Message-ID: <805acaca-3b91-438d-b842-25c055fd898c@amd.com>
Date: Thu, 14 Aug 2025 12:22:52 +0200
From: Christian König <christian.koenig@....com>
To: Thomas Zimmermann <tzimmermann@...e.de>,
 Shixiong Ou <oushixiong1025@....com>
Cc: Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
 Maxime Ripard <mripard@...nel.org>, David Airlie <airlied@...il.com>,
 Simona Vetter <simona@...ll.ch>, Sean Paul <sean@...rly.run>,
 Jocelyn Falempe <jfalempe@...hat.com>, dri-devel@...ts.freedesktop.org,
 linux-kernel@...r.kernel.org, Shixiong Ou <oushixiong@...inos.cn>
Subject: Re: [PATCH v5 RESEND 1/3] drm/shmem-helper: Import dmabuf without
 mapping its sg_table

On 13.08.25 16:06, Thomas Zimmermann wrote:
> Hi
> 
> Am 13.08.25 um 15:35 schrieb Christian König:
>> On 13.08.25 14:16, Thomas Zimmermann wrote:
>>> In addition:
>>>
>>>>> Here udl just silently assumes that it got a vaddr, but that isn't necessarily the case. Drivers amdgpu (or radeon, nouveau etc...) can return an io addr as well.
>>>> That's a union of the two pointer types. [1] Using them interchangeably has worked so far (*) and should work here as well. As I said in the other mail, ttm_bo_vmap() never bothers to pin the pages in place and amdgpu might relocated them shortly after.
>>> AFAICT with the old code we ran dma_buf_pin_on_map() at some point. Could we do this in the new version as well?
>> Ah, yes that could be it. So basically the buffer moves and because of that the vmap is unmapped.
>>
>> It would probably be a good idea to call dma_buf_pin() from drm_gem_dmabuf_vmap() before calling drm_gem_vmap_locked().
>>
>> Should be trivial since we already have the dma_buf object and the necessary lock is held as well.
> 
> The attached patch fixes the problem on my test system. It further integrates well with the existing logic, so changes are minimal. Any comments before I send it out for review?

Mhm that looks more or less like what I had in mind initially as well, but I was convinced that this is probably not the best idea.

We have exposed the dma_buf_pin()/dma_buf_unpin() functions for exactly that use case. So we should probably make use of that.

Either by pinning in the exporter, e.g. drm_gem_dmabuf_vmap()/drm_gem_dmabuf_vunmap() or even lower level (amdgpu/ttm) or in the importer.

Not doing it in TTM was intentional since we have use cases where kmap/vmap is only temporary valid until you drop the dma_resv lock again.

Doing it in the importer would be the cleanest approach, but that means changing the semantics of the DMA-buf API including audition all imports using dma_buf_vmap()... so a lot of work.

So that leaves either drm_gem_dmabuf_vmap()/drm_gem_dmabuf_vunmap() or amdgpu. I think drm_gem_dmabuf_vmap()/drm_gem_dmabuf_vunmap() is the cleanest and schould have no impact on other drivers.

Regards,
Christian.

> 
> Best regards
> Thomas
> 
>>
>> Regards,
>> Christian.
>>
>>> [1] https://elixir.bootlin.com/linux/v6.16/source/drivers/dma-buf/dma-buf.c#L1115
>>>
>>> Best regards
>>> Thomas
>>>
>>>> [1] https://elixir.bootlin.com/linux/v6.16/source/include/linux/iosys-map.h#L110
>>>>
>>>> Best regards
>>>> Thomas
>>>>
>>>> (*) White lie: we had problems on exotic architectures, such as sparc, but never on x86.
>>>>
>>>>
>>>>> Regards,
>>>>> Christian.
>>>>>
>>>>>> Best regards
>>>>>> Thomas
>>>>>>
>>>>>>> Best regards,
>>>>>>> Shixiong
>>>>>>>
>>>>>>>
>>>>>>>> [  168.785445] BUG: unable to handle page fault for address: ffffc9012b800000
>>>>>>>> [  168.792311] #PF: supervisor read access in kernel mode
>>>>>>>> [  168.797452] #PF: error_code(0x0000) - not-present page
>>>>>>>> [  168.802586] PGD 100000067 P4D 100000067 PUD 0
>>>>>>>> [  168.807042] Oops: Oops: 0000 [#1] SMP KASAN PTI
>>>>>>>> [  168.811573] CPU: 2 UID: 1000 PID: 2380 Comm: KMS thread Tainted: G            E       6.16.0-rc5-1-default+ #4080 PREEMPT(voluntary)
