linux-2.6-microblaze.git
8 months agoMerge tag 'drm-misc-next-2024-02-29' into msm-next
Rob Clark [Mon, 4 Mar 2024 02:31:21 +0000 (18:31 -0800)]
Merge tag 'drm-misc-next-2024-02-29' into msm-next

Merge to pick up commit 47f419e07111 ("drm/dp: move
intel_dp_vsc_sdp_pack() to generic helper")

drm-misc-next for v6.9:

UAPI Changes:

Cross-subsystem Changes:

backlight:
- corgi: include backlight header

fbdev:
- Cleanup includes in public header file
- fbtft: Include backlight header

Core Changes:

edid:
- Remove built-in EDID data

dp:
- Avoid AUX transfers on powered-down displays
- Add VSC SDP helpers

modesetting:
- Add sanity checks for polling
- Cleanups

scheduler:
- Cleanups

tests:
- Add helpers for mode-setting tests

Driver Changes:

i915:
- Use shared VSC SDP helper

mgag200:
- Work around PCI write bursts

mxsfb:
- Use managed mode config

nouveau:
- Include backlight header where necessary

qiac:
- Cleanups

sun4:
- HDMI: updates to atomic mode setting

tegra:
- Fix GEM refounting in error paths

tidss:
- Fix multi display
- Fix initial Z position

v3d:
- Support display MMU page size

Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/dp: Don't attempt AUX transfers when eDP panels are not powered
Douglas Anderson [Fri, 2 Feb 2024 22:11:16 +0000 (14:11 -0800)]
drm/dp: Don't attempt AUX transfers when eDP panels are not powered

If an eDP panel is not powered on then any attempts to talk to it over
the DP AUX channel will timeout. Unfortunately these attempts may be
quite slow. Userspace can initiate these attempts either via a
/dev/drm_dp_auxN device or via the created i2c device.

Making the DP AUX drivers timeout faster is a difficult proposition.
In theory we could just poll the panel's HPD line in the AUX transfer
function and immediately return an error there. However, this is
easier said than done. For one thing, there's no hard requirement to
hook the HPD line up for eDP panels and it's OK to just delay a fixed
amount. For another thing, the HPD line may not be fast to probe. On
parade-ps8640 we need to wait for the bridge chip's firmware to boot
before we can get the HPD line and this is a slow process.

The fact that the transfers are taking so long to timeout is causing
real problems. The open source fwupd daemon sometimes scans DP busses
looking for devices whose firmware need updating. If it happens to
scan while a panel is turned off this scan can take a long time. The
fwupd daemon could try to be smarter and only scan when eDP panels are
turned on, but we can also improve the behavior in the kernel.

Let's let eDP panels drivers specify that a panel is turned off and
then modify the common AUX transfer code not to attempt a transfer in
this case.

Tested-by: Steev Klimaszewski <steev@kali.org>
Reviewed-by: Hsin-Yi Wang <hsinyi@chromium.org>
Tested-by: Eizan Miyamoto <eizan@chromium.org>
Acked-by: Neil Armstrong <neil.armstrong@linaro.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240202141109.1.I24277520ac754ea538c9b14578edc94e1df11b48@changeid
8 months agodrm/tests: connector: Add tests for drmm_connector_init
Maxime Ripard [Thu, 22 Feb 2024 18:13:51 +0000 (19:13 +0100)]
drm/tests: connector: Add tests for drmm_connector_init

drmm_connector_init is the preferred function to initialize a
drm_connector structure. Let's add a bunch of unit tests for it.

Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-5-8f4af575fce2@kernel.org
8 months agodrm/tests: Add helper to create mock crtc
Maxime Ripard [Thu, 22 Feb 2024 18:13:50 +0000 (19:13 +0100)]
drm/tests: Add helper to create mock crtc

We're going to need a full-blown, functional, KMS device to test more
components of the atomic modesetting infrastructure.

Let's add a new helper to create a dumb, mocked, CRTC. By default it
will create a CRTC relying only on the default helpers, but drivers are
free to deviate from that.

Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-4-8f4af575fce2@kernel.org
8 months agodrm/tests: Add helper to create mock plane
Maxime Ripard [Thu, 22 Feb 2024 18:13:49 +0000 (19:13 +0100)]
drm/tests: Add helper to create mock plane

We're going to need a full-blown, functional, KMS device to test more
components of the atomic modesetting infrastructure.

Let's add a new helper to create a dumb, mocked, primary plane. By
default, it will create a linear XRGB8888 plane, using the default
helpers.

Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-3-8f4af575fce2@kernel.org
8 months agodrm/tests: helpers: Add atomic helpers
Maxime Ripard [Thu, 22 Feb 2024 18:13:48 +0000 (19:13 +0100)]
drm/tests: helpers: Add atomic helpers

The mock device we were creating was missing any of the driver-wide
helpers. That was fine before since we weren't testing the atomic state
path, but we're going to start, so let's use the default
implementations.

Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-2-8f4af575fce2@kernel.org
8 months agodrm/tests: helpers: Include missing drm_drv header
Maxime Ripard [Thu, 22 Feb 2024 18:13:47 +0000 (19:13 +0100)]
drm/tests: helpers: Include missing drm_drv header

We have a few functions declared in our kunit helpers header, some of
them dereferencing the struct drm_driver.

However, we don't include the drm_drv.h header file defining that
structure, leading to compilation errors if we don't include both
headers.

Fixes: d98780310719 ("drm/tests: helpers: Allow to pass a custom drm_driver")
Reviewed-by: Maíra Canal <mcanal@igalia.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-1-8f4af575fce2@kernel.org
8 months agodrm/scheduler: Simplify the allocation of slab caches in drm_sched_fence_slab_init
Kunwu Chan [Wed, 21 Feb 2024 08:55:58 +0000 (16:55 +0800)]
drm/scheduler: Simplify the allocation of slab caches in drm_sched_fence_slab_init

Use the new KMEM_CACHE() macro instead of direct kmem_cache_create
to simplify the creation of SLAB caches.

Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20240221085558.166774-1-chentao@kylinos.cn
8 months agodrm: Check polling initialized before enabling in drm_helper_probe_single_connector_modes
Shradha Gupta [Fri, 2 Feb 2024 06:43:44 +0000 (22:43 -0800)]
drm: Check polling initialized before enabling in drm_helper_probe_single_connector_modes

In function drm_helper_probe_single_connector_modes() when we enable
polling again, if it is already uninitialized, a warning is reported.
This patch fixes the warning message by checking if poll is initialized
before enabling it.

Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202401191128.db8423f1-oliver.sang@intel.com
Signed-off-by: Shradha Gupta <shradhagupta@linux.microsoft.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1706856224-9725-1-git-send-email-shradhagupta@linux.microsoft.com
8 months agodrm: Check output polling initialized before disabling
Shradha Gupta [Fri, 2 Feb 2024 06:43:28 +0000 (22:43 -0800)]
drm: Check output polling initialized before disabling

