this patch separates fimd specific power on/off function from pm function
and the pm interfaces will call that function for power on or off.
and also removes unnecessary codes of resume function.
Signed-off-by: NInki Dae <inki.dae@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>

I'd like to add my colleagues who dedicated to developing and
improving our driver to maintainer entry.
Signed-off-by: NInki Dae <inki.dae@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>

FB based FIMD and DRM based FIMD drivers use same hardware
so with this patch, only one of them would be selected.
Signed-off-by: NInki Dae <inki.dae@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>

DRM_EXYNOS_HDMI driver and VIDEO_SAMSUNG_S5P_TV driver should be
not enabled at once because they use same HW blocks. So dependency
for DRM_EXYNOS_HDMI is fixed to check VIDEO_SAMSUNG_S5P_TV=n.
Signed-off-by: NSeung-Woo Kim <sw0312.kim@samsung.com>
Signed-off-by: NInki Dae <inki.dae@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>

To make a api pair of request_mem_region and release_mem_region,
release_mem_region is used instead of release_resource.
Signed-off-by: NSeung-Woo Kim <sw0312.kim@samsung.com>
Signed-off-by: NInki Dae <inki.dae@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>

In error cleanup of agp_backend_initialize() and in agp_backend_cleanup(),
agp_destroy_page() is passed virtual address of the scratch page. This
leads to a kernel warning if the initialization fails (or upon regular
cleanup) as pointer to struct page should be passed instead.
Signed-off-by: NMichal Kubecek <mkubecek@suse.cz>
Signed-off-by: NDave Airlie <airlied@redhat.com>

* 'drm-intel-fixes' of git://people.freedesktop.org/~keithp/linux: (24 commits)
  drm/i915: fixup forcewake spinlock fallout in drpc debugfs function
  drm/i915: debugfs: show semaphore registers also on gen7
  drm/i915: allow userspace forcewake references also on gen7
  drm/i915: Re-enable gen7 RC6 and GPU turbo after resume.
  drm/i915: Correct debugfs printout for RC1e.
  Revert "drm/i915: Work around gen7 BLT ring synchronization issues."
  drm/i915: rip out the HWSTAM missed irq workaround
  drm/i915: paper over missed irq issues with force wake voodoo
  drm/i915: Hold gt_lock across forcewake register reads
  drm/i915: Hold gt_lock during reset
  drm/i915: Move reset forcewake processing to gen6_do_reset
  drm/i915: protect force_wake_(get|put) with the gt_lock
  drm/i915: convert force_wake_get to func pointer in the gpu reset code
  drm/i915: sprite init failure on pre-SNB is not a failure
  drm/i915: VBT Parser cleanup for eDP block
  drm/i915: mask transcoder select bits before setting them on LVDS
  drm/i915: Add Clientron E830 to the ignore LVDS list
  CHROMIUM: i915: Add DMI override to skip CRT initialization on ZGB
  drm/i915: handle 3rd pipe
  drm/i915: simplify pipe checking
  ...

Both changes in dc97b340 cause serious
regressions in the nouveau driver.

move_notify() was originally able to presume that bo->mem is the old node,
and new_mem is the new node.  The above commit moves the call to
move_notify() to after move() has been done, which means that now, sometimes,
new_mem isn't the new node at all, bo->mem is, and new_mem points at a
stale, possibly-just-been-killed-by-move node.

This is clearly not a good situation.  This patch reverts this change, and
replaces it with a cleanup in the move() failure path instead.

The second issue is that the call to move_notify() from cleanup_memtype_use()
causes the TTM ghost objects to get passed into the driver.  This is clearly
bad as the driver knows nothing about these "fake" TTM BOs, and ends up
accessing uninitialised memory.

