It has way too much potential for driver writers to do stupid things
like delayed hw setup because the load sequence is somehow racy (e.g.
the imx driver in staging). So don't call it for modesetting drivers,
which reduces the complexity of the drm core -> driver interface a
notch.

v2: Don't forget to update DocBook.

v3: Go with Laurent's slightly more elaborate proposal for the DocBook
update. Add a few words on top of his diff to elaborate a bit on what
KMS drivers should and shouldn't do in lastclose. There was already a
paragraph present talking about restoring properties, I've simply
extended that one.

Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

So if we survey kms drivers there's a bunch of things they commonly do
in ->lastclose
- delayed processing of vga switcheroo requests (i915, nouveau,
  radeon)
- force-restoring the fbcon (most)
- resetting a bunch properties to make fbcon work better (omap)
- disabling all outputs (vmwgfx)

In short besides the semantically important vga switcheroo stuff they
all try very hard to keep fbcon working in case X dies.

But none of them try to not do this at driver unload time safe for
vmwgfx, and digging through logs I couldn't find any reason for why
vmwgfx is special.

Since ->firstopen has lots of potential for abuse with kms drivers
(like delaying driver setup to pamper over races in the load sequence)
it's imo very much worth it to remove this logic so that we can
stop using the ->firstopen callback for kms drivers.

Also module unloading is rather a debug feature and developers should
know how to restore the display to a sane configuration.

Cc: Jakob Bornecrantz <jakob@vmware.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This thing seems to do some kind of delayed setup. Really, real kms
drivers shouldn't do that at all. Either stuff needs to be dynamically
hotplugged or the driver setup sequence needs to be fixed.

This patch here just moves the setup at the very end of the driver
load callback, with the locking adjusted accordingly.

v2: Also move the corresponding put from ->lastclose to ->unload.

Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Currently, both ranges overlap. Fix the limits so both ranges are mutually
exclusive. Also use the occasion to convert whitespaces to tabs.
Signed-off-by: NKristian Høgsberg <krh@bitplanet.net>
(fixed up tabs and adjust commit-msg accordingly)
Signed-off-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

The idr is protected with our spinlock, if we don't hold that nothing
prevents the gem objects from disappearing from under us.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

We might as well have a real ioctl function which checks for the
callbacks. This seems to be a remnant from back in the days when each
drm driver had their own complete ioctl table, with no shared core
drm table at all.

To make really sure no mis-guided user in a kms driver pops up again
explicitly check for that in the new ioctl implementation.

v2: Drop the unused variable I've accidentally left in the code,
spotted by David Herrmann.

Cc: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

I've forgotten this and shuffling all the little pieces into the
respective patches is rather cumbersome ...
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

The new arch_phys_wc_add/del functions do the right thing both with
and without MTRR support in the kernel. So we can drop these
additional checks.

David Herrmann suggest to also kill the DRIVER_USE_MTRR flag since
it's now unused, which spurred me to do a bit a better audit of the
affected drivers. David helped a lot in that. Quoting our mail
discussion:

