For awhile now, many of the GEM code paths have allocated page or
object arrays with the slab allocator.  This is nice and fast, but
won't work well if memory is fragmented, since the slab allocator works
with physically contiguous memory (i.e. order > 2 allocations are
likely to fail fairly early after booting and doing some work).

This patch works around the issue by falling back to vmalloc for
>PAGE_SIZE allocations.  This is ugly, but much less work than chaining
a bunch of pages together by hand (suprisingly there's not a bunch of
generic kernel helpers for this yet afaik).  vmalloc space is somewhat
precious on 32 bit kernels, but our allocations shouldn't be big enough
to cause problems, though they're routinely more than a page.

Note that this patch doesn't address the unchecked
alloc-based-on-ioctl-args in GEM; that needs to be fixed in a separate
patch.

Also, I've deliberately ignored the DRM's "area" junk.  I don't think
anyone actually uses it anymore and I'm hoping it gets ripped out soon.

[Updated: removed size arg to new free function.  We could unify the
free functions as well once the DRM mem tracking is ripped out.]

fd.o bug #20152 (part 1/3)
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Commit 79e53945 introduced a
regression where you cannot use sysrq 'g' to enter kgdb.  The solution
is to move the intel fb sysrq over to V for video instead of G for
graphics.  The SMP VOYAGER code to register for the sysrq-v is not
anywhere to be found in the mainline kernel, so the comments in the
code were cleaned up as well.

This patch also cleans up the sysrq definitions for kgdb to make it
generic for the kernel debugger, such that the sysrq 'g' can be used
in the future to enter a gdbstub or another kernel debugger.
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
Acked-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Acked-by: NRandy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>

This allows userlevel code to discover the pipe number corresponding
to a given CRTC ID. This is necessary for doing pipe-specific
operations such as waiting for vblank on a given CRTC.  Failure to use
the right pipe mapping can result in GPU hangs, or at least failure
to actually sync to vblank.
Signed-off-by: NCarl Worth <cworth@cworth.org>
[anholt: Style touchups from review]
Signed-off-by: NEric Anholt <eric@anholt.net>

We detect HDMI output connection status by writing to HOT Plug Interrupt
Detect Enable bit in PORT_HOTPLUG_EN.  The behavior will generate a specified
interrupt, which is caught by audio driver, but during one detection driver
set all Detect Enable bits of HDMIB, HDMIC HDMID, and generate wrong
interrupt signals for current output, according to the signals audio driver
misunderstand device status. The patch intends to handle corresponding
output precisely.

It fixed freedesktop.org bug #21371
Signed-off-by: NMa Ling <ling.ma@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

It fixed bug #21659
Signed-off-by: NMa Ling <ling.ma@intel.com>
[anholt: hand-applied because git-am is too picky]
Signed-off-by: NEric Anholt <eric@anholt.net>

Although spec say CRT_HOTPLUG_ACTIVATION_PERIOD_64 is only useful for
mobile platform, it is also required to detect vga on G4x desktops correctly.
Tested on G45/G43/Q45 platforms with no regressions.

It fixed freedesktop.org bug #21120 and part of bug #21210
Signed-off-by: NMa Ling <ling.ma@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

There are a number of small form factor desktop systems with Intel mobile
graphics chips that lie and say they have an LVDS. With kernel mode-setting,
this becomes a problem, and makes native resolution boot go haywire -- for
example, my Dell Studio Hybrid, hooked to a 1920x1080 display claims to
have a 1024x768 LVDS, and the resulting graphical boot on the 1920x1080
display uses only the top left 1024x768, and auto-configured X will end
up only 1024x768 as well. With this change, graphical boot and X
both do 1920x1080 as expected.

Note that we're simply embracing and extending the early bail-out code
in place for the Mac Mini here. The xorg intel driver uses pci subsystem
device and vendor id for matching, while we're using dmi lookups here.
The MSI addition is courtesy of and tested by Bill Nottingham.
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Tested-by: NBill Nottingham <notting@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

We might sleep here anyway so I hope an extra uncached read is ok to
add.

In #20896 we found that vbetool clobbers the IER.  In KMS mode this is
particularly bad since we don't set the interrupt regs late (in
EnterVT), so we'd fail to get *any* interrupts at all after X started
(since some distros have scripts that call vbetool at X startup
apparently).