I worked around this in nouveau's move_notify() hook by ensuring the BO
destructor was nouveau's.  I don't particularly like this solution, and
would rather TTM never pass the driver these objects.  However, I don't
clearly understand the reason why we're calling move_notify() here anyway
and am happy to work around the problem in nouveau instead of breaking the
behaviour expected by other drivers.
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>
Reviewed-by: NThomas Hellstrom <thellstrom@vmware.com>
Cc: Jerome Glisse <j.glisse@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

My forcewake spinlock patches have a functional conflict with Ben
Widawsky's gen6 drpc support for debugfs. Result was a benign warning
about trying to read an non-atomic variabla with atomic_read.

Note that the entire check is racy anyway and purely informational.
Also update it to reflect the forcewake voodoo changes, the kernel can
now also hold onto a forcewake reference for longer times.
Signed-Off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NBen Widawsky <ben@bwidawsk.net>
Signed-off-by: NKeith Packard <keithp@keithp.com>

If GPU lockup is detected in ib_pool get we are holding the ib_pool
mutex that will be needed by the GPU reset code. As ib_pool code is
safe to be reentrant from GPU reset code we should not block if we
are trying to get the ib pool lock on the behalf of the same userspace
caller, thus use the radeon_mutex_lock helper.
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Silence out the lock dependency warning by moving bo allocation out
of ib mutex protected section. Might lead to useless temporary
allocation but it's not harmful as such things only happen at
initialization.
Signed-off-by: NJerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

If the master tries to authenticate a client using drm_authmagic and
that client has already closed its drm file descriptor,
either wilfully or because it was terminated, the
call to drm_authmagic will dereference a stale pointer into kmalloc'ed memory
and corrupt it.

Typically this results in a hard system hang.

This patch fixes that problem by removing any authentication tokens
(struct drm_magic_entry) open for a file descriptor when that file
descriptor is closed.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Cc: stable@vger.kernel.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

Corresponding changes to improve our error_state are pending
some other patches to clean up things first.
Signed-Off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NKeith Packard <keithp@keithp.com>

We need this to correctly access registers in the gt power well from
userspace.
Signed-Off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: NKeith Packard <keithp@keithp.com>
Signed-off-by: NKeith Packard <keithp@keithp.com>

Signed-off-by: NEric Anholt <eric@anholt.net>
Cc: stable@vger.kernel.org
Reviewed-by: NKeith Packard <keithp@keithp.com>
Reviewed-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: NKenneth Graunke <kenneth@whitecape.org>
Signed-off-by: NKeith Packard <keithp@keithp.com>

We had two things in a row claiming to be RC6.
Signed-off-by: NEric Anholt <eric@anholt.net>
Reviewed-by: NKeith Packard <keithp@keithp.com>
Reviewed-by: NBen Widawsky <ben@bwidawsk.net>
Reviewed-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: NKenneth Graunke <kenneth@whitecape.org>
Signed-off-by: NKeith Packard <keithp@keithp.com>

GMA500 did it the old way and it's been on the TODO list to fix. Current kernels
now blow up if we use the old way so we'd better do the work !
Signed-off-by: NAlan Cox <alan@linux.intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

HDMI 1.3 defines single link clocks up to 340 Mhz.
Refine the current dual link checks to only enable
dual link for DVI > 165 Mhz or HDMI > 340 Mhz if the
hw supports HDMI 1.3 (DCE3+).

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=44755Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

We were previously just checking for audio.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

dig transmitter control table only has ENABLE/DISABLE actions
on DCE4.1/DCE5.

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=44955Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

Needs to happen earlier in the mode set.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

We really only need to set it up once on init or resume
rather than on every mode set.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Return a number of bytes read in radeon_atrm_get_bios_chunk() and
properly check this value in radeon_atrm_get_bios().
If radeon_atrm_get_bios_chunk() read less bytes then were requested,
it means that it finished reading bios data.

Prior to this patch, condition in radeon_atrm_get_bios() was always
equivalent to "if (ATRM_BIOS_PAGE <= 0)", so it was always false,
thus radeon_atrm_get_bios() was trying to read past the bios data
wasting boot time.