>>>>>>>> [  168.823537] Tainted: [E]=UNSIGNED_MODULE
>>>>>>>> [  168.827458] Hardware name: System manufacturer System Product Name/Z170-A, BIOS 3802 03/15/2018
>>>>>>>> [  168.836125] RIP: 0010:udl_compress_hline16+0x219/0x940 [udl]
>>>>>>>> [  168.841779] Code: 0f b6 34 28 4c 89 d8 49 d3 e5 83 e0 07 4d 01 dd 83 f9 01 0f 84 4a 03 00 00 83 c0 03 40 38 f0 7c 09 40 84 f6 0f 85 82 05 00 00 <41> 8b 03 4c 63 7c 24 78 4c 89 5c 24 08 89 c6 41 89 c4 c1 e8 08
>>>>>>>> c1
>>>>>>>> [  168.860476] RSP: 0018:ffff88811c7e75c0 EFLAGS: 00010246
>>>>>>>> [  168.865697] RAX: 0000000000000003 RBX: 0000000000000000 RCX: 0000000000000002
>>>>>>>> [  168.872815] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000100
>>>>>>>> [  168.879934] RBP: dffffc0000000000 R08: ffff8881082efe00 R09: ffff8881082e0000
>>>>>>>> [  168.887046] R10: 0000000000000002 R11: ffffc9012b800000 R12: ffff88811c7e76f8
>>>>>>>> [  168.894155] R13: ffffc9012b800400 R14: ffff8881082e0007 R15: 0000000000000000
>>>>>>>> [  168.901266] FS:  00007f4685f3b6c0(0000) GS:ffff88846c690000(0000) knlGS:0000000000000000
>>>>>>>> [  168.909330] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>>>>>>>> [  168.915058] CR2: ffffc9012b800000 CR3: 0000000117944004 CR4: 00000000003706f0
>>>>>>>> [  168.922170] Call Trace:
>>>>>>>> [  168.924616]  <TASK>
>>>>>>>> [  168.926714]  ? validate_chain+0x24e/0x5e0
>>>>>>>> [  168.930718]  ? __lock_acquire+0x568/0xae0
>>>>>>>> [  168.934725]  udl_render_hline+0x165/0x33b [udl]
>>>>>>>> [  168.939256]  ? __pfx_udl_render_hline+0x10/0x10 [udl]
>>>>>>>> [  168.944297]  ? local_clock_noinstr+0xb/0x100
>>>>>>>> [  168.948557]  ? __lock_release.isra.0+0x16c/0x2e0
>>>>>>>> [  168.953162]  ? mark_held_locks+0x40/0x70
>>>>>>>> [  168.957077]  ? lockdep_hardirqs_on_prepare.part.0+0x92/0x170
>>>>>>>> [  168.962721] udl_primary_plane_helper_atomic_update+0x432/0x670 [udl]
>>>>>>>> [  168.969145]  ? __pfx_udl_primary_plane_helper_atomic_update+0x10/0x10 [udl]
>>>>>>>> [  168.976089]  ? __pfx___drm_dev_dbg+0x10/0x10
>>>>>>>> [  168.980357]  ? drm_atomic_helper_calc_timestamping_constants+0x141/0x200
>>>>>>>> [  168.987044]  ? drm_atomic_helper_commit_planes+0x3b6/0x1030
>>>>>>>> [  168.992599] drm_atomic_helper_commit_planes+0x3b6/0x1030
>>>>>>>> [  168.997987]  drm_atomic_helper_commit_tail+0x41/0xb0
>>>>>>>> [  169.002943]  commit_tail+0x204/0x330
>>>>>>>> [  169.006513]  drm_atomic_helper_commit+0x242/0x2e0
>>>>>>>> [  169.011203]  ? __pfx_drm_atomic_helper_commit+0x10/0x10
>>>>>>>> [  169.016413]  drm_atomic_commit+0x1e1/0x290
>>>>>>>> [  169.020500]  ? prepare_signaling+0x355/0xda0
>>>>>>>> [  169.024769]  ? __pfx_drm_atomic_commit+0x10/0x10
>>>>>>>> [  169.029372]  ? __pfx___drm_printfn_info+0x10/0x10
>>>>>>>> [  169.034069]  drm_mode_atomic_ioctl+0x8ff/0xe40
>>>>>>>> [  169.038510]  ? __pfx_drm_mode_atomic_ioctl+0x10/0x10
>>>>>>>> [  169.043466]  ? find_held_lock+0x2b/0x80
>>>>>>>> [  169.047295]  ? __lock_acquire+0x568/0xae0
>>>>>>>> [  169.051293]  ? mark_usage+0x65/0x180
>>>>>>>> [  169.054870]  ? __pfx_drm_mode_atomic_ioctl+0x10/0x10
>>>>>>>> [  169.059823]  ? do_raw_spin_unlock+0x55/0x230
>>>>>>>> [  169.064081]  ? drm_is_current_master+0x26/0x30
>>>>>>>> [  169.068517]  ? __pfx_drm_mode_atomic_ioctl+0x10/0x10
>>>>>>>> [  169.073465]  drm_ioctl_kernel+0x141/0x2b0
>>>>>>>> [  169.077468]  ? __pfx_drm_ioctl_kernel+0x10/0x10
>>>>>>>> [  169.081987]  ? lock_release.part.0+0x47/0x90
>>>>>>>> [  169.086249]  drm_ioctl+0x481/0xb50
>>>>>>>> [  169.089653]  ? __pfx_drm_mode_atomic_ioctl+0x10/0x10
>>>>>>>> [  169.094610]  ? __pfx_drm_ioctl+0x10/0x10
>>>>>>>> [  169.098525]  ? find_held_lock+0x2b/0x80
>>>>>>>> [  169.102356]  ? lock_release.part.0+0x47/0x90
>>>>>>>> [  169.106621]  ? __fget_files+0x1aa/0x2f0
