Re: [PATCH] Revert "ACPI: CPPC: Fix remaining for_each_possible_cpu() to use online CPUs"

[Sean Kelley <skelley@nvidia.com>]

On 4/16/26 8:46 PM, Jinjie Ruan wrote:
> 
> 
> On 4/17/2026 11:00 AM, Sean Kelley wrote:
>> On 4/16/26 7:00 PM, Jinjie Ruan wrote:
>>>
>>>
>>> On 4/17/2026 3:36 AM, Sean Kelley wrote:
>>>> On 4/16/26 1:52 AM, Jinjie Ruan wrote:
>>>>> This reverts commit 56eb0c0ed345da7815274aa821a8546a073d7e97, because
>>>>> this commit cause warning call trace below when concurrently
>>>>> bringing up
>>>>> and down two SMT threads of a physical core.
>>>>>
>>>>> The issue timeline is as follows:
>>>>>
>>>>> 1. when the system starts,
>>>>>       cpufreq: cpu: 220, policy->related_cpus: 220-221, policy->cpus:
>>>>> 220-221
>>>>>
>>>>> 2. Offline cpu 220 and cpu 221.
>>>>>
>>>>> 3. Online cpu 220
>>>>> - cpu 221 is now offline, as acpi_get_psd_map() use
>>>>> for_each_online_cpu(),
>>>>>       so the cpu_data->shared_cpu_map, policy->cpus, and related_cpus
>>>>> has only
>>>>>       cpu 220.
>>>>>       cpufreq: cpu: 220, related_cpus: 220, cpus: 220
>>>>>
>>>>> 4. offline cpu 220
>>>>>
>>>>> 5. online cpu 221, the below call trace occurs:
>>>>> - Because cpu 220 and cpu 221 share one policy, and policy-
>>>>>> related_cpus
>>>>>      = 220 after step 3, so cpu 221 is not in policy->related_cpus
>>>>>      but per_cpu(cpufreq_cpu_data, cpu221) is not NULL.
>>>>>
>>>>> The _PSD (P-State Dependency) defines the hardware-level dependency of
>>>>> frequency control across CPU cores. Since this relationship is a
>>>>> physical
>>>>> attribute of the hardware topology, it remains constant regardless
>>>>> of the
>>>>> online or offline status of the CPUs.
>>>>>
>>>>> Using for_each_online_cpu() in acpi_get_psd_map() is problematic. If a
>>>>> CPU is offline, it will be excluded from the shared_cpu_map.
>>>>> Consequently, if that CPU is brought online later, the kernel will
>>>>> fail to
>>>>> recognize it as part of any shared frequency domain.
>>>>>
>>>>> Switch back to for_each_possible_cpu() to ensure that all cores defined
>>>>> in the ACPI tables are correctly mapped into their respective
>>>>> performance
>>>>> domains from the start. This aligns with the logic of policy-
>>>>>> related_cpus,
>>>>> which must encompass all potentially available cores in the domain to
>>>>> prevent logic gaps during CPU hotplug operations.
>>>>
>>>>
>>>> Yep, agree that using for_each_online_cpu() in acpi_get_psd_map()
>>>> drops valid domain members and breaks the hotplug case you described.
>>>>
>>>> But a plain revert also re-exposes the nosmt bug. On systems where a
>>>> possible CPU is never probed, per_cpu(cpc_desc_ptr, i) is NULL and
>>>> acpi_get_psd_map() currently hits goto err_fault instead of just
>>>> skipping that CPU.
>>>
>>> I have a question regarding the original issue where it states 'This
>>> breaks systems booted with "nosmt" or "nosmt=force"'. As far as I know,
>>> the 'nosmt' parameter is only supported on x86. However, it seems the
>>> current x86 kernel does not support enabling CONFIG_ACPI_CPPC_CPUFREQ
>>> (only supported on arm64/arm/riscv). Could you please share the details
>>> of your testing environment?
>>
>> Yeah, good question. However, nosmt is actually defined generically in
>> kernel/cpu.c
>> (early_param("nosmt", smt_cmdline_disable)), not in any
>> arch-specific code, so it works on anything that selects HOTPLUG_SMT.
>>
>> arm64 does:
>>    arch/arm64/Kconfig    select HOTPLUG_SMT if HOTPLUG_CPU
>>
>> And acpi_get_psd_map() is called from drivers/cpufreq/cppc_cpufreq.c,
>> which has: depends on ARM || ARM64 || RISCV.
>>
>> So the overlap where this bug actually triggers is arm64. Testing was
>> on an NVIDIA Vera (Olympus) platform booted with "nosmt". That's the
>> environment where the original -EFAULT failure in acpi_get_psd_map()
>> was reproduced.
> 
> After reverting commit 56eb0c0ed345 ("ACPI: CPPC: Fix remaining
> for_each_possible_cpu() to use online CPUs"), I tried to reproduce your
> original issue on my local ARM64 machine with the 'nosmt' boot parameter
> enabled and I have confirmed that it works., but I couldn't trigger it
> and the acpi cppc cpufreq driver successfully registered. Could you
> please help test if the new fix resolves the problem you encountered
> previously?
> 
> Best regards,
> Jinjie
> 


Hi Jinjie,

Applied v2 on current master and built for arm64 (NVIDIA Vera, has SMT).

