Proposal for improving vgaarb arbitration method
It appears that after session is up, in most cases, drivers only do
non-legacy accesses.
Non-legacy accesses do not need to block each other. Blocking
arbitration is needed
mostly for session initialization and exiting. To improve
Jamey Sharp wrote:
On Wed, Jun 9, 2010 at 1:01 PM, Alan Coopersmith
alan.coopersm...@oracle.com wrote:
Alan Coopersmith wrote:
-#if defined(__GNUC__) \
-((__GNUC__ 2) || ((__GNUC__ == 2) (__GNUC_MINOR__ 4)))
-__attribute((format(printf,3,4)))
BTW, one thing I've noticed, but will
On Mon, Jun 07, 2010 at 01:20:39AM -0400, Rafi Rubin wrote:
On 06/07/2010 12:26 AM, Peter Hutterer wrote:
On Wed, Jun 02, 2010 at 04:40:34PM +0200, Carlos Garnacho wrote:
I've been discussing with Peter Hutterer about the convenience of the
touchpoints as multiple valuators approach, and how
On Mon, Jun 07, 2010 at 01:33:05AM -0400, Rafi Rubin wrote:
Some more targeted comments.
* Confines multitouch to a single client/Window: This is a well
known problem, to my knowledge only done to have proper
interaction with the paired keyboard.
There are
2010/6/9 Michel Dänzer mic...@daenzer.net:
On Die, 2010-06-08 at 12:57 -0700, Keith Packard wrote:
On Tue, 08 Jun 2010 09:57:44 +0200, Michel Dänzer mic...@daenzer.net wrote:
FWIW, I would have proposed the patch below instead.
That looks better to me; I was after 'not crashing' while your
Hi experts:
In Xvfb implementation, the frame buffer is allocated with IPC_PRIVATE
if uses SHM.
Given that IPC_PRIVATE is just useful to parent/child process, and given
X is a single process,
I don't realize what is the use.
Why not add a param which allows user pass a string to
On Wed, Jun 09, 2010 at 01:20:41PM -0700, Alan Coopersmith wrote:
That sounds like we could almost just call it #ifdef __GNUC__ then.
The oldest gcc version I remember us claiming to support in X.Org
was 2.95, since that's what one of the BSD's was using (and possibly
a previous RHEL?).
TV == Tiago Vignatti tiago.vigna...@nokia.com writes:
TV X server suffers in startup time when relying on the pciaccess's linear
search
TV to fetch vendor and video device name from PCI ID file (when existent). Such
TV names are only used to write the log, which may be superfluous. Users can
On Thu, Jun 10, 2010 at 02:31:48PM +1000, Peter Hutterer wrote:
It's still being pulled in by the HAL CFLAGS but the requirement to define
this was dropped from DBus pre 1.0 (November 2006).
This means we require dbus 1.0 now.
Signed-off-by: Peter Hutterer peter.hutte...@who-t.net
On Wed, Jun 09, 2010 at 11:15:33PM -0700, Alan Coopersmith wrote:
I don't think the device initialization code is involved. What I was
thinking
of was replacing the code currently in xf86AutoConfig() in
common/xf86AutoConfig.c and videoPtrToDriverList() in xf86pciBus.c (and I
guess
On Wed, Jun 09, 2010 at 04:10:02PM -0700, Alan Coopersmith wrote:
Gaetan Nadon wrote:
Because it is available on other platforms, it only fails on Linux.
*sigh* is GNU libc ever going to get with the program?
Nope; Ulrich Drepper thinks it's stupid, so it's not ever going to get
merged.
On Wed, Jun 09, 2010 at 12:26:34PM -0700, Richard Barnette wrote:
And for my next trick...
Initialization for PS/2 compatible mice uses a number of explicit calls
to usleep(). The code mostly lives in src/pnp.c, under
xorg/driver/xf86-input-mouse.