On my lenovo ideapad u455 laptop this patch drops bios reading time
from ~5.5s to ~1.5s.
Signed-off-by: NIgor Murzov <e-mail@date.by>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

At a boot time I observed following bug:

 BUG: unable to handle kernel paging request at ffff8800a4244000
 IP: [<ffffffff81275b5b>] memcpy+0xb/0x120
 PGD 1816063 PUD 1fe7d067 PMD 1ff9f067 PTE 80000000a4244160
 Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
 CPU 0
 Modules linked in: btusb bluetooth brcmsmac brcmutil crc8 cordic b43 radeon(+)
  mac80211 cfg80211 ttm ohci_hcd drm_kms_helper rfkill drm ssb agpgart mmc_core
  sp5100_tco video battery ac thermal processor rtc_cmos thermal_sys snd_hda_codec_hdmi
  joydev snd_hda_codec_conexant button bcma pcmcia snd_hda_intel snd_hda_codec
  snd_hwdep snd_pcm shpchp pcmcia_core k8temp snd_timer atl1c snd psmouse hwmon
  i2c_piix4 i2c_algo_bit soundcore evdev i2c_core ehci_hcd sg serio_raw snd_page_alloc
  loop btrfs

 Pid: 1008, comm: modprobe Not tainted 3.3.0-rc1 #21 LENOVO 20046                           /AMD CRB
 RIP: 0010:[<ffffffff81275b5b>]  [<ffffffff81275b5b>] memcpy+0xb/0x120
 RSP: 0018:ffff8800aa72db00  EFLAGS: 00010246
 RAX: ffff8800a4150000 RBX: 0000000000001000 RCX: 0000000000000087
 RDX: 0000000000000000 RSI: ffff8800a4244000 RDI: ffff8800a4150bc8
 RBP: ffff8800aa72db78 R08: 0000000000000010 R09: ffffffff8174bbec
 R10: ffffffff812ee010 R11: 0000000000000001 R12: 0000000000001000
 R13: 0000000000010000 R14: ffff8800a4140000 R15: ffff8800aaba1800
 FS:  00007ff9a3bd4720(0000) GS:ffff8800afa00000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
 CR2: ffff8800a4244000 CR3: 00000000a9c18000 CR4: 00000000000006f0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
 Process modprobe (pid: 1008, threadinfo ffff8800aa72c000, task ffff8800aa0e4000)
 Stack:
  ffffffffa04e7c7b 0000000000000001 0000000000010000 ffff8800aa72db28
  ffffffff00000001 0000000000001000 ffffffff8113cbef 0000000000000020
  ffff8800a4243420 ffff880000000002 ffff8800aa72db08 ffff8800a9d42000
 Call Trace:
  [<ffffffffa04e7c7b>] ? radeon_atrm_get_bios_chunk+0x8b/0xd0 [radeon]
  [<ffffffff8113cbef>] ? kmalloc_order_trace+0x3f/0xb0
  [<ffffffffa04a9298>] radeon_get_bios+0x68/0x2f0 [radeon]
  [<ffffffffa04c7a30>] rv770_init+0x40/0x280 [radeon]
  [<ffffffffa047d740>] radeon_device_init+0x560/0x600 [radeon]
  [<ffffffffa047ef4f>] radeon_driver_load_kms+0xaf/0x170 [radeon]
  [<ffffffffa043cdde>] drm_get_pci_dev+0x18e/0x2c0 [drm]
  [<ffffffffa04e7e95>] radeon_pci_probe+0xad/0xb5 [radeon]
  [<ffffffff81296c5f>] local_pci_probe+0x5f/0xd0
  [<ffffffff81297418>] pci_device_probe+0x88/0xb0
  [<ffffffff813417aa>] ? driver_sysfs_add+0x7a/0xb0
  [<ffffffff813418d8>] really_probe+0x68/0x180
  [<ffffffff81341be5>] driver_probe_device+0x45/0x70
  [<ffffffff81341cb3>] __driver_attach+0xa3/0xb0
  [<ffffffff81341c10>] ? driver_probe_device+0x70/0x70
  [<ffffffff813400ce>] bus_for_each_dev+0x5e/0x90
  [<ffffffff8134172e>] driver_attach+0x1e/0x20
  [<ffffffff81341298>] bus_add_driver+0xc8/0x280
  [<ffffffff813422c6>] driver_register+0x76/0x140
  [<ffffffff812976d6>] __pci_register_driver+0x66/0xe0
  [<ffffffffa043d021>] drm_pci_init+0x111/0x120 [drm]
  [<ffffffff8133c67a>] ? vga_switcheroo_register_handler+0x3a/0x60
  [<ffffffffa0229000>] ? 0xffffffffa0228fff
  [<ffffffffa02290ec>] radeon_init+0xec/0xee [radeon]
  [<ffffffff810002f2>] do_one_initcall+0x42/0x180
  [<ffffffff8109d8d2>] sys_init_module+0x92/0x1e0
  [<ffffffff815407a9>] system_call_fastpath+0x16/0x1b
 Code: 58 2a 43 50 88 43 4e 48 83 c4 08 5b c9 c3 66 90 e8 cb fd ff ff eb
  e6 90 90 90 90 90 90 90 90 90 48 89 f8 89 d1 c1 e9 03 83 e2 07 <f3> 48
  a5 89 d1 f3 a4 c3 20 48 83 ea 20 4c 8b 06 4c 8b 4e 08 4c
 RIP  [<ffffffff81275b5b>] memcpy+0xb/0x120
  RSP <ffff8800aa72db00>
 CR2: ffff8800a4244000
 ---[ end trace fcffa1599cf56382 ]---

