Steven Price [Thu, 29 Oct 2020 17:00:47 +0000 (17:00 +0000)]
drm/panfrost: Don't corrupt the queue mutex on open/close
The mutex within the panfrost_queue_state should have the lifetime of
the queue, however it was erroneously initialised/destroyed during
panfrost_job_{open,close} which is called every time a client
opens/closes the drm node.
Move the initialisation/destruction to panfrost_job_{init,fini} where it
belongs.
Robin Murphy [Tue, 22 Sep 2020 14:16:50 +0000 (15:16 +0100)]
arm64: dts: meson: Describe G12b GPU as coherent
According to a downstream commit I found in the Khadas vendor kernel,
the GPU on G12b is wired up for ACE-lite, so (now that Panfrost knows
how to handle this properly) we should describe it as such. Otherwise
the mismatch leads to all manner of fun with mismatched attributes and
inadvertently snooping stale data from caches, which would account for
at least some of the brokenness observed on this platform.
Robin Murphy [Tue, 22 Sep 2020 14:16:49 +0000 (15:16 +0100)]
drm/panfrost: Support cache-coherent integrations
When the GPU's ACE-Lite interface is fully wired up and capable of
snooping CPU caches, it may be described as "dma-coherent" in
devicetree, which will already inform the DMA layer not to perform
unnecessary cache maintenance. However, we still need to ensure that
the GPU uses the appropriate cacheable outer-shareable attributes in
order to generate the requisite snoop signals, and that CPU mappings
don't create a mismatch by using a non-cacheable type either.
Robin Murphy [Tue, 22 Sep 2020 14:16:48 +0000 (15:16 +0100)]
iommu/io-pgtable-arm: Support coherency for Mali LPAE
Midgard GPUs have ACE-Lite master interfaces which allows systems to
integrate them in an I/O-coherent manner. It seems that from the GPU's
viewpoint, the rest of the system is its outer shareable domain, and so
even when snoop signals are wired up, they are only emitted for outer
shareable accesses. As such, setting the TTBR_SHARE_OUTER bit does
indeed get coherent pagetable walks working nicely for the coherent
T620 in the Arm Juno SoC.
Chris Wilson [Thu, 29 Oct 2020 21:30:42 +0000 (21:30 +0000)]
drm: Quieten [zero] EDID carping
We have a few displays in CI that always report their EDID as a bunch of
zeroes. This is consistent behaviour, so one assumes intentional
indication of an "absent" EDID. Flagging these consistent warnings
detracts from CI.
One option would be to ignore the zero EDIDs as intentional behaviour,
but Ville would like to keep the information available for debugging.
The simple alternative then is to reduce the loglevel for all the EDID
dumping from WARN to DEBUG so the information is present but not annoy
CI. Note that the bad EDID dumping is already only shown if
drm.debug=KMS, it's just the loglevel chosen was set to be caught by CI
if it ever occurred as it was expected to be an internal error not
external.
Christian König [Thu, 22 Oct 2020 16:26:58 +0000 (18:26 +0200)]
drm/ttm: new TT backend allocation pool v3
This replaces the spaghetti code in the two existing page pools.
First of all depending on the allocation size it is between 3 (1GiB) and
5 (1MiB) times faster than the old implementation.
It makes better use of buddy pages to allow for larger physical contiguous
allocations which should result in better TLB utilization at least for
amdgpu.
Instead of a completely braindead approach of filling the pool with one
CPU while another one is trying to shrink it we only give back freed
pages.
This also results in much less locking contention and a trylock free MM
shrinker callback, so we can guarantee that pages are given back to the
system when needed.
Downside of this is that it takes longer for many small allocations until
the pool is filled up. We could address this, but I couldn't find an use
case where this actually matters. We also don't bother freeing large
chunks of pages any more since the CPU overhead in that path isn't really
that important.
The sysfs files are replaced with a single module parameter, allowing
users to override how many pages should be globally pooled in TTM. This
unfortunately breaks the UAPI slightly, but as far as we know nobody ever
depended on this.
Zeroing memory coming from the pool was handled inconsistently. The
alloc_pages() based pool was zeroing it, the dma_alloc_attr() based one
wasn't. For now the new implementation isn't zeroing pages from the pool
either and only sets the __GFP_ZERO flag when necessary.
The implementation has only 768 lines of code compared to the over 2600
of the old one, and also allows for saving quite a bunch of code in the
drivers since we don't need specialized handling there any more based on
kernel config.