In drm_kms_helper_poll_disable() check if output polling
support is initialized before disabling polling. If not flag
this as a warning.
Additionally in drm_mode_config_helper_suspend() and
drm_mode_config_helper_resume() calls, that re the callers of these
functions, avoid invoking them if polling is not initialized.
For drivers like hyperv-drm, that do not initialize connector
polling, if suspend is called without this check, it leads to
suspend failure with following stack
[  770.719392] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done.
[  770.720592] printk: Suspending console(s) (use no_console_suspend to debug)
[  770.948823] ------------[ cut here ]------------
[  770.948824] WARNING: CPU: 1 PID: 17197 at kernel/workqueue.c:3162 __flush_work.isra.0+0x212/0x230
[  770.948831] Modules linked in: rfkill nft_counter xt_conntrack xt_owner udf nft_compat crc_itu_t nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ip_set nf_tables nfnetlink vfat fat mlx5_ib ib_uverbs ib_core mlx5_core intel_rapl_msr intel_rapl_common kvm_amd ccp mlxfw kvm psample hyperv_drm tls drm_shmem_helper drm_kms_helper irqbypass pcspkr syscopyarea sysfillrect sysimgblt hv_balloon hv_utils joydev drm fuse xfs libcrc32c pci_hyperv pci_hyperv_intf sr_mod sd_mod cdrom t10_pi sg hv_storvsc scsi_transport_fc hv_netvsc serio_raw hyperv_keyboard hid_hyperv crct10dif_pclmul crc32_pclmul crc32c_intel hv_vmbus ghash_clmulni_intel dm_mirror dm_region_hash dm_log dm_mod
[  770.948863] CPU: 1 PID: 17197 Comm: systemd-sleep Not tainted 5.14.0-362.2.1.el9_3.x86_64 #1
[  770.948865] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 05/09/2022
[  770.948866] RIP: 0010:__flush_work.isra.0+0x212/0x230
[  770.948869] Code: 8b 4d 00 4c 8b 45 08 89 ca 48 c1 e9 04 83 e2 08 83 e1 0f 83 ca 02 89 c8 48 0f ba 6d 00 03 e9 25 ff ff ff 0f 0b e9 4e ff ff ff <0f> 0b 45 31 ed e9 44 ff ff ff e8 8f 89 b2 00 66 66 2e 0f 1f 84 00
[  770.948870] RSP: 0018:ffffaf4ac213fb10 EFLAGS: 00010246
[  770.948871] RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffff8c992857
[  770.948872] RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffff9aad82b00330
[  770.948873] RBP: ffff9aad82b00330 R08: 0000000000000000 R09: ffff9aad87ee3d10
[  770.948874] R10: 0000000000000200 R11: 0000000000000000 R12: ffff9aad82b00330
[  770.948874] R13: 0000000000000001 R14: 0000000000000000 R15: 0000000000000001
[  770.948875] FS:  00007ff1b2f6bb40(0000) GS:ffff9aaf37d00000(0000) knlGS:0000000000000000
[  770.948878] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  770.948878] CR2: 0000555f345cb666 CR3: 00000001462dc005 CR4: 0000000000370ee0
[  770.948879] Call Trace:
[  770.948880]  <TASK>
[  770.948881]  ? show_trace_log_lvl+0x1c4/0x2df
[  770.948884]  ? show_trace_log_lvl+0x1c4/0x2df
[  770.948886]  ? __cancel_work_timer+0x103/0x190
[  770.948887]  ? __flush_work.isra.0+0x212/0x230
[  770.948889]  ? __warn+0x81/0x110
[  770.948891]  ? __flush_work.isra.0+0x212/0x230
[  770.948892]  ? report_bug+0x10a/0x140
[  770.948895]  ? handle_bug+0x3c/0x70
[  770.948898]  ? exc_invalid_op+0x14/0x70
[  770.948899]  ? asm_exc_invalid_op+0x16/0x20
[  770.948903]  ? __flush_work.isra.0+0x212/0x230
[  770.948905]  __cancel_work_timer+0x103/0x190
[  770.948907]  ? _raw_spin_unlock_irqrestore+0xa/0x30
[  770.948910]  drm_kms_helper_poll_disable+0x1e/0x40 [drm_kms_helper]
[  770.948923]  drm_mode_config_helper_suspend+0x1c/0x80 [drm_kms_helper]
[  770.948933]  ? __pfx_vmbus_suspend+0x10/0x10 [hv_vmbus]
[  770.948942]  hyperv_vmbus_suspend+0x17/0x40 [hyperv_drm]
[  770.948944]  ? __pfx_vmbus_suspend+0x10/0x10 [hv_vmbus]
[  770.948951]  dpm_run_callback+0x4c/0x140
[  770.948954]  __device_suspend_noirq+0x74/0x220
[  770.948956]  dpm_noirq_suspend_devices+0x148/0x2a0
[  770.948958]  dpm_suspend_end+0x54/0xe0
[  770.948960]  create_image+0x14/0x290
[  770.948963]  hibernation_snapshot+0xd6/0x200
[  770.948964]  hibernate.cold+0x8b/0x1fb
[  770.948967]  state_store+0xcd/0xd0
[  770.948969]  kernfs_fop_write_iter+0x124/0x1b0
[  770.948973]  new_sync_write+0xff/0x190
[  770.948976]  vfs_write+0x1ef/0x280
[  770.948978]  ksys_write+0x5f/0xe0
[  770.948979]  do_syscall_64+0x5c/0x90
[  770.948981]  ? syscall_exit_work+0x103/0x130
[  770.948983]  ? syscall_exit_to_user_mode+0x12/0x30
[  770.948985]  ? do_syscall_64+0x69/0x90
[  770.948986]  ? do_syscall_64+0x69/0x90
[  770.948987]  ? do_user_addr_fault+0x1d6/0x6a0
[  770.948989]  ? do_syscall_64+0x69/0x90
[  770.948990]  ? exc_page_fault+0x62/0x150
[  770.948992]  entry_SYSCALL_64_after_hwframe+0x72/0xdc
[  770.948995] RIP: 0033:0x7ff1b293eba7
[  770.949010] Code: 0b 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 48 89 54 24 18 48 89 74 24
[  770.949011] RSP: 002b:00007ffde3912128 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
[  770.949012] RAX: ffffffffffffffda RBX: 0000000000000005 RCX: 00007ff1b293eba7
[  770.949013] RDX: 0000000000000005 RSI: 00007ffde3912210 RDI: 0000000000000004
[  770.949014] RBP: 00007ffde3912210 R08: 000055d7dd4c9510 R09: 00007ff1b29b14e0
[  770.949014] R10: 00007ff1b29b13e0 R11: 0000000000000246 R12: 0000000000000005
[  770.949015] R13: 000055d7dd4c53e0 R14: 0000000000000005 R15: 00007ff1b29f69e0
[  770.949016]  </TASK>
[  770.949017] ---[ end trace e6fa0618bfa2f31d ]---

Built-on: Rhel9, Ubuntu22
Signed-off-by: Shradha Gupta <shradhagupta@linux.microsoft.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1706856208-9617-1-git-send-email-shradhagupta@linux.microsoft.com
8 months agodrm: Remove drm_num_crtcs() helper
Thierry Reding [Tue, 27 Feb 2024 11:20:38 +0000 (12:20 +0100)]
drm: Remove drm_num_crtcs() helper

The drm_num_crtcs() helper determines the number of CRTCs by iterating
over the list of CRTCs that have been registered with the mode config.
However, we already keep track of that number in the mode config's
num_crtcs field, so we can simply retrieve the value from that and
remove the extra helper function.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240227112038.411846-1-thierry.reding@gmail.com
8 months agofbdev: Clean up include statements in header file
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:34 +0000 (10:37 +0100)]
fbdev: Clean up include statements in header file

Include mutex.h, printk.h and types.h, remove several unnecessary
include statements, and sort the list alphabetically.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-10-tzimmermann@suse.de
8 months agofbdev: Clean up forward declarations in header file
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:33 +0000 (10:37 +0100)]
fbdev: Clean up forward declarations in header file

Add forward declarations for struct i2c_adapter and struct module, and
sort the list alphabetically.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-9-tzimmermann@suse.de
8 months agofbdev: Do not include <linux/slab.h> in header
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:32 +0000 (10:37 +0100)]
fbdev: Do not include <linux/slab.h> in header

Forward declare struct page and remove the include statement.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-8-tzimmermann@suse.de
8 months agofbdev: Do not include <linux/notifier.h> in header
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:31 +0000 (10:37 +0100)]
fbdev: Do not include <linux/notifier.h> in header

Forward declare struct notifier_block and remove the include
statement.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-7-tzimmermann@suse.de
8 months agofbdev: Do not include <linux/fs.h> in header
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:30 +0000 (10:37 +0100)]
fbdev: Do not include <linux/fs.h> in header

Forward declare struct inode and remove the include statement.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-6-tzimmermann@suse.de
8 months agofbdev: Do not include <linux/backlight.h> in header
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:29 +0000 (10:37 +0100)]
fbdev: Do not include <linux/backlight.h> in header

Forward declare struct backlight_device and remove the include
statement.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-5-tzimmermann@suse.de
8 months agostaging/fbtft: Include <linux/backlight.h>
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:28 +0000 (10:37 +0100)]
staging/fbtft: Include <linux/backlight.h>

Resolves the proxy include via <linux/fb.h>, which does not require the
backlight header.

v3:
* fix grammar in commit message

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-4-tzimmermann@suse.de
8 months agodrm/nouveau: Include <linux/backlight.h>
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:27 +0000 (10:37 +0100)]
drm/nouveau: Include <linux/backlight.h>

Resolves the proxy include via <linux/fb.h>, which does not require the
backlight header.

v3:
* fix grammar in commit message

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Helge Deller <deller@gmx.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-3-tzimmermann@suse.de
8 months agobacklight/corgi-lcd: Include <linux/backlight.h>
Thomas Zimmermann [Mon, 19 Feb 2024 09:37:26 +0000 (10:37 +0100)]
backlight/corgi-lcd: Include <linux/backlight.h>

Resolves the proxy include via <linux/fb.h>, which does not require the
backlight header.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Lee Jones <lee@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240219093941.3684-2-tzimmermann@suse.de
8 months agodrm/mgag200: Add a workaround for low-latency
Jocelyn Falempe [Thu, 8 Feb 2024 09:51:10 +0000 (10:51 +0100)]
drm/mgag200: Add a workaround for low-latency

We found a regression in v5.10 on real-time server, using the
rt-kernel and the mgag200 driver. It's some really specialized
workload, with <10us latency expectation on isolated core.
After the v5.10, the real time tasks missed their <10us latency
when something prints on the screen (fbcon or printk)

The regression has been bisected to 2 commits:
commit 0b34d58b6c32 ("drm/mgag200: Enable caching for SHMEM pages")
commit 4862ffaec523 ("drm/mgag200: Move vmap out of commit tail")

The first one changed the system memory framebuffer from Write-Combine
to the default caching.
Before the second commit, the mgag200 driver used to unmap the
framebuffer after each frame, which implicitly does a cache flush.
Both regressions are fixed by this commit, which restore WC mapping
for the framebuffer in system memory, and add a cache flush.
This is only needed on x86_64, for low-latency workload,
so the new kconfig DRM_MGAG200_IOBURST_WORKAROUND depends on
PREEMPT_RT and X86.

For more context, the whole thread can be found here [1]

Signed-off-by: Jocelyn Falempe <jfalempe@redhat.com>
Link: https://lore.kernel.org/dri-devel/20231019135655.313759-1-jfalempe@redhat.com/
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240208095125.377908-1-jfalempe@redhat.com
8 months agodrm/msm/adreno: Add A702 support
Konrad Dybcio [Fri, 23 Feb 2024 21:21:41 +0000 (22:21 +0100)]
drm/msm/adreno: Add A702 support

The A702 is a weird mix of 600 and 700 series.. Perhaps even a
testing ground for some A7xx features with good ol' A6xx silicon.
It's basically A610 that's been beefed up with some new registers
and hw features (like APRIV!), that was then cut back in size,
memory bus and some other ways.

Add support for it, tested with QCM2290 / RB1.

Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/579752/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm: Fix page fault client detection on a660 family and a7xx
Connor Abbott [Thu, 25 Jan 2024 13:10:58 +0000 (13:10 +0000)]
drm/msm: Fix page fault client detection on a660 family and a7xx

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/575918/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm: More fully implement devcoredump for a7xx
Connor Abbott [Thu, 25 Jan 2024 13:10:57 +0000 (13:10 +0000)]
drm/msm: More fully implement devcoredump for a7xx

Use the vendor-provided snapshot headers to dump the contextless
registers, shader blocks, and cluster registers. Still unimplemented are
the GMU registers and "external core" registers, which would require
more work because they use register spaces we don't have described in
devicetree and dump registers from multiple spaces in a single list.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/575919/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm: Fix snapshotting a7xx indexed regs
Connor Abbott [Thu, 25 Jan 2024 13:10:56 +0000 (13:10 +0000)]
drm/msm: Fix snapshotting a7xx indexed regs

We were overwriting the last indexed reg (CP_ROQ) and we were
snapshotting the same CP_MEMPOOL block twice instead of snapshotting
CP_BV_MEMPOOL as intended.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/575920/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm: Import a7xx crashdump register lists from kgsl
Connor Abbott [Thu, 25 Jan 2024 13:10:54 +0000 (13:10 +0000)]
drm/msm: Import a7xx crashdump register lists from kgsl

This imports these files as-is, the following commits will have to make
slight changes to get them to compile because downstream uses
un-namespaced enums that conflict with a6xx. However we should try as
much as possible to stick to downstream's format to make importing new
gens easier.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/575921/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm: add support for A750 GPU
Neil Armstrong [Fri, 16 Feb 2024 11:03:52 +0000 (12:03 +0100)]
drm/msm: add support for A750 GPU

Add support for the A750 GPU found on the SM8650 platform

Unlike the the very close A740 GPU on the SM8550 SoC, the A750 GPU
doesn't have an HWCFG block but a separate register set.

The A750 GPU info are added under the adreno_is_a750() macro and
the ADRENO_7XX_GEN3 family id.

Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/578693/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodt-bindings: arm-smmu: Document SM8650 GPU SMMU
Neil Armstrong [Fri, 16 Feb 2024 11:03:50 +0000 (12:03 +0100)]
dt-bindings: arm-smmu: Document SM8650 GPU SMMU

Document the GPU SMMU found on the SM8650 platform.

Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/578685/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodt-bindings: arm-smmu: fix SM8[45]50 GPU SMMU if condition
Neil Armstrong [Fri, 16 Feb 2024 11:03:49 +0000 (12:03 +0100)]
dt-bindings: arm-smmu: fix SM8[45]50 GPU SMMU if condition

The if condition for the SM8[45]50 GPU SMMU is too large,
add the other compatible strings to the condition to only
allow the clocks for the GPU SMMU nodes.

Fixes: 4fff78dc2490 ("dt-bindings: arm-smmu: Document SM8[45]50 GPU SMMU")
Suggested-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/578686/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodt-bindings: display/msm/gmu: Document Adreno 750 GMU
Neil Armstrong [Fri, 16 Feb 2024 11:03:48 +0000 (12:03 +0100)]
dt-bindings: display/msm/gmu: Document Adreno 750 GMU

Document the Adreno 750 GMU found on the SM8650 platform.

Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/578684/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm/adreno: Add A305B support
Luca Weiss [Sun, 21 Jan 2024 10:40:39 +0000 (11:40 +0100)]
drm/msm/adreno: Add A305B support

Add support for the Adreno 305B GPU that is found in MSM8226(v2) SoC.
Previously this was mistakenly claimed to be supported but using wrong
a configuration.

In MSM8226v1 there's also a A305B but with chipid 0x03000510 which
should work with the same configuration but due to lack of hardware for
testing this is not added.

Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Luca Weiss <luca@z3ntu.xyz>
Reviewed-by: David Heidelberg <david@ixit.cz>
Patchwork: https://patchwork.freedesktop.org/patch/575274/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodt-bindings: display/msm: gpu: Allow multiple digits for patchid
Luca Weiss [Sun, 21 Jan 2024 10:40:38 +0000 (11:40 +0100)]
dt-bindings: display/msm: gpu: Allow multiple digits for patchid

Some GPUs like the Adreno A305B has a patchid higher than 9, in this
case 18. Make sure the regexes can account for that.

Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Luca Weiss <luca@z3ntu.xyz>
Reviewed-by: David Heidelberg <david@ixit.cz>
Patchwork: https://patchwork.freedesktop.org/patch/575272/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm/a7xx: Fix LLC typo
Rob Clark [Tue, 2 Jan 2024 19:33:45 +0000 (11:33 -0800)]
drm/msm/a7xx: Fix LLC typo

We'd miss actually activating LLC.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Fixes: af66706accdf ("drm/msm/a6xx: Add skeleton A7xx support")
Patchwork: https://patchwork.freedesktop.org/patch/573043/

8 months agodrm/msm/adreno: Update generated headers
Rob Clark [Wed, 17 Jan 2024 20:37:18 +0000 (12:37 -0800)]
drm/msm/adreno: Update generated headers

This updates the GPU headers to latest from mesa, using gen_header.py
(which is used to generate headers at bulid time for mesa), rather than
headergen2 (which doesn't have proper support for A6XX vs A7XX register
variants).

Mostly just uninteresting churn, but there are a couple spots in a7xx
paths which update REG_A6XX_foo to REG_A7XX_foo for registers which are
a7xx specific.

Cc: Connor Abbott <cwabbott0@gmail.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/574880/

8 months agodrm/msm/adreno: Add support for SM7150 SoC machine
Danila Tikhonov [Sun, 18 Feb 2024 19:50:07 +0000 (22:50 +0300)]
drm/msm/adreno: Add support for SM7150 SoC machine

SM7150 has 5 power levels which correspond to 5 speed-bin values: 0,
128, 146, 167, 172. Speed-bin value is calulated as FMAX/4.8MHz round up
to zero decimal places.

Also a618 on SM7150 uses a615 zapfw. Add a squashed version (.mbn).

Add this as machine = "qcom,sm7150", because speed-bin values are
different from atoll (sc7180/sm7125).

Signed-off-by: Danila Tikhonov <danila@jiaxyga.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/578902/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agodrm/msm/a6xx: specify UBWC config for sc7180
Dmitry Baryshkov [Tue, 20 Feb 2024 17:12:10 +0000 (19:12 +0200)]
drm/msm/a6xx: specify UBWC config for sc7180

Historically the Adreno driver has not been updating memory
configuration registers on a618 (SC7180 platform) implying that the
default configuration is fine. After the rework performed in the commit
8814455a0e54 ("drm/msm: Refactor UBWC config setting") the function
a6xx_calc_ubwc_config() still contained this shortcut and did not
calculate UBWC configuration. However the function which now actually
updates hardware registers, a6xx_set_ubwc_config(), doesn't contain such
check.

Rather than adding the check to a6xx_set_ubwc_config(), fill in the
UBWC config for a618 (based on readings from SC7180).

Reported-by: Leonard Lausen <leonard@lausen.nl>
Link: https://gitlab.freedesktop.org/drm/msm/-/issues/49
Fixes: 8814455a0e54 ("drm/msm: Refactor UBWC config setting")
Cc: Connor Abbott <cwabbott0@gmail.com>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/579113/
Signed-off-by: Rob Clark <robdclark@chromium.org>
8 months agoMerge drm/drm-next into drm-misc-next
Thomas Zimmermann [Mon, 26 Feb 2024 13:20:50 +0000 (14:20 +0100)]
Merge drm/drm-next into drm-misc-next

Backmerging to get drm-misc-next up to v6.8-rc6.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
8 months agodrm/edid/firmware: Remove built-in EDIDs
Maxime Ripard [Wed, 21 Feb 2024 09:26:36 +0000 (10:26 +0100)]
drm/edid/firmware: Remove built-in EDIDs

The EDID firmware loading mechanism introduced a few built-in EDIDs that
could be forced on any connector, bypassing the EDIDs it exposes.

While convenient, this limited set of EDIDs doesn't take into account
the connector type, and we can end up with an EDID that is completely
invalid for a given connector.

For example, the edid/800x600.bin file matches the following EDID:

  edid-decode (hex):

  00 ff ff ff ff ff ff 00 31 d8 00 00 00 00 00 00
  05 16 01 03 6d 1b 14 78 ea 5e c0 a4 59 4a 98 25
  20 50 54 01 00 00 45 40 01 01 01 01 01 01 01 01
  01 01 01 01 01 01 a0 0f 20 00 31 58 1c 20 28 80
  14 00 15 d0 10 00 00 1e 00 00 00 ff 00 4c 69 6e
  75 78 20 23 30 0a 20 20 20 20 00 00 00 fd 00 3b
  3d 24 26 05 00 0a 20 20 20 20 20 20 00 00 00 fc
  00 4c 69 6e 75 78 20 53 56 47 41 0a 20 20 00 c2

  ----------------

  Block 0, Base EDID:
    EDID Structure Version & Revision: 1.3
    Vendor & Product Identification:
      Manufacturer: LNX
      Model: 0
      Made in: week 5 of 2012
    Basic Display Parameters & Features:
      Analog display
      Signal Level Standard: 0.700 : 0.000 : 0.700 V p-p
      Blank level equals black level
      Sync: Separate Composite Serration
      Maximum image size: 27 cm x 20 cm
      Gamma: 2.20
      DPMS levels: Standby Suspend Off
      RGB color display
      First detailed timing is the preferred timing
    Color Characteristics:
      Red  : 0.6416, 0.3486
      Green: 0.2919, 0.5957
      Blue : 0.1474, 0.1250
      White: 0.3125, 0.3281
    Established Timings I & II:
      DMT 0x09:   800x600    60.316541 Hz   4:3     37.879 kHz     40.000000 MHz
    Standard Timings:
      DMT 0x09:   800x600    60.316541 Hz   4:3     37.879 kHz     40.000000 MHz
    Detailed Timing Descriptors:
      DTD 1:   800x600    60.316541 Hz   4:3     37.879 kHz     40.000000 MHz (277 mm x 208 mm)
                   Hfront   40 Hsync 128 Hback   88 Hpol P
                   Vfront    1 Vsync   4 Vback   23 Vpol P
      Display Product Serial Number: 'Linux #0'
      Display Range Limits:
        Monitor ranges (GTF): 59-61 Hz V, 36-38 kHz H, max dotclock 50 MHz
      Display Product Name: 'Linux SVGA'
  Checksum: 0xc2

So, an analog monitor EDID. However, if the connector was an HDMI
monitor for example, it breaks the HDMI specification that requires,
among other things, a digital display, the VIC 1 mode and an HDMI Forum
Vendor Specific Data Block in an CTA-861 extension.