The pattern of sleep calls is
On Thu, Jun 10, 2010 at 12:21:36PM +1000, Peter Hutterer wrote:
The modifier key count is maintained by the XKB layer and
increased/decreased for all modifiers that set state.
Test case, MD/SD modifier key count in comment:
1. keyboard 1: press and hold Shift_L# SD:1 MD:1
2.
On Thu, Jun 10, 2010 at 03:10:52PM +0800, xin zhang wrote:
Hi experts:
In Xvfb implementation, the frame buffer is allocated with IPC_PRIVATE
if uses SHM.
Given that IPC_PRIVATE is just useful to parent/child process, and given
X is a single process,
I don't realize what is the
On Thu, Jun 10, 2010 at 01:04:34PM +0200, ext Daniel Stone wrote:
On Wed, Jun 09, 2010 at 12:26:34PM -0700, Richard Barnette wrote:
And for my next trick...
Initialization for PS/2 compatible mice uses a number of explicit calls
to usleep(). The code mostly lives in src/pnp.c, under
-Original Message-
From: xorg-devel-bounces+shashi.shekar=ti@lists.x.org
[mailto:xorg-devel-bounces+shashi.shekar=ti@lists.x.org]
On Behalf Of Tiago Vignatti
Sent: Thursday, June 10, 2010 4:38 PM
To: ext Daniel Stone
Cc: xorg-devel@lists.x.org; Richard Barnette
On Thu, Jun 10, 2010 at 14:08:14 +0300, Tiago Vignatti wrote:
Does it makes sense to set a trap on autoconf to disables Linux on those
drivers then?
IMO no. If nothing else there are still people using serial mice.
Cheers,
Julien
___
Twas brillig at 13:50:27 10.06.2010 UTC+02 when jcris...@debian.org did
gyre and gimble:
Does it makes sense to set a trap on autoconf to disables Linux on those
drivers then?
JC If nothing else there are still people using serial mice.
Aren't those handled by input driver in kernel? If
On Wed, Jun 9, 2010 at 9:48 PM, Peter Hutterer peter.hutte...@who-t.net wrote:
On Mon, Jun 07, 2010 at 08:39:54PM -0700, Dan Nicholson wrote:
Serial input devices lack properties such as product or vendor name. This
makes matching InputClass sections difficult. Add a MatchPnPID entry to
test
Any module (drivers) depending on xserver also depends on some of the
server private dependencies. Any driver including xf86.h depends on
xext, kbproto, inputproto and randr.
These dependencies are in separate packages, so anything can happen,
removal, wrong version, etc... and the driver fails
On Thu, Jun 10, 2010 at 02:48:48PM +1000, Peter Hutterer wrote:
On Mon, Jun 07, 2010 at 08:39:54PM -0700, Dan Nicholson wrote:
Serial input devices lack properties such as product or vendor name. This
makes matching InputClass sections difficult. Add a MatchPnPID entry to
test against the
On Thu, Jun 10, 2010 at 08:54:38 -0400, Gaetan Nadon wrote:
diff --git a/xorg-server.pc.in b/xorg-server.pc.in
index 44f886a..898bed2 100644
--- a/xorg-server.pc.in
+++ b/xorg-server.pc.in
@@ -16,5 +16,6 @@ Name: xorg-server
Description: Modular X.Org X Server
Version: @PACKAGE_VERSION@
On Mon, Jun 07, 2010 at 08:39:55PM -0700, Dan Nicholson wrote:
Sometimes the vendor and product names aren't specific enough to target
a USB device, so expose the numeric codes in the ID. A MatchUSBID entry
has been added that supports shell pattern matching when fnmatch(3) is
available. For
On Wed, Jun 9, 2010 at 11:15 PM, Alan Coopersmith
alan.coopersm...@oracle.com wrote:
Dan Nicholson wrote:
On Wed, Jun 9, 2010 at 7:27 AM, Alan Coopersmith
alan.coopersm...@oracle.com wrote:
Tiago Vignatti wrote:
Vendor and board naming were never used to create the configure file a
device.
xin zhang wrote:
Hi experts:
In Xvfb implementation, the frame buffer is allocated with
IPC_PRIVATE if uses SHM.
