If we can't find a free fence register to handle a fault in the GMADR
range just return VM_FAULT_NOPAGE without populating the PTE so that
userspace will retry the access and trigger another fault. Eventually
we should find a free fence and the fault will get properly handled.
A further improvement idea might be to reserve a fence (or one per CPU?)
for the express purpose of handling faults without having to retry. But
that would require some additional work.
Looks like this may have gotten broken originally by
commit 39965b3766 ("drm/i915: don't trash the gtt when running out of fences")
as that changed the errno to -EDEADLK which wasn't handle by the gtt
fault code either. But later in commit 2feeb52859 ("drm/i915/gt: Fix
-EDEADLK handling regression") I changed it again to -ENOBUFS as -EDEADLK
was now getting used for the ww mutex dance. So this fix only makes
sense after that last commit.
Cc: stable@vger.kernel.org
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9479
Fixes: 2feeb52859 ("drm/i915/gt: Fix -EDEADLK handling regression")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231012132801.16292-1-ville.syrjala@linux.intel.com
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
The previous patches converted all users of the DSC DPCD caps to look
these up from the connector, so remove the version stored in intel_dp.
A follow-up patchset will read out the MST connector specific
capabilities in intel_dp_add_mst_connector() ->
intel_dp_mst_read_decompression_port_dsc_caps().
v2:
- Rebased on intel_edp_get_dsc_sink_cap() addition in the patchset.
v3:
- Rebased on read-out fix for eDP in the patchset.
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> (v1)
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231011171606.2540078-3-imre.deak@intel.com
In an MST topology the DSC capabilities are specific to each connector,
retrieved either from the sink if it decompresses the stream, or from a
branch device between the source and the sink in case this branch device
does the decompression. Accordingly each connector needs to cache its
own DSC DPCD and FEC capabilities, along with the AUX device through
which the decompression can be enabled. This patch prepares for that by
storing the capabilities and the DSC AUX device in the connector, for
now these just matching the version stored in intel_dp. The follow-up
patches will convert all users to look up these in the connector instead
of intel_dp, after which the intel_dp copies are removed.
v2:
- Rebased on intel_edp_get_dsc_sink_cap() addition in previous patch.
v3:
- Rebased on read-out fix for eDP in previous patch.
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> (v1)
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231011171606.2540078-2-imre.deak@intel.com
Check only the eDP or the DP specific DPCD revision depending on the
sink type. Pass the corresponding revision to the function, which allows
getting the DSC caps of a branch device (in an MST topology, which has
its own DPCD and so DPCD revision).
While at it use DP_DPCD_REV_14 instead of open coding it and for clarity
add a separate function to read out the DSC capability on eDP.
v2:
- Use DP_DPCD_REV_14 instead of open coding it. (Stan)
- Check EDP_DCPD_REV/DPCD_REV in a clearer way. (Ville)
v3:
- Fix the read-out for eDP in intel_dp_detect().
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Stanislav Lisovskiy <stanislav.lisovskiy@intel.com> (v1)
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> (v2)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231011171606.2540078-1-imre.deak@intel.com
drm/i915 feature pull #2 for v6.7:
Features and functionality:
- Preparation for i915 display code reuse in upcoming Xe driver (Jani)
- Drop the fastboot module parameter and use the platform defaults (Arun)
- Enable new LNL FBC features (Vinod)
- Add LNL display feature capability reads (Vinod)
Refactoring and cleanups:
- Locally enable W=1 warnings by default in i915 (Jani)
- Move HDCP GSC message code to a separate file (Suraj)
- GVT include cleanups (Jani)
- Move more display init under display/ (Jani)
- DPLL ID refactoring (Ville)
- Better abstraction of GT0 (Jani)
- Move VGA decode function to GMCH code (Uma)
- Use local64_try_cmpxchg() to optimize PMU event read (Uros Bizjak)
- Clean up FBC checks (Ville)
- Constify and unify state checker calling conventions (Ville)
- Add display step name helper (Chaitanya)
Documentation:
- Update CCS and GSC CS documentation (Rodrigo)
- Fix a number of documentation typos (Randy Dunlap)
Fixes:
- VLV DSI fixes and quirks (Hans)
- Fix crtc state memory leaks (Suraj)
- Increase LSPCON mode settle timeout (Niko Tsirakis)
- Stop clobbering old crtc state during state check (Ville)
- Fix VLV color state readout (Ville)
- Fix cx0 PHY pipe reset to allow S0iX (Khaled)
- Ensure DP MST pbn_div is up-to-date after sink reconnect (Imre)
- Drop an unnecessary NULL check to fix static analyzer warning (Suraj)
- Use an explicit rather than implicit include for frontbuffer tracking (Jouni)
Merges:
- Backmerge drm-next to fix a conflict (Jani)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87r0m00xew.fsf@intel.com
The display engine does not snoop the caches so we should mark
the DSB command buffer as I915_CACHE_NONE.