We thus end up with a completely invalid EDID, which thus might confuse
HDMI-related code that could parse it.

After some discussions on IRC, we identified mainly two ways to fix
this:

  - We can either create more EDIDs for each connector type to provide
    a built-in EDID that matches the resolution passed in the name, and
    still be a sensible EDID for that connector type;

  - Or we can just prevent the EDID to be exposed to userspace if it's
    built-in.

Or possibly both.

However, the conclusion was that maybe we just don't need the built-in
EDIDs at all and we should just get rid of them. So here we are.

Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240221092636.691701-1-mripard@kernel.org
8 months agoMerge v6.8-rc6 into drm-next
Daniel Vetter [Mon, 26 Feb 2024 10:41:07 +0000 (11:41 +0100)]
Merge v6.8-rc6 into drm-next

Thomas Zimmermann asked to backmerge -rc6 for drm-misc branches,
there's a few same-area-changed conflicts (xe and amdgpu mostly) that
are getting a bit too annoying.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
8 months agoMerge tag 'drm-habanalabs-next-2024-02-26' of https://git.kernel.org/pub/scm/linux...
Daniel Vetter [Mon, 26 Feb 2024 10:06:19 +0000 (11:06 +0100)]
Merge tag 'drm-habanalabs-next-2024-02-26' of https://git./linux/kernel/git/ogabbay/linux into drm-next

This tag contains habanalabs driver and accel changes for v6.9.

The notable changes are:

- New features and improvements:
  - Configure interrupt affinity according to NUMA nodes for the MSI-X interrupts that are
    assigned to the userspace application which acquires the device.
  - Move the HBM MMU page tables to reside inside the HBM to minimize latency when doing
    page-walks.
  - Improve the device reset mechanism when consecutive heartbeat failures occur (firmware
    fails to ack on heartbeat message).
  - Check also extended errors in the PCIe addr_dec interrupt information.
  - Rate limit the error messages that can be printed to dmesg log by userspace actions.

- Firmware related fixes:
  - Handle requests from firmware to reserve device memory

- Bug fixes and code cleanups:
  - constify the struct device_type usage in accel (accel_sysfs_device_minor).
  - Fix the PCI health check by reading uncached register.
  - Fix reporting of drain events.
  - Fix debugfs files permissions.
  - Fix calculation of DRAM BAR base address.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Oded Gabbay <ogabbay@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/ZdxJprop0EniVQtf@ogabbay-vm-u22.habana-labs.com
8 months agoMerge tag 'drm-xe-next-2024-02-25' of ssh://gitlab.freedesktop.org/drm/xe/kernel...
Daniel Vetter [Mon, 26 Feb 2024 09:49:09 +0000 (10:49 +0100)]
Merge tag 'drm-xe-next-2024-02-25' of ssh://gitlab.freedesktop.org/drm/xe/kernel into drm-next

drm/xe feature pull for v6.9:

UAPI Changes:

- New query to the GuC firmware submission version. (José Roberto de Souza)
- Remove unused persistent exec_queues (Thomas Hellström)
- Add vram frequency sysfs attributes (Sujaritha Sundaresan, Rodrigo Vivi)
- Add the flag XE_VM_BIND_FLAG_DUMPABLE to notify devcoredump that mapping
  should be dumped (Maarten Lankhorst)

Cross-drivers Changes:

- Make sure intel_wakeref_t is treated as opaque type on i915-display
  and fix its type on xe

Driver Changes:

- Drop pre-production workarounds (Matt Roper)
- Drop kunit tests for unsuported platforms: PVC and pre-production DG2 (Lucas De Marchi)
- Start pumbling SR-IOV support with memory based interrupts
  for VF (Michal Wajdeczko)
- Allow to map BO in GGTT with PAT index corresponding to
  XE_CACHE_UC to work with memory based interrupts (Michal Wajdeczko)
- Improve logging with GT-oriented drm_printers (Michal Wajdeczko)
- Add GuC Doorbells Manager as prep work SR-IOV during
  VF provisioning ((Michal Wajdeczko)
- Refactor fake device handling in kunit integration ((Michal Wajdeczko)
- Implement additional workarounds for xe2 and MTL (Tejas Upadhyay,
  Lucas De Marchi, Shekhar Chauhan, Karthik Poosa)
- Program a few registers according to perfomance guide spec for Xe2 (Shekhar Chauhan)
- Add error handling for non-blocking communication with GuC (Daniele Ceraolo Spurio)
- Fix remaining 32b build issues and enable it back (Lucas De  Marchi)
- Fix build with CONFIG_DEBUG_FS=n (Jani Nikula)
- Fix warnings from GuC ABI headers (Matthew Brost)
- Introduce Relay Communication for SR-IOV for VF <-> GuC <-> PF (Michal Wajdeczko)
- Add mocs reset kunit (Ruthuvikas Ravikumar)
- Fix spellings (Colin Ian King)
- Disable mid-thread preemption when not properly supported by hardware (Nirmoy Das)
- Release mmap mappings on rpm suspend (Badal Nilawar)
- Fix BUG_ON on xe_exec by moving fence reservation to the validate stage (Matthew Auld)
- Fix xe_exec by reserving extra fence slot for CPU bind (Matthew Brost)
- Fix xe_exec with full long running exec queue, now returning
  -EWOULDBLOCK to userspace (Matthew Brost)
- Fix CT irq handler when CT is disabled (Matthew Brost)
- Fix VM_BIND_OP_UNMAP_ALL without any bound vmas (Thomas Hellström)
- Fix missing __iomem annotations (Thomas Hellström)
- Fix exec queue priority handling with GuC (Brian Welty)
- Fix setting SLPC flag to GuC when it's not supported (Vinay Belgaumkar)
- Fix C6 disabling without SLPC (Matt Roper)
- Drop -Wstringop-overflow to fix build with GCC11 (Paul E. McKenney)
- Circumvent bogus -Wstringop-overflow in one case (Arnd Bergmann)
- Refactor exec_queue user extensions handling and fix USM attributes
  being applied too late (Brian Welty)
- Use circ_buf head/tail convention (Matthew Brost)
- Fail build if circ_buf-related defines are modified with incompatible values
  (Matthew Brost)
- Fix several error paths (Dan Carpenter)
- Fix CCS copy for small VRAM copy chunks (Thomas Hellström)
- Rework driver initialization order and paths to account for driver running
  in VF mode (Michal Wajdeczko)
- Initialize GuC earlier during probe to handle driver in VF mode (Michał Winiarski)
- Fix migration use of MI_STORE_DATA_IMM to write PTEs (Matt Roper)
- Fix bounds checking in __xe_bo_placement_for_flags (Brian Welty)
- Drop display dependency on CONFIG_EXPERT (Jani Nikula)
- Do not hand-roll kstrdup when creating snapshot (Michal Wajdeczko)
- Stop creating one kunit module per kunit suite (Lucas De Marchi)
- Reduce scope and constify variables (Thomas Hellström, Jani Nikula, Michal Wajdeczko)
- Improve and document xe_guc_ct_send_recv() (Michal Wajdeczko)
- Add proxy communication between CSME and GSC uC (Daniele Ceraolo Spurio)
- Fix size calculation when writing pgtable (Fei Yang)
- Make sure cfb is page size aligned in stolen memory (Vinod Govindapillai)
- Stop printing guc log to dmesg when waiting for GuC fails (Rodrigo Vivi)
- Use XE_CACHE_WB instead of XE_CACHE_NONE for cpu coherency on migration
  (Himal Prasad Ghimiray)
- Fix error path in xe_vm_create (Moti Haimovski)
- Fix warnings in doc generation (Thomas Hellström, Badal Nilawar)
- Improve devcoredump content for mesa debugging (José Roberto de Souza)
- Fix crash in trace_dma_fence_init() (José Roberto de Souza)
- Improve CT state change handling (Matthew Brost)
- Toggle USM support for Xe2 (Lucas De Marchi)
- Reduces code duplication to emit PIPE_CONTROL (José Roberto de Souza)
- Canonicalize addresses where needed for Xe2 and add to devcoredump
  (José Roberto de Souza)
- Only allow 1 ufence per exec / bind IOCTL (Matthew Brost)
- Move all display code to display/ (Jani Nikula)
- Fix sparse warnings by correctly using annotations (Thomas Hellström)
- Warn on job timeouts instead of using asserts (Matt Roper)
- Prefix macros to avoid clashes with sparc (Matthew Brost)
- Fix -Walloc-size by subclassing instead of allocating size smaller than struct (Thomas Hellström)
- Add status check during gsc header readout (Suraj Kandpal)
- Fix infinite loop in vm_bind_ioctl_ops_unwind() (Matthew Brost)
- Fix fence refcounting (Matthew Brost)
- Fix picking incorrect userptr VMA (Matthew Brost)
- Fix USM on integrated by mapping both mem.kernel_bb_pool and usm.bb_pool (Matthew Brost)
- Fix double initialization of display power domains (Xiaoming Wang)
- Check expected uC versions by major.minor.patch instead of just major.minor (John Harrison)
- Bump minimum GuC version to 70.19.2 for all platforms under force-probe
  (John Harrison)
- Add GuC firmware loading for Lunar Lake (John Harrison)
- Use kzalloc() instead of hand-rolled alloc + memset (Nirmoy Das)
- Fix max page size of VMA during a REMAP (Matthew Brost)
- Don't ignore error when pinning pages in kthread (Matthew Auld)
- Refactor xe hwmon (Karthik Poosa)
- Add debug logs for D3cold (Riana Tauro)
- Remove broken TEST_VM_ASYNC_OPS_ERROR (Matthew Brost)
- Always allow to override firmware blob with module param and improve
  log when no firmware is found (Lucas De Marchi)
- Fix shift-out-of-bounds due to xe_vm_prepare_vma() accepting zero fences (Thomas Hellström)
- Fix shift-out-of-bounds by distinguishing xe_pt/xe_pt_dir subclass (Thomas Hellström)
- Fail driver bind if platform supports MSIX, but fails to allocate all of them (Dani Liberman)
- Fix intel_fbdev thinking memory is backed by shmem (Matthew Auld)
- Prefer drm_dbg() over dev_dbg() (Jani Nikula)
- Avoid function cast warnings with clang-16 (Arnd Bergmann)
- Enhance xe_bo_move trace (Priyanka Dandamudi)
- Fix xe_vma_set_pte_size() not setting the right gpuva.flags for 4K size (Matthew Brost)
- Add XE_VMA_PTE_64K VMA flag (Matthew Brost)
- Return 2MB page size for compact 64k PTEs (Matthew Brost)
- Remove usage of the deprecated ida_simple_xx() API (Christophe JAILLET)
- Fix modpost warning on xe_mocs live kunit module (Ashutosh Dixit)
- Drop extra newline in from sysfs files (Ashutosh Dixit)
- Implement VM snapshot support for BO's and userptr (Maarten Lankhorst)
- Add debug logs when skipping rebinds (Matthew Brost)
- Fix code generation when mixing build directories (Dafna Hirschfeld)
- Prefer struct_size over open coded arithmetic (Erick Archer)