Given that IPC_PRIVATE is just useful to parent/child process, and
given X is a single process,
I don't realize what is the use.
Why not add a param which
On Thu, Jun 10, 2010 at 6:14 AM, Julien Cristau jcris...@debian.org wrote:
On Thu, Jun 10, 2010 at 08:54:38 -0400, Gaetan Nadon wrote:
diff --git a/xorg-server.pc.in b/xorg-server.pc.in
index 44f886a..898bed2 100644
--- a/xorg-server.pc.in
+++ b/xorg-server.pc.in
@@ -16,5 +16,6 @@ Name:
On Wed, 2010-06-09 at 16:10:02 -0700, Alan Coopersmith wrote:
Gaetan Nadon wrote:
Because it is available on other platforms, it only fails on Linux.
*sigh* is GNU libc ever going to get with the program?
On GNU based systems libbsd [0] provides such function (among others).
regards,
X server doesn't need to understand fpe internals, so let it transparent
turning all registration functions in a single one. For that, fill the already
existent register_fpe_functions().
Some X servers don't want font server support, so this patch also sets font
server support to be configured in
Signed-off-by: Tiago Vignatti tiago.vigna...@nokia.com
---
configure.ac|2 +-
src/Makefile.am |2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/configure.ac b/configure.ac
index 35ac80b..49f1d06 100644
--- a/configure.ac
+++ b/configure.ac
@@ -23,7 +23,7 @@ dnl
It's required to get new register_fpe_functions dependency.
Signed-off-by: Tiago Vignatti tiago.vigna...@nokia.com
---
configure.ac |8 ++--
1 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/configure.ac b/configure.ac
index 48ee44d..be524be 100644
--- a/configure.ac
+++
It's already defined inside REQUIRED_LIBS and all DDX are getting it already.
No semantical changes.
Signed-off-by: Tiago Vignatti tiago.vigna...@nokia.com
---
This is just clean-up.
configure.ac |8
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/configure.ac
On Wed, 2010-06-09 at 07:27:32 -0700, Alan Coopersmith wrote:
Tiago Vignatti wrote:
Vendor and board naming were never used to create the configure file a
device.
This patch remove their references.
I have been wanting for a while to move the autoconfiguration matching for
video cards
Twas brillig at 16:41:12 10.06.2010 UTC+03 when tiago.vigna...@nokia.com did
gyre and gimble:
TV X server doesn't need to understand fpe internals, so let it
TV transparent turning all registration functions in a single one. For
TV that, fill the already existent register_fpe_functions().
On Thu, Jun 10, 2010 at 9:41 AM, Tiago Vignatti
tiago.vigna...@nokia.com wrote:
X server doesn't need to understand fpe internals, so let it transparent
turning all registration functions in a single one. For that, fill the already
existent register_fpe_functions().
Some X servers don't want
On Thu, Jun 10, 2010 at 9:41 AM, Tiago Vignatti
tiago.vigna...@nokia.com wrote:
X server doesn't need to understand fpe internals, so use
register_fpe_functions from libXfont.
Signed-off-by: Tiago Vignatti tiago.vigna...@nokia.com
Seems sane to me. For the series:
Reviewed-by: Alex Deucher
On Thu, Jun 10, 2010 at 16:41:14 +0300, Tiago Vignatti wrote:
diff --git a/include/dixfont.h b/include/dixfont.h
index cf86f54..617b94a 100644
--- a/include/dixfont.h
+++ b/include/dixfont.h
@@ -141,10 +141,7 @@ extern _X_EXPORT void InitGlyphCaching(void);
extern _X_EXPORT void
On Thu, Jun 10, 2010 at 16:41:16 +0300, Tiago Vignatti wrote:
It's required to get new register_fpe_functions dependency.
