Re: [PATCH] staging: media: atomisp: Fix premature setting of HMM_BO_DEVICE_INITED flag

[Abdelrahman Fekry]

Hi andy ,
On Tue, Jul 1, 2025 at 3:45 PM Andy Shevchenko
<andriy.shevchenko@xxxxxxxxx> wrote:

> > > Nice. Can you make some fault injection (temporary by modifying the
> > > code to always fail, for example) and actually prove this in practice?
> > > If so, the few (important) lines from the given Oops would be nice to
> > > have here.
>
> > I have been trying to test it without having any intel atomisp
> > hardware and failed continuously, do you have any tips or maybe some
> > resources on how i can test this driver.
>
> So, the easiest way as I can see it is to ask people who possess the HW to
> test, but you need to provide a testing patch (which can be applied on top
> of this one, for example).
>

Well, after several hours of trial and error, I finally managed to
find a workaround that allowed me to test the scenario. As expected,
the system crashed exactly at the point we discussed. I was able to
capture the kernel panic log, which is shown below.

To simulate the issue, I injected a failure right after setting the
HMM_BO_DEVICE_INITED flag, this mimics a failure in one of the
subsequent initialization steps. Then, I wrote a test module that
calls the hmm_init() function directly. As anticipated, the kernel
panicked at the hmm_alloc(1) call inside hmm_init().