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/dbdkrwmcoqqlwftuc3olbauazc3pbamj26wa34puztowsnauoh@i3zms7ut4yuw
8 months agodrm/sun4i: hdmi: Consolidate atomic_check and mode_valid
Maxime Ripard [Thu, 22 Feb 2024 18:14:21 +0000 (19:14 +0100)]
drm/sun4i: hdmi: Consolidate atomic_check and mode_valid

atomic_check and mode_valid do not check for the same things which can
lead to surprising result if the userspace commits a mode that didn't go
through mode_valid. Let's merge the two implementations into a function
called by both.

Acked-by: Sui Jingfeng <sui.jingfeng@linux.dev>
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-35-8f4af575fce2@kernel.org
8 months agodrm/sun4i: hdmi: Switch to container_of_const
Maxime Ripard [Thu, 22 Feb 2024 18:14:20 +0000 (19:14 +0100)]
drm/sun4i: hdmi: Switch to container_of_const

container_of_const() allows to preserve the pointer constness and is
thus more flexible than inline functions.

Let's switch all our instances of container_of() to container_of_const().

Reviewed-by: Sui Jingfeng <sui.jingfeng@linux.dev>
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-34-8f4af575fce2@kernel.org
8 months agodrm/sun4i: hdmi: Move mode_set into enable
Maxime Ripard [Thu, 22 Feb 2024 18:14:19 +0000 (19:14 +0100)]
drm/sun4i: hdmi: Move mode_set into enable

We're not doing anything special in atomic_mode_set so we can simply
merge it into atomic_enable.

Acked-by: Sui Jingfeng <sui.jingfeng@linux.dev>
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-33-8f4af575fce2@kernel.org
8 months agodrm/sun4i: hdmi: Convert encoder to atomic
Maxime Ripard [Thu, 22 Feb 2024 18:14:18 +0000 (19:14 +0100)]
drm/sun4i: hdmi: Convert encoder to atomic

The sun4i_hdmi driver still uses the non-atomic variants of the encoder
hooks, so let's convert to their atomic equivalents.

Acked-by: Sui Jingfeng <sui.jingfeng@linux.dev>
Reviewed-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222-kms-hdmi-connector-state-v7-32-8f4af575fce2@kernel.org
8 months agoMerge tag 'drm-misc-next-2024-02-22' of git://anongit.freedesktop.org/drm/drm-misc...
Daniel Vetter [Mon, 26 Feb 2024 08:51:48 +0000 (09:51 +0100)]
Merge tag 'drm-misc-next-2024-02-22' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for v6.9:

UAPI Changes:

- changes to fdinfo stats

Cross-subsystem Changes:

agp:
- remove unused type field from struct agp_bridge_data

Core Changes:

ci:
- update test names
- cleanups

gem:
- add stats for shared buffers plus updates to amdgpu, i915, xe

Documentation:
- fixes

syncobj:
- fixes to waiting and sleeping

Driver Changes:

bridge:
- adv7511: fix crash on irq during probe
- dw_hdmi: set bridge type

host1x:
- cleanups

ivpu:
- updates to firmware API
- refactor BO allocation

meson:
- fix error handling in probe

panel:
- revert "drm/panel-edp: Add auo_b116xa3_mode"
- add Himax HX83112A plus DT bindings
- ltk500hd1829: add support for ltk101b4029w and admatec 9904370
- simple: add BOE BP082WX1-100 8.2" panel plus DT bindungs

renesas:
- add RZ/G2L DU support plus DT bindings

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240222135841.GA6677@localhost.localdomain
8 months agodrm/tidss: Fix sync-lost issue with two displays
Tomi Valkeinen [Tue, 13 Feb 2024 08:16:37 +0000 (10:16 +0200)]
drm/tidss: Fix sync-lost issue with two displays

A sync lost issue can be observed with two displays, when moving a plane
from one disabled display to an another disabled display, and then
enabling the display to which the plane was moved to. The exact
requirements for this to trigger are not clear.

It looks like the issue is that the layers are left enabled in the first
display's OVR registers. Even if the corresponding VP is disabled, it
still causes an issue, as if the disabled VP and its OVR would still be
in use, leading to the same VID being used by two OVRs. However, this is
just speculation based on testing the DSS behavior.

Experimentation shows that as a workaround, we can disable all the
layers in the OVR when disabling a VP. There should be no downside to
this, as the OVR is anyway effectively disabled if its VP is disabled,
and it seems to solve the sync lost issue.

However, there may be a bigger issue in play here, related to J721e
erratum i2097 ("DSS: Disabling a Layer Connected to Overlay May Result
in Synclost During the Next Frame"). Experimentation also shows that the
OVR's CHANNELIN field has similar issue. So we may need to revisit this
when we find out more about the core issue.

Fixes: 32a1795f57ee ("drm/tidss: New driver for TI Keystone platform Display SubSystem")
Reviewed-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240213-tidss-fixes-v1-2-d709e8dfa505@ideasonboard.com
8 months agodrm/tidss: Fix initial plane zpos values
Tomi Valkeinen [Tue, 13 Feb 2024 08:16:36 +0000 (10:16 +0200)]
drm/tidss: Fix initial plane zpos values

When the driver sets up the zpos property it sets the default zpos value
to the HW id of the plane. That is fine as such, but as on many DSS
versions the driver arranges the DRM planes in a different order than
the HW planes (to keep the non-scalable planes first), this leads to odd
initial zpos values. An example is J721e, where the initial zpos values
for DRM planes are 1, 3, 0, 2.

In theory the userspace should configure the zpos values properly when
using multiple planes, and in that sense the initial zpos values
shouldn't matter, but there's really no reason not to fix this and help
the userspace apps which don't handle zpos perfectly. In particular,
some versions of Weston seem to have issues dealing with the planes
with the current default zpos values.

So let's change the zpos values for the DRM planes to 0, 1, 2, 3.

Another option would be to configure the planes marked as primary planes
to zpos 0. On a two display system this would give us plane zpos values
of 0, 0, 1, 2. The end result and behavior would be very similar in this
option, and I'm not aware that this would actually help us in any way.
So, to keep the code simple, I opted for the 0, 1, 2, 3 values.

Fixes: 32a1795f57ee ("drm/tidss: New driver for TI Keystone platform Display SubSystem")
Reviewed-by: Aradhya Bhatia <a-bhatia1@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240213-tidss-fixes-v1-1-d709e8dfa505@ideasonboard.com
8 months agoaccel: constify the struct device_type usage
Ricardo B. Marliere [Mon, 19 Feb 2024 19:48:28 +0000 (16:48 -0300)]
accel: constify the struct device_type usage

Since commit aed65af1cc2f ("drivers: make device_type const"), the driver
core can properly handle constant struct device_type. Move the
accel_sysfs_device_minor variable to be a constant structure as well,
placing it into read-only memory which can not be modified at runtime.

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Ricardo B. Marliere <ricardo@marliere.net>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: modify pci health check
Ofir Bitton [Mon, 12 Feb 2024 12:35:24 +0000 (14:35 +0200)]
accel/habanalabs: modify pci health check

Today we read PCI VENDOR-ID in order to make sure PCI link is
healthy. Apparently the VENDOR-ID might be stored on host and
hence, when we read it we might not access the PCI bus.
In order to make sure PCI health check is reliable, we will start
checking the DEVICE-ID instead.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: keep explicit size of reserved memory for FW
Tomer Tayar [Tue, 30 Jan 2024 07:57:32 +0000 (09:57 +0200)]
accel/habanalabs: keep explicit size of reserved memory for FW

The reserved memory for FW is currently saved in an ASIC property in
units of MB, just like the value that comes from FW.
Except the fact that it is not clear from the property's name, it means
also that a calculation to actual size is required everywhere that it is
used.
Modify the property to hold the size in bytes.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: handle reserved memory request when working with full FW
Tomer Tayar [Mon, 29 Jan 2024 15:26:17 +0000 (17:26 +0200)]
accel/habanalabs: handle reserved memory request when working with full FW

Currently the reserved memory request from FW is handled when running
with preboot only, but this request is relevant also when running with
full FW.
Modify to always handle this reservation request.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/hwmon: rate limit errors user can generate
Ofir Bitton [Tue, 6 Feb 2024 19:12:21 +0000 (21:12 +0200)]
accel/habanalabs/hwmon: rate limit errors user can generate

Fetching sensor data can fail due to various reasons. In order
not to pollute the kernel log, those error prints must be
rate limited.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/gaudi2: drain event lacks rd/wr indication
Ofir Bitton [Mon, 5 Feb 2024 15:36:38 +0000 (17:36 +0200)]
accel/habanalabs/gaudi2: drain event lacks rd/wr indication

Due to a H/W issue, AXI drain event does not include a read/write
indication, hence we remove this print.

