Re: [PATCH 1/3] lib/stackdepot: Add a refcount field in stack_record

From: Andrey Konovalov <andreyknvl@gmail.com>
To: Marco Elver <elver@google.com>, Oscar Salvador <osalvador@suse.de>
Cc: Michal Hocko <mhocko@suse.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Linux Memory Management List <linux-mm@kvack.org>,
	Vlastimil Babka <vbabka@suse.cz>,
	Eric Dumazet <edumazet@google.com>,
	Waiman Long <longman@redhat.com>,
	Suren Baghdasaryan <surenb@google.com>,
	Alexander Potapenko <glider@google.com>,
	Dmitry Vyukov <dvyukov@google.com>,
	kasan-dev <kasan-dev@googlegroups.com>
Subject: Re: [PATCH 1/3] lib/stackdepot: Add a refcount field in stack_record
Date: Mon, 5 Sep 2022 22:53:47 +0200	[thread overview]
Message-ID: <CA+fCnZckFNqDA2SJSMjM7gpUF_U7Ps_3u+JzvN_cKvskz0FuOQ@mail.gmail.com> (raw)
In-Reply-To: <CANpmjNNjkgibnBcp7ZOWGC5CcBJ=acgrRKo0cwZG0xOB5OCpLw@mail.gmail.com>

On Thu, Sep 1, 2022 at 11:18 AM Marco Elver <elver@google.com> wrote:
>
> On Thu, 1 Sept 2022 at 10:38, Michal Hocko <mhocko@suse.com> wrote:
> >
> > On Thu 01-09-22 10:24:58, Marco Elver wrote:
> > > On Thu, Sep 01, 2022 at 06:42AM +0200, Oscar Salvador wrote:
> > [...]
> > > > diff --git a/lib/stackdepot.c b/lib/stackdepot.c
> > > > index 5ca0d086ef4a..aeb59d3557e2 100644
> > > > --- a/lib/stackdepot.c
> > > > +++ b/lib/stackdepot.c
> > > > @@ -63,6 +63,7 @@ struct stack_record {
> > > >     u32 hash;                       /* Hash in the hastable */
> > > >     u32 size;                       /* Number of frames in the stack */
> > > >     union handle_parts handle;
> > > > +   refcount_t count;               /* Number of the same repeated stacks */
> > >
> > > This will increase stack_record size for every user, even if they don't
> > > care about the count.
> >
> > Couldn't this be used for garbage collection?
>
> Only if we can precisely figure out at which point a stack is no
> longer going to be needed.
>
> But more realistically, stack depot was designed to be simple. Right
> now it can allocate new stacks (from an internal pool), but giving the
> memory back to that pool isn't supported. Doing garbage collection
> would effectively be a redesign of stack depot. And for the purpose
> for which stack depot was designed (debugging tools), memory has never
> been an issue (note that stack depot also has a fixed upper bound on
> memory usage).
>
> We had talked (in the context of KASAN) about bounded stack storage,
> but the preferred solution is usually a cache-based design which
> allows evictions (in the simplest case a ring buffer), because
> figuring out (and relying on) where precisely a stack will
> definitively no longer be required in bug reports is complex and does
> not guarantee the required bound on memory usage. Andrey has done the
> work on this for tag-based KASAN modes:
> https://lore.kernel.org/all/cover.1658189199.git.andreyknvl@google.com/

To be clear, the stack ring buffer implementation for the KASAN
tag-based modes still uses the stack depot as a back end to store
stack traces.

I plan to explore redesigning the stack depot implementation to allow
evicting unneeded stack traces as the next step. (The goal is to have
a memory-bounded stack depot that doesn't just stop collecting stack
traces once the memory limit is reached.) Having a refcount for each
saved stack trace will likely be a part of this redesign.