Call to acpi_evaluate_object() not always returns 4096 bytes chunks,
on my system it can return 2048 bytes chunk, so pass the length of
retrieved chunk to memcpy(), not the length of the recieving buffer.
Signed-off-by: NIgor Murzov <e-mail@date.by>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Interrupts only work with MSIs.
https://bugs.freedesktop.org/show_bug.cgi?id=37679Reported-by: NDmitry Podgorny <pasis.uax@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

In sis_driver_load(), the only use of 'ret' is as the return value
from the function, unfortunately it is never initialized, so the
function just returns garbage when it succeeds.
To fix that, remove the variable and just return 0 directly on success.
Signed-off-by: NJesper Juhl <jj@chaosbits.net>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This reverts commit 42ff6572.

New forcewake voodoo makes this no longer necessary.
Acked-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NKeith Packard <keithp@keithp.com>

With the new ducttape of much finer quality, this seems to be no
longer necessary.

Tested on my ivb and snb machine with the usual suspects of testcases.

(v2 by keithp -- limited change to IVB only for now)
Signed-Off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: NBen Widawsky <ben@bwidawsk.net>
Signed-off-by: NKeith Packard <keithp@keithp.com>

Merge branches 'sched-urgent-for-linus', 'perf-urgent-for-linus' and 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/accounting, proc: Fix /proc/stat interrupts sum

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  tracepoints/module: Fix disabling tracepoints with taint CRAP or OOT
  x86/kprobes: Add arch/x86/tools/insn_sanity to .gitignore
  x86/kprobes: Fix typo transferred from Intel manual

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, syscall: Need __ARCH_WANT_SYS_IPC for 32 bits
  x86, tsc: Fix SMI induced variation in quick_pit_calibrate()
  x86, opcode: ANDN and Group 17 in x86-opcode-map.txt
  x86/kconfig: Move the ZONE_DMA entry under a menu
  x86/UV2: Add accounting for BAU strong nacks
  x86/UV2: Ack BAU interrupt earlier
  x86/UV2: Remove stale no-resources test for UV2 BAU
  x86/UV2: Work around BAU bug
  x86/UV2: Fix BAU destination timeout initialization
  x86/UV2: Fix new UV2 hardware by using native UV2 broadcast mode
  x86: Get rid of dubious one-bit signed bitfield

gpio bug fixes for v3.3

* tag 'gpio-for-linus' of git://git.secretlab.ca/git/linux-2.6:
  gpio: tps65910: Use correct offset for gpio initialization
  gpio/it8761e: Restrict it8761e gpio driver to x86.
  gpio-ml-ioh: cleanup __iomem annotation usage
  gpio-ml-ioh: cleanup NULL pointer checking
  gpio-pch: cleanup __iomem annotation usage
  gpio-pch: cleanup NULL pointer checking

