I caught a few errors in our current PHY/CDCLK programming by sanity
checking the actual programmed state, so I thought it would be also
useful for the future. In addition to verifying the state after
programming it also verify it after exiting DC5, to make sure DMC
restored/kept intact everything related.
v2:
- Inlining __phy_reg_verify_state() doesn't make sense and also
incorrect, so don't do it (PW/CI gcc)
v3:
- Rebase on latest -nightly
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: David Weinehall <david.weinehall@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1459780030-15781-1-git-send-email-imre.deak@intel.com
Power well 1 is managed by the DMC firmware so don't toggle it on-demand
from the driver. This means we need to follow the BSpec display
initialization sequence during driver loading and resuming (both system
and runtime) and enable power well 1 only once there. Afterwards DMC
will toggle power well 1 whenever entering/exiting DC5.
For this to work we also need to do away getting the PLL power domain,
since that just kept runtime PM disabled for good.
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1459515767-29228-12-git-send-email-imre.deak@intel.com
On Broxton we need to enable/disable power well 1 during the init/unit
display sequence similarly to Skylake/Kabylake. The code for this will
be added in a follow-up patch, but to prepare for that unexport
skl_pw1_misc_io_init(). It's a simple function called only from a single
place and having it inlined in the Skylake display core init/unit
functions will make it easier to compare it with its Broxton
counterpart.
This also flips the order of Misc IO and power well 1 disabling which
matches the enabling order. The specification doesn't prescribe the
disabling order, so this should be fine.
v2:
- Fix incorrect enable vs. disable power well call in
skl_display_core_uninit() (Patrik)
- Add commit comment about chaning the order of PW1 and Misc IO power
well disabling (Patrik)
CC: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1459773777-10701-1-git-send-email-imre.deak@intel.com
DMC forces on power well 1 and the misc IO power well by setting the
corresponding request bits both in the BIOS and the DEBUG power well
request registers. This is somewhat unexpected since the firmware should
really just save and restore state but not alter it. We also depend on
being able to disable power well 1, and the misc IO power well before
entering S3/S4 on BXT and SKL or entering DC9 on BXT. To fix this make
sure these request bits are cleared whenever we want to disable the
given power wells.
On SKL there is another twist where the firmware also clears the power
well 1 request bit in HSW_POWER_WELL_DRIVER (but not that of the misc IO
power well). This happens to not cause a problem due to the forced-on
request bits in the other request registers.
I've filed a bug about all this, but fixing that may take a while and
having this sanity check in place makes sense even for future firmware
versions.
At the same time also check the KVMR request bits. I haven't seen this
being altered, but we don't expect any request bits in here either, so
sanitize this register as well.
v2:
- Apply the workaround on SKL as well. I noticed the related failure
from the CI report, later Patrik also reported seeing it on his
machine.
CC: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Patrik Jakobsson <patrik.jakobsson@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1459851965-6137-1-git-send-email-imre.deak@intel.com
On VLV/CHV the master interrupt enable bit only affects GT/PM
interrupts. Display interrupts are not affected by the master
irq control.
Also it seems that the CPU interrupt will only be generated when
the combined result of all GT/PM/display interrupts has a 0->1
edge. We already use the master interrupt enable bit to make sure
GT/PM interrupt can generate such an edge if we don't end up clearing
all IIR bits. We must do the same for display interrupts, and for
that we can simply clear out VLV_IER, and restore after we've acked
all the interrupts we are about to process.
So with both master interrupt enable and VLV_IER cleared out, we will
guarantee that there will be a 0->1 edge if any IIR bits are still set
at the end, and thus another CPU interrupt will be generated.
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Fixes: 579de73b04 ("drm/i915: Exit cherryview_irq_handler() after one pass")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1460571598-24452-6-git-send-email-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Allow for the MOCS to be programmed for all engines.
Currently we program the MOCS when the first render batch
goes through. This works on most platforms but fails on
platforms that do not run a render batch early,
i.e. headless servers. The patch now programs all initialised engines
on init and the RCS is programmed again within the initial batch. This
is done for predictable consistency with regards to the hardware
context.
Hardware context loading sets the values of the MOCS for RCS
and L3CC. Programming them from within the batch makes sure that
the render context is valid, no matter what the previous state of
the saved-context was.
v2: posted correct version to the mailing list.
v3: moved programming to within engine->init_hw() (Chris Wilson)
v4: code formatting and white-space changes. (Chris Wilson)
Testcase: igt/gem_mocs_settings
Signed-off-by: Peter Antoine <peter.antoine@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1460556205-6644-1-git-send-email-chris@chris-wilson.co.uk
Conceptually, each request is a record of a hardware transaction - we
build up a list of pending commands and then either commit them to
hardware, or cancel them. However, whilst building up the list of
pending commands, we may modify state outside of the request and make
references to the pending request. If we do so and then cancel that
request, external objects then point to the deleted request leading to
both graphical and memory corruption.
The easiest example is to consider object/VMA tracking. When we mark an
object as active in a request, we store a pointer to this, the most
recent request, in the object. Then we want to free that object, we wait
for the most recent request to be idle before proceeding (otherwise the
hardware will write to pages now owned by the system, or we will attempt
to read from those pages before the hardware is finished writing). If
the request was cancelled instead, that wait completes immediately. As a
result, all requests must be committed and not cancelled if the external
state is unknown.
All that remains of i915_gem_request_cancel() users are just a couple of
extremely unlikely allocation failures, so remove the API entirely.