Additional to all of that there was a neat bug with IOMMU, coherent DMA
mappings and huge pages which is now fixed in the new code as well.
v2: make ttm_pool_apply_caching static as reported by the kernel bot, add
some more checks
v3: fix some more checkpatch.pl warnings
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-by: Madhav Chauhan <madhav.chauhan@amd.com> Tested-by: Huang Rui <ray.huang@amd.com> Link: https://patchwork.freedesktop.org/patch/397080/?series=83051&rev=1
Peilin Ye [Wed, 28 Oct 2020 10:56:47 +0000 (06:56 -0400)]
Fonts: Make font size unsigned in font_desc
`width` and `height` are defined as unsigned in our UAPI font descriptor
`struct console_font`. Make them unsigned in our kernel font descriptor
`struct font_desc`, too.
Also, change the corresponding printk() format identifiers from `%d` to
`%u`, in sti_select_fbfont().
Peilin Ye [Tue, 27 Oct 2020 16:31:08 +0000 (12:31 -0400)]
fbdev/atafb: Remove unused extern variables
Remove 6 unused extern variables to reduce confusion. It is worth
mentioning that lib/fonts/font_8x8.c and lib/fonts/font_8x16.c also
declare `fontdata_8x8` and `fontdata_8x16` respectively, and this file
has nothing to do with them.
Daniel Vetter [Fri, 23 Oct 2020 12:39:25 +0000 (14:39 +0200)]
drm/doc: Document legacy_cursor_update better
It's the horror and shouldn't be used. Realized we're not clear on
this in a discussion with Rob about what msm is doing to better
support async commits.
v2: Refine existing todo item to include this (Thomas)
Cc: Sean Paul <sean@poorly.run> Cc: Rob Clark <robdclark@gmail.com> Acked-by: Rob Clark <robdclark@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Maxime Ripard <mripard@kernel.org> Cc: Thomas Zimmermann <tzimmermann@suse.de> Cc: David Airlie <airlied@linux.ie> Cc: Daniel Vetter <daniel@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20201023123925.2374863-3-daniel.vetter@ffwll.ch
Arnd Bergmann [Mon, 26 Oct 2020 19:39:55 +0000 (20:39 +0100)]
matroxfb: avoid -Warray-bounds warning
The open-coded list_for_each_entry() causes a harmless warning:
drivers/video/fbdev/matrox/matroxfb_base.c: In function 'matroxfb_register_driver':
include/linux/kernel.h:856:3: warning: array subscript -98 is outside array bounds of 'struct list_head[1]' [-Warray-bounds]
Simon Ser [Mon, 26 Oct 2020 13:12:30 +0000 (14:12 +0100)]
drm: deprecate DRM_FORMAT_MOD_NONE
DRM_FORMAT_MOD_NONE is in the list of vendors, which is pretty
confusing. We already have DRM_FORMAT_MOD_VENDOR_NONE. Move it down in
the list of format modifiers.
DRM_FORMAT_MOD_NONE is an alias for DRM_FORMAT_MOD_LINEAR, however the
name is confusing: NONE doesn't mean that the modifier is implicit,
instead it means that the layout is linear. Deprecate it.
Christian König [Mon, 19 Oct 2020 16:49:27 +0000 (18:49 +0200)]
drm/ttm: nuke ttm_tt_set_(un)populated again
Neither page allocation backend nor the driver should mess with that.
Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-by: Madhav Chauhan <madhav.chauhan@amd.com> Link: https://patchwork.freedesktop.org/patch/396948/
The error happens when the pixels value is calculated before performing the sanity checks on bits_per_pixel.
A bits_per_pixel set to zero causes divide by zero error.
This patch moves the calculation after the sanity check.
Maxime Ripard [Tue, 7 Jul 2020 10:19:12 +0000 (12:19 +0200)]
drm/vc4: dsi: Only register our component once a DSI device is attached
If the DSI driver is the last to probe, component_add will try to run all
the bind callbacks straight away and return the error code.
However, since we depend on a power domain, we're pretty much guaranteed to
be in that case on the BCM2711, and are just lucky on the previous SoCs
since the v3d also depends on that power domain and is further in the probe
order.
In that case, the DSI host will not stick around in the system: the DSI
bind callback will be executed, will not find any DSI device attached and
will return EPROBE_DEFER, and we will then remove the DSI host and ask to
be probed later on.