>>>>>>>> [  169.110450]  ? __fget_files+0x1b4/0x2f0
>>>>>>>> [  169.114281]  __x64_sys_ioctl+0x135/0x1c0
>>>>>>>> [  169.118201]  do_syscall_64+0x68/0x2a0
>>>>>>>> [  169.121856]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
>>>>>>>> [  169.126892] RIP: 0033:0x7f469391a53f
>>>>>>>> [  169.130460] Code: 00 48 89 44 24 18 31 c0 48 8d 44 24 60 c7 04 24 10 00 00 00 48 89 44 24 08 48 8d 44 24 20 48 89 44 24 10 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 18 48 8b 44 24 18 64 48 2b 04 25 28 00
>>>>>>>> 00
>>>>>>>> [  169.149151] RSP: 002b:00007f4685f39a80 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
>>>>>>>> [  169.156703] RAX: ffffffffffffffda RBX: 00007f465c002da0 RCX: 00007f469391a53f
>>>>>>>> [  169.163868] RDX: 00007f4685f39b20 RSI: 00000000c03864bc RDI: 0000000000000010
>>>>>>>> [  169.171029] RBP: 00007f4685f39b20 R08: 0000000000001310 R09: 00007f465c044cf0
>>>>>>>> [  169.178194] R10: 0000000000000002 R11: 0000000000000246 R12: 00000000c03864bc
>>>>>>>> [  169.185358] R13: 0000000000000010 R14: 00007f465c0284b0 R15: 00007f45d402b110
>>>>>>>> [  169.192525]  </TASK>
>>>>>>>> [  169.194732] Modules linked in: udl(E) snd_seq_dummy(E) snd_hrtimer(E) snd_seq(E) snd_seq_device(E) af_packet(E) nf_tables(E) iptable_filter(E) binfmt_misc(E) intel_rapl_msr(E) nls_iso8859_1(E) eeepc_wmi(E) nls
>>>>>>>> _cp437(E) intel_rapl_common(E) snd_hda_codec_realtek(E) asus_wmi(E) iTCO_wdt(E) vfat(E) ee1004(E) snd_hda_codec_generic(E) sparse_keymap(E) x86_pkg_temp_thermal(E) iTCO_vendor_support(E) snd_hda_scodec_component(
>>>>>>>> E) intel_powerclamp(E) snd_hda_codec_hdmi(E) platform_profile(E) fat(E) e1000e(E) i2c_i801(E) ptp(E) battery(E) snd_hda_intel(E) i2c_smbus(E) snd_intel_dspcfg(E) mxm_wmi(E) rfkill(E) wmi_bmof(E) intel_wmi_thunder
>>>>>>>> bolt(E) coretemp(E) pps_core(E) i2c_mux(E) pcspkr(E) snd_hda_codec(E) xfs(E) snd_hda_core(E) snd_hwdep(E) snd_pcm(E) snd_timer(E) snd(E) soundcore(E) mei_me(E) acpi_pad(E) button(E) mei(E) joydev(E) nvme_fabrics(
>>>>>>>> E) loop(E) fuse(E) efi_pstore(E) dm_mod(E) configfs(E) nfnetlink(E) ip_tables(E) x_tables(E) amdgpu(E) amdxcp(E) i2c_algo_bit(E) drm_ttm_helper(E) ttm(E) drm_exec(E) hid_generic(E)
>>>>>>>> [  169.194874]  gpu_sched(E) drm_suballoc_helper(E) ghash_clmulni_intel(E) sha512_ssse3(E) video(E) sha1_ssse3(E) aesni_intel(E) usbhid(E) drm_panel_backlight_quirks(E) drm_buddy(E) drm_display_helper(E) cec(E) w
>>>>>>>> mi(E) btrfs(E) blake2b_generic(E) xor(E) raid6_pq(E) msr(E) i2c_dev(E) efivarfs(E) dmi_sysfs(E)
>>>>>>>> [  169.311501] CR2: ffffc9012b800000
>>>>>>>> [  169.314835] ---[ end trace 0000000000000000 ]---
>>>>>>>> [  169.434549] RIP: 0010:udl_compress_hline16+0x219/0x940 [udl]
>>>>>>>> [  169.440237] Code: 0f b6 34 28 4c 89 d8 49 d3 e5 83 e0 07 4d 01 dd 83 f9 01 0f 84 4a 03 00 00 83 c0 03 40 38 f0 7c 09 40 84 f6 0f 85 82 05 00 00 <41> 8b 03 4c 63 7c 24 78 4c 89 5c 24 08 89 c6 41 89 c4 c1 e8 08
>>>>>>>> c1
>>>>>>>> [  169.459062] RSP: 0018:ffff88811c7e75c0 EFLAGS: 00010246
>>>>>>>> [  169.464309] RAX: 0000000000000003 RBX: 0000000000000000 RCX: 0000000000000002
>>>>>>>> [  169.471474] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000100
>>>>>>>> [  169.478635] RBP: dffffc0000000000 R08: ffff8881082efe00 R09: ffff8881082e0000
>>>>>>>> [  169.485800] R10: 0000000000000002 R11: ffffc9012b800000 R12: ffff88811c7e76f8
>>>>>>>> [  169.492962] R13: ffffc9012b800400 R14: ffff8881082e0007 R15: 0000000000000000
>>>>>>>> [  169.500126] FS:  00007f4685f3b6c0(0000) GS:ffff88846c690000(0000) knlGS:0000000000000000
>>>>>>>> [  169.508246] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>>>>>>>> [  169.514014] CR2: ffffc9012b800000 CR3: 0000000117944004 CR4: 00000000003706f0
>>>>>>>> [  169.521180] note: KMS thread[2380] exited with irqs disabled
