Re: [PATCH] Revert "driver core: Use mod_delayed_work to prevent lost deferred probe work"

[Danilo Krummrich]

On Tue May 26, 2026 at 2:46 PM CEST, Geert Uytterhoeven wrote:
> Hi Danilo,
>
> On Tue, 26 May 2026 at 14:09, Danilo Krummrich <dakr@xxxxxxxxxx> wrote:
>> On 5/26/26 2:03 PM, Biju wrote:
>> > From: Biju Das <biju.das.jz@xxxxxxxxxxxxxx>
>> >
>> > Commit 1137838865bf ("driver core: Use mod_delayed_work to prevent lost
>> > deferred probe work") introduced below regression on Renesas SMARC RZ/G2L
>> > EVK:
>> >  class_for_each_device called for class 'devlink' before it was registered
>> >  WARNING: drivers/base/class.c:415 at class_for_each_device+0x12c/0x13c, CPU#1: kworker/1:1/26
>> >  class_for_each_device+0x12c/0x13c (P)
>> >  fw_devlink_probing_done+0x58/0xa0
>> >  deferred_probe_timeout_work_func+0x5c/0xb8
>> >  process_one_work+0x150/0x290
>> >  worker_thread+0x18c/0x300
>> >  kthread+0x114/0x120
>> >
>> > Reverting the commit fixes the issue.
>> There's already [1], which should also fix the issue.
>
> Thanks!
>
> FTR, that patch is not sufficient to fix the WARNING (on RZ/Five)
> for me: I need both patches from the series to get rid of the both
> the warning thousands of "sync_state() pending due to" and
> hundreds of "deferred probe pending" messages on other Renesas systems.

Yes, with linking to [1] I meant both patches.

Thanks,
Danilo

>> [1]
>> https://lore.kernel.org/driver-core/20260525012340.3860581-1-dakr@xxxxxxxxxx/t/#u