various updates to build against xorg-server 21.1.3.

- avoid a const issue with an expanded __UNCONST().
- cope with renamed member names
- deal with "xf86RamDac.h" going away (now use "xf86Cursor.h")

Avoid multiple definitions of the same variable.

Use UXA (not SNA) by default on Intel Gen3/4/5 chipsets (except 915).

This works around PR/54995 (screen corruption on N455)
and PR/55198 (X freezes/hangs on 965GM). Ok'ed by mrg@.

Should be pulled up to netbsd-9.

fix merge botches.

re-add files autoconf files that the build sort of needs at
least configure for. command used:

% cd /usr/xsrc/external/mit/xf86-video-intel/dist
% autoreconf -I m4 -I /usr/xsrc/external/mit/xorg-server/dist -i

tested on sandy bridge and kaby lake.

merge xf86-video-intel 6afed33b2d673d88674f0c76efe500ae414e8e1b.

initial import of xf86-video-intel git version 6afed33b2d673d88674f0c76efe500ae414e8e1b

merge changes from xorg-server 1.20 list.h into this older version.

fixes sandy bridge crashing problems for me.

make the intel driver work again with xorg-server 1.20:

- BlockHandler() timeout value changed from struct timeval ** to int **
that has a millisecond value.

- avoid a NULL ptr deref in ProcXF86BigfontQueryFont() and make some
code actually able to be run.

avoid using an uninitialised variable.

fix various errors and warnings with xorg-server 1.20:

- no Error() in scope
- more wakeup/blockhandler ABI changes
- more notify fd API changes
- more xf86CursorResetCursor changes
- lots of char* -> const char*
- fix ati kms config.h now HAVE_XF86_CURSOR_RESET_CURSOR is available
- another xf86GetOS() is removed fix
- OsBlockSIGIO() etc, -> old, but still around API until properly converted
- don't free const char * data owned by some other system

mostly fix amd64 xorg-server 1.20 builds:

- avoid bswap_16 etc defines for now. ends up with code-in-decl
- more ABI_VERSION 23 fixes (BlockHandler/WakeHandler) including re-fix
from previous
- more Notify API
- more shadowUpdatePackedWeak() removal
- remove fd_set handling from intel driver (part of ABI_VERSION 23)
- cope better with PixMap/Drawable separation in new server
- remove support for maxHValue and maxVValue (XXX?)
- hide drmmode_udev_notify() when not building udev
- comment out nouveau 'wfb' support, xorg-server 1.20 only supplies
either 'wfb' or 'fb' interface, not both, and we choose 'fb'
- define 'pointer' in nv_proto.h rather than relying upon Xdefs.h
- normalise the value of _XTYPEDEF_POINTER
- EGL is broken (missing) in xsrc currently

various updates for xorg-server 1.20. some of these are work-in-progress,
the xsrc build is broken for most ports currently.

- xf86DisableRandR() is gone
- use HAVE_NOTIFY_FD like other modern drivers in compat-api.h
- pRotatedPixmap is done
- some PixmanPtr is now DrawablePtr (this one is all over problem with many
consumers, and perhaps the major breakage point currently.)

set mmap_arg.flags before calling DRM_IOCTL_I915_GEM_MMAP.

s/8/0x8/ and s/9/0x9/ to catch up with libdrm update.

cope with HAS_DIRTYTRACKING_ROTATION issues

Remove differences with upstream xf86-video-intel which are no longer
necessary. (Use do_ioctl and mmap rather than drmIoct and drmMap.)

http://mail-index.netbsd.org/tech-x11/2015/01/25/msg001486.html

apply upstream revision e143ce600af286ed9ad9d47afc22dbfbb66628dc to fix
building with x servers as ancient as ours

merge conflicts

initial import of xf86-video-intel-2.99.917

convert to drmMap. say hello to a visible cursor!

merge conflicts

Import the latest from x.org; no improvement over the current screen corruption.