>>>>>>>> [  175.343111] ==================================================================
>>>>>>>> [  175.350342] BUG: KASAN: slab-use-after-free in mutex_can_spin_on_owner+0x1a6/0x1c0
>>>>>>>> [  175.357886] Read of size 4 at addr ffff8881172daff4 by task kworker/5:0/49
>>>>>>>> [  175.364738]
>>>>>>>> [  175.366235] CPU: 5 UID: 0 PID: 49 Comm: kworker/5:0 Tainted: G      D     E       6.16.0-rc5-1-default+ #4080 PREEMPT(voluntary)
>>>>>>>> [  175.366240] Tainted: [D]=DIE, [E]=UNSIGNED_MODULE
>>>>>>>> [  175.366242] Hardware name: System manufacturer System Product Name/Z170-A, BIOS 3802 03/15/2018
>>>>>>>> [  175.366244] Workqueue: events output_poll_execute
>>>>>>>> [  175.366249] Call Trace:
>>>>>>>> [  175.366251]  <TASK>
>>>>>>>> [  175.366254]  dump_stack_lvl+0x68/0x90
>>>>>>>> [  175.366259]  ? mutex_can_spin_on_owner+0x1a6/0x1c0
>>>>>>>> [  175.366261] print_address_description.constprop.0+0x88/0x380
>>>>>>>> [  175.366266]  ? lock_acquire+0xf2/0x140
>>>>>>>> [  175.366269]  ? mutex_can_spin_on_owner+0x1a6/0x1c0
>>>>>>>> [  175.366272]  print_report+0xf8/0x1e2
>>>>>>>> [  175.366275]  ? __virt_addr_valid+0x22e/0x500
>>>>>>>> [  175.366279]  ? kasan_addr_to_slab+0x9/0x90
>>>>>>>> [  175.366282]  ? mutex_can_spin_on_owner+0x1a6/0x1c0
>>>>>>>> [  175.366284]  kasan_report+0xd8/0x190
>>>>>>>> [  175.366288]  ? mutex_can_spin_on_owner+0x1a6/0x1c0
>>>>>>>> [  175.366294]  mutex_can_spin_on_owner+0x1a6/0x1c0
>>>>>>>> [  175.366297]  __ww_mutex_lock.constprop.0+0x2f8/0x34d0
>>>>>>>> [  175.366301]  ? do_raw_spin_trylock+0xa2/0x160
>>>>>>>> [  175.366304]  ? __pfx_do_raw_spin_trylock+0x10/0x10
>>>>>>>> [  175.366308]  ? get_nohz_timer_target+0x28/0x3d0
>>>>>>>> [  175.366311]  ? modeset_lock+0x3c6/0x640
>>>>>>>> [  175.366316]  ? __pfx___ww_mutex_lock.constprop.0+0x10/0x10
>>>>>>>> [  175.366320]  ? rcu_is_watching+0x11/0xb0
>>>>>>>> [  175.366323]  ? timerqueue_add+0x154/0x3c0
>>>>>>>> [  175.366328]  ? __hrtimer_start_range_ns+0x2e1/0x750
>>>>>>>> [  175.366331]  ? rcu_is_watching+0x11/0xb0
>>>>>>>> [  175.366334]  ? lock_acquired+0xb6/0xf0
>>>>>>>> [  175.366337]  ? rcu_is_watching+0x11/0xb0
>>>>>>>> [  175.366340]  ? rcu_is_watching+0x11/0xb0
>>>>>>>> [  175.366342]  ? drm_helper_probe_detect_ctx+0x6d/0x1a0
>>>>>>>> [  175.366345]  ? lock_acquire+0xf2/0x140
>>>>>>>> [  175.366349]  ? ww_mutex_lock+0x27/0x150
>>>>>>>> [  175.366352]  ? drm_helper_probe_detect_ctx+0x6d/0x1a0
>>>>>>>> [  175.366355]  ww_mutex_lock+0x27/0x150
>>>>>>>> [  175.366358]  modeset_lock+0x3c6/0x640
>>>>>>>> [  175.366362]  drm_helper_probe_detect_ctx+0xa6/0x1a0
>>>>>>>> [  175.366366]  ? __pfx_drm_helper_probe_detect_ctx+0x10/0x10
>>>>>>>> [  175.366375]  ? __pfx_drm_connector_list_iter_next+0x10/0x10
>>>>>>>> [  175.366381]  output_poll_execute+0x29b/0x760
>>>>>>>> [  175.366387]  ? trace_hardirqs_on+0x14/0x150
>>>>>>>> [  175.366391]  ? __pfx_output_poll_execute+0x10/0x10
>>>>>>>> [  175.366396]  process_one_work+0x7b5/0x1390
>>>>>>>> [  175.366404]  ? __pfx_process_one_work+0x10/0x10
>>>>>>>> [  175.366409]  ? assign_work+0x156/0x390
>>>>>>>> [  175.366413]  worker_thread+0x58d/0xf60
>>>>>>>> [  175.366420]  ? __pfx_worker_thread+0x10/0x10
>>>>>>>> [  175.366422]  kthread+0x370/0x720
>>>>>>>> [  175.366425]  ? __pfx_kthread+0x10/0x10
>>>>>>>> [  175.366428]  ? local_clock_noinstr+0x56/0x100
>>>>>>>> [  175.366431]  ? local_clock+0x11/0x30
>>>>>>>> [  175.366433]  ? __lock_release.isra.0+0x16c/0x2e0
>>>>>>>> [  175.366437]  ? rcu_is_watching+0x11/0xb0
>>>>>>>> [  175.366439]  ? lockdep_hardirqs_on_prepare.part.0+0x92/0x170
>>>>>>>> [  175.366442]  ? __pfx_kthread+0x10/0x10
>>>>>>>> [  175.366445]  ret_from_fork+0x1f4/0x2f0
>>>>>>>> [  175.366448]  ? __pfx_kthread+0x10/0x10
>>>>>>>> [  175.366450]  ret_from_fork_asm+0x1a/0x30
>>>>>>>> [  175.366459]  </TASK>
>>>>>>>> [  175.366460]
