netdev - Re: [PATCH RFC bpf-next 2/2] bpf: Use kmalloc_nolock() in local storage unconditionally

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAADnVQKWKC3oh6ycxE+tstYupwVsdbhYHOncnfTOFWLL2DmJjw@mail.gmail.com>
Date: Wed, 12 Nov 2025 12:04:47 -0800
From: Alexei Starovoitov <alexei.starovoitov@...il.com>
To: Amery Hung <ameryhung@...il.com>
Cc: bpf <bpf@...r.kernel.org>, Network Development <netdev@...r.kernel.org>, 
	Andrii Nakryiko <andrii@...nel.org>, Daniel Borkmann <daniel@...earbox.net>, 
	Martin KaFai Lau <martin.lau@...nel.org>, Kumar Kartikeya Dwivedi <memxor@...il.com>, KP Singh <kpsingh@...nel.org>, 
	Yonghong Song <yonghong.song@...ux.dev>, Song Liu <song@...nel.org>, 
	Kernel Team <kernel-team@...a.com>
Subject: Re: [PATCH RFC bpf-next 2/2] bpf: Use kmalloc_nolock() in local
 storage unconditionally

On Wed, Nov 12, 2025 at 11:51 AM Amery Hung <ameryhung@...il.com> wrote:
>
> On Wed, Nov 12, 2025 at 11:35 AM Alexei Starovoitov
> <alexei.starovoitov@...il.com> wrote:
> >
> > On Wed, Nov 12, 2025 at 9:59 AM Amery Hung <ameryhung@...il.com> wrote:
> > >
> > > @@ -80,23 +80,12 @@ bpf_selem_alloc(struct bpf_local_storage_map *smap, void *owner,
> > >         if (mem_charge(smap, owner, smap->elem_size))
> > >                 return NULL;
> > >
> > > -       if (smap->bpf_ma) {
> > > -               selem = bpf_mem_cache_alloc_flags(&smap->selem_ma, gfp_flags);
> > > -               if (selem)
> > > -                       /* Keep the original bpf_map_kzalloc behavior
> > > -                        * before started using the bpf_mem_cache_alloc.
> > > -                        *
> > > -                        * No need to use zero_map_value. The bpf_selem_free()
> > > -                        * only does bpf_mem_cache_free when there is
> > > -                        * no other bpf prog is using the selem.
> > > -                        */
> > > -                       memset(SDATA(selem)->data, 0, smap->map.value_size);
> > > -       } else {
> > > -               selem = bpf_map_kzalloc(&smap->map, smap->elem_size,
> > > -                                       gfp_flags | __GFP_NOWARN);
> > > -       }
> > > +       selem = bpf_map_kmalloc_nolock(&smap->map, smap->elem_size, gfp_flags, NUMA_NO_NODE);
> >
> >
> > Pls enable CONFIG_DEBUG_VM=y then you'll see that the above triggers:
> > void *kmalloc_nolock_noprof(size_t size, gfp_t gfp_flags, int node)
> > {
> >         gfp_t alloc_gfp = __GFP_NOWARN | __GFP_NOMEMALLOC | gfp_flags;
> > ...
> >         VM_WARN_ON_ONCE(gfp_flags & ~(__GFP_ACCOUNT | __GFP_ZERO |
> >                                       __GFP_NO_OBJ_EXT));
> >
> > and benchmarking numbers have to be redone, since with
> > unsupported gfp flags kmalloc_nolock() is likely doing something wrong.
>
> I see. Thanks for pointing it out. Currently the verifier determines
> the flag and rewrites the program based on if the caller of
> storage_get helpers is sleepable. I will remove it and redo the
> benchmark.

yes. that part of the verifier can be removed too.
First I would redo the benchmark numbers with s/gfp_flags/0 in the above line.