rust: drm: gpuvm: Add a missing lock#433
Open
WhatAmISupposedToPutHere wants to merge 1468 commits intoAsahiLinux:asahi-wipfrom
Open
rust: drm: gpuvm: Add a missing lock#433WhatAmISupposedToPutHere wants to merge 1468 commits intoAsahiLinux:asahi-wipfrom
WhatAmISupposedToPutHere wants to merge 1468 commits intoAsahiLinux:asahi-wipfrom
Conversation
DCP best scored color mode might result in an HDR mode. As long as the driver (and DRM) is not ready for HDR try to avoid such modes. Signed-off-by: Janne Grunau <j@jannau.net>
Probably not important but avoids an unnecessary difference compared to macOS. Signed-off-by: Janne Grunau <j@jannau.net>
Can be used on devices with camera notch to use the full display height and thus show the notch. Signed-off-by: Janne Grunau <j@jannau.net>
This works on both 8-bit and 10-bit modes without any weirdness, and gives us the native colorspace without any conversion. Color correction should probably be handled in software anyway. However, we need to use surface 1 (at least on t600x), since 0 seems stuck in bg-sRGB mode for some reason... Signed-off-by: Hector Martin <marcan@marcan.st>
With "drm/apple: Enable 10-bit mode & set colorspace to native" kernel log messages are shown in an Apple logo shaped region in the middle of the display when using BGRA. The field currently identified as "opaque" is mislabeled and has to be investigated further. Signed-off-by: Janne Grunau <j@jannau.net>
- opaque -> is_premultiplied - swap_enabled BIT(31) seems to be update background with dcp_swap.bg_color - add unused fields is_tearing_allowed, ycbcr_matrix, protection_opts, unk_num, unk_denom Changes: use is_premultiplied only for XRGB8/XBGR8, Update background only when necessary. Signed-off-by: Janne Grunau <j@jannau.net>
This does not seem to be as racy as drm_aperture_remove_framebuffers() and seems to reliably takes over simpledrm's device node. Signed-off-by: Janne Grunau <j@jannau.net>
This check for the "nomodeset" kernel command line parameter in its register method. Signed-off-by: Janne Grunau <j@jannau.net>
drm's documentaion explicitly tells us not to use devm_kzalloc(). drm device structures might out live the device when they are in use by userspace while the device vanishes.
There was a report of a race between DRM device registration (and removal of the simpledrm device) and GDM startup. The component based device binding ensures that all necessary devices are bind in the probe method of the last missing component. Technically the piodma-mapper should be a component of dcp but since it is only used for its iommu it can be a component of the display subsystem. Signed-off-by: Janne Grunau <j@jannau.net>
Avoids "[drm] Cannot find any crtc or sizes" during fbdev initialization if a display is connected. Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Asahi Lina <lina@asahilina.net>
Signed-off-by: Asahi Lina <lina@asahilina.net>
Fixes following warning when systemd-backlight restores the backlight level on boot before a mode is set: Call trace: drm_atomic_helper_crtc_duplicate_state+0x58/0x74 drm_atomic_get_crtc_state+0x84/0x120 dcp_set_brightness+0xd8/0x21c [apple_dcp] backlight_device_set_brightness+0x78/0x130 ... Signed-off-by: Janne Grunau <j@jannau.net>
Backlight drivers are expected to use this instead of accessing backlight properties. Signed-off-by: Janne Grunau <j@jannau.net>
Fixes overwriting the panel's physical dimensions on poweroff/sleep. Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Asahi Lina <lina@asahilina.net>
Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Janne Grunau <j@jannau.net>