>>>>>>>> [  175.632772] Allocated by task 2342:
>>>>>>>> [  175.636282]  kasan_save_stack+0x1c/0x40
>>>>>>>> [  175.640137]  kasan_save_track+0x10/0x30
>>>>>>>> [  175.643992]  __kasan_slab_alloc+0x5f/0x70
>>>>>>>> [  175.648023]  kmem_cache_alloc_node_noprof+0x13a/0x380
>>>>>>>> [  175.653097]  dup_task_struct+0x32/0x730
>>>>>>>> [  175.656952]  copy_process+0x2d8/0x5380
>>>>>>>> [  175.660720]  kernel_clone+0x9f/0x5e0
>>>>>>>> [  175.664318]  __do_sys_clone3+0x135/0x180
>>>>>>>> [  175.668258]  do_syscall_64+0x68/0x2a0
>>>>>>>> [  175.671940]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
>>>>>>>> [  175.677014]
>>>>>>>> [  175.678520] Freed by task 0:
>>>>>>>> [  175.681418]  kasan_save_stack+0x1c/0x40
>>>>>>>> [  175.685273]  kasan_save_track+0x10/0x30
>>>>>>>> [  175.689130]  kasan_save_free_info+0x37/0x70
>>>>>>>> [  175.693333]  __kasan_slab_free+0x33/0x40
>>>>>>>> [  175.697278]  kmem_cache_free+0x10b/0x4d0
>>>>>>>> [  175.701221]  delayed_put_task_struct+0x15e/0x1e0
>>>>>>>> [  175.705858]  rcu_do_batch+0x2e3/0xb30
>>>>>>>> [  175.709542]  rcu_core+0x51d/0xb60
>>>>>>>> [  175.712873]  handle_softirqs+0x1a2/0x6b0
>>>>>>>> [  175.716817]  __irq_exit_rcu+0xf7/0x160
>>>>>>>> [  175.720585]  irq_exit_rcu+0xa/0x30
>>>>>>>> [  175.724006]  sysvec_apic_timer_interrupt+0x9d/0xc0
>>>>>>>> [  175.728820]  asm_sysvec_apic_timer_interrupt+0x16/0x20
>>>>>>>> [  175.733980]
>>>>>>>> [  175.735486] Last potentially related work creation:
>>>>>>>> [  175.740385]  kasan_save_stack+0x1c/0x40
>>>>>>>> [  175.744243]  kasan_record_aux_stack+0x88/0xa0
>>>>>>>> [  175.748618]  __call_rcu_common.constprop.0+0x77/0x850
>>>>>>>> [  175.753693]  __schedule+0x887/0x1d00
>>>>>>>> [  175.757287]  schedule+0xd0/0x260
>>>>>>>> [  175.760533]  smpboot_thread_fn+0x583/0x7a0
>>>>>>>> [  175.764650]  kthread+0x370/0x720
>>>>>>>> [  175.767896]  ret_from_fork+0x1f4/0x2f0
>>>>>>>> [  175.771665]  ret_from_fork_asm+0x1a/0x30
>>>>>>>> [  175.775608]
>>>>>>>> [  175.777113] Second to last potentially related work creation:
>>>>>>>> [  175.782886]  kasan_save_stack+0x1c/0x40
>>>>>>>> [  175.786739]  kasan_record_aux_stack+0x88/0xa0
>>>>>>>> [  175.791118]  task_work_add+0x1b1/0x270
>>>>>>>> [  175.794886]  sched_tick+0x226/0x6f0
>>>>>>>> [  175.798394]  update_process_times+0xe9/0x1f0
>>>>>>>> [  175.802685]  tick_nohz_handler+0x1a6/0x4b0
>>>>>>>> [  175.806801]  __hrtimer_run_queues+0x161/0x960
>>>>>>>> [  175.811181]  hrtimer_interrupt+0x33e/0x880
>>>>>>>> [  175.815295]  __sysvec_apic_timer_interrupt+0xf6/0x390
>>>>>>>> [  175.820370]  sysvec_apic_timer_interrupt+0x98/0xc0
>>>>>>>> [  175.825183]  asm_sysvec_apic_timer_interrupt+0x16/0x20
>>>>>>>> [  175.830343]
>>>>>>>> [  175.831849] The buggy address belongs to the object at ffff8881172dafc0
>>>>>>>> [  175.831849]  which belongs to the cache task_struct of size 11968
>>>>>>>> [  175.844582] The buggy address is located 52 bytes inside of
>>>>>>>> [  175.844582]  freed 11968-byte region [ffff8881172dafc0, ffff8881172dde80)
>>>>>>>> [  175.856969]
>>>>>>>> [  175.858474] The buggy address belongs to the physical page:
>>>>>>>> [  175.864070] page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1172d8
>>>>>>>> [  175.872104] head: order:3 mapcount:0 entire_mapcount:0 nr_pages_mapped:0 pincount:0
>>>>>>>> [  175.879788] memcg:ffff88811068a301
>>>>>>>> [  175.883209] anon flags: 0x2ffff800000040(head|node=0|zone=2|lastcpupid=0x1ffff)
>>>>>>>> [  175.890547] page_type: f5(slab)
>>>>>>>> [  175.893706] raw: 002ffff800000040 ffff888100930500 0000000000000000 0000000000000001
>>>>>>>> [  175.901479] raw: 0000000000000000 0000000000020002 00000000f5000000 ffff88811068a301
>>>>>>>> [  175.909249] head: 002ffff800000040 ffff888100930500 0000000000000000 0000000000000001
>>>>>>>> [  175.917109] head: 0000000000000000 0000000000020002 00000000f5000000 ffff88811068a301
