switch to recent devPrivates ABI (as used in X.org's Xserver) #560
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
sunweaver
changed the title
uli42
reviewed
Nov 21, 2017
| @@ -266,17 +264,15 @@ XvScreenInit(ScreenPtr pScreen) | |||
| ErrorF("XvScreenInit: Unable to allocate resource types\n"); | |||
| return BadAlloc; | |||
| } | |||
| if (XvScreenIndex < 0) | |||
There was a problem hiding this comment.
is that variable also removed? I have not seen it at a first glance..
|
|
Running |
|
On Wednesday, November 22, 2017, Ulrich Sibiller wrote:
Program received signal SIGBUS, Bus error.
fixup_visuals (screen=0) at xf86glx.c:551
551 const int vis_class = _gl_convert_to_x_visual_type( modes->visualType );
(gdb) bt
#0 fixup_visuals (screen=0) at xf86glx.c:551
#1 __MESA_screenProbe (screen=0) at xf86glx.c:679
#2 0x000000000059b10c in __glXScreenInit (numscreens=1) at glxscreens.c:337
#3 0x0000000000443103 in GlxExtensionInit () at ../../GL/glx/glxext.c:282
#4 0x000000000040f795 in InitExtensions ***@***.***=2, ***@***.***=0x7fffffffdcd8) at miinitext.c:447
#5 0x000000000040edc4 in main (argc=2, argv=0x7fffffffdcd8, envp=<optimized out>) at main.c:280
Yep, seeing that, too. Have a bunch of local commits since last night, work in progress...
…--
Sent from my Fairphone 2 (running Sailfish OS)
|
sunweaver
force-pushed
the
pr/switch-devPrivates-ABI
branch
from
5ae81d1
to
09c2d80
Compare
|
I played a bit with this. After removing one typo it compiles but crashes in XkbSetExtension. This is because of xorg-xserver's commit 1987af8c498a1bf394a8951ca6d5b0b7f7a35188 missing here (which is pre 1.2.99). Our current 3.6.x has dix from 1.3.0.0 ans should include that commit. So please rebase this! |
|
Cool! Can you push the typo to the PR branch? Thanks!
…On Sunday, July 29, 2018, Ulrich Sibiller wrote:
I played a bit with this. After removing one typo it compiles but crashes in XkbSetExtension. This is because of xorg-xserver's commit 1987af8c498a1bf394a8951ca6d5b0b7f7a35188 missing here (which is pre 1.2.99). Our current 3.6.x has dix from 1.3.0.0 ans should include that commit. So please rebase this!
--
You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub:
#560 (comment)
--
Sent from my Fairphone (powered by SailfishOS)
|
|
not today, but I made comment in the PR
I have already tried rebasing but this is a bit more work so I ran out of
time.
On Sun, Jul 29, 2018 at 10:20 AM, Mike Gabriel <[email protected]>
wrote:
… Cool! Can you push the typo to the PR branch? Thanks!
On Sunday, July 29, 2018, Ulrich Sibiller wrote:
> I played a bit with this. After removing one typo it compiles but
crashes in XkbSetExtension. This is because of xorg-xserver's commit
1987af8c498a1bf394a8951ca6d5b0b7f7a35188 missing here (which is pre
1.2.99). Our current 3.6.x has dix from 1.3.0.0 ans should include that
commit. So please rebase this!
>
> --
> You are receiving this because you authored the thread.
> Reply to this email directly or view it on GitHub:
> #560 (comment)
--
Sent from my Fairphone (powered by SailfishOS)
—
You are receiving this because your review was requested.
Reply to this email directly, view it on GitHub
<#560 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AGBw5BBhoT7cA47F0wgTFnW6eMFdIzVgks5uLXAzgaJpZM4QlniS>
.
|
Backported from X.org: commit b6b26560d65d9468c89dd046396d9aeb14e1b227 Author: Peter Hutterer <[email protected]> Date: Fri Oct 31 16:23:34 2008 +1030 Move EXTENSION_BASE and EXTENSION_EVENT_BASE to misc.h.