This adds support for multiple incompatible DCP firmware versions. The approach taken here duplicates more than necessary. Unmodified calls do not need to be templated. For simplicity and in the expectation that more calls and callbacks are modified in the future everything is templated. Signed-off-by: Janne Grunau <j@jannau.net>
MacOS 13.2 does the same and it is unclear if surf[3] can be used at all. PRobably not necessary but found during debugging to firmware 13.2. Signed-off-by: Janne Grunau <j@jannau.net>
kwin 5.27.3 adds support for "Night Color" via drm "CTM" properties. Wire CTM support up via the "set_matrix" iomfb call. Link: https://bugs.kde.org/show_bug.cgi?id=455720 Signed-off-by: Janne Grunau <j@jannau.net>
Depends on https://gitlab.freedesktop.org/asahi/mesa/-/merge_requests/5 Signed-off-by: Janne Grunau <j@jannau.net>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Janne Grunau <j@jannau.net>
Under certain conditions (more prevalent after a suspend/resume cycle), the touchscreen controller can send the "boot complete" interrupt before it actually finished booting. In those cases, attempting to read touch data resuls in a stream of "not ready" messages being read and interpreted as a touch report. Check that the response is in fact a touch report and discard it otherwise. Reported-by: pitust <piotr@stelmaszek.com> Closes: https://oftc.catirclogs.org/asahi/2025-12-17#34878715; Fixes: 471a92f ("Input: apple_z2 - add a driver for Apple Z2 touchscreens") Signed-off-by: Sasha Finkelstein <fnkl.kernel@gmail.com>
The next display might not the same and the EDID will not necessarily be replaced on the next connect. This results in confused user space with hilarious but broken results. In kwin it resulted in a gap between two displays makling it impossible to move the mouse pointer to the other display. Signed-off-by: Janne Grunau <j@jannau.net>
This works around missing (or incomplete) suspend/resume handling in atc/dcp for the the HDMI output on 14 and 16-inch Macbook Pros. Signed-off-by: Janne Grunau <j@jannau.net>
Lockdep complains otherwise: ------------[ cut here ]------------ WARNING: CPU: 5 PID: 885 at drivers/gpu/drm/drm_gpuvm.c:1620 drm_gpuvm_bo_put+0x1b4/0x254 Modules linked in: brcmfmac_wcc uhid overlay squashfs zlib_inflate brcmfmac hci_bcm4377 brcmutil spi_nor aop_las aop_als industrialio cfg80211 fuse nfn> CPU: 5 UID: 1000 PID: 885 Comm: kwin_wayland Tainted: G S W 6.18.2+ AsahiLinux#5 PREEMPTLAZY Tainted: [S]=CPU_OUT_OF_SPEC, [W]=WARN Hardware name: Apple MacBook Pro (14-inch, M1 Pro, 2021) (DT) pstate: 61400009 (nZCv daif +PAN -UAO -TCO +DIT -SSBS BTYPE=--) pc : drm_gpuvm_bo_put+0x1b4/0x254 lr : drm_gpuvm_bo_put+0x14c/0x254 sp : ffff8000893d7c10 x29: ffff8000893d7c10 x28: ffff00001b8ef9c0 x27: 0000000000000000 x26: 0000000000000002 x25: ffff800081451000 x24: dead000000000100 x23: ffff800080ee82d0 x22: ffff0000108f9d50 x21: ffff0000108f9c00 x20: ffff0000492e0700 x19: ffff000048700000 x18: 0000000000000000 x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000 x14: 0000000000000008 x13: 0000000000000000 x12: ffff8000815e34d0 x11: 0000000000000001 x10: 00000000ffffffff x9 : 0000000100000000 x8 : 0000000000000000 x7 : 0000000000000000 x6 : ffff8000807d2bcc x5 : 0000000000000000 x4 : 0000000000000000 x3 : 0000000000000000 x2 : 0000000000000000 x1 : ffff000048700258 x0 : 0000000000000000 Call trace: drm_gpuvm_bo_put+0x1b4/0x254 (P) _RINvNtCs1tcwcP3FgYC_4core3ptr13drop_in_placeINtNtNtCsgIauPoi8ikU_6kernel3drm2mm8NodeDatauNtNtCshTJcMxhWd7O_5asahi3mmu18KernelMappingInnerEEB1t_+0xb0/> _RINvNtCs1tcwcP3FgYC_4core3ptr13drop_in_placeNtNtCshTJcMxhWd7O_5asahi4file2VmEBK_+0xcc/0xf0 _RNvMNtNtCsgIauPoi8ikU_6kernel3drm4fileINtB2_4FileNtNtCshTJcMxhWd7O_5asahi4file4FileE18postclose_callbackBP_+0xac/0x2f4 drm_file_free+0x1b8/0x210 drm_release+0xb8/0x140 __fput+0xf8/0x2e4 fput_close_sync+0x44/0x114 __arm64_sys_close+0xb0/0xfc invoke_syscall+0x48/0xc8 do_el0_svc+0x7c/0xa8 el0_svc+0x3c/0xd8 el0t_64_sync_handler+0x68/0xdc el0t_64_sync+0x198/0x19c ---[ end trace 0000000000000000 ]--- Signed-off-by: Sasha Finkelstein <fnkl.kernel@gmail.com>