But since that host doesn't stick around, DSI devices like the RaspberryPi
touchscreen whose probe is not linked to the DSI host (unlike the usual DSI
devices that will be probed through the call to mipi_dsi_host_register)
cannot attach to the DSI host, and we thus end up in a situation where the
DSI host cannot probe because the panel hasn't probed yet, and the panel
cannot probe because the DSI host hasn't yet.
In order to break this cycle, let's wait until there's a DSI device that
attaches to the DSI host to register the component and allow to progress
further.
Simon Ser [Thu, 22 Oct 2020 10:34:44 +0000 (10:34 +0000)]
drm: document that user-space should avoid parsing EDIDs
User-space should avoid parsing EDIDs for metadata already exposed via
other KMS interfaces and properties. For instance, user-space should not
try to extract a list of modes from the EDID: the kernel might mutate
the mode list (because of link capabilities or quirks for instance).
Other metadata not exposed by KMS can be parsed by user-space. This
includes for instance monitor identification (make/model/serial) and
supported color-spaces.
v2: add short explanation why user-space shouldn't do this (Brian)
Dave Airlie [Wed, 21 Oct 2020 04:40:29 +0000 (14:40 +1000)]
drm/ttm: replace last move_notify with delete_mem_notify
The move notify callback is only used in one place, this should
be removed in the future, but for now just rename it to the use
case which is to notify the driver that the GPU memory is to be
deleted.
Matteo Franchin [Mon, 12 Oct 2020 16:40:43 +0000 (17:40 +0100)]
drm/fourcc: Add AXBXGXRX106106106106 format
Add ABGR format with 10-bit components packed in 64-bit per pixel.
This format can be used to handle
VK_FORMAT_R10X6G10X6B10X6A10X6_UNORM_4PACK16 on little-endian
architectures.
Robin Murphy [Thu, 3 Sep 2020 20:36:13 +0000 (21:36 +0100)]
drm/komeda: Drop local dma_parms
Since commit 75a877d1eaa6 ("driver core: platform: Initialize dma_parms
for platform devices"), struct platform_device already provides a
dma_parms structure, so we can save allocating another one.
Daniel Vetter [Fri, 9 Oct 2020 23:21:56 +0000 (01:21 +0200)]
drm/vkms: fbdev emulation support
Hooray for generic fbdev support, making this a oneliner. We just
needed to fix preferred_depth fixed and the vmap support added first.
This is useful for testing fbdev testcases in igt. Right now we only
have a simple one in the fbdev testcases, which passes both info and
mmap subtests.
v2: Augment commit message as suggested by Melissa.
Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Reviewed-by: Melissa Wen <melissa.srw@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com> Cc: Melissa Wen <melissa.srw@gmail.com> Cc: Haneen Mohammed <hamohammed.sa@gmail.com> Cc: Daniel Vetter <daniel@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20201009232156.3916879-3-daniel.vetter@ffwll.ch
Neil Armstrong [Tue, 15 Sep 2020 12:19:12 +0000 (14:19 +0200)]
drm: panel: add TDO tl070wsh30 panel driver
This adds support for the TDO TL070WSH30 TFT-LCD panel module.
The panel has a 1024×600 resolution and uses 24 bit RGB per pixel.
It provides a MIPI DSI interface to the host, a built-in LED backlight
and touch controller.
Bernard Zhao [Fri, 14 Aug 2020 08:30:19 +0000 (01:30 -0700)]
drm/via: reduce no need mutex_lock area
In function via_mem_alloc`s error branch, DRM_ERROR is protected
in the mutex_lock(&dev->struct_mutex) area.
>From the code, we see that DRM_ERROR is just an error log print
without any struct element, there is no need to protect this.
drm/bridge: ti-sn65dsi86: Add retries for link training
On some panels hooked up to the ti-sn65dsi86 bridge chip we found that
link training was failing. Specifically, we'd see:
ti_sn65dsi86 2-002d: [drm:ti_sn_bridge_enable] *ERROR* Link training failed, link is off (-5)
The panel was hooked up to a logic analyzer and it was found that, as
part of link training, the bridge chip was writing a 0x1 to DPCD
address 00600h and the panel responded NACK. As can be seen in header
files, the write of 0x1 to DPCD address 0x600h means we were trying to
write the value DP_SET_POWER_D0 to the register DP_SET_POWER. The
panel vendor says that a NACK in this case is not unexpected and means
"not ready, try again".