>>>>>>>> [  175.924968] head: 002ffff800000003 ffffea00045cb601 00000000ffffffff 00000000ffffffff
>>>>>>>> [  175.932827] head: ffffffffffffffff 0000000000000000 00000000ffffffff 0000000000000008
>>>>>>>> [  175.940685] page dumped because: kasan: bad access detected
>>>>>>>> [  175.946283]
>>>>>>>> [  175.947787] Memory state around the buggy address:
>>>>>>>> [  175.952603]  ffff8881172dae80: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
>>>>>>>> [  175.959854]  ffff8881172daf00: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>>>>>>>> [  175.967101] >ffff8881172daf80: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb
>>>>>>>> [ 175.974353] ^
>>>>>>>> [  175.981256]  ffff8881172db000: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>>>>>>>> [  175.988504]  ffff8881172db080: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>>>>>>>> [  175.995755] ==================================================================
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Am 22.05.25 um 09:07 schrieb oushixiong1025@....com:
>>>>>>>>> From: Shixiong Ou <oushixiong@...inos.cn>
>>>>>>>>>
>>>>>>>>> [WHY]
>>>>>>>>> 1. Drivers using DRM_GEM_SHADOW_PLANE_HELPER_FUNCS and
>>>>>>>>>       DRM_GEM_SHMEM_DRIVER_OPS (e.g., udl, ast) do not require
>>>>>>>>>       sg_table import.
>>>>>>>>>       They only need dma_buf_vmap() to access the shared buffer's
>>>>>>>>>       kernel virtual address.
>>>>>>>>>
>>>>>>>>> 2. On certain Aspeed-based boards, a dma_mask of 0xffff_ffff may
>>>>>>>>>       trigger SWIOTLB during dmabuf import. However, IO_TLB_SEGSIZE
>>>>>>>>>       restricts the maximum DMA streaming mapping memory, resulting in
>>>>>>>>>       errors like:
>>>>>>>>>
>>>>>>>>>       ast 0000:07:00.0: swiotlb buffer is full (sz: 3145728 bytes), total 32768 (slots), used 0 (slots)
>>>>>>>>>
>>>>>>>>> [HOW]
>>>>>>>>> Provide a gem_prime_import implementation without sg_table mapping
>>>>>>>>> to avoid issues (e.g., "swiotlb buffer is full"). Drivers that do not
>>>>>>>>> require sg_table can adopt this.
>>>>>>>>>
>>>>>>>>> Signed-off-by: Shixiong Ou <oushixiong@...inos.cn>
>>>>>>>>> ---
>>>>>>>>> v1->v2:
>>>>>>>>>       Patch rebase.
>>>>>>>>> v2->v3:
>>>>>>>>>       Rename the import callback function.
>>>>>>>>>       Remove drm_gem_shmem_prime_export() and separate some codes
>>>>>>>>>       to drm_gem_prime_import_self().
>>>>>>>>> v3->v4:
>>>>>>>>>       Separate the test from the policy.
>>>>>>>>>       Rename the macro.
>>>>>>>>> v4->v5:
>>>>>>>>>       Rename some functions.
>>>>>>>>>
>>>>>>>>>     drivers/gpu/drm/drm_gem_shmem_helper.c | 57 ++++++++++++++++++++++++++
>>>>>>>>>     drivers/gpu/drm/drm_prime.c            | 36 ++++++++++++----
>>>>>>>>>     include/drm/drm_gem_shmem_helper.h     | 15 +++++++
>>>>>>>>>     include/drm/drm_prime.h                |  3 ++
>>>>>>>>>     4 files changed, 102 insertions(+), 9 deletions(-)
>>>>>>>>>
>>>>>>>>> diff --git a/drivers/gpu/drm/drm_gem_shmem_helper.c b/drivers/gpu/drm/drm_gem_shmem_helper.c
>>>>>>>>> index aa43265f4f4f..126aa79042ad 100644
>>>>>>>>> --- a/drivers/gpu/drm/drm_gem_shmem_helper.c
>>>>>>>>> +++ b/drivers/gpu/drm/drm_gem_shmem_helper.c
>>>>>>>>> @@ -800,6 +800,63 @@ drm_gem_shmem_prime_import_sg_table(struct drm_device *dev,
>>>>>>>>>     }
>>>>>>>>> EXPORT_SYMBOL_GPL(drm_gem_shmem_prime_import_sg_table);
>>>>>>>>>     +/**
>>>>>>>>> + * drm_gem_shmem_prime_import_no_map - Import dmabuf without mapping its sg_table
>>>>>>>>> + * @dev: Device to import into
>>>>>>>>> + * @dma_buf: dma-buf object to import
>>>>>>>>> + *
>>>>>>>>> + * Drivers that use the shmem helpers but also wants to import dmabuf without
>>>>>>>>> + * mapping its sg_table can use this as their &drm_driver.gem_prime_import
>>>>>>>>> + * implementation.
