Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230722074753.568696-1-arnd@kernel.org> <CALOAHbCV2v3X7g1TD42yve0juhRD2vhq=DMtDz9P6+mX8Dae_w@mail.gmail.com>
 <CAADnVQKGe8DN+Zs387UVwpij3ROGqNEnc5r940h5ueqQYHTYCA@mail.gmail.com>
 <fa5e9098-d6f9-48a2-bb77-2620b6bb6556@app.fastmail.com> <CAADnVQ+p4wpd=tKJAiwB34O1y5vv4mibtkt9D-F7sG=rQapcew@mail.gmail.com>
 <679d8d63-ce92-4294-8620-e98c82365b2c@app.fastmail.com> <39444a4e-70da-4d17-a40a-b51e05236d23@app.fastmail.com>
In-Reply-To: <39444a4e-70da-4d17-a40a-b51e05236d23@app.fastmail.com>
From:   Alexei Starovoitov <alexei.starovoitov@gmail.com>
Date:   Mon, 24 Jul 2023 12:15:45 -0700
Message-ID: <CAADnVQ+zdV9+UNV9NeEzY2rWd8qvW3cvHxS9mYwfhnqZOV+9=A@mail.gmail.com>
Subject: Re: [PATCH] bpf: force inc_active()/dec_active() to be inline functions
To:     Arnd Bergmann <arnd@kernel.org>
Cc:     Arnd Bergmann <arnd@arndb.de>, Yafang Shao <laoar.shao@gmail.com>,
        Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Andrii Nakryiko <andrii@kernel.org>,
        Hou Tao <houtao1@huawei.com>,
        Martin KaFai Lau <martin.lau@linux.dev>,
        Song Liu <song@kernel.org>, Yonghong Song <yhs@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        KP Singh <kpsingh@kernel.org>,
        Stanislav Fomichev <sdf@google.com>,
        Hao Luo <haoluo@google.com>, Jiri Olsa <jolsa@kernel.org>,
        Kumar Kartikeya Dwivedi <memxor@gmail.com>,
        bpf <bpf@vger.kernel.org>, LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Mon, Jul 24, 2023 at 11:30=E2=80=AFAM Arnd Bergmann <arnd@kernel.org> wr=
ote:
>
> On Mon, Jul 24, 2023, at 20:13, Arnd Bergmann wrote:
>
> >>> One difference between gcc and clang is that gcc tries to
> >>> be smart about warnings by using information from inlining
> >>> to produce better warnings, while clang never uses information
> >>> across function boundaries for generated warnings, so it won't
> >>> find this one, but also would ignore an unconditional use
> >>> of the uninitialized variable.
> >>>
> >>> >> If we have to change the kernel, what about the change below?
> >>> >
> >>> > To workaround the compiler bug we can simply init flag=3D0 to silen=
ce
> >>> > the warn, but even that is silly. Passing flag=3D0 into irqrestore =
is buggy.
> >>>
> >>> Maybe inc_active() could return the flags instead of modifying
> >>> the stack variable? that would also result in slightly better
> >>> code when it's not inlined.
> >>
> >> Which gcc are we talking about here that is so buggy?
> >
> > I think I only tried versions 8 through 13 for this one, but
> > can check others as well.
>
> I have a minimized test case at https://godbolt.org/z/hK4ev17fv
> that shows the problem happening with all versions of gcc
> (4.1 through 14.0) if I force the dec_active() function to be
> inline and force inc_active() to be non-inline.

That's a bit of cheating, but I see your point now.
How about we do:
diff --git a/kernel/bpf/memalloc.c b/kernel/bpf/memalloc.c
index 51d6389e5152..3fa0944cb975 100644
--- a/kernel/bpf/memalloc.c
+++ b/kernel/bpf/memalloc.c
@@ -183,11 +183,11 @@ static void inc_active(struct bpf_mem_cache *c,
unsigned long *flags)
        WARN_ON_ONCE(local_inc_return(&c->active) !=3D 1);
 }

-static void dec_active(struct bpf_mem_cache *c, unsigned long flags)
+static void dec_active(struct bpf_mem_cache *c, unsigned long *flags)
 {
        local_dec(&c->active);
        if (IS_ENABLED(CONFIG_PREEMPT_RT))
-               local_irq_restore(flags);
+               local_irq_restore(*flags);
 }

 static void add_obj_to_free_list(struct bpf_mem_cache *c, void *obj)
@@ -197,7 +197,7 @@ static void add_obj_to_free_list(struct
bpf_mem_cache *c, void *obj)
        inc_active(c, &flags);
        __llist_add(obj, &c->free_llist);
        c->free_cnt++;
-       dec_active(c, flags);
+       dec_active(c, &flags);


It's symmetrical and there is no 'flags =3D 0' ugliness.