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  qnx4: don't leak ->BitMap on late failure exits
  qnx4: reduce the insane nesting in qnx4_checkroot()
  qnx4: di_fname is an array, for crying out loud...
  vfs: remove printk from set_nlink()
  wake up s_wait_unfrozen when ->freeze_fs fails

In checkin

  303395ac x86: Generate system call tables and unistd_*.h from tables

the feature macros in <asm/unistd.h> were unified between 32 and 64
bits.  Unfortunately 32 bits requires __ARCH_WANT_SYS_IPC and this was
inadvertently dropped.
Reported-by: NDmitry Kasatkin <dmitry.kasatkin@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: NH. Peter Anvin <hpa@zytor.com>
Link: http://lkml.kernel.org/r/CALLzPKbeXN5gdngo8uYYU8mAow=XhrwBFBhKfG811f37BubQOg@mail.gmail.com

Two things seem to do the trick on my ivb machine here:
- prevent the gt from powering down while waiting for seqno
  notification interrupts by grabbing the force_wake in get_irq (and
  dropping it in put_irq again).
- ordering writes from the ring's CS by reading a CS register, ACTHD
  seems to work.

Only the blt&bsd ring on ivb seem to be massively affected by this,
but for paranoia do this dance also on the render ring and on snb
(i.e. all gpus with forcewake).

Tested with Eric's glCopyPixels loop which without this patch scores a
missed irq every few seconds.

This patch needs my forcewake rework to use a spinlock instead of
dev->struct_mutex.

After crawling through docs a lot I've found the following nugget:

Internal doc "SNB GT PM Programming Guide", Section 4.3.1:

"GT does not generate interrupts while in RC6 (by design)"

So it looks like rc6 and irq generation are indeed related.

v2: Improve the comment per Eugeni Dodonov's suggestion.

v3: Add the documentation snipped. Also restrict the w/a to ivb only
for -fixes, as suggested by Keith Packard.

Cc: stable@kernel.org
Cc: Eric Anholt <eric@anholt.net>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Eugeni Dodonov <eugeni.dodonov@intel.com>
Tested-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Reviewed-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Signed-Off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NKeith Packard <keithp@keithp.com>

Along with the previous patch to make the reset operation protected by
the gt_lock as well, this ensures that all register read operations
will occur with the forcewake hardware enabled. As an added bonus,
this makes read operations more efficient by taking the spinlock only
once per read instead of twice.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This ensures that no register reads occur while the forcewake state of
the hardware is indeterminate during the reset operation.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

No reason to have half of the reset split from the other half.
Signed-off-by: NKeith Packard <keithp@keithp.com>
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The problem this patch solves is that the forcewake accounting
necessary for register reads is protected by dev->struct_mutex. But the
hangcheck and error_capture code need to access registers without
grabbing this mutex because we hold it while waiting for the gpu.
So a new lock is required. Because currently the error_state capture
is called from the error irq handler and the hangcheck code runs from
a timer, it needs to be an irqsafe spinlock (note that the registers
used by the irq handler (neglecting the error handling part) only uses
registers that don't need the forcewake dance).

We could tune this down to a normal spinlock when we rework the
error_state capture and hangcheck code to run from a workqueue.  But
we don't have any read in a fastpath that needs forcewake, so I've
decided to not care much about overhead.

This prevents tests/gem_hangcheck_forcewake from i-g-t from killing my
snb on recent kernels - something must have slightly changed the
timings. On previous kernels it only trigger a WARN about the broken
locking.

v2: Drop the previous patch for the register writes.

v3: Improve the commit message per Chris Wilson's suggestions.
Signed-Off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NEugeni Dodonov <eugeni.dodonov@intel.com>
Signed-off-by: NKeith Packard <keithp@keithp.com>