i915_gem_object_create_internal() always gives us I915_CACHE_LLC
on LLC platforms. And to make things 100% correct we should also
clflush at the end, if necessary.
Note that currently this is a non-issue as we always write the
command buffer through a WC mapping, so a cache flush is not actually
needed. But we might actually want to consider a WB mapping since
we also end up reading from the command buffer (in the indexed
reg write handling). Either that or we should do something else
to avoid those reads (might actually be even more sensible on DGFX
since we end up reading over PCIe). But we should measure the overhead
first...
Anyways, no real harm in adding the belts and suspenders here so
that the code will work correctly regardless of how we map the
buffer. If we do get a WC mapping (as we request)
i915_gem_object_flush_map() will be a nop. Well, apart form
a wmb() which may just flush the WC buffer a bit earlier
than would otherwise happen (at the latest the mmio accesses
would trigger the WC flush).
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009132204.15098-2-ville.syrjala@linux.intel.com
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
Using system memory for the DSB command buffer doesn't appear to work.
On DG2 it seems like the hardware internally replaces the actual memory
reads with zeroes, and so we end up executing a bunch of NOOPs instead
of whatever commands we put in the buffer. To determine that I measured
the time it takes to execute the instructions, and the results are
always more or less consistent with executing a buffer full of NOOPs
from local memory.
Another theory I considered was some kind of cache coherency issue.
Looks like i915_gem_object_pin_map_unlocked() will in fact give you a
WB mapping for system memory on DGFX regardless of what mapping mode
was requested (WC in case of the DSB code). But clflush did not
change the behaviour at all, so that theory seems moot.
On DG1 it looks like the hardware might actually be fetching data from
system memory as the logs indicate that we just get underruns. But that
is equally bad, so doesn't look like we can really use system memory on
DG1 either.
Thus always allocate the DSB command buffer from local memory on
discrete GPUs.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231009132204.15098-1-ville.syrjala@linux.intel.com
Reviewed-by: Uma Shankar <uma.shankar@intel.com>
On the Lenovo Yoga Tab 3 Pro YT3-X90F there are 2 issues with the backlight
on/off MIPI sequences:
1. The backlight on sequence has an I2C MIPI sequence element which uses
bus 0, but there is a bogus I2cSerialBus resource under the GPU in
the DSDT which causes i2c_acpi_find_adapter() to pick the wrong bus.
2. There is no backlight off sequence, causing the backlight to stay on.
Add a DMI quirk fixing both issues.
v2:
- Add Closes tag to gitlab issue with drm.debug=0xe, VBT info
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9380
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-4-hdegoede@redhat.com
On the Lenovo Yoga Tablet 2 830 / 1050 there are 2 problems:
1. The I2C MIPI sequence elements reference bus 3. ACPI has I2C1 - I2C7
which under Linux become bus 0 - 6. And the MIPI sequence reference
to bus 3 is indented for I2C3 which is bus 2 under Linux.
This leads to errors like these:
[ 178.244049] i2c_designware 80860F41:03: controller timed out
[ 178.245703] i915 0000:00:02.0: [drm] *ERROR* Failed to xfer payload of size (1) to reg (169)
There are 3 timeouts when the panel is on, delaying
waking up the screen on a key press by 3 seconds.
Note mipi_exec_i2c() cannot just subtract 1 from the bus
given in the I2C MIPI sequence element. Since on other
devices the I2C bus-numbers used in the MIPI sequences do
actually start at 0.
2. width_/height_mm contain a bogus 192mm x 120mm size. This is
especially a problem on the 8" 830 version which uses a 10:16
portrait screen where as the bogus size is 16:10.