Member
|
manually merged to my local bits/210-gpu iirc I tried to replace the resv lock here with the gpuva lock which resulting in the object being gone at unlock time. The dma_resv_lock probably prevents this. The safety comment is non-sense but this is going to change anyway when we rebase onto the upstream gpuvm bindings. |
Member
|
This still causea NULL pointer derefs under some conditions. Probably the same why it was missing in the first place. I hope this will be fixed by switching to Alice upstream submitted GPUVM abstractions. In the meantime I'll revert. |
jannau
force-pushed
the
asahi-wip
branch
2 times, most recently
from
d9f615d to
33ef1b6
Compare
herrnst
pushed a commit
to herrnst/linux-asahi
that referenced
this pull request
Feb 5, 2026
This reverts commit 0475333. due to NULL ptr deref, see AsahiLinux/linux#433 Signed-off-by: Janne Grunau <j@jannau.net>
jannau
force-pushed
the
asahi-wip
branch
2 times, most recently
from
0031cb6 to
79a307d
Compare
jannau
force-pushed
the
asahi-wip
branch
2 times, most recently
from
7457799 to
89c915d
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
14 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Lockdep complains otherwise:
------------[ cut here ]------------
WARNING: CPU: 5 PID: 885 at drivers/gpu/drm/drm_gpuvm.c:1620 drm_gpuvm_bo_put+0x1b4/0x254
Modules linked in: brcmfmac_wcc uhid overlay squashfs zlib_inflate brcmfmac hci_bcm4377 brcmutil spi_nor aop_las aop_als industrialio cfg80211 fuse nfn>
CPU: 5 UID: 1000 PID: 885 Comm: kwin_wayland Tainted: G S W 6.18.2+ #5 PREEMPTLAZY
Tainted: [S]=CPU_OUT_OF_SPEC, [W]=WARN
Hardware name: Apple MacBook Pro (14-inch, M1 Pro, 2021) (DT)
pstate: 61400009 (nZCv daif +PAN -UAO -TCO +DIT -SSBS BTYPE=--)
pc : drm_gpuvm_bo_put+0x1b4/0x254
lr : drm_gpuvm_bo_put+0x14c/0x254
sp : ffff8000893d7c10
x29: ffff8000893d7c10 x28: ffff00001b8ef9c0 x27: 0000000000000000
x26: 0000000000000002 x25: ffff800081451000 x24: dead000000000100
x23: ffff800080ee82d0 x22: ffff0000108f9d50 x21: ffff0000108f9c00
x20: ffff0000492e0700 x19: ffff000048700000 x18: 0000000000000000
x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000
x14: 0000000000000008 x13: 0000000000000000 x12: ffff8000815e34d0
x11: 0000000000000001 x10: 00000000ffffffff x9 : 0000000100000000
x8 : 0000000000000000 x7 : 0000000000000000 x6 : ffff8000807d2bcc
x5 : 0000000000000000 x4 : 0000000000000000 x3 : 0000000000000000
x2 : 0000000000000000 x1 : ffff000048700258 x0 : 0000000000000000
Call trace:
drm_gpuvm_bo_put+0x1b4/0x254 (P)
RINvNtCs1tcwcP3FgYC_4core3ptr13drop_in_placeINtNtNtCsgIauPoi8ikU_6kernel3drm2mm8NodeDatauNtNtCshTJcMxhWd7O_5asahi3mmu18KernelMappingInnerEEB1t+0xb0/>
RINvNtCs1tcwcP3FgYC_4core3ptr13drop_in_placeNtNtCshTJcMxhWd7O_5asahi4file2VmEBK+0xcc/0xf0
RNvMNtNtCsgIauPoi8ikU_6kernel3drm4fileINtB2_4FileNtNtCshTJcMxhWd7O_5asahi4file4FileE18postclose_callbackBP+0xac/0x2f4
drm_file_free+0x1b8/0x210
drm_release+0xb8/0x140
__fput+0xf8/0x2e4
fput_close_sync+0x44/0x114
__arm64_sys_close+0xb0/0xfc
invoke_syscall+0x48/0xc8
do_el0_svc+0x7c/0xa8
el0_svc+0x3c/0xd8
el0t_64_sync_handler+0x68/0xdc
el0t_64_sync+0x198/0x19c
---[ end trace 0000000000000000 ]---