Em Mon, 18 Oct 2021 01:23:33 +0900
Tsuchiya Yuto <kitakar@xxxxxxxxx> escreveu:
> This is almost a BUG report with a patch to just avoid kernel error
> message. Thus, prefixed with [BUG][RFC].
>
> RFC:
> 1. Regarding the NULL check in hmm_init(), it was added for the case
> when probe failed [1]. So, need to address that case with the other
> means? But I'm not sure how we can do for the case.
> 2. Regarding the WARN_ON() in hmm_free(), there is a patch from Alan
> ("atomisp: hmm gives a bogus warning on unload") [1] but I haven't
> looked into it yet.
>
> [1] https://lore.kernel.org/linux-media/151001141201.77201.10725942741811192730.stgit@alans-desktop/
>
> >8-----------------------------------------------------------------8<
>
> The NULL check for dummy_ptr in hmm_init() [1] results in the following
> "hmm_init Failed to create sysfs" error exactly once every
> two times on atomisp reload by rmmod/insmod (although atomisp module
> loads and works fine regardless of this error):
>
> kern :warn : [ 140.230662] sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:03.0/active_bo'
> kern :warn : [ 140.230668] CPU: 1 PID: 2502 Comm: insmod Tainted: G C OE 5.15.0-rc4-1-surface-mainline #1 b8acf6eb64994414b2e20bad312a7a2c45f748f9
> kern :warn : [ 140.230675] Hardware name: OEMB OEMB/OEMB, BIOS 1.51116.238 03/09/2015
> kern :warn : [ 140.230678] Call Trace:
> kern :warn : [ 140.230687] dump_stack_lvl+0x46/0x5a
> kern :warn : [ 140.230702] sysfs_warn_dup.cold+0x17/0x24
> kern :warn : [ 140.230710] sysfs_add_file_mode_ns+0x160/0x170
> kern :warn : [ 140.230717] internal_create_group+0x126/0x390
> kern :warn : [ 140.230723] hmm_init+0x5c/0x70 [atomisp 7a6a680bf400629363d2a6f58fd10e7299678b99]
> kern :warn : [ 140.230811] atomisp_pci_probe.cold+0x1136/0x148e [atomisp 7a6a680bf400629363d2a6f58fd10e7299678b99]
> kern :warn : [ 140.230875] local_pci_probe+0x45/0x80
> kern :warn : [ 140.230882] ? pci_match_device+0xd7/0x130
> kern :warn : [ 140.230887] pci_device_probe+0xfa/0x1b0
> kern :warn : [ 140.230892] really_probe+0x1f5/0x3f0
> kern :warn : [ 140.230899] __driver_probe_device+0xfe/0x180
> kern :warn : [ 140.230903] driver_probe_device+0x1e/0x90
> kern :warn : [ 140.230908] __driver_attach+0xc0/0x1c0
> kern :warn : [ 140.230912] ? __device_attach_driver+0xe0/0xe0
> kern :warn : [ 140.230915] ? __device_attach_driver+0xe0/0xe0
> kern :warn : [ 140.230919] bus_for_each_dev+0x89/0xd0
> kern :warn : [ 140.230924] bus_add_driver+0x12b/0x1e0
> kern :warn : [ 140.230929] driver_register+0x8f/0xe0
> kern :warn : [ 140.230933] ? 0xffffffffc153f000
> kern :warn : [ 140.230937] do_one_initcall+0x57/0x220
> kern :warn : [ 140.230945] do_init_module+0x5c/0x260
> kern :warn : [ 140.230952] load_module+0x24bd/0x26a0
> kern :warn : [ 140.230962] ? __do_sys_finit_module+0xae/0x110
> kern :warn : [ 140.230966] __do_sys_finit_module+0xae/0x110
> kern :warn : [ 140.230972] do_syscall_64+0x5c/0x80
> kern :warn : [ 140.230979] ? syscall_exit_to_user_mode+0x23/0x40
> kern :warn : [ 140.230983] ? do_syscall_64+0x69/0x80
> kern :warn : [ 140.230988] ? exc_page_fault+0x72/0x170