Backported from X.org: commit 854c1fa4a1e90288d1e2f3777fac3e1a9acb1235 Author: Daniel Stone <[email protected]> Date: Tue Jul 10 02:02:50 2012 +0100 Add a common ARRAY_SIZE macro to dix.h Does what it says on the box, replacing those from Xi/ and glx/. Signed-off-by: Daniel Stone <[email protected]> Reviewed-by: Cyril Brulebois <[email protected]> Reviewed-by: Jamey Sharp <[email protected]> Reviewed-by: Peter Hutterer <[email protected]> Signed-off-by: Peter Hutterer <[email protected]> Signed-off-by: Keith Packard <[email protected]>
Backports this X.org commit:
Revert "Bug #6924: Restore the ABI for DrawableRec and ColormapRec to the state"
This reverts commit bc0c56c407117d1545e20d21f7d30eb3472d618b since we're
breaking ABI anyway.
Conflicts:
ChangeLog
dix/colormap.c
dix/pixmap.c
dix/window.c
include/pixmapstr.h
Signed-off-by: Jamey Sharp <[email protected]>
Reviewed-by: Aaron Plattner <[email protected]>
Backported-to-NX-by: Mike Gabriel <[email protected]>
Backported from X.org: commit 48b47928133bc462947104bd0f4204f00b993862 Author: Adam Jackson <[email protected]> Date: Fri Oct 3 17:16:28 2008 -0400 mi{Create,Destroy}GCOps are unused, nuke 'em.
Backported from X.org: commit 0ce687634fa5e0d955e896a0d0d5f584a4cfd1a4 Author: Adam Jackson <[email protected]> Date: Wed Aug 13 17:28:08 2008 -0400 Eviscerate save-under support. Use a compositing manager already. Plus I really wanted to use 'eviscerate' in a commit message.
Backported from X.org: commit 0214d0b96a9b900a8cf5c7f2183add3a411441f3 Author: Peter Hutterer <[email protected]> Date: Tue May 1 15:49:41 2007 +0930 Start devices after windows are initialized. This fixes the hotplug segfault introduced with the multihead changes and cleans up the code a bit as well.
commit f38b2b628322f4d763b0c7f8387d5ab0472a7d2b Author: Keith Packard <[email protected]> Date: Thu Jul 5 11:15:18 2012 -0700 mi: Don't save midispcur rendering resources for each cursor Instead, only save them for the most recently drawn cursor. This saves a bunch of storage for idle cursors. Signed-off-by: Keith Packard <[email protected]> Reviewed-by: Dave Airlie <[email protected]> Backported to NX by: Mike Gabriel <[email protected]>n
Backported from X.org: commit 957bf959fb577b292a3e4f6bb67740ca09e7aeb9 Author: Dave Airlie <[email protected]> Date: Wed Jun 27 14:36:04 2012 +0100 dix/render: consolidate window format matching code. This code existed in 3 different forms, perhaps it should be consolidated. Reviewed-by: Keith Packard <[email protected]> Reviewed-by: Alex Deucher <[email protected]> Signed-off-by: Dave Airlie <[email protected]> Signed-off-by: Keith Packard <[email protected]> Backported to NX by: Mike Gabriel <[email protected]>
Backported from X.org: commit f308cbea8c9fa5c5c6705bc301178b167de450d5 Author: Jamey Sharp <[email protected]> Date: Mon May 24 18:41:54 2010 -0700 mi: Delete unused flicker-free MoveCursor code. It's been commented-out for three and a half years and nobody seems to be missing it enough to resurrect it. Besides deleting code that is untested and therefore buggy, this saves a little memory for each pointer device on each screen. Signed-off-by: Jamey Sharp <[email protected]> Reviewed-by: Peter Hutterer <[email protected]> Backported to NX by: Mike Gabriel <[email protected]>
|
@uli42: I have just rebased this PR's update branch and it now builds. But it also segfaults when starting it with nxagent -ac :1. Need to attend to other stuff now, but maybe this helps. |
sunweaver
force-pushed
the
pr/switch-devPrivates-ABI
branch
from
09c2d80
to
d1cedfb
Compare
|