So this patch checks IER at wait_request time, and re-enables
interrupts if it's been clobbered.  In a proper config this check
should never be triggered.

This is really a distro issue, but having a sanity check is nice, as
long as it doesn't have a real performance hit.
Tested-by: NMateusz Kaduk <mateusz.kaduk@gmail.com>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
[anholt: Moved the check inside of the sleeping case to avoid perf cost]
Signed-off-by: NEric Anholt <eric@anholt.net>

In IGD, DPCUNIT_CLOCK_GATE_DISABLE bit should be set, otherwise i2c
access will be wrong.

v2: Disable CLOCK_GATE_DISABLE bit after bit bashing as suggested by Eric.
Signed-off-by: NShaohua Li <shaohua.li@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

This should avoid a class of bugs where the hardware prefetches past the
end of the object, and walks into unallocated memory when the object is
bound to the last page of the aperture.

fd.o bug #21488
Signed-off-by: NEric Anholt <eric@anholt.net>

This makes software fallbacks not do tiling wrong on i965 and later after
resume. It also should fix 945 performance reduction after resume which
would have disabled tiling without causing any visible effect.
Signed-off-by: NKeith Packard <keithp@keithp.com>
[anholt: Fixed up the 915 case to not save/restore the new regs]
Signed-off-by: NEric Anholt <eric@anholt.net>

i386 has inline code for writeq and readq, so just use those instead of ugly
macros which evaluate arguments multiple times.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

The ACPI video driver defers registration to the i915 driver if the
system supports opregion-mediated backlight control. This registration
was only being performed in the KMS case. Ensure it's done even if we
don't have modesetting enabled.

http://bugzilla.kernel.org/show_bug.cgi?id=13048Signed-off-by: NMatthew Garrett <mjg@redhat.com>
Signed-off-by: NLen Brown <len.brown@intel.com>

Commit 201361a5 introduces a leak when unwinding on error.  Reorder
unwind, and eliminate leak.

Cc: Eric Anholt <eric@anholt.net>
Cc: Keith Packard <keithp@keithp.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NChris Wright <chrisw@sous-sol.org>
[anholt: fixed uninit variable use introduced in original patch]
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NWu Fengguang <fengguang.wu@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

This had been delayed for some time due to failure to work on the one piece
of G41 hardware we had, and lack of success reports from anybody else.
Current hardware appears to be OK.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
[anholt: hand-applied due to conflicts with IGD patches]
Signed-off-by: NEric Anholt <eric@anholt.net>

The changes to opregion initialisation order meant that the ASLE setup
code might not be run at the correct time. Ensure that the interrupts are
set up.
Signed-off-by: NMatthew Garrett <mjg@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

The i915 DRM triggers registration of the ACPI video driver on load. It
should unregister it at unload in order to avoid generating backtraces on
being reloaded.
Signed-off-by: NMatthew Garrett <mjg@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

The ACPI video driver defers registration to the i915 driver if the
system supports opregion-mediated backlight control. This registration
was only being performed in the KMS case. Ensure it's done even if we
don't have modesetting enabled.
Signed-off-by: NMatthew Garrett <mjg@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Transitions to TILING_NONE skipped the call to unbind the object, which left
the fence register set and caused future CPU access through the GTT to
access the object in tiled mode.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Userland is broken if it's trying this, but we also shouldn't allow oopses.
Signed-off-by: NEric Anholt <eric@anholt.net>

This patch corrects a pretty big oversight in the KMS code for 965+
chips.  The current code is missing tiled surface register programming,
so userland can allocate a tiled surface and use it for mode setting,
resulting in corruption.  This patch fixes that, allowing for tiled
front buffers on 965+.

