Bspec lists different VT-d guard numbers (the number of dummy
padding PTEs) for different platforms and plane types. Use those
instead of just assuming the max glk+ number for everything.
This could avoid a bit of overhead on older platforms due to
reduced padding, and it makes it easier to cross check with the
spec.
Note that VLV/CHV do not document this w/a at all, so not sure
if it's actually needed or not. Nor do we actually know how much
padding is required if it is needed. For now use the same 128
PTEs that we use for snb-bdw primary planes.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250122151755.6928-5-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
We need to be able to do both MMIO and DSB based pipe/plane
programming. To that end plumb the 'dsb' all way from the top
into the plane commit hooks.
The compiler appears smart enough to combine the branches from
all the back-to-back register writes into a single branch.
So the generated asm ends up looking more or less like this:
plane_hook()
{
if (dsb) {
intel_dsb_reg_write();
intel_dsb_reg_write();
...
} else {
intel_de_write_fw();
intel_de_write_fw();
...
}
}
which seems like a reasonably efficient way to do this.
An alternative I was also considering is some kind of closure
(register write function + display vs. dsb pointer passed to it).
That does result is smaller code as there are no branches anymore,
but having each register access go via function pointer sounds
less efficient.
Not that I actually measured the overhead of either approach yet.
Also the reg_rw tracepoint seems to be making a huge mess of the
generated code for the mmio path. And additionally there's some
kind of IS_GSI_REG() hack in __raw_uncore_read() which ends up
generating a pointless branch for every mmio register access.
So looks like there might be quite a bit of room for improvement
in the mmio path still.
Reviewed-by: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240930170415.23841-12-ville.syrjala@linux.intel.com
In case of legacy cursor update, the cursor VMA needs to be unpinned
only after vblank. This exceeds the lifetime of the whole atomic commit.
Any trick I attempted to keep the atomic commit alive didn't work, as
drm_atomic_helper_setup_commit() force throttles on any old commit that
wasn't cleaned up.
The only option remaining is to remove the plane from the atomic commit,
and use the same path as the legacy cursor update to clean the state
after vblank.
Changes since previous version:
- Call the memset for plane state immediately when scheduling vblank,
this prevents a use-after-free in cursor cleanup.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240522053341.137592-4-maarten.lankhorst@linux.intel.com
The cursor hardware only does sync updates, and thus the hardware
will be scanning out from the old fb until the next start of vblank.
So in order to make the legacy cursor fastpath actually safe we
should not unpin the old fb until we're sure the hardware has
ceased accessing it. The simplest approach is to just use a vblank
work here to do the delayed unpin.
Not 100% sure it's a good idea to put this onto the same high
priority vblank worker as eg. our timing critical gamma updates.
But let's keep it simple for now, and it we later discover that
this is causing problems we can think about adding a lower
priority worker for such things.
This patch is slightly reworked by Maarten
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240522053341.137592-3-maarten.lankhorst@linux.intel.com
Different hardware generations have different scanout alignment
requirements. Introduce a new vfunc that will allow us to
make that distinction without horrible if-ladders.
For now we directly plug in the existing intel_surf_alignment()
and intel_cursor_alignment() functions.
For fbdev we (temporarily) introduce intel_fbdev_min_alignment()
that simply queries the alignment from the primary plane of
the first crtc.
TODO: someone will need to fix xe's alignment handling
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240612204712.31404-4-ville.syrjala@linux.intel.com
Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Split the cursor stuff from the rest of the selective fetch
plane registers so that we can collect all cursor registers
in intel_cursor_regs.h. Also take the opportunity to rename
the registers to match the spec.
v2: Pass the correct register offset fpr pipe B (Jani)
s/mtl+/tgl+/ as that's where this was introduced
Drop the bogus SEL_FETCH_CUR_CTL_ENABLE bit, the contents
actually match the normal CUR_CTL register
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240520171459.9661-1-ville.syrjala@linux.intel.com
Having the plane WM/DDB regitster write functions in skl_watermarks.c
is rather annoying when trying to implement DSB based plane updates.
Move them into the respective files that handle all other plane
register writes. Less places where I need to worry about the DSB
vs. MMIO decisions.
The downside is that we spread the wm struct details a bit further
afield. But if that becomes too annoying we can probably abstract
things a bit more with a few extra functions.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240510152329.24098-17-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
drm-misc-next for v6.10-rc1:
UAPI Changes:
- Add SIZE_HINTS property for cursor planes.
Cross-subsystem Changes:
Core Changes:
- Document the requirements and expectations of adding new
driver-specific properties.
- Assorted small fixes to ttm.
- More Kconfig fixes.
- Add struct drm_edid_product_id and helpers.
- Use drm device based logging in more drm functions.
- Fixes for drm-panic, and option to test it.
- Assorted small fixes and updates to edid.
- Add drm_crtc_vblank_crtc and use it in vkms, nouveau.
Driver Changes:
- Assorted small fixes and improvements to bridge/imx8mp-hdmi-tx, nouveau, ast, qaic, lima, vc4, bridge/anx7625, mipi-dsi.