A consequence of committing all incomplete requests is that we generate
excess breadcrumbs and fill the ring much more often with dummy work. We
have completely undone the outstanding_last_seqno optimisation.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93907
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-16-git-send-email-chris@chris-wilson.co.uk
For reasons unknown Sandybridge GT1 (at least) will eventually hang when
it encounters a ring wraparound at offset 0. The test case that
reproduces the bug reliably forces a large number of interrupted context
switches, thereby causing very frequent ring wraparounds, but there are
similar bug reports in the wild with the same symptoms, seqno writes
stop just before the wrap and the ringbuffer at address 0. It is also
timing crucial, but adding various delays hasn't helped pinpoint where
the window lies.
Whether the fault is restricted to the ringbuffer itself or the GTT
addressing is unclear, but moving the ringbuffer fixes all the hangs I
have been able to reproduce.
References: (e.g.) https://bugs.freedesktop.org/show_bug.cgi?id=93262
Testcase: igt/gem_exec_whisper/render-contexts-interruptible #snb-gt1
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@vger.kernel.org
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-12-git-send-email-chris@chris-wilson.co.uk
Two concurrent writes into the same register cacheline has the chance of
killing the machine on Ivybridge and other gen7. This includes LRI
emitted from the command parser. The MI_SET_CONTEXT itself serves as
serialising barrier and prevents the pair of register writes in the first
packet from triggering the fault. However, if a second switch-context
immediately occurs then we may have two adjacent blocks of LRI to the
same registers which may then trigger the hang. To counteract this we
need to insert a delay after the second register write using SRM.
This is easiest to reproduce with something like
igt/gem_ctx_switch/interruptible that triggers back-to-back context
switches (with no operations in between them in the command stream,
which requires the execbuf operation to be interrupted after the
MI_SET_CONTEXT) but can be observed sporadically elsewhere when running
interruptible igt. No reports from the wild though, so it must be of low
enough frequency that no one has correlated the random machine freezes
with i915.ko
The issue was introduced with
commit 2c55018347 [v3.19]
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Dec 16 10:02:27 2014 +0000
drm/i915: Disable PSMI sleep messages on all rings around context switches
Testcase: igt/gem_ctx_switch/render-interruptible #ivb
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-11-git-send-email-chris@chris-wilson.co.uk
Reporting -EIO from i915_wait_request() has proven very troublematic
over the years, with numerous hard-to-reproduce bugs cropping up in the
corner case of where a reset occurs and the code wasn't expecting such
an error.
If the we reset the GPU or have detected a hang and wish to reset the
GPU, the request is forcibly complete and the wait broken. Currently, we
report either -EAGAIN or -EIO in order for the caller to retreat and
restart the wait (if appropriate) after dropping and then reacquiring
the struct_mutex (essential to allow the GPU reset to proceed). However,
if we take the view that the request is complete (no further work will
be done on it by the GPU because it is dead and soon to be reset), then
we can proceed with the task at hand and then drop the struct_mutex
allowing the reset to occur. This transfers the burden of checking
whether it is safe to proceed to the caller, which in all but one
instance it is safe - completely eliminating the source of all spurious
-EIO.
Of note, we only have two API entry points where we expect that
userspace can observe an EIO. First is when submitting an execbuf, if
the GPU is terminally wedged, then the operation cannot succeed and an
-EIO is reported. Secondly, existing userspace uses the throttle ioctl
to detect an already wedged GPU before starting using HW acceleration
(or to confirm that the GPU is wedged after an error condition). So if
the GPU is wedged when the user calls throttle, also report -EIO.
v2: Split more carefully the change to i915_wait_request() and assorted
ABI from the reset handling.
v3: Add a couple of WARN_ON(EIO) to the interruptible modesetting code
so that we don't start to leak EIO there in future (and break our hang
resistant modesetting).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-9-git-send-email-chris@chris-wilson.co.uk
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-1-git-send-email-chris@chris-wilson.co.uk
As the request is only valid during the same global reset epoch, we can
record the current reset_counter when constructing the request and reuse
it when waiting upon that request in future. This removes a very hairy
atomic check serialised by the struct_mutex at the time of waiting and
allows us to transfer those waits to a central dispatcher for all
waiters and all requests.
PS: With per-engine resets, we obviously cannot assume a global reset
epoch for the requests - a per-engine epoch makes the most sense. The
challenge then is how to handle checking in the waiter for when to break
the wait, as the fine-grained reset may also want to requeue the
request (i.e. the assumption that just because the epoch changes the
request is completed may be broken - or we just avoid breaking that
assumption with the fine-grained resets).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-7-git-send-email-chris@chris-wilson.co.uk
In the reset_counter, we use two bits to track a GPU hang and reset. The
low bit is a "reset-in-progress" flag that we set to signal when we need
to break waiters in order for the recovery task to grab the mutex. As
soon as the recovery task has the mutex, we can clear that flag (which
we do by incrementing the reset_counter thereby incrementing the gobal
reset epoch). By clearing that flag when the recovery task holds the
struct_mutex, we can forgo a second flag that simply tells GEM to ignore
the "reset-in-progress" flag.
The second flag we store in the reset_counter is whether the
reset failed and we consider the GPU terminally wedged. Whilst this flag
is set, all access to the GPU (at least through GEM rather than direct mmio
access) is verboten.
PS: Fun is in store, as in the future we want to move from a global
reset epoch to a per-engine reset engine with request recovery.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-6-git-send-email-chris@chris-wilson.co.uk
This is principally a little bit of syntatic sugar to hide the
atomic_read()s throughout the code to retrieve the current reset_counter.
It also provides the other utility functions to check the reset state on the
already read reset_counter, so that (in later patches) we can read it once
and do multiple tests rather than risk the value changing between tests.
v2: Be more strict on converting existing i915_reset_in_progress() over to
the more verbose i915_reset_in_progress_or_wedged().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1460565315-7748-4-git-send-email-chris@chris-wilson.co.uk