Signed-off-by: Ofir Bitton <obitton@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: fix error print
Dani Liberman [Mon, 5 Feb 2024 07:19:30 +0000 (09:19 +0200)]
accel/habanalabs: fix error print

The unmasking is for event and it can be other event than RAZWI.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: initialize maybe-uninitialized variables
Tal Risin [Wed, 31 Jan 2024 09:08:33 +0000 (11:08 +0200)]
accel/habanalabs: initialize maybe-uninitialized variables

Prevent static analysis warning.

Signed-off-by: Tal Risin <trisin@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: fix debugfs files permissions
Avri Kehat [Tue, 16 Jan 2024 15:54:36 +0000 (17:54 +0200)]
accel/habanalabs: fix debugfs files permissions

debugfs files are created with permissions that don't align
with the access requirements.

Signed-off-by: Avri Kehat <akehat@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: fix glbl error cause handling
Tomer Tayar [Thu, 25 Jan 2024 20:59:02 +0000 (22:59 +0200)]
accel/habanalabs: fix glbl error cause handling

The glbl error cause handling has a wrong assumption that all error
bits are consecutive.
Fix the handling to check all relevant error bits per ASIC.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/gaudi2: check extended errors according to PCIe addr_dec interrupt...
Tomer Tayar [Thu, 18 Jan 2024 12:29:02 +0000 (14:29 +0200)]
accel/habanalabs/gaudi2: check extended errors according to PCIe addr_dec interrupt info

The FW interrupt info for a PCIe addr_dec event is set correctly, so
check for either global errors or razwi according to the indications
there.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: modify print for skip loading linux FW to debug log
Tomer Tayar [Thu, 18 Jan 2024 17:18:43 +0000 (19:18 +0200)]
accel/habanalabs: modify print for skip loading linux FW to debug log

Skip loading a linux FW image into the device with the current supported
ASICs is done for test purposes only.
Moreover, for future supported ASICs it is possible that there won't be
a need to load such an image.
The print in such a case is therefore not needed in most cases, so
replace the used dev_info() with dev_dbg().

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: remove hop size from asic properties
Farah Kassabri [Mon, 15 Jan 2024 10:49:24 +0000 (12:49 +0200)]
accel/habanalabs: remove hop size from asic properties

The hop size related properties is a MMU properties and not
asic properties.
As for PMMU and HMMU we could have different sizes.

Signed-off-by: Farah Kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Reviewed-by: Carl Vanderlip <quic_carlv@quicinc.com>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agodrm: lcdif: Switch to drmm_mode_config_init
Marek Vasut [Mon, 27 Nov 2023 21:26:25 +0000 (22:26 +0100)]
drm: lcdif: Switch to drmm_mode_config_init

Switch from deprecated unmanaged drm_mode_config_init() to
managed drmm_mode_config_init(). No functional change.

Signed-off-by: Marek Vasut <marex@denx.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231127212638.77688-1-marex@denx.de
9 months agodrm/mxsfb: Switch to drmm_mode_config_init
Marek Vasut [Mon, 27 Nov 2023 21:25:38 +0000 (22:25 +0100)]
drm/mxsfb: Switch to drmm_mode_config_init

Switch from deprecated unmanaged drm_mode_config_init() to
managed drmm_mode_config_init(). No functional change.

Signed-off-by: Marek Vasut <marex@denx.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20231127212612.77667-1-marex@denx.de
9 months agoaccel/habanalabs: use kcalloc() instead of kzalloc()
Erick Archer [Sat, 20 Jan 2024 15:10:28 +0000 (16:10 +0100)]
accel/habanalabs: use kcalloc() instead of kzalloc()

As noted in the "Deprecated Interfaces, Language Features, Attributes,
and Conventions" documentation [1], size calculations (especially
multiplication) should not be performed in memory allocator (or similar)
function arguments due to the risk of them overflowing. This could lead
to values wrapping around and a smaller allocation being made than the
caller was expecting. Using those allocations could lead to linear
overflows of heap memory and other misbehaviors.

So, use the purpose specific kcalloc() function instead of the argument
size * count in the kzalloc() function.

Link: https://www.kernel.org/doc/html/next/process/deprecated.html#open-coded-arithmetic-in-allocator-arguments
Link: https://github.com/KSPP/linux/issues/162
Signed-off-by: Erick Archer <erick.archer@gmx.com>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/goya: remove redundant assignment to pointer 'input'
Colin Ian King [Sat, 6 Jan 2024 12:42:13 +0000 (12:42 +0000)]
accel/habanalabs/goya: remove redundant assignment to pointer 'input'

The pointer input is assigned a value that is not read, it is
being re-assigned again later with the same value. Resolve this
by moving the declaration to input into the if block.

Cleans up clang scan build warning:
warning: Value stored to 'input' during its initialization is never
read [deadcode.DeadStores]

Signed-off-by: Colin Ian King <colin.i.king@intel.com>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/gaudi2: fail memory memset when failing to copy QM packet to device
Tomer Tayar [Tue, 2 Jan 2024 14:51:09 +0000 (16:51 +0200)]
accel/habanalabs/gaudi2: fail memory memset when failing to copy QM packet to device

gaudi2_memset_memory_chunk_using_edma_qm() calls the access_dev_mem()
ASIC function, but ignores its return value.
Add this missing check.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: remove call to deprecated function
Dani Liberman [Mon, 1 Jan 2024 20:37:43 +0000 (22:37 +0200)]
accel/habanalabs: remove call to deprecated function

In newer kernel versions, irq_set_affinity_hint() is deprecated.
Instead, use the newer version which is irq_set_affinity_and_hint().

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: Remove unnecessary braces from if statement
Malkoot Khan [Thu, 28 Dec 2023 21:08:58 +0000 (21:08 +0000)]
accel/habanalabs: Remove unnecessary braces from if statement

The coding style in the Linux kernel prefers not to use
braces for single-statement if conditions.
This patch removes the unnecessary braces from an if statement
in the file drivers/accel/habanalabs/common/command_submission.c,
which also resolves a coding style warning.

Signed-off-by: Malkoot Khan <engr.mkhan1990@gmail.com>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/gaudi2: move HMMU page tables to device memory
Farah Kassabri [Thu, 2 Nov 2023 09:53:29 +0000 (11:53 +0200)]
accel/habanalabs/gaudi2: move HMMU page tables to device memory

Currently the HMMU page tables reside in the host memory,
which will cause host access from the device for every page walk.
This can affect PCIe bandwidth in certain scenarios.

To prevent that problem, HMMU page tables will be moved to the device
memory so the miss transaction will read the hops from there instead of
going to the host.

Signed-off-by: Farah Kassabri <fkassabri@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: abort device reset for consecutive heartbeat failures
Tomer Tayar [Sun, 24 Dec 2023 22:28:36 +0000 (00:28 +0200)]
accel/habanalabs: abort device reset for consecutive heartbeat failures

The mechanism of aborting device reset for consecutive fatal errors is
currently only for fatal errors that are reported by FW.
A non-responsive FW and consecutive heartbeat failures is also
considered fatal, so add them as well to this mechanism to avoid
recurring device reset in such a case.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: fix DRAM BAR base address calculation
Tomer Tayar [Thu, 14 Dec 2023 08:38:06 +0000 (10:38 +0200)]
accel/habanalabs: fix DRAM BAR base address calculation

When the DRAM region size in the BAR is not a power of 2, calculating
the corresponding BAR base address should be done using the offset from
the DRAM start address, and not using directly the DRAM address.

Signed-off-by: Tomer Tayar <ttayar@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs: increase HL_MAX_STR to 64 bytes to avoid warnings
Koby Elbaz [Mon, 11 Dec 2023 08:03:29 +0000 (10:03 +0200)]
accel/habanalabs: increase HL_MAX_STR to 64 bytes to avoid warnings

Fix a warning of a buffer overflow:
‘snprintf’ output between 38 and 47 bytes into a destination of size 32

Signed-off-by: Koby Elbaz <kelbaz@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoaccel/habanalabs/gaudi2: add interrupt affinity for user interrupts
Dani Liberman [Thu, 21 Sep 2023 14:02:33 +0000 (17:02 +0300)]
accel/habanalabs/gaudi2: add interrupt affinity for user interrupts

User interrupts are MSIx interrupts coming from Gaudi2, that have
specific range of IDs and are assigned to the sole use of the user
process that opened the Gaudi2 device (reminder: there can be only
a single user process running on Gaudi2 at any given time).

The interrupts are allocated and managed by the driver and therefore,
the user expects the driver to initialize them properly, which also
includes setting the affinity to the related CPU cores of the
device's NUMA node to get maximum performance.