Signed-off-by: Tiago Vignatti tiago.vigna...@nokia.com
---
configure.ac |8 ++--
1 files changed, 6 insertions(+), 2 deletions(-)
This should probably be
On Thu, Jun 10, 2010 at 03:59:51PM +0200, Guillem Jover wrote:
On Wed, 2010-06-09 at 16:10:02 -0700, Alan Coopersmith wrote:
Gaetan Nadon wrote:
Because it is available on other platforms, it only fails on Linux.
*sigh* is GNU libc ever going to get with the program?
On GNU based
On Thu, Jun 10, 2010 at 04:41:13PM +0300, Tiago Vignatti wrote:
diff --git a/src/Makefile.am b/src/Makefile.am
index d658667..c870585 100644
--- a/src/Makefile.am
+++ b/src/Makefile.am
@@ -65,4 +65,4 @@ libXfont_la_LIBADD = \
libXfont_la_SOURCES = dummy.c
-libXfont_la_LDFLAGS =
On Thu, Jun 10, 2010 at 04:41:16PM +0300, Tiago Vignatti wrote:
It's required to get new register_fpe_functions dependency.
This should come with or before patch #1, which actually introduces the
dependency.
Cheers,
Daniel
signature.asc
Description: Digital signature
On Thu, 2010-06-10 at 15:53:52 +0100, Daniel Stone wrote:
On Thu, Jun 10, 2010 at 03:59:51PM +0200, Guillem Jover wrote:
On Wed, 2010-06-09 at 16:10:02 -0700, Alan Coopersmith wrote:
Gaetan Nadon wrote:
Because it is available on other platforms, it only fails on Linux.
*sigh* is
Hi,
Even if there is an interesting thread concerning the use of multitouch on
xorg-devel,
I still send the up-to-date patches to enable multitouch as valuators.
Changes since last time:
* Remove the detection of the maximum number of touches based on the tracking ID
* changes in comments
The driver has to know the correct labels of the ABS_MT_* events.
However, to keep compatibility with older xserver that don't define
the corresponding labels, this patch does not inject them in the list.
Signed-off-by: Benjamin Tissoires tisso...@cena.fr
---
src/evdev.c | 29
The step one in implementing multitouch in evdev is to report all the
touches in different valuators.
This patch detects multitouch devices and creates the extra valuators
required for multitouch.
Bonus point: this patch also sort the multitouch valuators to have
ABS_MT_POSITION_X and
The MT-protocol A is the one that sends SYN_MT_REPORT. It is opposed to
the new one based on MT_SLOTS (protocol B).
(see kernel_src/Documentation/input/multi-touch-protocol.txt)
The processing is dummy since there is no computation while receiving
mt events: it just stores it at the first
On Thu, 2010-06-10 at 07:01 -0700, Dan Nicholson wrote:
On Thu, Jun 10, 2010 at 6:14 AM, Julien Cristau jcris...@debian.org wrote:
On Thu, Jun 10, 2010 at 08:54:38 -0400, Gaetan Nadon wrote:
diff --git a/xorg-server.pc.in b/xorg-server.pc.in
index 44f886a..898bed2 100644
---
On Thu, 10 Jun 2010 15:10:52 +0800, xin zhang xing5...@gmail.com wrote:
Why not add a param which allows user pass a string to produce IPC key,
and create share memory
with the key, then other process will really access to the frame buffer.
Xvfb used to do precisely this, and it was
Using type == DRAWABLE_WINDOW to differentiate between pixmaps and
windows isn't sufficient as input-only windows will end up in the
pixmap case. This patch changes a few more code paths to use
WindowDrawable instead.
Signed-off-by: Keith Packard kei...@keithp.com
---
The xinerama patch
On Mon, 07 Jun 2010 00:52:11 -0700, Jeremy Huddleston jerem...@apple.com
wrote:
Ok, that seems to get past the problem in XQuartz, but we're still
failing in xtest ...
