Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   David Moses <mosesster@gmail.com>
Mime-Version: 1.0 (1.0)
Subject: Re: [PATCH] x86/hyper-v: guard against cpu mask changes in hyperv_flush_tlb_others()
Date:   Sat, 7 Aug 2021 08:00:13 +0300
Message-Id: <FD8265E6-895E-45CF-9AE3-787FAD669FC8@gmail.com>
References: <MWHPR21MB15935468547C25294A253E0AD7F39@MWHPR21MB1593.namprd21.prod.outlook.com>
Cc:     =?utf-8?B?16rXldee16gg15DXkdeV15jXkdeV15w=?= 
        <tomer432100@gmail.com>, David Mozes <david.mozes@silk.us>,
        linux-hyperv@vger.kernel.org, linux-kernel@vger.kernel.org
In-Reply-To: <MWHPR21MB15935468547C25294A253E0AD7F39@MWHPR21MB1593.namprd21.prod.outlook.com>
To:     Michael Kelley <mikelley@microsoft.com>
Precedence: bulk


Sent from my iPhone

> On Aug 7, 2021, at 12:51 AM, Michael Kelley <mikelley@microsoft.com> wrote=
:
>=20
> =EF=BB=BFFrom: =D7=AA=D7=95=D7=9E=D7=A8 =D7=90=D7=91=D7=95=D7=98=D7=91=D7=95=
=D7=9C <tomer432100@gmail.com>  Sent: Friday, August 6, 2021 11:03 AM
>=20
>> Attaching the patches Michael asked for debugging=20
>> 1) Print the cpumask when < num_possible_cpus():
>> diff --git a/arch/x86/hyperv/mmu.c b/arch/x86/hyperv/mmu.c
>> index e666f7eaf32d..620f656d6195 100644
>> --- a/arch/x86/hyperv/mmu.c
>> +++ b/arch/x86/hyperv/mmu.c
>> @@ -60,6 +60,7 @@ static void hyperv_flush_tlb_others(const struct cpumas=
k *cpus,
>>         struct hv_tlb_flush *flush;
>>         u64 status =3D U64_MAX;
>>         unsigned long flags;
>> +       unsigned int cpu_last;
>>=20
>>         trace_hyperv_mmu_flush_tlb_others(cpus, info);
>>=20
>> @@ -68,6 +69,11 @@ static void hyperv_flush_tlb_others(const struct cpuma=
sk *cpus,
>>=20
>>         local_irq_save(flags);
>>=20
>> +       cpu_last =3D cpumask_last(cpus);
>> +       if (cpu_last > num_possible_cpus()) {
>=20
> I think this should be ">=3D" since cpus are numbered starting at zero.
> In your VM with 64 CPUs, having CPU #64 in the list would be error.
>=20
>> +               pr_emerg("ERROR_HYPERV: cpu_last=3D%*pbl", cpumask_pr_arg=
s(cpus));
>> +       }
>> +
>>         /*
>>          * Only check the mask _after_ interrupt has been disabled to avo=
id the
>>          * mask changing under our feet.
>>=20
>> 2) disable the Hyper-V specific flush routines:
>> diff --git a/arch/x86/hyperv/mmu.c b/arch/x86/hyperv/mmu.c
>> index e666f7eaf32d..8e77cc84775a 100644
>> --- a/arch/x86/hyperv/mmu.c
>> +++ b/arch/x86/hyperv/mmu.c
>> @@ -235,6 +235,7 @@ static u64 hyperv_flush_tlb_others_ex(const struct cp=
umask *cpus,
>>=20
>> void hyperv_setup_mmu_ops(void)
>>  {
>> +  return;
>>         if (!(ms_hyperv.hints & HV_X64_REMOTE_TLB_FLUSH_RECOMMENDED))
>>                 return;
>=20
> Otherwise, this code looks good to me and matches what I had in mind.
>=20
> Note that the function native_flush_tlb_others() is used when the Hyper-V s=
pecific
> flush function is disabled per patch #2 above, or when hv_cpu_to_vp_index(=
) returns
> VP_INVALID.  In a quick glance through the code, it appears that native_fl=
ush_tlb_others()
> will work even if there's a non-existent CPU in the cpumask that is passed=
 as an argument.
> So perhaps an immediate workaround is Patch #2 above.

The current code of hv_cpu_to_vp_index (where I generated the warning ) is r=
eturning VP_INVALID in this case (see previous mail) and look like it is not=
 completely workaround the issue.
the cpu is hanging even not panic Will continue watching .
>  =20
>=20
> Perhaps hyperv_flush_tlb_others() should be made equally tolerant of a non=
-existent
> CPU being in the list. But if you are willing, I'm still interested in the=
 results of an
> experiment with just Patch #1.  I'm curious about what the CPU list looks l=
ike when
> it has a non-existent CPU.  Is it complete garbage, or is there just one n=
on-existent
> CPU?
>=20
 We will do my be not next week since vacation but the week after

> The other curiosity is that I haven't seen this Linux panic reported by ot=
her users,
> and I think it would have come to our attention if it were happening with a=
ny frequency.
> You see the problem fairly regularly.  So I'm wondering what the differenc=
e is.
>=20
> Michael