In testing, we found that this panel would respond with a NACK in
about 1/25 times. Adding the retry logic worked fine and the most
number of tries needed was 3. Just to be safe, we'll add 10 tries
here and we'll add a little blurb to the logs if we ever need more
than 5.
Guido Günther [Tue, 13 Oct 2020 10:32:47 +0000 (12:32 +0200)]
drm/panel: mantix: Fix panel reset
The mantix panel needs two reset lines (RESX and TP_RSTN) deasserted to
output an image. Only deasserting RESX is not enough and the display
will stay blank. Deassert in prepare() and assert in unprepare() to keep
device held in reset when off.
Xu Wang [Wed, 14 Oct 2020 08:57:22 +0000 (08:57 +0000)]
video: fbdev: sh_mobile_lcdcfb: Remove redundant null check before clk_prepare_enable/clk_disable_unprepare
Because clk_prepare_enable() and clk_disable_unprepare() already checked
NULL clock parameter, so the additional checks are unnecessary, just
remove them.
Xu Wang [Wed, 14 Oct 2020 08:49:20 +0000 (08:49 +0000)]
omapfb/dss: Remove redundant null check before clk_prepare_enable/clk_disable_unprepare
Because clk_prepare_enable() and clk_disable_unprepare() already checked
NULL clock parameter, so the additional checks are unnecessary, just
remove them.
Jianxin Xiong [Wed, 14 Oct 2020 16:16:01 +0000 (09:16 -0700)]
dma-buf: Clarify that dma-buf sg lists are page aligned
The dma-buf API have been used under the assumption that the sg lists
returned from dma_buf_map_attachment() are fully page aligned. Lots of
stuff can break otherwise all over the place. Clarify this in the
documentation and add a check when DMA API debug is enabled.
Zwane Mwaikambo [Tue, 13 Oct 2020 05:59:14 +0000 (22:59 -0700)]
drm/dp_aux_dev: check aux_dev before use in drm_dp_aux_dev_get_by_minor()
I observed this when unplugging a DP monitor whilst a computer is asleep
and then waking it up. This left DP chardev nodes still being present on
the filesystem and accessing these device nodes caused an oops because
drm_dp_aux_dev_get_by_minor() assumes a device exists if it is opened.
This can also be reproduced by creating a device node with mknod(1) and
issuing an open(2)
Looking at the caller, checks on the minor are pushed down to
drm_dp_aux_dev_get_by_minor()
static int auxdev_open(struct inode *inode, struct file *file)
{
unsigned int minor = iminor(inode);
struct drm_dp_aux_dev *aux_dev;
aux_dev = drm_dp_aux_dev_get_by_minor(minor); <====
if (!aux_dev)
return -ENODEV;
file->private_data = aux_dev;
return 0;
}
Fixes: 520434106a9f ("drm/dp: Add a drm_aux-dev module for reading/writing dpcd registers.") Cc: <stable@vger.kernel.org> # v4.6+ Signed-off-by: Zwane Mwaikambo <zwane@yosper.io> Reviewed-by: Lyude Paul <lyude@redhat.com>
[added Cc to stable] Signed-off-by: Lyude Paul <lyude@redhat.com> Link: https://patchwork.freedesktop.org/patch/msgid/alpine.DEB.2.21.2010122231070.38717@montezuma.home
Novatek NT36672a is a generic DSI IC that drives command and video mode
panels. Add the driver for it.
Right now adding support for some Poco F1 phones that have an LCD panel
from Tianma connected with this IC, with a resolution of 1080x2246 that
operates in DSI video mode.
During testing, Benni Steini <bennisteinir@gmail.com> helped us fix
the reset sequence timing (from 10ms to 20ms), to get the bootanimation
to work on Android.
With current AOSP, we need to increase it to 200ms - this seems to be a
safe high value to avoid a white screen occasionally.
dt-bindings: display: panel: Add bindings for Novatek nt36672a
Novatek nt36672a is a display driver IC that can drive DSI panel. It
is also present in the Tianma video mode panel, which is a FHD+ panel
with a resolution of 1080x2246 and 6.18 inches size. It is found in
some of the Poco F1 phones.
This patch adds the display driver for the IC, with support added for
this tianma fhd video mode panel.
projects / kernel.git / log