On Wed, 6 Mar 2024 10:01:53 +0100
Petr Tesařík <petr@xxxxxxxxxxx> wrote:
> On Wed, 6 Mar 2024 09:23:53 +0100
> "Linux regression tracking (Thorsten Leemhuis)" <regressions@xxxxxxxxxxxxx> wrote:
>
> > On 28.02.24 12:03, Petr Tesařík wrote:
> > > On Wed, 28 Feb 2024 07:19:56 +0100
> > > "Linux regression tracking (Thorsten Leemhuis)" <regressions@xxxxxxxxxxxxx> wrote:
> > >
> > >> Net maintainers, chiming in here, as it seems handling this regression
> > >> stalled.
> > > Indeed, I was too busy with sandbox mode...
> >
> > Hmm, no reply in the past week to Petr's request for help from someone
> > with more knowledge about the field. :-/
> >
> > So I guess this means that this won't be fixed for 6.8? Unfortunate, but
> > well, that's how it it sometimes.
>
> For the record, I _can_ reproduce lockdep splats on my device, but they
> don't make any sense to me. They seem to confirm Jisheng Zhang's
> conclusion that lockdep conflates two locks which should have different
> lock-classes.
>
> So far I have noticed only one issue: the per-cpu syncp's are not
> initialized. I'll recompile and see if that's what confuses lockdep.
That wasn't the issue. FTR the syncp was in fact initialized, because
devm_netdev_alloc_pcpu_stats() is a macro that also takes care of the
initialization of the syncp struct field.
The problem is u64_stats_init().
Commit 9464ca650008 ("net: make u64_stats_init() a function") changed
it to an inline function. But that's wrong. It uses seqcount_init(),
which in turn declares:
static struct lock_class_key __key;
This assumes that each lock gets its own instance. But if
u64_stats_init() is a function (albeit an inline one), all calls
within the same file end up using the same instance.
Eric, would it be OK to revert the above-mentioned commit?
Petr T