I don't have this patch in master yet; can you ship me what you've
tested so I can get it merged today?
--
The following changes since commit 8e97e5f9425639ad0a084150d0b232cad417595d:
Keith Packard (1):
If XTest is always required, then eliminate the XTest devPrivate
are available in the git repository at:
git://people.freedesktop.org/~jeremyhu/xserver master
Jeremy Huddleston (2):
On Thu, Jun 10, 2010 at 8:51 AM, Gaetan Nadon mems...@videotron.ca wrote:
On Thu, 2010-06-10 at 07:01 -0700, Dan Nicholson wrote:
On Thu, Jun 10, 2010 at 6:14 AM, Julien Cristau jcris...@debian.org wrote:
On Thu, Jun 10, 2010 at 08:54:38 -0400, Gaetan Nadon wrote:
diff --git
On Thu, Jun 10, 2010 at 7:04 AM, Guillem Jover guil...@hadrons.org wrote:
On Wed, 2010-06-09 at 07:27:32 -0700, Alan Coopersmith wrote:
Tiago Vignatti wrote:
Vendor and board naming were never used to create the configure file a
device.
This patch remove their references.
I have been
On Thu, 2010-06-10 at 07:38 -0700, Keith Packard wrote:
Checking timestamps in post 1.1 randr requests was never a good idea,
let's ignore them and just make the configuration changes.
Signed-off-by: Keith Packard kei...@keithp.com
---
Because the new randr code uses persistent resource
On Wed, Jun 9, 2010 at 9:31 PM, Peter Hutterer peter.hutte...@who-t.net wrote:
It's still being pulled in by the HAL CFLAGS but the requirement to define
this was dropped from DBus pre 1.0 (November 2006).
This means we require dbus 1.0 now.
Signed-off-by: Peter Hutterer
On Thu, Jun 10, 2010 at 7:31 AM, Keith Packard kei...@keithp.com wrote:
Using type == DRAWABLE_WINDOW to differentiate between pixmaps and
windows isn't sufficient as input-only windows will end up in the
pixmap case. This patch changes a few more code paths to use
WindowDrawable instead.
Dan Nicholson wrote:
On Thu, Jun 10, 2010 at 7:04 AM, Guillem Jover guil...@hadrons.org wrote:
On Wed, 2010-06-09 at 07:27:32 -0700, Alan Coopersmith wrote:
Tiago Vignatti wrote:
Vendor and board naming were never used to create the configure file a
device.
This patch remove their
A window's DrawableRec's x and y fields contain the offsets to be
applied to protocol-visible window-relative coordinates in order to make
those coordinates screen-relative instead. When Xinerama is active,
coordinates relative to the root window need to be adjusted for each
screen by the position
Hello,
This patch originally written by Yves fixes this bug:
https://bugs.freedesktop.org/show_bug.cgi?id=28077
Could someone review and apply it?
Yves, don't hesitate to add your Signed-off-by.
Cheers,
Eric
From f958be1d48b7a75a4044a3db8a2b25c7c44662a3 Mon Sep 17 00:00:00 2001
From:
On Thu, 10 Jun 2010 13:49:52 -0400, Chase Douglas chase.doug...@canonical.com
wrote:
On Thu, 2010-06-10 at 07:38 -0700, Keith Packard wrote:
Checking timestamps in post 1.1 randr requests was never a good idea,
let's ignore them and just make the configuration changes.
Signed-off-by:
On Thu, Jun 10, 2010 at 02:08:14PM +0300, Tiago Vignatti wrote:
On Thu, Jun 10, 2010 at 01:04:34PM +0200, ext Daniel Stone wrote:
On Wed, Jun 09, 2010 at 12:26:34PM -0700, Richard Barnette wrote:
And for my next trick...
Initialization for PS/2 compatible mice uses a number of
Keith,
please pull the set of input fixes below from my repo.