>>>>>>>>> + */
>>>>>>>>> +struct drm_gem_object *drm_gem_shmem_prime_import_no_map(struct drm_device *dev,
>>>>>>>>> +                             struct dma_buf *dma_buf)
>>>>>>>>> +{
>>>>>>>>> +    struct dma_buf_attachment *attach;
>>>>>>>>> +    struct drm_gem_shmem_object *shmem;
>>>>>>>>> +    struct drm_gem_object *obj;
>>>>>>>>> +    size_t size;
>>>>>>>>> +    int ret;
>>>>>>>>> +
>>>>>>>>> +    if (drm_gem_is_prime_exported_dma_buf(dev, dma_buf)) {
>>>>>>>>> +        /*
>>>>>>>>> +         * Importing dmabuf exported from our own gem increases
>>>>>>>>> +         * refcount on gem itself instead of f_count of dmabuf.
>>>>>>>>> +         */
>>>>>>>>> +        obj = dma_buf->priv;
>>>>>>>>> +        drm_gem_object_get(obj);
>>>>>>>>> +        return obj;
>>>>>>>>> +    }
>>>>>>>>> +
>>>>>>>>> +    attach = dma_buf_attach(dma_buf, dev->dev);
>>>>>>>>> +    if (IS_ERR(attach))
>>>>>>>>> +        return ERR_CAST(attach);
>>>>>>>>> +
>>>>>>>>> +    get_dma_buf(dma_buf);
>>>>>>>>> +
>>>>>>>>> +    size = PAGE_ALIGN(attach->dmabuf->size);
>>>>>>>>> +
>>>>>>>>> +    shmem = __drm_gem_shmem_create(dev, size, true, NULL);
>>>>>>>>> +    if (IS_ERR(shmem)) {
>>>>>>>>> +        ret = PTR_ERR(shmem);
>>>>>>>>> +        goto fail_detach;
>>>>>>>>> +    }
>>>>>>>>> +
>>>>>>>>> +    drm_dbg_prime(dev, "size = %zu\n", size);
>>>>>>>>> +
>>>>>>>>> +    shmem->base.import_attach = attach;
>>>>>>>>> +    shmem->base.resv = dma_buf->resv;
>>>>>>>>> +
>>>>>>>>> +    return &shmem->base;
>>>>>>>>> +
>>>>>>>>> +fail_detach:
>>>>>>>>> +    dma_buf_detach(dma_buf, attach);
>>>>>>>>> +    dma_buf_put(dma_buf);
>>>>>>>>> +
>>>>>>>>> +    return ERR_PTR(ret);
>>>>>>>>> +}
>>>>>>>>> +EXPORT_SYMBOL_GPL(drm_gem_shmem_prime_import_no_map);
>>>>>>>>> +
>>>>>>>>>     MODULE_DESCRIPTION("DRM SHMEM memory-management helpers");
>>>>>>>>>     MODULE_IMPORT_NS("DMA_BUF");
>>>>>>>>>     MODULE_LICENSE("GPL v2");
>>>>>>>>> diff --git a/drivers/gpu/drm/drm_prime.c b/drivers/gpu/drm/drm_prime.c
>>>>>>>>> index d828502268b8..b825b71038d6 100644
>>>>>>>>> --- a/drivers/gpu/drm/drm_prime.c
>>>>>>>>> +++ b/drivers/gpu/drm/drm_prime.c
>>>>>>>>> @@ -910,6 +910,26 @@ struct dma_buf *drm_gem_prime_export(struct drm_gem_object *obj,
>>>>>>>>>     }
>>>>>>>>>     EXPORT_SYMBOL(drm_gem_prime_export);
>>>>>>>>>     +
>>>>>>>>> +/**
>>>>>>>>> + * drm_gem_is_prime_exported_dma_buf -
>>>>>>>>> + * checks if the DMA-BUF was exported from a GEM object belonging to @dev.
>>>>>>>>> + * @dev: drm_device to check against
>>>>>>>>> + * @dma_buf: dma-buf object to import
>>>>>>>>> + *
>>>>>>>>> + * Return: true if the DMA-BUF was exported from a GEM object belonging
>>>>>>>>> + * to @dev, false otherwise.
