linux-kernel - Re: [next] arm64: boot failed

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: <YqGC5NxDm1WyOHcw@carbon>
Date:   Wed, 8 Jun 2022 22:19:32 -0700
From:   Roman Gushchin <roman.gushchin@...ux.dev>
To:     Kefeng Wang <wangkefeng.wang@...wei.com>
Cc:     Vasily Averin <vvs@...nvz.org>,
        Naresh Kamboju <naresh.kamboju@...aro.org>,
        Shakeel Butt <shakeelb@...gle.com>,
        Linux ARM <linux-arm-kernel@...ts.infradead.org>,
        Stephen Rothwell <sfr@...b.auug.org.au>,
        Linux-Next Mailing List <linux-next@...r.kernel.org>,
        open list <linux-kernel@...r.kernel.org>,
        regressions@...ts.linux.dev, lkft-triage@...ts.linaro.org,
        linux-mm <linux-mm@...ck.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Ard Biesheuvel <ardb@...nel.org>,
        Arnd Bergmann <arnd@...db.de>,
        Catalin Marinas <catalin.marinas@....com>,
        Raghuram Thammiraju <raghuram.thammiraju@....com>,
        Mark Brown <broonie@...nel.org>, Will Deacon <will@...nel.org>,
        Qian Cai <quic_qiancai@...cinc.com>
Subject: Re: [next] arm64: boot failed - next-20220606

On Thu, Jun 09, 2022 at 12:43:00PM +0800, Kefeng Wang wrote:
> 
> On 2022/6/9 11:44, Kefeng Wang wrote:
> > 
> > On 2022/6/9 10:49, Vasily Averin wrote:
> > > Dear ARM developers,
> > > could you please help me to find the reason of this problem?
> > Hi,
> > > mem_cgroup_from_obj():
> > > ffff80000836cf40:       d503245f        bti     c
> > > ffff80000836cf44:       d503201f        nop
> > > ffff80000836cf48:       d503201f        nop
> > > ffff80000836cf4c:       d503233f        paciasp
> > > ffff80000836cf50:       d503201f        nop
> > > ffff80000836cf54:       d2e00021        mov     x1,
> > > #0x1000000000000            // #281474976710656
> > > ffff80000836cf58:       8b010001        add     x1, x0, x1
> > > ffff80000836cf5c:       b25657e4        mov     x4,
> > > #0xfffffc0000000000         // #-4398046511104
> > > ffff80000836cf60:       d34cfc21        lsr     x1, x1, #12
> > > ffff80000836cf64:       d37ae421        lsl     x1, x1, #6
> > > ffff80000836cf68:       8b040022        add     x2, x1, x4
> > > ffff80000836cf6c:       f9400443        ldr     x3, [x2, #8]
> > > 
> > > x5 : ffff80000a96f000 x4 : fffffc0000000000 x3 : ffff80000ad5e680
> > > x2 : fffffe00002bc240 x1 : 00000200002bc240 x0 : ffff80000af09740
> > > 
> > > x0 = 0xffff80000af09740 is an argument of mem_cgroup_from_obj()
> > > according to System.map it is init_net
> > > 
> > > This issue is caused by calling virt_to_page() on address of static
> > > variable init_net.
> > > Arm64 consider that addresses of static variables are not valid
> > > virtual addresses.
> > > On x86_64 the same API works without any problem.
> > > 
> > > Unfortunately I do not understand the cause of the problem.
> > > I do not see any bugs in my patch.
> > > I'm using an existing API, mem_cgroup_from_obj(), to find the memory
> > > cgroup used
> > > to account for the specified object.
> > > In particular, in the current case, I wanted to get the memory
> > > cgroup of the
> > > specified network namespace by the name taken from for_each_net().
> > > The first object in this list is the static structure unit_net
> > 
> > root@...t:~# cat /proc/kallsyms |grep -w _data
> > ffff80000a110000 D _data
> > root@...t:~# cat /proc/kallsyms |grep -w _end
> > ffff80000a500000 B _end
> > root@...t:~# cat /proc/kallsyms |grep -w init_net
> > ffff80000a4eb980 B init_net
> > 
> > the init_net is located in data section, on arm64, it is allowed by
> > vmalloc, see
> > 
> >     map_kernel_segment(pgdp, _data, _end, PAGE_KERNEL, &vmlinux_data, 0,
> > 0);
> > 
> > and the arm has same behavior.
> > 
> > We could let init_net be allocated dynamically, but I think it could
> > change a lot.
> > 
> > Any better sugguestion, Catalin?
> 
> or  add vmalloc check in mem_cgroup_from_obj()?
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 27cebaa53472..fb817e5da5f0 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -2860,7 +2860,10 @@ struct mem_cgroup *mem_cgroup_from_obj(void *p)
>         if (mem_cgroup_disabled())
>                 return NULL;
> 
> -       folio = virt_to_folio(p);
> +       if (unlikely(is_vmalloc_addr(p)))
> +               folio = page_folio(vmalloc_to_page(p));
> +       else
> +               folio = virt_to_folio(p);
> 
>         /*
>          * Slab objects are accounted individually, not per-page.
> 

It sounds right. Later we can add something like mem_cgroup_from_slab_obj()
to use on hot paths and avoid this check.