On Wed, Jul 10, 2013 at 5:41 PM, David Herrmann <dh.herrmann@gmail.com> wrote:
> On Wed, Jul 10, 2013 at 5:22 PM, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
>> On Wed, Jul 10, 2013 at 3:51 PM, David Herrmann <dh.herrmann@gmail.com> wrote:
>>>> -#if __OS_HAS_MTRR
>>>> -static inline int drm_core_has_MTRR(struct drm_device *dev)
>>>> -{
>>>> -       return drm_core_check_feature(dev, DRIVER_USE_MTRR);
>>>> -}
>>>> -#else
>>>> -#define drm_core_has_MTRR(dev) (0)
>>>> -#endif
>>>> -
>>>
>>> That was the last user of DRIVER_USE_MTRR (apart from drivers setting
>>> it in .driver_features). Any reason to keep it around?
>>
>> Yeah, I guess we could rip things out. Which will also force me to
>> properly audit drivers for the eventual behaviour change this could
>> entail (in case there's an x86 driver which did not ask for an mtrr,
>> but iirc there isn't).
>
> david@david-mb ~/dev/kernel/linux $ for i in drivers/gpu/drm/* ; do if
> test -d "$i" ; then if ! grep -q USE_MTRR -r $i ; then echo $i ; fi ;
> fi ; done
> drivers/gpu/drm/exynos
> drivers/gpu/drm/gma500
> drivers/gpu/drm/i2c
> drivers/gpu/drm/nouveau
> drivers/gpu/drm/omapdrm
> drivers/gpu/drm/qxl
> drivers/gpu/drm/rcar-du
> drivers/gpu/drm/shmobile
> drivers/gpu/drm/tilcdc
> drivers/gpu/drm/ttm
> drivers/gpu/drm/udl
> drivers/gpu/drm/vmwgfx
> david@david-mb ~/dev/kernel/linux $
>
> So for x86 gma500,nouveau,qxl,udl,vmwgfx don't set DRIVER_USE_MTRR.
> But I cannot tell whether they break if we call arch_phys_wc_add/del,
> anyway. At least nouveau seemed to work here, but it doesn't use AGP
> or drm_bufs, I guess.

Cool, thanks a lot for stitching together the list of drivers to look
at. So for real KMS drivers it's the drives responsibility to add an
mtrr if it needs one. nouvea, radeon, mgag200, i915 and vmwgfx do that
already. Somehow the savage driver also ends up doing that, I have no
idea why.

Note that gma500 as a pure KMS driver doesn't need MTRR setup since
the platforms that it supports all support PAT. So no MTRRs needed to
get wc iomappings.

The mtrr support in the drm core is all for legacy mappings of garts,
framebuffers and registers. All legacy drivers set the USE_MTRR flag,
so we're good there.

All in all I think we can really just ditch this

/endquote

v2: Also kill DRIVER_USE_MTRR as suggested by David Herrmann

v3: Rebase on top of David Herrmann's agp setup/cleanup changes.

Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: NAndy Lutomirski <luto@amacapital.net>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Trying to drop a reference we don't have is a pretty serious bug.
Trying to paper over it is an even worse offense.

So scream into dmesg with a big WARN in case that ever happens.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Calling this function with a NULL object is simply a bug, so papering
over a NULL object not a good idea.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

We have three callers of this function now and it's neither
performance critical nor really small. So an inline function feels
like overkill and unecessarily separates the different parts of the
code.

Since all callers of drm_gem_object_handle_free are now in drm_gem.c
we can make that static (and remove the unused EXPORT_SYMBOL). To
avoid a forward declaration move it (and drm_gem_object_free_bug) up a
bit.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Lifetime rules seem to be solid around ->import_attach. So this patch
just properly documents them.

Note that pointing directly at the attachment might have issues for
devices that have multiple struct device *dev parts constituting the
logical gpu and so might need multiple attachment points. Similarly
for drm devices which don't need a dma attachment at all (like udl).

But fixing that up is material for different patches.
Reviewed-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

I've checked both implementations (radeon/nouveau) and they both grab
the page array from ttm simply by dereferencing it and then wrapping
it up with drm_prime_pages_to_sg in the callback and map it with
dma_map_sg (in the helper).

Only the grabbing of the underlying page array is anything we need to
be concerned about, and either those pages are pinned independently,
or we're screwed no matter what.

And indeed, nouveau/radeon pin the backing storage in their
attach/detach functions.

Since I've created this patch cma prime support for dma_buf was added.
drm_gem_cma_prime_get_sg_table only calls kzalloc and the creates&maps
the sg table with dma_get_sgtable. It doesn't touch any gem object
state otherwise. So the cma helpers also look safe.

The only thing we might claim it does is prevent concurrent mapping of
dma_buf attachments. But a) that's not allowed and b) the current code
is racy already since it checks whether the sg mapping exists _before_
grabbing the lock.

So the dev->struct_mutex locking here does absolutely nothing useful,
but only distracts. Remove it.

This should also help Maarten's work to eventually pin the backing
storage more dynamically by preventing locking inversions around
dev->struct_mutex.

v2: Add analysis for recently added cma helper prime code.

Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NMaarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NInki Dae <inki.dae@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Note that this is slightly tricky since both drivers store their
native objects in dma_buf->priv. But both also embed the base
drm_gem_object at the first position, so the implicit cast is ok.

To use the release helper we need to export it, too.

Cc: Inki Dae <inki.dae@samsung.com>
Cc: Intel Graphics Development <intel-gfx@lists.freedesktop.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

VMA offsets are 64bit so do not cast them to "unsigned int". Also remove
the (now useless) offset-retrieval helper. The VMA manager provides simple
enough helpers.

Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: "Terje Bergström" <tbergstrom@nvidia.com>
Cc: Arto Merilainen <amerilainen@nvidia.com>
Signed-off-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This makes it so that reloading a module does not cause all the
connector ids to change, which are user-visible and sometimes used
for configuration.
Signed-off-by: NIlia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Basically just extracting some code duplicated in gma500, omapdrm, udl,
and upcoming msm driver.
Signed-off-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Variant of drm_gem_create_mmap_offset() which doesn't make the
assumption that virtual size and physical size (obj->size) are the same.
This is needed in omapdrm to deal with tiled buffers.  And lets us get
rid of a duplicated and slightly modified version of
drm_gem_create_mmap_offset() in omapdrm.
Signed-off-by: NRob Clark <robdclark@gmail.com>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

And simplify how we hold a ref+pin to what is being scanned out by using
fb refcnt'ing.  The previous logic pre-dated fb refcnt, and as a result
was less straightforward than it could have been.  By holding a ref to
the fb, we don't have to care about how many plane's there are and
holding a ref to each color plane's bo.
Signed-off-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NRob Clark <robdclark@gmail.com>
Tested-by: NDarren Etheridge <detheridge@ti.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

A small helper to queue up work to do, from workqueue context, after a
flip.  Typically useful to defer unreffing buffers that may be read by
the display controller until vblank.

v1: original
v2: wire up docbook + couple docbook fixes
Signed-off-by: NRob Clark <robdclark@gmail.com>
Acked-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This function is unused.
Signed-off-by: NStéphane Marchesin <marcheu@chromium.org>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Again only used by a tests in libdrm and by dristat. Nowadays we have
much better tracing tools to get detailed insights into what a drm
driver is doing. And for a simple "does it work" kind of question that
these stats could answer we have plenty of dmesg debug log spew.

So I don't see any use for this stat gathering complexity at all.

To be able to gradually drop things start with ripping out the
interfaces to it, here the ioctl.

To prevent dristat from eating its own stack garbage we can't use the
drm_noop ioctl though, since we need to clear the return data with a
memset.

Cc: Eric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDave Airlie <airlied@redhat.com>

We not only have debugfs files to do pretty much the equivalent of
lsof, we also have an ioctl. Not that compared to lsof this dumps a
wee bit more information, but we can still get at that from debugfs
easily.

I've dug around in mesa, libdrm and ddx histories and the only users
seem to be drm/tests/dristat.c and drm/tests/getclients.c. The later
is a testcase for the ioctl itself since up to

commit b018fcda
Author: Eric Anholt <eric@anholt.net>
Date:   Thu Nov 22 18:46:54 2007 +1000

    drm: Make DRM_IOCTL_GET_CLIENT return EINVAL when it can't find client #idx

there was actually no way at all for userspace to enumerate all
clients since the kernel just wouldn't tell it when to stop. Which
completely broke it's only user, dristat -c.

So obviously that ioctl wasn't much use for debugging. Hence I don't
see any point in keeping support for a tool which was pretty obviously
never really used, and while we have good replacements in the form of
equivalent debugfs files.

Still, to keep dristat -c from looping forever again stop it early by
returning an unconditional -EINVAL. Also add a comment in the code
about why.

v2: Slightly less hollowed-out implementation. libva uses GET_CLIENTS
to figure out whether the fd it has is already authenticated or not.
So we need to keep that part of things working. Simplest way is to
just return one entry to keep va_drm_is_authenticated in
libva/va/drm/va_drm_auth.c working.

