lkml.org 
[lkml]   [2016]   [Oct]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH] arch/x86: Fix kdump on x86 with physically hotadded CPUs


On 10/04/2016 06:58 AM, Thomas Gleixner wrote:
> On Mon, 3 Oct 2016, Prarit Bhargava wrote:
>> BUG: unable to handle kernel paging request at 0000000000841f1f
>> IP: [<ffffffff81014ec4>] uncore_change_context+0xd4/0x180
> ...
>> [<ffffffff81015a60>] ? uncore_cpu_starting+0x130/0x130
>> [<ffffffff81015acc>] uncore_event_cpu_online+0x6c/0x80
>> [<ffffffff8108e819>] cpuhp_invoke_callback+0x49/0x100
>> [<ffffffff8108ead1>] cpuhp_thread_fun+0x41/0x100
>> [<ffffffff810b054f>] smpboot_thread_fn+0x10f/0x160
>> [<ffffffff810b0440>] ? sort_range+0x30/0x30
>> [<ffffffff810accd8>] kthread+0xd8/0xf0
>> [<ffffffff816ff4bf>] ret_from_fork+0x1f/0x40
>> [<ffffffff810acc00>] ? kthread_park+0x60/0x60
>
>> arch/x86/events/intel/uncore.c:
>> 1137 static void uncore_change_type_ctx(struct intel_uncore_type *type, int old_ cpu,
>> 1138 int new_cpu)
>> 1139 {
>> 1140 struct intel_uncore_pmu *pmu = type->pmus;
>> 1141 struct intel_uncore_box *box;
>> 1142 int i, pkg;
>> 1143
>> 1144 pkg = topology_logical_package_id(old_cpu < 0 ? new_cpu : old_cpu);
>> 1145 for (i = 0; i < type->num_boxes; i++, pmu++) {
>> 1146 box = pmu->boxes[pkg];
>>
>> pmu->boxes[pkg] is garbage because pkg was returned as 0xffff.
>
> And that's what needs to be fixed in the first place.
>
>> This patch adds the missing generic_processor_info() to
>> prefill_possible_map() to ensure the initialization of the boot cpu is
>> correct.
>
>> This results in smp_init_package_map() having correct data and
>> properly setting the package map for the hotplugged boot cpu, which in
>> turn resolves the kdump kernel panic on physically hotplugged cpus.
>
> While it is the right thing to initialize the package map in that case, it
> still papers over a robustness issue in the uncore code, which needs to be
> fixed first.

I will include a separate patch with an error check for pkg == 0xffff in the
uncore code.

>
>> [2] prefill_possible_map() is called before smp_store_boot_cpu_info().
>> The comment beside the call to smp_store_boot_cpu_info() states that the
>> completed call results in "Final full version of the data".
>
> I'm not sure what that [2] here means and I cannot figure out the meaning
> of this sentence either.

My understanding is that after the call to smp_store_boot_cpu_info(), that for
the rest of the initial bringup the cpu_data structs, etc., are complete.

>
> This changelog is incomprehensible in general and more a "oh look how I
> decoded this problem" report than something which clearly describes the
> problem at hand, the root cause and the fix.

That wasn't my intention.

I had to figure it out and it took quite a while to get through it. I had
doubts that all of the others on the cc list would understand this and I
expected questions of "Can you prove that is the case?".

My changelog addresses those expected questions and I have no problem with a
shortened changelog.

The latter wants a
> understandable explanation why prefill_possible_map() is the right place to
> do this.

>
>> diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
>> index 4296beb8fdd3..d1272febc13b 100644
>> --- a/arch/x86/kernel/smpboot.c
>> +++ b/arch/x86/kernel/smpboot.c
>> @@ -1406,9 +1406,18 @@ __init void prefill_possible_map(void)
>> {
>> int i, possible;
>>
>> - /* no processor from mptable or madt */
>> - if (!num_processors)
>> - num_processors = 1;
>> + /* No boot processor was found in mptable or ACPI MADT */
>> + if (!num_processors) {
>> + /* Make sure boot cpu is enumerated */
>> + if (apic->cpu_present_to_apicid(0) == BAD_APICID &&
>> + apic->apic_id_valid(boot_cpu_physical_apicid))
>> + generic_processor_info(boot_cpu_physical_apicid,
>> + apic_version[boot_cpu_physical_apicid]);
>> + if (!num_processors) {
>> + pr_warn("CPU 0 not enumerated in mptable or ACPI MADT\n");
>> + num_processors = 1;
>
> And in this case we end up with the same problem, right?

It occurs to me that I over thought this: I was thinking that there might exist
a pre-ACPI (or at least a system without an MADT) x86 system that wold boot such
that num_processors = 0. But in that case, the cpu should be listed in the
mptables so the above should not happen. I'll change that to a BUG().

P.

>
> Thanks,
>
> tglx
>

\
 
 \ /
  Last update: 2016-10-04 14:10    [W:0.135 / U:0.868 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site