I have tried several hours to getting this fixed but I failed. My main problem is that I am absolutely unsure about the expected state of this code. Which is the xorg upstream commit we are trying to achieve? The are so many commits titled "devPrivates rework" with huge differences to our code base... |
|
The expected state of code is what we find in X.org these days regarding the privates ABI. There have been various rethinkings of this ABI, there is not a range of commits that we can backport and then we have it. Unfortunately. With some looks at X.org Xserver git log, I'd say that we want to get to a stage as found in a817271d461e2f95dd7dc62cd1c7d123ce92f555 (X.org Xserver). |
|
For convenience: https://cgit.freedesktop.org/xorg/xserver/commit/?id=a817271d461e2f95dd7dc62cd1c7d123ce92f555 I'd did not know about this document until today. |
|
That xorg document states "devPrivates have been revised substantially for X.Org X server release 1.5, updated again for the 1.9 release and extended again for the 1.13 relealse." So maybe we should step back and limit ourselves to 1.5 for now. |
|
Hi Uli,
On Tuesday, August 7, 2018, Ulrich Sibiller wrote:
That xorg document states "devPrivates have been revised substantially for X.Org X server release 1.5, updated again for the 1.9 release and extended again for the 1.13 relealse."
So maybe we should step back and limit ourselves to 1.5 for now.
I disagree. I'd say let's have the porting effort once and bump devPrivates to the state of X.org 1.13 immediately and now. Saves us the hassle of working on devPrivates bumps multiple times...
Mike
…--
Sent from my Fairphone (powered by SailfishOS)
|
|
In principle I'd agree, but we are facing two problems here:
The easier way is to implement the privates stuff as it was in - say - 1.4.2 and lift all modules to that very level. Then we can upgrade to 1.5 quite easily (compared to the current approach). I don't think implementing later reworks of devPrivates will be a hassle then. |
|
Hi Uli,
On Mi 08 Aug 2018 12:38:01 CEST, Ulrich Sibiller wrote:
In principle I'd agree, but we are facing two problems here:
1. it complicates upgrading to newer (not newest!) versions of xorg
significantly until we are par with 1.13 (which I don't expect
happening soon)
This is true.
2. currenty XkbSetExtension() crashes. I suppose this is because some
It crashes in RegisterPointerDevice() (or what the function name is).
The Register*Device() code has been fully removed from X.org later, so
my recent attempt was to remove those code portions altogether.
init code is missing in dix/devices.c. I have been trying to
implement that code but I failed because it requires a lot of
yak-shaving.
Do you have a raw and unshaved patch?
The easier way is to implement the privates stuff as it was in - say
- 1.4.2 and lift all modules to that very level. Then we can upgrade
to 1.5 quite easily (compared to the current approach).
This, however, does sound reasonable indeed. The devPrivates ABI
attempts have been bogus in the course of their different states. So,
I'd like to avoid having broken states as stable release in nx-libs.
I don't think implementing later reworks of devPrivates will be hassle then.
Ok... Please go through X.org's git log and check if the version (1.5)
state you propose does not have follow-up commits complaining about
things having been broken or bogus before. Thanks. (I'll chime in once
when back from VAC).
Mike
--
DAS-NETZWERKTEAM
mike gabriel, herweg 7, 24357 fleckeby
mobile: +49 (1520) 1976 148
landline: +49 (4354) 8390 139
GnuPG Fingerprint: 9BFB AEE8 6C0A A5FF BF22 0782 9AF4 6B30 2577 1B31
mail: [email protected], http://das-netzwerkteam.de
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.