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: <CA+QYu4rPbfH-4wNR06Vn=31RCGKFFEB-KRskn52Jtig_UugYzg@mail.gmail.com>
 <20220805135955.5to2bm6ou35p5zsy@bogus> <20220805150820.3q4kwrlh7qrt7poa@bogus>
In-Reply-To: <20220805150820.3q4kwrlh7qrt7poa@bogus>
From:   Bruno Goncalves <bgoncalv@redhat.com>
Date:   Fri, 5 Aug 2022 20:38:39 +0200
Message-ID: <CA+QYu4qq3Op4=dJcRWGDTqyt6Md4vQfRBaafLtC+h7n5sFYM9A@mail.gmail.com>
Subject: Re: [5.19.0] [aarch64] WARNING: CPU: 0 PID: 1 at mm/page_alloc.c:5407 __alloc_pages+0x1a0/0x290
To:     Sudeep Holla <sudeep.holla@arm.com>
Cc:     linux-arm-kernel@lists.infradead.org,
        LKML <linux-kernel@vger.kernel.org>,
        CKI Project <cki-project@redhat.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, 5 Aug 2022 at 17:08, Sudeep Holla <sudeep.holla@arm.com> wrote:
>

Thanks for the fix, it works well.

[    0.003840] ACPI PPTT: No PPTT table found, CPU and cache topology
may be inaccurate
[    0.004588] cblist_init_generic: Setting adjustable number of
callback queues.
[    0.004595] cblist_init_generic: Setting shift to 2 and lim to 1.
[    0.004707] cblist_init_generic: Setting shift to 2 and lim to 1.
[    0.004779] cblist_init_generic: Setting shift to 2 and lim to 1.
[    0.004908] rcu: Hierarchical SRCU implementation.

Bruno

> On Fri, Aug 05, 2022 at 02:59:55PM +0100, Sudeep Holla wrote:
> > On Fri, Aug 05, 2022 at 11:47:57AM +0200, Bruno Goncalves wrote:
> > > Hello,
> > >
> > > Since commit "c1c76700a0d6" we started to hit the following call trace
> > > on aarch64 VMs when looking at journalctl log
> > >
> > > kernel: Mountpoint-cache hash table entries: 16384 (order: 5, 131072
> > > bytes, linear)
> > > kernel: ACPI PPTT: No PPTT table found, CPU and cache topology may be inaccurate
> > > kernel: ------------[ cut here ]------------
> > > kernel: WARNING: CPU: 0 PID: 1 at mm/page_alloc.c:5407 __alloc_pages+0x1a0/0x290
> > > kernel: Modules linked in:
> > > kernel: CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.19.0 #1
> > > kernel: pstate: 20400005 (nzCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> > > kernel: pc : __alloc_pages+0x1a0/0x290
> > > kernel: lr : alloc_page_interleave+0x24/0xbc
> > > kernel: sp : ffff80000803bb90
> > > kernel: x29: ffff80000803bb90 x28: 0000000000000000 x27: 0000000000000000
> > > kernel: x26: 0000000000000000 x25: 0000000000000000 x24: 0000000000000000
> > > kernel: x23: 000000000000001e x22: 0000000000000000 x21: 000000000000001e
> > > kernel: x20: 000000000000001e x19: 0000000000040b20 x18: 0000000000000014
> > > kernel: x17: 6e69206562207961 x16: 6d2079676f6c6f70 x15: 6f74206568636163
> > > kernel: x14: 20646e6120555043 x13: 6574617275636361 x12: 6e69206562207961
> > > kernel: x11: 6d2079676f6c6f70 x10: 6f74206568636163 x9 : ffffa4f6617fd5e8
> > > kernel: x8 : ffffa4f663ed0440 x7 : 0000000000000007 x6 : 0000000000000000
> > > kernel: x5 : 0000000000000000 x4 : 0000000000000001 x3 : 0000000000000000
> > > kernel: x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffffa4f663c53532
> > > kernel: Call trace:
> > > kernel:  __alloc_pages+0x1a0/0x290
> > > kernel:  alloc_page_interleave+0x24/0xbc
> > > kernel:  alloc_pages+0x10c/0x16c
> > > kernel:  kmalloc_order+0x3c/0xc0
> > > kernel:  kmalloc_order_trace+0x38/0x130
> > > kernel:  __kmalloc+0x2e8/0x350
> > > kernel:  detect_cache_attributes+0x9c/0x110
> > > kernel:  update_siblings_masks+0x34/0x270
> > > kernel:  store_cpu_topology+0x7c/0xc0
> > > kernel:  smp_prepare_cpus+0x34/0x108
> > > kernel:  kernel_init_freeable+0x108/0x1b8
> > > kernel:  kernel_init+0x30/0x150
> > > kernel:  ret_from_fork+0x10/0x20
> > > kernel: ---[ end trace 0000000000000000 ]---
> > > kernel: Early cacheinfo failed, ret = -12
> > > kernel: cblist_init_generic: Setting adjustable number of callback queues.
> > > kernel: cblist_init_generic: Setting shift to 2 and lim to 1.
> > >
> >
> > Thanks for the report. I will see if I can reproduce by forcing the absence
> > of PPTT which is something I haven't tested after I fixed hotplug issues
> > reported with my changes.
>
> OK I was able to reproduce this easily by just saying PPTT doesn't exist
> on my system. Not sure if my recent changes made this issue show up easily
> or it existed even w/o my changes but it is stupid issue :(.
>
> Can you check if this fixes it for you ? I can post the patch once
> you confirm. We assign negative/error value to unsigned int and then
> the number of leaves ends up with that value because of wrong comparison.
>
> Regards,
> Sudeep
>
> -->8
> diff --git i/arch/arm64/kernel/cacheinfo.c w/arch/arm64/kernel/cacheinfo.c
> index 587543c6c51c..97c42be71338 100644
> --- i/arch/arm64/kernel/cacheinfo.c
> +++ w/arch/arm64/kernel/cacheinfo.c
> @@ -45,7 +45,8 @@ static void ci_leaf_init(struct cacheinfo *this_leaf,
>
>  int init_cache_level(unsigned int cpu)
>  {
> -       unsigned int ctype, level, leaves, fw_level;
> +       unsigned int ctype, level, leaves;
> +       int fw_level;
>         struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
>
>         for (level = 1, leaves = 0; level <= MAX_CACHE_LEVEL; level++) {
> @@ -63,6 +64,9 @@ int init_cache_level(unsigned int cpu)
>         else
>                 fw_level = acpi_find_last_cache_level(cpu);
>
> +       if (fw_level < 0)
> +               return fw_level;
> +
>         if (level < fw_level) {
>                 /*
>                  * some external caches not specified in CLIDR_EL1
>