>>>>>>>>> + */
>>>>>>>>> +
>>>>>>>>> +bool drm_gem_is_prime_exported_dma_buf(struct drm_device *dev,
>>>>>>>>> +                       struct dma_buf *dma_buf)
>>>>>>>>> +{
>>>>>>>>> +    struct drm_gem_object *obj = dma_buf->priv;
>>>>>>>>> +
>>>>>>>>> +    return (dma_buf->ops == &drm_gem_prime_dmabuf_ops) && (obj->dev == dev);
>>>>>>>>> +}
>>>>>>>>> +EXPORT_SYMBOL(drm_gem_is_prime_exported_dma_buf);
>>>>>>>>> +
>>>>>>>>>     /**
>>>>>>>>>      * drm_gem_prime_import_dev - core implementation of the import callback
>>>>>>>>>      * @dev: drm_device to import into
>>>>>>>>> @@ -933,16 +953,14 @@ struct drm_gem_object *drm_gem_prime_import_dev(struct drm_device *dev,
>>>>>>>>>         struct drm_gem_object *obj;
>>>>>>>>>         int ret;
>>>>>>>>>     -    if (dma_buf->ops == &drm_gem_prime_dmabuf_ops) {
>>>>>>>>> +    if (drm_gem_is_prime_exported_dma_buf(dev, dma_buf)) {
>>>>>>>>> +        /*
>>>>>>>>> +         * Importing dmabuf exported from our own gem increases
>>>>>>>>> +         * refcount on gem itself instead of f_count of dmabuf.
>>>>>>>>> +         */
>>>>>>>>>             obj = dma_buf->priv;
>>>>>>>>> -        if (obj->dev == dev) {
>>>>>>>>> -            /*
>>>>>>>>> -             * Importing dmabuf exported from our own gem increases
>>>>>>>>> -             * refcount on gem itself instead of f_count of dmabuf.
>>>>>>>>> -             */
>>>>>>>>> -            drm_gem_object_get(obj);
>>>>>>>>> -            return obj;
>>>>>>>>> -        }
>>>>>>>>> +        drm_gem_object_get(obj);
>>>>>>>>> +        return obj;
>>>>>>>>>         }
>>>>>>>>>           if (!dev->driver->gem_prime_import_sg_table)
>>>>>>>>> diff --git a/include/drm/drm_gem_shmem_helper.h b/include/drm/drm_gem_shmem_helper.h
>>>>>>>>> index b4f993da3cae..35f7466dca84 100644
>>>>>>>>> --- a/include/drm/drm_gem_shmem_helper.h
>>>>>>>>> +++ b/include/drm/drm_gem_shmem_helper.h
>>>>>>>>> @@ -287,6 +287,8 @@ drm_gem_shmem_prime_import_sg_table(struct drm_device *dev,
>>>>>>>>>                         struct sg_table *sgt);
>>>>>>>>>     int drm_gem_shmem_dumb_create(struct drm_file *file, struct drm_device *dev,
>>>>>>>>>                       struct drm_mode_create_dumb *args);
>>>>>>>>> +struct drm_gem_object *drm_gem_shmem_prime_import_no_map(struct drm_device *dev,
>>>>>>>>> +                             struct dma_buf *buf);
>>>>>>>>>       /**
>>>>>>>>>      * DRM_GEM_SHMEM_DRIVER_OPS - Default shmem GEM operations
>>>>>>>>> @@ -298,4 +300,17 @@ int drm_gem_shmem_dumb_create(struct drm_file *file, struct drm_device *dev,
>>>>>>>>>         .gem_prime_import_sg_table = drm_gem_shmem_prime_import_sg_table, \
>>>>>>>>>         .dumb_create           = drm_gem_shmem_dumb_create
>>>>>>>>>     +/**
>>>>>>>>> + * DRM_GEM_SHMEM_DRIVER_OPS_NO_MAP_SGT - shmem GEM operations
>>>>>>>>> + *                                       without mapping sg_table on
>>>>>>>>> + *                                       imported buffer.
>>>>>>>>> + *
>>>>>>>>> + * This macro provides a shortcut for setting the shmem GEM operations in
>>>>>>>>> + * the &drm_driver structure for drivers that do not require a sg_table on
>>>>>>>>> + * imported buffers.
>>>>>>>>> + */
>>>>>>>>> +#define DRM_GEM_SHMEM_DRIVER_OPS_NO_MAP_SGT \
>>>>>>>>> +    .gem_prime_import       = drm_gem_shmem_prime_import_no_map, \
>>>>>>>>> +    .dumb_create            = drm_gem_shmem_dumb_create
>>>>>>>>> +
>>>>>>>>>     #endif /* __DRM_GEM_SHMEM_HELPER_H__ */
>>>>>>>>> diff --git a/include/drm/drm_prime.h b/include/drm/drm_prime.h
>>>>>>>>> index fa085c44d4ca..f50f862f0d8b 100644
>>>>>>>>> --- a/include/drm/drm_prime.h
>>>>>>>>> +++ b/include/drm/drm_prime.h
>>>>>>>>> @@ -100,6 +100,9 @@ struct dma_buf *drm_gem_prime_export(struct drm_gem_object *obj,
>>>>>>>>>     unsigned long drm_prime_get_contiguous_size(struct sg_table *sgt);
>>>>>>>>>       /* helper functions for importing */
>>>>>>>>> +bool drm_gem_is_prime_exported_dma_buf(struct drm_device *dev,
>>>>>>>>> +                       struct dma_buf *dma_buf);
>>>>>>>>> +
>>>>>>>>>     struct drm_gem_object *drm_gem_prime_import_dev(struct drm_device *dev,
>>>>>>>>>                             struct dma_buf *dma_buf,
>>>>>>>>>                             struct device *attach_dev);
> 

Powered by blists - more mailing lists