Add a DMI quirk to override the I2C bus and the panel size with
the correct values.
Note both the 10" 1050 models as well as the 8" 830 models use the same
mainboard and thus the same DMI strings. The 10" 1050 uses a 1920x1200
landscape screen, where as the 8" 830 uses a 1200x1920 portrait screen,
so the quirk handling uses the display resolution to detect the model.
v2:
- Also override i2c_bus_num to fix mipi_exec_i2c() timeouts
v3:
- Add Closes tag to gitlab issue with drm.debug=0xe, VBT info
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9379
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-3-hdegoede@redhat.com
Vtotal is wrong in the BIOS supplied modeline for the DSI panel on
the Asus TF103C leading to the last line of the display being shown
as the first line.
Original: "1280x800": 60 67700 1280 1312 1328 1376 800 808 812 820 0x8 0xa
Fixed: "1280x800": 60 67700 1280 1312 1328 1376 800 808 812 816 0x8 0xa
The factory installed Android has a hardcoded modeline in its kernel,
causing it to not suffer from this BIOS bug;
and the Android boot-splash which uses the EFI FB which does have this bug
has the last line all black causing the bug to not be visible.
This commit introduces a generic DMI based quirk mechanism to vlv_dsi for
doing various fixups, and uses this to correct the modeline.
v2:
- s/mode_fixup/dmi_quirk/ to make the new DMI quirk mechanism more generic
- Add a comment with the old and new modelines to the patch and commit msg
v3:
- Add Closes tag to gitlab issue with drm.debug=0xe, VBT info
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/9381
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230920195613.304091-2-hdegoede@redhat.com
The functions drm_framebuffer_plane_{width,height} and
fb_plane_{width,height} do exactly the same job of its
equivalents drm_format_info_plane_{width,height} from drm_fourcc.
The only reason to have these functions on drm_framebuffer
would be if they would added a abstraction layer to call it just
passing a drm_framebuffer pointer and the desired plane index,
which is not the case, where these functions actually implements
just part of it. In the actual implementation, every call to both
drm_framebuffer_plane_{width,height} and fb_plane_{width,height} should
pass some drm_framebuffer attribute, which is the same as calling the
drm_format_info_plane_{width,height} functions.
The drm_format_info_pane_{width,height} functions are much more
consistent in both its implementation and its location on code. The
kind of calculation that they do is intrinsically derivated from the
drm_format_info struct and has not to do with drm_framebuffer, except
by the potential motivation described above, which is still not a good
justification to have drm_framebuffer functions to calculate it.
So, replace each drm_framebuffer_plane_{width,height} and
fb_plane_{width,height} call to drm_format_info_plane_{width,height}
and remove them.
Signed-off-by: Carlos Eduardo Gallo Filho <gcarlos@disroot.org>
Reviewed-by: André Almeida <andrealmeid@igalia.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230926141519.9315-3-gcarlos@disroot.org
We enable a bunch more compiler warnings than the kernel
defaults. However, they've drifted to become a unique set of warnings,
and have increasingly fallen behind from the W=1 set.
Align with the W=1 warnings from scripts/Makefile.extrawarn for clarity,
by copy-pasting them with s/KBUILD_CFLAGS/subdir-ccflags-y/ to make it
easier to compare in the future.
Some of the -Wextra warnings do need to be disabled, just like in
Makefile.extrawarn, but take care to not disable them for W=2 or W=3
builds, depending on the warning.
v2: Add back some -Wextra warning disables (Nathan)
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Masahiro Yamada <masahiroy@kernel.org>
[Final s/KBUILD_CFLAGS/subdir-ccflags-y/ fix while applying]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/69a812273091b6535ddc7f9346289d71bb30f43d.1697009258.git.jani.nikula@intel.com
This commit comes at the tail end of a greater effort to remove the
empty elements at the end of the ctl_table arrays (sentinels) which
will reduce the overall build time size of the kernel and run time
memory bloat by ~64 bytes per sentinel (further information Link :
https://lore.kernel.org/all/ZO5Yx5JFogGi%2FcBo@bombadil.infradead.org/)
Remove sentinel from oa_table
Signed-off-by: Joel Granados <j.granados@samsung.com>
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