Signed-off-by: Dani Liberman <dliberman@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
9 months agoLinux 6.8-rc6
Linus Torvalds [Sun, 25 Feb 2024 23:46:06 +0000 (15:46 -0800)]
Linux 6.8-rc6

9 months agoMerge tag 'bcachefs-2024-02-25' of https://evilpiepirate.org/git/bcachefs
Linus Torvalds [Sun, 25 Feb 2024 23:31:57 +0000 (15:31 -0800)]
Merge tag 'bcachefs-2024-02-25' of https://evilpiepirate.org/git/bcachefs

Pull bcachefs fixes from Kent Overstreet:
 "Some more mostly boring fixes, but some not

  User reported ones:

   - the BTREE_ITER_FILTER_SNAPSHOTS one fixes a really nasty
     performance bug; user reported an untar initially taking two
     seconds and then ~2 minutes

   - kill a __GFP_NOFAIL in the buffered read path; this was a leftover
     from the trickier fix to kill __GFP_NOFAIL in readahead, where we
     can't return errors (and have to silently truncate the read
     ourselves).

     bcachefs can't use GFP_NOFAIL for folio state unlike iomap based
     filesystems because our folio state is just barely too big, 2MB
     hugepages cause us to exceed the 2 page threshhold for GFP_NOFAIL.

     additionally, the flags argument was just buggy, we weren't
     supplying GFP_KERNEL previously (!)"

* tag 'bcachefs-2024-02-25' of https://evilpiepirate.org/git/bcachefs:
  bcachefs: fix bch2_save_backtrace()
  bcachefs: Fix check_snapshot() memcpy
  bcachefs: Fix bch2_journal_flush_device_pins()
  bcachefs: fix iov_iter count underflow on sub-block dio read
  bcachefs: Fix BTREE_ITER_FILTER_SNAPSHOTS on inodes btree
  bcachefs: Kill __GFP_NOFAIL in buffered read path
  bcachefs: fix backpointer_to_text() when dev does not exist

9 months agobcachefs: fix bch2_save_backtrace()
Kent Overstreet [Sun, 25 Feb 2024 20:45:34 +0000 (15:45 -0500)]
bcachefs: fix bch2_save_backtrace()

Missed a call in the previous fix.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
9 months agoMerge tag 'docs-6.8-fixes3' of git://git.lwn.net/linux
Linus Torvalds [Sun, 25 Feb 2024 18:58:12 +0000 (10:58 -0800)]
Merge tag 'docs-6.8-fixes3' of git://git.lwn.net/linux

Pull two documentation build fixes from Jonathan Corbet:

 - The XFS online fsck documentation uses incredibly deeply nested
   subsection and list nesting; that broke the PDF docs build. Tweak a
   parameter to tell LaTeX to allow the deeper nesting.

 - Fix a 6.8 PDF-build regression

* tag 'docs-6.8-fixes3' of git://git.lwn.net/linux:
  docs: translations: use attribute to store current language
  docs: Instruct LaTeX to cope with deeper nesting

9 months agoMerge tag 'usb-6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 25 Feb 2024 18:41:57 +0000 (10:41 -0800)]
Merge tag 'usb-6.8-rc6' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB fixes for 6.8-rc6 to resolve some reported
  problems. These include:

   - regression fixes with typec tpcm code as reported by many

   - cdnsp and cdns3 driver fixes

   - usb role setting code bugfixes

   - build fix for uhci driver

   - ncm gadget driver bugfix

   - MAINTAINERS entry update

  All of these have been in linux-next all week with no reported issues
  and there is at least one fix in here that is in Thorsten's regression
  list that is being tracked"

* tag 'usb-6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: typec: tpcm: Fix issues with power being removed during reset
  MAINTAINERS: Drop myself as maintainer of TYPEC port controller drivers
  usb: gadget: ncm: Avoid dropping datagrams of properly parsed NTBs
  Revert "usb: typec: tcpm: reset counter when enter into unattached state after try role"
  usb: gadget: omap_udc: fix USB gadget regression on Palm TE
  usb: dwc3: gadget: Don't disconnect if not started
  usb: cdns3: fix memory double free when handle zero packet
  usb: cdns3: fixed memory use after free at cdns3_gadget_ep_disable()
  usb: roles: don't get/set_role() when usb_role_switch is unregistered
  usb: roles: fix NULL pointer issue when put module's reference
  usb: cdnsp: fixed issue with incorrect detecting CDNSP family controllers
  usb: cdnsp: blocked some cdns3 specific code
  usb: uhci-grlib: Explicitly include linux/platform_device.h

9 months agoMerge tag 'tty-6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 25 Feb 2024 18:35:41 +0000 (10:35 -0800)]
Merge tag 'tty-6.8-rc6' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial driver fixes from Greg KH:
 "Here are three small serial/tty driver fixes for 6.8-rc6 that resolve
  the following reported errors:

   - riscv hvc console driver fix that was reported by many

   - amba-pl011 serial driver fix for RS485 mode

   - stm32 serial driver fix for RS485 mode

  All of these have been in linux-next all week with no reported
  problems"

* tag 'tty-6.8-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  serial: amba-pl011: Fix DMA transmission in RS485 mode
  serial: stm32: do not always set SER_RS485_RX_DURING_TX if RS485 is enabled
  tty: hvc: Don't enable the RISC-V SBI console by default

9 months agoMerge tag 'x86_urgent_for_v6.8_rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Feb 2024 18:22:21 +0000 (10:22 -0800)]
Merge tag 'x86_urgent_for_v6.8_rc6' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:

 - Make sure clearing CPU buffers using VERW happens at the latest
   possible point in the return-to-userspace path, otherwise memory
   accesses after the VERW execution could cause data to land in CPU
   buffers again

* tag 'x86_urgent_for_v6.8_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  KVM/VMX: Move VERW closer to VMentry for MDS mitigation
  KVM/VMX: Use BT+JNC, i.e. EFLAGS.CF to select VMRESUME vs. VMLAUNCH
  x86/bugs: Use ALTERNATIVE() instead of mds_user_clear static key
  x86/entry_32: Add VERW just before userspace transition
  x86/entry_64: Add VERW just before userspace transition
  x86/bugs: Add asm helpers for executing VERW

9 months agoMerge tag 'irq_urgent_for_v6.8_rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Feb 2024 18:14:12 +0000 (10:14 -0800)]
Merge tag 'irq_urgent_for_v6.8_rc6' of git://git./linux/kernel/git/tip/tip

Pull irq fixes from Borislav Petkov:

 - Make sure GICv4 always gets initialized to prevent a kexec-ed kernel
   from silently failing to set it up

 - Do not call bus_get_dev_root() for the mbigen irqchip as it always
   returns NULL - use NULL directly

 - Fix hardware interrupt number truncation when assigning MSI
   interrupts

 - Correct sending end-of-interrupt messages to disabled interrupts
   lines on RISC-V PLIC

* tag 'irq_urgent_for_v6.8_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3-its: Do not assume vPE tables are preallocated
  irqchip/mbigen: Don't use bus_get_dev_root() to find the parent
  PCI/MSI: Prevent MSI hardware interrupt number truncation
  irqchip/sifive-plic: Enable interrupt if needed before EOI

9 months agoMerge tag 'erofs-for-6.8-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Feb 2024 17:53:13 +0000 (09:53 -0800)]
Merge tag 'erofs-for-6.8-rc6-fixes' of git://git./linux/kernel/git/xiang/erofs

Pull erofs fix from Gao Xiang:

 - Fix page refcount leak when looking up specific inodes
   introduced by metabuf reworking

* tag 'erofs-for-6.8-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
  erofs: fix refcount on the metabuf used for inode lookup

9 months agoMerge tag 'pull-fixes.pathwalk-rcu-2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Feb 2024 17:29:05 +0000 (09:29 -0800)]
Merge tag 'pull-fixes.pathwalk-rcu-2' of git://git./linux/kernel/git/viro/vfs

Pull RCU pathwalk fixes from Al Viro:
 "We still have some races in filesystem methods when exposed to RCU
  pathwalk. This series is a result of code audit (the second round of
  it) and it should deal with most of that stuff.

  Still pending: ntfs3 ->d_hash()/->d_compare() and ceph_d_revalidate().
  Up to maintainers (a note for NTFS folks - when documentation says
  that a method may not block, it *does* imply that blocking allocations
  are to be avoided. Really)"

[ More explanations for people who aren't familiar with the vagaries of
  RCU path walking: most of it is hidden from filesystems, but if a
  filesystem actively participates in the low-level path walking it
  needs to make sure the fields involved in that walk are RCU-safe.

  That "actively participate in low-level path walking" includes things
  like having its own ->d_hash()/->d_compare() routines, or by having
  its own directory permission function that doesn't just use the common
  helpers.  Having a ->d_revalidate() function will also have this issue.

  Note that instead of making everything RCU safe you can also choose to
  abort the RCU pathwalk if your operation cannot be done safely under
  RCU, but that obviously comes with a performance penalty. One common
  pattern is to allow the simple cases under RCU, and abort only if you
  need to do something more complicated.