Here’s the relevant panic log:
[  161.802542] atomisp: loading out-of-tree module taints kernel.
[  161.823358] ===== HMM BO DEVICE TEST =====
[  161.823666] (NULL device *): Simulated failure for testing purposes.
[  161.824064] (NULL device *): invalid L1PT: pte = 0x7fffffff
[  161.824427] (NULL device *): hmm_bo_device_init failed.
[  161.824818] BUG: kernel NULL pointer dereference, address: 0000000000000020
[  161.825309] #PF: supervisor read access in kernel mode
[  161.825693] #PF: error_code(0x0000) - not-present page
[  161.826100] PGD 0 P4D 0
[  161.826237] Oops: Oops: 0000 [#1] SMP PTI
[  161.826482] CPU: 2 UID: 0 PID: 3688 Comm: modprobe Kdump: loaded
Tainted: G           O        6.16.0-rc4+ #2 PREEMPT(voluntary)
[  161.827445] Tainted: [O]=OOT_MODULE
[  161.827650] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS
VirtualBox 12/01/2006
[  161.828273] RIP:
0010:__bo_search_and_remove_from_free_rbtree+0xf/0xd0 [atomisp]
[  161.828977] Code: 2e 0f 1f 84 00 00 00 00 00 66 90 90 90 90 90 90
90 90 90 90 90 90 90 90 90 90 90 0f 1f 44 00 00 55 48 89 e5 41 54 49
89 fc 53 <8b> 47 20 39 f0 74 46 89 f3 76 07 48 83 7f 10 00 74 3b 39 d8
73 1f
[  161.830239] RSP: 0018:ffffb28104a2e970 EFLAGS: 00010246
[  161.830588] RAX: 0000000000000000 RBX: ffffffffc0a868e0 RCX: ffff8d6141e1ce88
[  161.831071] RDX: ffff8d5f47601980 RSI: 0000000000000001 RDI: 0000000000000000
[  161.831524] RBP: ffffb28104a2e980 R08: 0000000000000003 R09: 0000000000000001
[  161.831977] R10: 6369766564204c4c R11: 6564204c4c554e28 R12: 0000000000000000
[  161.832422] R13: 0000000000000000 R14: ffffffffc0a87950 R15: 0000000000000001
[  161.833019] FS:  00007f04fce83740(0000) GS:ffff8d619f0c4000(0000)
knlGS:0000000000000000
[  161.833527] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  161.833868] CR2: 0000000000000020 CR3: 000000010625a003 CR4: 00000000000706f0
[  161.834307] Call Trace:
[  161.834434]  <TASK>
[  161.834545]  hmm_bo_alloc+0x5c/0x2c0 [atomisp]
[  161.834959]  __hmm_alloc+0x48/0xf0 [atomisp]
[  161.835267]  hmm_init+0x98/0xd0 [atomisp]
[  161.835561]  ? __pfx_test_init+0x10/0x10 [atomisp]
[  161.835863]  test_init+0x42/0xff0 [atomisp]
[  161.836174]  do_one_initcall+0x4b/0x320
[  161.836446]  do_init_module+0x6a/0x2b0
[  161.836675]  load_module+0x24f7/0x25c0
[  161.836905]  ? kernel_read_file+0x226/0x2d0
[  161.837160]  init_module_from_file+0x9b/0xe0
[  161.837413]  ? init_module_from_file+0x9b/0xe0
[  161.837687]  idempotent_init_module+0x170/0x270
[  161.837958]  __x64_sys_finit_module+0x6f/0xe0
[  161.838225]  x64_sys_call+0x1b7a/0x2150
[  161.838454]  do_syscall_64+0x74/0x1d0
[  161.838701]  ? ksys_mmap_pgoff+0x1b7/0x240
[  161.838950]  ? __x64_sys_mmap+0x37/0x50
[  161.839176]  ? x64_sys_call+0x2008/0x2150
[  161.839429]  ? do_syscall_64+0xa3/0x1d0
[  161.839640]  ? __x64_sys_read+0x1e/0x30
[  161.839863]  ? x64_sys_call+0x1b90/0x2150
[  161.840098]  ? do_syscall_64+0xa3/0x1d0
[  161.840315]  ? do_syscall_64+0x199/0x1d0
[  161.840538]  ? x64_sys_call+0x1b90/0x2150
[  161.840775]  ? do_syscall_64+0xa3/0x1d0
[  161.841007]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
[  161.841289] RIP: 0033:0x7f04fc92695d
[  161.841490] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e
fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24
08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 03 35 0d 00 f7 d8 64 89
01 48
[  161.842992] RSP: 002b:00007ffd12ffbb88 EFLAGS: 00000246 ORIG_RAX:
0000000000000139
[  161.843500] RAX: ffffffffffffffda RBX: 0000557fdea491a0 RCX: 00007f04fc92695d
[  161.843968] RDX: 0000000000000000 RSI: 0000557fd288c358 RDI: 000000000000000c
[  161.844401] RBP: 0000000000040000 R08: 0000000000000000 R09: 0000000000000000
[  161.844857] R10: 000000000000000c R11: 0000000000000246 R12: 0000557fd288c358
[  161.845285] R13: 0000000000000000 R14: 0000557fdea492b0 R15: 0000557fdea491a0
[  161.845740]  </TASK>
[  161.845844] Modules linked in: atomisp(O+) ipu_bridge v4l2_fwnode
videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 videobuf2_common
v4l2_async videodev mc isofs vboxsf vboxguest vboxvideo
drm_vram_helper nls_iso8859_1 intel_rapl_msr intel_rapl_common
intel_uncore_frequency_common ghash_clmulni_intel sha512_ssse3
sha1_ssse3 aesni_intel snd_intel8x0 rapl snd_ac97_codec ac97_bus
snd_pcm binfmt_misc joydev snd_seq_midi snd_seq_midi_event snd_rawmidi
snd_seq vga16fb snd_seq_device vgastate input_leds sch_fq_codel
snd_timer snd mac_hid soundcore serio_raw vmwgfx drm_ttm_helper ttm
drm_client_lib drm_kms_helper drm msr parport_pc ppdev lp parport
ramoops pstore_blk reed_solomon efi_pstore pstore_zone ip_tables
x_tables autofs4 hid_generic usbhid hid e1000 video psmouse wmi ahci
libahci i2c_piix4 pata_acpi i2c_smbus [last unloaded: vboxguest]
[  161.851072] CR2: 0000000000000020


> With Best Regards,
> Andy Shevchenko

Best Regards,
Abelrahman Fekry