This is exercised by igt/drm_get_client_auth which contains a
copypasta of the libva auth check code.

Cc: Gwenole Beauchesne <gwenole.beauchesne@intel.com>
Cc: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

They're only used by the agpgart support code in drm_agpgart.c,
not by any drivers.

I think long-term we should create a drm_internal.h include file with
all the various functions only used by the drm core and not exported
to drivers, and remove them from drmP.h. Oh, and someone should kill
that upper-case P sometimes ;-) But that's all stuff for future patch
bombs.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

The gma500 driver somehow set the DRIVER_IRQ_VBL flag, but since
there's no code at all to check for this we can kill it. The other two
are completely unused.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

No driver ever sets that flag, so good riddance!
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

So I've stumbled over drm_fasync and wondered what it does. Digging
that up is quite a story.

First I've had to read up on what this does and ended up being rather
bewildered why peopled loved signals so much back in the days that
they've created SIGIO just for that ...

Then I wondered how this ever works, and what that strange "No-op."
comment right above it should mean. After all calling the core fasync
helper is pretty obviously not a noop. After reading through the
kernels FASYNC implementation I've noticed that signals are only sent
out to the processes attached with FASYNC by calling kill_fasync.

No merged drm driver has ever done that.

After more digging I've found out that the only driver that ever used
this is the so called GAMMA driver. I've frankly never heard of such a
gpu brand ever before. Now FASYNC seems to not have been the only bad
thing with that driver, since Dave Airlie removed it from the drm
driver with prejudice:

commit 1430163b4bbf7b00367ea1066c1c5fe85dbeefed
Author: Dave Airlie <airlied@linux.ie>
Date:   Sun Aug 29 12:04:35 2004 +0000

    Drop GAMMA DRM from a great height ...

Long story short, the drm fasync support seems to be doing absolutely
nothing. And the only user of it was never merged into the upstream
kernel. And we don't need any fops->fasync callback since the fcntl
implementation in the kernel already implements the noop case
correctly.

So stop this particular cargo-cult and rip it all out.

v2: Kill drm_fasync assignments in rcar (newly added) and imx drivers
(somehow I've missed that one in staging). Also drop the reference in
the drm DocBook. ARM compile-fail reported by Rob Clark.

v3: Move the removal of dev->buf_asnyc assignment in drm_setup to this
patch here.

v4: Actually git add ... tsk.

Cc: Dave Airlie <airlied@linux.ie>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Rob Clark <robdclark@gmail.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

It's kzalloced ...
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

So after a lot of digging around in git histories it looks like this
has only ever be used by dri1 render clients. Hence we can fully
disable the entire thing for modesetting drivers and so greatly reduce
the attack surface for potential exploits (or at least tools like
trinity ...).

Also add the drm_legacy prefix for functions which are called from
common code. To further reduce the impact on common code also extract
all the ctx release handling into a function (instead of only
releasing individual handles) and make ctxbitmap_cleanup return void -
it can never fail.
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Now only legacy ums drivers have the DRIVER_HAVE_DMA driver feature
flag set, so strictly speaking the modesetting check is redundant. But
adding it has the upside that it makes it very clear that the dma
support is legacy stuff.
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

And hide the checks a bit better. This was already disallowed for
modesetting drivers, so no functinal change here.
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Only the radeon/r128/ati ums drivers use this. Furthermore the cleanup
was already only done for UMS drivers. Also a quick check of the ATI
ddx git history shows that only the UMS code ever used this facility.

So we can safely disallow these pair of ioctls for modesetting
drivers.
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

I've decided that some clear markers for what's legacy dri1/non-gem
code is useful. I've opted to use the drm_legacy prefix and then hide
all the checks in that function for better readability in the common
code.
Reviewed-by: NEric Anholt <eric@anholt.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Totally unused, so just rip it out. Anyway, we want drivers to be
fully backwards compatible, allowing them to change behaviour is just
a recipe for them to break badly.
Reviewed-by: NEric Anholt <eric@anholt.net>
Reviewed-by: NRob Clark <robdclark@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NDave Airlie <airlied@redhat.com>