use O_NOFOLLOW to open the existing lock file, so symbolic links
aren't followed, thus avoid revealing if it point to an existing file.
Note that xserver on OpenBSD isn't affected by CVE-2011-4029.
is just noise and now happens while the X autoconfiguration
code probes all /dev/wsmouse<n> devices.
If the error matters, the driver will emit a proper error
nevertheless.
hotplug framework.This also takes care of the configuration of the
keyboard layout from the wscons layout that was hacked inside the
keyboard driver.
Tested by many, ok oga@.
differ. Cache rep.length so that it can be used after returning from
WriteReplyToClient().
Fixes xinput --list when client and server have different endianess.
gtk3-demo now fails a bit further in this case.
commit e06fa804009798ea95efa8babaabb0228dfdfe65
EXA: Fix crash with fill using 1x1 tile of depth < 8 (bug #24703).
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=24703
Reported by Barry Grumbine on ports@
ok matthieu@
This hasn't been used for a very, very, very long time, (since before
OpenBSD had dri support, for example) and it causes segfaults on dri
drivers when sigio is disabled. Now we don't need to do context swaps on sigio
nor are we trying to do interrupts in userland (thank fuck for that) this
function can die the death that I intended it to die about two years ago, may it
burn.
The kernel support the the sigio ioctl will be removed in a couple of
weeks to give people time to update (right now it accepts it, then
ignores it).
ok kettenis@, matthieu@.
attempt to open that first before trying /dev/ttyC[0-7].
This makes X autoconfiguration a tad bit more intuitive on machines
with multiple SBus or UPA framebuffers, where wsdisplay0 isn't the
console. PCI framebuffers are still busted though.
ok matthieu@
The code to add sunffb unconditionally on !solaris for __sparc__
systems is incorrect for openbsd. More specifically, due to interactions
between hardware drivers and wsfb in preinit we can't unconditionally
add wsfb to the list of fallbacks, so we only add wsfb if no other
option was found. Additionally sunffb does not need to be
unconditionally added because the bus probing code will find these
devices already.
So, long story short: make that code chunk conditional on __sparc__ &&
defined(__linux__) instead.
change from !openbsd to __linux__ requested by kettenis@.
Tested by at least myself and stsp@.
ok matthieu@, kettenis@.