Revert stop-gap in 1.2 now that Mesa is updated.

Disable DRI for Ivy Bridge (and later) until we update Mesa.

This is a stop-gap for PR 48916.

Use drmMap rather than mmap directly, to invoke the ioctl on NetBSD.

merge xf86-video-intel-2.21.15

initial import of xf86-video-intel-2.21.15

CID 1107540: Make the code safe avoiding random stack access. In the first
loop where there is a singleton point to pptSrc, only access that singleton,
no matter what.

CID 1107544: Fix memory leak

Fix memset size argument, found by coverity.

Make xf86-video-intel driver work on Atom N455 + GMA3150 graphics chipset
by pulling the following change from upstream:
http://cgit.freedesktop.org/xorg/driver/xf86-video-intel/commit/src/i830.h?id=8f64837e56b2de0fb8a9100d1a844fd3f18d751c
> Disable FBC on IGD for UMS
> It appears the new chip doesn't support FBC currently.

Tested on MSI U135DX + NetBSD/i386 6.0_BETA2 with this patch.

Should be pulled up to netbsd-6.

Pull the similar fix from i915_render.c rev 1.2 and i965_render.c rev 1.2:

http://mail-index.NetBSD.org/source-changes/2010/12/27/msg016345.html
> XXX: Fix crash with Firefox, where drawable is NULL.
> There must be a better fix for this.

http://mail-index.NetBSD.org/source-changes/2011/01/21/msg017560.html
> Avoid a crash when pDrawable is null.

Fixes Xserver SEGV on starting Firefox on old i830/855GM machines
running 6.0_BETA2.

Should be pulled up to netbsd-6.

Use sysctl machdep.dmi.* instead of /sys/class/dmi/id/* for extracting DMI info on NetBSD

Fix memset usage.

i915 drivers needs the aperture prototype.

In case somebody wants to try to get UXA working, make it compile against
newer X servers

stop screaming: s/DEBUG/debug/g

add an ifndef __linux__ requested by mrg

dont negate error value passed to strerror
the linux libpciaccess returns negated value, but no other os does...

Avoid a crash when pDrawable is null.

workaround a lint bug

Revert previous hack and fix this properly. We need dpms to be turned on
to be able to load the colormap, since dpms off turns off the channel.
Guard against the lut being loaded with dpms off, and move the call to
load the lut after the dmps on one.

Disable loading the lut that crashes the system for now. Unfortunately this
causes butt ugly X, but at least we have X again. Wee!

XXX: Fix crash with Firefox, where drawable is NULL.
There must be a better fix for this.

apply the hack to avoid xf86LoaderReqSymLists() and LoaderRefSymLists().

should fix PR#44167.

fix the previous - use GET_ABI_MAJOR(ABI_VIDEODRV_VERSION) > 6.

copy a hack from the vmware driver, and avoid using xf86LoaderReqSymLists()
and LoaderRefSymLists() in xorg-server >= 1.7.

#if 0 the code that tries to:

- call xf86RegisterResources()
- use ResUnusedOpr, ResDisableOpr or ResNone
- use racMemFlags member of ScrnInfoPtr{}
- set/use CreateBuffers/DestroyBuffers members of DRI2InfoRec{}
- one of the two calls to i830_fill_colorkey().


for the other call to i830_fill_colorkey(), update it based upon code
seen in xf86-video-intel 2.11.0, and use xf86XVFillKeyHelperDrawable().
i can't see how to get a DrawablePtr for the other call, though.


XXX: i'm entirely unsure this will work with xorg-server 1.9.2. there's
XXX: a decent chance that everything but i830-based will work, and that
XXX: even i830 will work ok, but it really needs to be tested.

remove includes for xf86RAC.h and xf86Resources.h, they don't exist anymore.

initial import of xf86-video-intel-2.7.1

initial import of xf86-video-intel-2.4.0