, add SourceValidate comment,
add a comment addressing possible optimisation possibility
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Just add the interfaces to attach/detach output slaves, and
a linked list to keep track of them. Hook up the randr providers
list to include these slaves.
v1.1: add another assert to the add path.
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>