> kern :warn : [ 140.230991] entry_SYSCALL_64_after_hwframe+0x44/0xae
> kern :warn : [ 140.230997] RIP: 0033:0x7f7fd5d8718d
> kern :warn : [ 140.231003] Code: b4 0c 00 0f 05 eb a9 66 0f 1f 44 00 00 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 b3 6c 0c 00 f7 d8 64 89 01 48
> kern :warn : [ 140.231006] RSP: 002b:00007ffefc25f0e8 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
> kern :warn : [ 140.231012] RAX: ffffffffffffffda RBX: 000055ac3edcd7f0 RCX: 00007f7fd5d8718d
> kern :warn : [ 140.231015] RDX: 0000000000000000 RSI: 000055ac3d723270 RDI: 0000000000000003
> kern :warn : [ 140.231017] RBP: 0000000000000000 R08: 0000000000000000 R09: 00007f7fd5e52380
> kern :warn : [ 140.231019] R10: 0000000000000003 R11: 0000000000000246 R12: 000055ac3d723270
> kern :warn : [ 140.231021] R13: 0000000000000000 R14: 000055ac3edd06e0 R15: 0000000000000000
> kern :err : [ 140.231038] atomisp-isp2 0000:00:03.0: hmm_init Failed to create sysfs
>
> So, remove the NULL check to always call sysfs_remove_group() and fix
> the above error.
>
> At this point, atomisp now gives WARN_ON() in hmm_free() [2] on atomisp
> reload by rmmod/insmod. So, also remove it.
>
> [1] added on commit
> d9ab83953fa7 ("media: atomisp: don't cause a warn if probe failed")
> [2] added on commit
> b83cc378dfc4 ("atomisp: clean up the hmm init/cleanup indirections")
Interesting bug. As dummy_ptr == 0 is indeed a valid value, and
the driver should not call sysfs_remove_group() if sysfs_create_group(),
the right fix seem to use mmgr_EXCEPTION as the default value for
dummy_ptr. This way, the check before calling sys group removal logic
will still preserved.
On other words:
@@ -39,7 +39,7 @@
struct hmm_bo_device bo_device;
struct hmm_pool dynamic_pool;
struct hmm_pool reserved_pool;
-static ia_css_ptr dummy_ptr;
+static ia_css_ptr dummy_ptr = mmgr_EXCEPTION;
static bool hmm_initialized;
struct _hmm_mem_stat hmm_mem_stat;
@@ -209,7 +209,7 @@ int hmm_init(void)
void hmm_cleanup(void)
{
- if (!dummy_ptr)
+ if (dummy_ptr == mmgr_EXCEPTION)
return;
sysfs_remove_group(&atomisp_dev->kobj, atomisp_attribute_group);
The WARN_ON(!virt) should be removed anyway, as virt == 0 is a valid
value.
I'll do the above changes and apply the patch.
>
> Signed-off-by: Tsuchiya Yuto <kitakar@xxxxxxxxx>
> ---
> drivers/staging/media/atomisp/pci/hmm/hmm.c | 4 ----
> 1 file changed, 4 deletions(-)
>
> diff --git a/drivers/staging/media/atomisp/pci/hmm/hmm.c b/drivers/staging/media/atomisp/pci/hmm/hmm.c
> index 6a5ee4607089..47dc3df3b800 100644
> --- a/drivers/staging/media/atomisp/pci/hmm/hmm.c
> +++ b/drivers/staging/media/atomisp/pci/hmm/hmm.c
> @@ -209,8 +209,6 @@ int hmm_init(void)
>
> void hmm_cleanup(void)
> {
> - if (!dummy_ptr)
> - return;
> sysfs_remove_group(&atomisp_dev->kobj, atomisp_attribute_group);
>
> /* free dummy memory first */
> @@ -288,8 +286,6 @@ void hmm_free(ia_css_ptr virt)
>
> dev_dbg(atomisp_dev, "%s: free 0x%08x\n", __func__, virt);
>
> - WARN_ON(!virt);
> -
> bo = hmm_bo_device_search_start(&bo_device, (unsigned int)virt);
>
> if (!bo) {