Cc: stable@kernel.org
Tested-by: NArkadiusz Miskiewicz <arekm@maven.pl>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

regression caused by commit 5e118f41:
i915_gem_object_move_to_inactive() should be called in task context,
as it calls fput();

Signed-off-by: Shaohua Li<shaohua.li@intel.com>
[anholt: Add more detail to the comment about the lock break that's added]
Signed-off-by: NEric Anholt <eric@anholt.net>

Save the bit 17 state of the pages when freeing the page list, and
reswizzle them if necessary when rebinding the pages (in case they were
swapped out).  Since we have userland with expectations that the swizzle
enums let it pread and pwrite contents accurately, we can't expose a new
swizzle enum for bit 17 (which it would have to GTT map to handle), so we
handle it down in pread and pwrite by swizzling the copy when bit 17 of the
page address is set.
Signed-off-by: NEric Anholt <eric@anholt.net>

Otherwise, the results of our read didn't show up when we were faulting in
the page being read into (as happened with a testcase reading into a big
stack area).  Likely accounts for some conformance test failures.
Signed-off-by: NEric Anholt <eric@anholt.net>

i915_gem_put_relocs_to_user returned an uninitialized value which
got returned to userspace. This caused libdrm in my setup to never
get out of a do{}while() loop retrying i915_gem_execbuffer.

result was hanging X, overheating of cpu and 2-3gb of logfile-spam.

This patch adresses the issue by
 1. initializing vars in this file where necessary
 2. correcting wrongly interpreted return values of copy_[from/to]_user
Signed-off-by: NFlorian Mickler <florian@mickler.org>
[anholt: cleanups of unnecessary changes, consistency in APIs]
Signed-off-by: NEric Anholt <eric@anholt.net>

At the same time, bring the action message closer to the usual format.
Signed-off-by: NFerenc Wagner <wferi@niif.hu>
Acked-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Currently we detect HDMI monitor by hardware detection, but if an HDMI-DVI
adapter is used to connect a DVI monitor, hardware detection will incorrectly
take monitor as HDMI. HDMI spec says any device containing IEEE registration
identifier will be treated as HDMI device.  The patch intends to detect HDMI
monitor by drm_detect_hdmi_monitor function which follows that rule.
Signed-off-by: NMa Ling <ling.ma@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NShaohua Li <shaohua.li@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

We create a debugfs node (i915_ringbuffer_data) to expose a hex dump
of the ring buffer itself.  We also expose another debugfs node
(i915_ringbuffer_info) with information on the state (i.e. head, tail
addresses) of the ringbuffer.

For batchbuffer dumping, we look at the device's active_list, dumping
each object which has I915_GEM_DOMAIN_COMMAND in its read
domains. This is all exposed through the dri/i915_batchbuffers debugfs
file with a header for each object (giving the objects gtt_offset so
that it can be matched against the offset given in the
BATCH_BUFFER_START command.
Signed-off-by: NBen Gamari <bgamari@gmail.com>
Signed-off-by: NCarl Worth <cworth@cworth.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Cleanup some leftovers from the X port.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This is a baby-step in the direction of having finer-grained
locking than the struct_mutex. Specifically, this will enable
new debugging code to read the active list for printing out
GPU state when the GPU is wedged, (while the struct_mutex is
held, of course).
Signed-off-by: NCarl Worth <cworth@cworth.org>
[anholt: indentation fix]
Signed-off-by: NEric Anholt <eric@anholt.net>

This brings SDVO TV support from 2D driver, including origin
fix f1ca56e17d0 and later fix 2fcf4fcccfe. Also fix wrong modeline
definitions for SDVO TV.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

This brings fix commit acde0ef683 from 2D driver.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

This brings fix commit 62c0c2f5549a from 2D driver.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Fix compile error of intel_sdvo_debug_response(),
and explicit use KERN_DEBUG for printk.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Only set TV DAC in property change seems doesn't work, we have to
setup whole crtc pipe which assigned to TV alone.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
[anholt: Note that this should also fix the oops at startup with new 2D]
Signed-off-by: NEric Anholt <eric@anholt.net>

If there's no real property change, don't need to set TV mode again.
Signed-off-by: NZhenyu Wang <zhenyu.z.wang@intel.com>
[anholt: checkpatch.pl fix]
Signed-off-by: NEric Anholt <eric@anholt.net>

udelay() of 20 milliseconds really ought to just use mdelay(), that avoids
the various wrap scenarios and also is more readable
Signed-off-by: NArjan van de Ven <arjan@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Add VGA port hotplug detection to the i915 driver.  When KMS is enabled,
plugging in or removing a VGA cable from the VGA connector will
generate a uevent, which indicates to userspace that it should re-probe
outputs on this device (to determine modes, etc.).
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
[anholt: dropped extra PORT_HOTPLUG_STAT clear with ack from jbarnes]
Signed-off-by: NEric Anholt <eric@anholt.net>