- Add drm panic to simpledrm, mgag200, imx, ast.
- Use dev_err_probe in bridge/panel drivers.
- Add Innolux G121X1-L03, LG sw43408 panels.
- Use struct drm_edid in i915 bios parsing.
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/2dc1b7c6-1743-4ddd-ad42-36f700234fbe@linux.intel.com
Our legacy cursor updates are actually mailbox updates.
Ie. the hardware latches things once per frame on start of
vblank, but we issue an number of updates per frame,
withough any attempt to synchronize against the vblank
in software. So in theory only the last update issued
during the frame will latch, and the previous ones are
discarded.
However this can lead to problems with maintaining the
ggtt/iommu mappings as we have no idea which updates
will actually latch.
The problem is exacerbated by the hardware's annoying disarming
behaviour; any non-arming register write will disarm an already
armed update, only to be rearmed later by the arming register
(CURBASE in case of cursors). If a disarming write happens
just before the start of vblank, and the arming write happens
after start of vblank we have effectively prevented the hardware
from latching anything. And if we manage to straddle multiple
sequential vblank starts in this manner we effectively prevent
the hardware from latching any new registers for an arbitrary
amount of time. This provides more time for the (potentially
still in use by the hardware) gtt/iommu mappings to be torn
down.
A partial solution, of course, is to use vblank evasion to
avoid the register writes from spreading on both sides of
the start of vblank.
I've previously highlighted this problem as a general issue
affecting mailbox updates. I even added some notes to the
{i9xx,skl}_crtc_planes_update_arm() to remind us that the noarm
and arm phases both need to pulled into the vblank evasion
critical section if we actually decided to implement mailbox
updates in general. But as I never impelemented the noarm+arm
split for cursors we don't have to worry about that for the
moment.
We've been lucky enough so far that this hasn't really caused
problems. One thing that does help is that Xorg generally
sticks to the same cursor BO. But igt seems pretty good at
hitting this on MTL now, so apparently we have to start
thinking about this.
v2: Wait for PSR exit to avoid the vblank evasion timeout (1ms)
tripping due to PSR exit latency (~5ms typically)
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240116204927.23499-1-ville.syrjala@linux.intel.com
Reviewed-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Currently we are enabling selective fetch for all planes that are visible.
This is suboptimal as we might be fetching for memory for planes that are
not part of selective update.
Fix this by adding proper handling for disabling plane selective fetch:
If plane previously part of selective update is now not part of update:
Add it into updated planes and let the plane configuration to disable
selective fetch for it.
v3: Checkpatch warnings fixed
v2:
- Add setting sel_fetch_area->y1/y2 to -1
- Remove setting again local sel_fetch_area variable
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231120082606.3156488-3-jouni.hogander@intel.com
Rather than embeddeding the display's device info within the main device
info structure, just provide a pointer to the display-specific
structure. This is in preparation for moving the display device info
definitions into the display code itself and for eventually allowing the
pointer to be assigned at runtime on platforms that use GMD_ID for
device identification.
In the future, this will also eventually allow the same display device
info structures to be used outside the current i915 code (e.g., from the
Xe driver).
v2:
- Move introduction of DISPLAY_INFO() to this patch. (Andrzej)
v3:
- Also use DISPLAY_INFO() in intel_display_reg_defs.h. (Andrzej)
- Use "{}" instead of "{ 0 }" for empty struct init. (Jani)
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Acked-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230523195609.73627-3-matthew.d.roper@intel.com
SEL_FETCH_CTL registers are armed immediately when plane is disabled.
SEL_FETCH_* instances of plane configuration are used when doing
selective update and normal plane register instances for full updates.
Currently all SEL_FETCH_* registers are written as a part of noarm
plane configuration. If noarm and arm plane configuration are not
happening within same vblank we may end up having plane as a part of
selective update before it's PLANE_SURF register is written.
Fix this by splitting plane selective fetch configuration into arm and
noarm versions and call them accordingly. Write SEL_FETCH_CTL in arm
version.
v3:
- add arm suffix into intel_psr2_disable_plane_sel_fetch
v2:
- drop color_plane parameter from arm part
- dev_priv -> i915 in arm part
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Vinod Govindapillai <vinod.govindapillai@intel.com>
Cc: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com>
Cc: Luca Coelho <luciano.coelho@intel.com>
Signed-off-by: Jouni Högander <jouni.hogander@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Luca Coelho <luciano.coelho@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230130080651.3796929-1-jouni.hogander@intel.com
Turns out many of the files that need i915_reg.h get it implicitly via
{display/intel_de.h, gt/intel_context.h} -> i915_trace.h -> i915_irq.h
-> i915_reg.h. Since i915_trace.h doesn't actually need i915_irq.h,
makes sense to drop it, but that requires adding quite a few new
includes all over the place.
Prefer including i915_reg.h where needed instead of adding another
implicit include, because eventually we'll want to split up i915_reg.h
and only include the specific registers at each place.
Also some places actually needed i915_irq.h too.
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/6e78a2e0ac1bffaf5af3b5ccc21dff05e6518cef.1668008071.git.jani.nikula@intel.com