  So not everything needs to be RCU-safe, and things like the inode etc
  that the VFS itself maintains obviously already are. But these fixes
  tend to be about properly RCU-delaying things like ->s_fs_info that
  are maintained by the filesystem and that got potentially released too
  early.   - Linus ]

* tag 'pull-fixes.pathwalk-rcu-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  ext4_get_link(): fix breakage in RCU mode
  cifs_get_link(): bail out in unsafe case
  fuse: fix UAF in rcu pathwalks
  procfs: make freeing proc_fs_info rcu-delayed
  procfs: move dropping pde and pid from ->evict_inode() to ->free_inode()
  nfs: fix UAF on pathwalk running into umount
  nfs: make nfs_set_verifier() safe for use in RCU pathwalk
  afs: fix __afs_break_callback() / afs_drop_open_mmap() race
  hfsplus: switch to rcu-delayed unloading of nls and freeing ->s_fs_info
  exfat: move freeing sbi, upcase table and dropping nls into rcu-delayed helper
  affs: free affs_sb_info with kfree_rcu()
  rcu pathwalk: prevent bogus hard errors from may_lookup()
  fs/super.c: don't drop ->s_user_ns until we free struct super_block itself

9 months agoMerge tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sun, 25 Feb 2024 17:17:15 +0000 (09:17 -0800)]
Merge tag 'pull-fixes' of git://git./linux/kernel/git/viro/vfs

Pull vfs fixes from Al Viro:
 "A couple of fixes - revert of regression from this cycle and a fix for
  erofs failure exit breakage (had been there since way back)"

* tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  erofs: fix handling kern_mount() failure
  Revert "get rid of DCACHE_GENOCIDE"

9 months agoext4_get_link(): fix breakage in RCU mode
Al Viro [Sat, 3 Feb 2024 06:17:34 +0000 (01:17 -0500)]
ext4_get_link(): fix breakage in RCU mode

1) errors from ext4_getblk() should not be propagated to caller
unless we are really sure that we would've gotten the same error
in non-RCU pathwalk.
2) we leak buffer_heads if ext4_getblk() is successful, but bh is
not uptodate.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agocifs_get_link(): bail out in unsafe case
Al Viro [Wed, 20 Sep 2023 02:28:16 +0000 (22:28 -0400)]
cifs_get_link(): bail out in unsafe case

->d_revalidate() bails out there, anyway.  It's not enough
to prevent getting into ->get_link() in RCU mode, but that
could happen only in a very contrieved setup.  Not worth
trying to do anything fancy here unless ->d_revalidate()
stops kicking out of RCU mode at least in some cases.

Reviewed-by: Christian Brauner <brauner@kernel.org>
Acked-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agofuse: fix UAF in rcu pathwalks
Al Viro [Thu, 28 Sep 2023 04:19:39 +0000 (00:19 -0400)]
fuse: fix UAF in rcu pathwalks

->permission(), ->get_link() and ->inode_get_acl() might dereference
->s_fs_info (and, in case of ->permission(), ->s_fs_info->fc->user_ns
as well) when called from rcu pathwalk.

Freeing ->s_fs_info->fc is rcu-delayed; we need to make freeing ->s_fs_info
and dropping ->user_ns rcu-delayed too.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agoprocfs: make freeing proc_fs_info rcu-delayed
Al Viro [Wed, 20 Sep 2023 04:12:00 +0000 (00:12 -0400)]
procfs: make freeing proc_fs_info rcu-delayed

makes proc_pid_ns() safe from rcu pathwalk (put_pid_ns()
is still synchronous, but that's not a problem - it does
rcu-delay everything that needs to be)

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agoprocfs: move dropping pde and pid from ->evict_inode() to ->free_inode()
Al Viro [Wed, 20 Sep 2023 03:52:58 +0000 (23:52 -0400)]
procfs: move dropping pde and pid from ->evict_inode() to ->free_inode()

that keeps both around until struct inode is freed, making access
to them safe from rcu-pathwalk

Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agonfs: fix UAF on pathwalk running into umount
Al Viro [Thu, 28 Sep 2023 02:11:26 +0000 (22:11 -0400)]
nfs: fix UAF on pathwalk running into umount

NFS ->d_revalidate(), ->permission() and ->get_link() need to access
some parts of nfs_server when called in RCU mode:
server->flags
server->caps
*(server->io_stats)
and, worst of all, call
server->nfs_client->rpc_ops->have_delegation
(the last one - as NFS_PROTO(inode)->have_delegation()).  We really
don't want to RCU-delay the entire nfs_free_server() (it would have
to be done with schedule_work() from RCU callback, since it can't
be made to run from interrupt context), but actual freeing of
nfs_server and ->io_stats can be done via call_rcu() just fine.
nfs_client part is handled simply by making nfs_free_client() use
kfree_rcu().

Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agonfs: make nfs_set_verifier() safe for use in RCU pathwalk
Al Viro [Thu, 28 Sep 2023 01:50:25 +0000 (21:50 -0400)]
nfs: make nfs_set_verifier() safe for use in RCU pathwalk

nfs_set_verifier() relies upon dentry being pinned; if that's
the case, grabbing ->d_lock stabilizes ->d_parent and guarantees
that ->d_parent points to a positive dentry.  For something
we'd run into in RCU mode that is *not* true - dentry might've
been through dentry_kill() just as we grabbed ->d_lock, with
its parent going through the same just as we get to into
nfs_set_verifier_locked().  It might get to detaching inode
(and zeroing ->d_inode) before nfs_set_verifier_locked() gets
to fetching that; we get an oops as the result.

That can happen in nfs{,4} ->d_revalidate(); the call chain in
question is nfs_set_verifier_locked() <- nfs_set_verifier() <-
nfs_lookup_revalidate_delegated() <- nfs{,4}_do_lookup_revalidate().
We have checked that the parent had been positive, but that's
done before we get to nfs_set_verifier() and it's possible for
memory pressure to pick our dentry as eviction candidate by that
time.  If that happens, back-to-back attempts to kill dentry and
its parent are quite normal.  Sure, in case of eviction we'll
fail the ->d_seq check in the caller, but we need to survive
until we return there...

Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agoafs: fix __afs_break_callback() / afs_drop_open_mmap() race
Al Viro [Sat, 30 Sep 2023 00:24:34 +0000 (20:24 -0400)]
afs: fix __afs_break_callback() / afs_drop_open_mmap() race

In __afs_break_callback() we might check ->cb_nr_mmap and if it's non-zero
do queue_work(&vnode->cb_work).  In afs_drop_open_mmap() we decrement
->cb_nr_mmap and do flush_work(&vnode->cb_work) if it reaches zero.

The trouble is, there's nothing to prevent __afs_break_callback() from
seeing ->cb_nr_mmap before the decrement and do queue_work() after both
the decrement and flush_work().  If that happens, we might be in trouble -
vnode might get freed before the queued work runs.

__afs_break_callback() is always done under ->cb_lock, so let's make
sure that ->cb_nr_mmap can change from non-zero to zero while holding
->cb_lock (the spinlock component of it - it's a seqlock and we don't
need to mess with the counter).

Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agohfsplus: switch to rcu-delayed unloading of nls and freeing ->s_fs_info
Al Viro [Wed, 20 Sep 2023 00:18:59 +0000 (20:18 -0400)]
hfsplus: switch to rcu-delayed unloading of nls and freeing ->s_fs_info

->d_hash() and ->d_compare() use those, so we need to delay freeing
them.

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agoexfat: move freeing sbi, upcase table and dropping nls into rcu-delayed helper
Al Viro [Tue, 19 Sep 2023 19:53:32 +0000 (15:53 -0400)]
exfat: move freeing sbi, upcase table and dropping nls into rcu-delayed helper

That stuff can be accessed by ->d_hash()/->d_compare(); as it is, we have
a hard-to-hit UAF if rcu pathwalk manages to get into ->d_hash() on a filesystem
that is in process of getting shut down.

Besides, having nls and upcase table cleanup moved from ->put_super() towards
the place where sbi is freed makes for simpler failure exits.

Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agoaffs: free affs_sb_info with kfree_rcu()
Al Viro [Tue, 19 Sep 2023 23:36:07 +0000 (19:36 -0400)]
affs: free affs_sb_info with kfree_rcu()

one of the flags in it is used by ->d_hash()/->d_compare()

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agorcu pathwalk: prevent bogus hard errors from may_lookup()
Al Viro [Sat, 30 Sep 2023 01:11:41 +0000 (21:11 -0400)]
rcu pathwalk: prevent bogus hard errors from may_lookup()

If lazy call of ->permission() returns a hard error, check that
try_to_unlazy() succeeds before returning it.  That both makes
life easier for ->permission() instances and closes the race
in ENOTDIR handling - it is possible that positive d_can_lookup()
seen in link_path_walk() applies to the state *after* unlink() +
mkdir(), while nd->inode matches the state prior to that.

Normally seeing e.g. EACCES from permission check in rcu pathwalk
means that with some timings non-rcu pathwalk would've run into
the same; however, running into a non-executable regular file
in the middle of a pathname would not get to permission check -
it would fail with ENOTDIR instead.

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agofs/super.c: don't drop ->s_user_ns until we free struct super_block itself
Al Viro [Fri, 2 Feb 2024 02:10:01 +0000 (21:10 -0500)]
fs/super.c: don't drop ->s_user_ns until we free struct super_block itself

Avoids fun races in RCU pathwalk...  Same goes for freeing LSM shite
hanging off super_block's arse.

Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
9 months agobcachefs: Fix check_snapshot() memcpy
Kent Overstreet [Sat, 24 Feb 2024 06:18:45 +0000 (01:18 -0500)]
bcachefs: Fix check_snapshot() memcpy

check_snapshot() copies the bch_snapshot to a temporary to easily handle
older versions that don't have all the fields of the current version,
but it lacked a min() to correctly handle keys newer and larger than the
current version.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
9 months agobcachefs: Fix bch2_journal_flush_device_pins()
Kent Overstreet [Sun, 18 Feb 2024 01:38:47 +0000 (20:38 -0500)]
bcachefs: Fix bch2_journal_flush_device_pins()

If a journal write errored, the list of devices it was written to could
be empty - we're not supposed to mark an empty replicas list.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
9 months agobcachefs: fix iov_iter count underflow on sub-block dio read
Brian Foster [Thu, 15 Feb 2024 17:16:05 +0000 (12:16 -0500)]
bcachefs: fix iov_iter count underflow on sub-block dio read

bch2_direct_IO_read() checks the request offset and size for sector
alignment and then falls through to a couple calculations to shrink
the size of the request based on the inode size. The problem is that
these checks round up to the fs block size, which runs the risk of
underflowing iter->count if the block size happens to be large
enough. This is triggered by fstest generic/361 with a 4k block
size, which subsequently leads to a crash. To avoid this crash,
check that the shorten length doesn't exceed the overall length of
the iter.

Fixes:
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Su Yue <glass.su@suse.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>