Dan's patches add a few new matches for the InputClass matching system,
essentially bringing it up to what the HAL system used to support.
The Xi fix for 25480 is a rather important one, I'm pretty sure that's the
cause for many of
Any module (drivers) depending on xserver also depends on some of the
server private dependencies. Any driver including xf86.h depends on
xext, kbproto, inputproto and randr.
These dependencies are in separate packages, so anything can happen,
removal, wrong version, etc... and the driver fails
I've got a few DRI2 cherry-picks for 1.8 on
git://people.freedesktop.org/~whot/xserver.git server-1.8-branch
Cherry-picking them was more complicated than I hoped for since we had one
of the patches reverted, then others on top, etc. And I don't know what I'm
doing when it comes to DRI2.
So,
On Thu, 10 Jun 2010 15:37:40 -0700, Jamey Sharp ja...@minilop.net wrote:
A window's DrawableRec's x and y fields contain the offsets to be
applied to protocol-visible window-relative coordinates in order to make
those coordinates screen-relative instead. When Xinerama is active,
coordinates
The following changes since commit 07a093add0b7e40c4d9b9b59273e3ff9e14a88a7:
Keith Packard (1):
Merge remote branch 'whot/for-keith'
are available in the git repository at:
git://people.freedesktop.org/~alanc/xserver.git master
Alan Coopersmith (2):
Record some additional
On Mon, 07 Jun 2010 21:04:38 +0700, Mikhail Gusarov dotted...@dottedmag.net
wrote:
xfree86: Get rid of xstrdup when argument is definitely non-NULL
This one now conflicts in non-obvious ways. Please feel free to fix this
up and re-submit the pull request.
--
keith.pack...@intel.com
On Thu, Jun 10, 2010 at 6:59 PM, Keith Packard kei...@keithp.com wrote:
As you might imagine, there are a million corner cases with tile origins
that could be broken here. Certainly the basic idea of making
drawable.x/drawable.y correct for tiling has a lot of merit.
Fortunately, the test
On Thu, 10 Jun 2010 19:50:29 -0700, Jamey Sharp ja...@minilop.net wrote:
I guess I'd need to run the test suite on a Xinerama server, and
probably need to ensure that windows that XTS creates span two
screens. Or something? I'm not sure how to even hit these cases,
especially in XTS.
Yeah.
From: Ben Skeggs bske...@redhat.com
Without resetting the private's pointer to NULL, we can end up freeing
the struct twice:
==11188== Invalid free() / delete / delete[]
==11188==at 0x4C24D72: free (vg_replace_malloc.c:325)
==11188==by 0x42D8A3: dixFreePrivates (privates.c:217)
==11188==
On Fri, Jun 11, 2010 at 02:26:57PM +1000, Ben Skeggs wrote:
From: Ben Skeggs bske...@redhat.com
Without resetting the private's pointer to NULL, we can end up freeing
the struct twice:
==11188== Invalid free() / delete / delete[]
==11188==at 0x4C24D72: free (vg_replace_malloc.c:325)
Argh! A recent commit to the xf86 config parsing code added list.h to
xf86Parser.h, which is included by drivers to look stuff up. Because of
this, the intel driver no longer builds against master (would that the
drivers were in the server tree...). I have an RC1 tar file sitting here
and decided
On Thu, Jun 10, 2010 at 10:13:16PM -0700, Keith Packard wrote:
Argh! A recent commit to the xf86 config parsing code added list.h to
xf86Parser.h, which is included by drivers to look stuff up. Because of
this, the intel driver no longer builds against master (would that the
drivers were in
thank you you all! now i understand the reason and solution.
2010/6/10 Daniel Stone dan...@fooishbar.org
On Thu, Jun 10, 2010 at 03:10:52PM +0800, xin zhang wrote:
Hi experts:
In Xvfb implementation, the frame buffer is allocated with IPC_PRIVATE
if uses SHM.
Given that
73 matches
Mail list logo