Tested both scenarios:

1. nosmt boot -- cppc_cpufreq registers cleanly, no -EFAULT, no
    warnings in dmesg.

2. Concurrent hotplug of SMT siblings (your original reproducer
    adapted to valid CPU indices on my system). Ran the dual
    offline/online loop for 30s. No WARN, no cpufreq_online call
    trace.

Both pass.

Thanks!

Sean




>>
>>>
>>>>
>>>> So I think the fix is to restore for_each_possible_cpu() for the PSD
>>>> map, but change the NULL case to continue:
>>>
>>> I agree.
>>>
>>>>
>>>> --- a/drivers/acpi/cppc_acpi.c
>>>> +++ b/drivers/acpi/cppc_acpi.c
>>>> @@ -530,7 +530,7 @@
>>>>
>>>>                   match_cpc_ptr = per_cpu(cpc_desc_ptr, i);
>>>>                   if (!match_cpc_ptr)
>>>> -                       goto err_fault;
>>>> +                       continue;
>>>>
>>>>                   match_pdomain = &(match_cpc_ptr->domain_info);
>>>>
>>>> That way offline CPUs with valid descriptors remain in shared_cpu_map
>>>> (fixing the hotplug trace), while never-probed CPUs are skipped
>>>> instead of failing map construction.
>>>>
>>>> The send_pcc_cmd() hunk already does if (!desc) continue, so reverting
>>>> that loop back to for_each_possible_cpu() looks fine as-is.
>>>>
>>>> Happy to send the continue fix as a patch on top, or please feel free
>>>> to fold it into yours if that makes sense.
>>>
>>> Thanks! I will fold your fix into my patch and add your Co-developed-by
>>> tag in the next version.
>>>
>>
>> Sounds good, thanks.
>>
>> Sean
>>
>>>>
>>>> Sean
>>>>
>>>>>
>>>>> How to reproduce, on arm64 machine with SMT support which use acpi cppc
>>>>> cpufreq driver:
>>>>>
>>>>>       bash test.sh 220 & bash test.sh 221 &
>>>>>
>>>>>       The test.sh is as below:
>>>>>           while true
>>>>>               do
>>>>>               echo 0 > /sys/devices/system/cpu/cpu${1}/online
>>>>>               sleep 0.5
>>>>>               cat /sys/devices/system/cpu/cpu${1}/cpufreq/related_cpus
>>>>>               echo 1 >  /sys/devices/system/cpu/cpu${1}/online
>>>>>               cat /sys/devices/system/cpu/cpu${1}/cpufreq/related_cpus
>>>>>           done
>>>>>
>>>>>       CPU: 221 PID: 1119 Comm: cpuhp/221 Kdump: loaded Not tainted
>>>>> 6.6.0debug+ #5
>>>>>       Hardware name: To be filled by O.E.M. S920X20/BC83AMDA01-7270Z,
>>>>> BIOS 20.39 09/04/2024
>>>>>       pstate: a1400009 (NzCv daif +PAN -UAO -TCO +DIT -SSBS BTYPE=--)
>>>>>       pc : cpufreq_online+0x8ac/0xa90
>>>>>       lr : cpuhp_cpufreq_online+0x18/0x30
>>>>>       sp : ffff80008739bce0
>>>>>       x29: ffff80008739bce0 x28: 0000000000000000 x27: ffff28400ca32200
>>>>>       x26: 0000000000000000 x25: 0000000000000003 x24: ffffd483503ff000
>>>>>       x23: ffffd483504051a0 x22: ffffd48350024a00 x21: 00000000000000dd
>>>>>       x20: 000000000000001d x19: ffff28400ca32000 x18: 0000000000000000
>>>>>       x17: 0000000000000020 x16: ffffd4834e6a3fc8 x15: 0000000000000020
>>>>>       x14: 0000000000000008 x13: 0000000000000001 x12: 00000000ffffffff
>>>>>       x11: 0000000000000040 x10: ffffd48350430728 x9 : ffffd4834f087c78
>>>>>       x8 : 0000000000000001 x7 : ffff2840092bdf00 x6 : ffffd483504264f0
>>>>>       x5 : ffffd48350405000 x4 : ffff283f7f95cc60 x3 : 0000000000000000
>>>>>       x2 : ffff53bc2f94b000 x1 : 00000000000000dd x0 : 0000000000000000
>>>>>       Call trace:
>>>>>        cpufreq_online+0x8ac/0xa90
>>>>>        cpuhp_cpufreq_online+0x18/0x30
>>>>>        cpuhp_invoke_callback+0x128/0x580
>>>>>        cpuhp_thread_fun+0x110/0x1b0
>>>>>        smpboot_thread_fn+0x140/0x190
>>>>>        kthread+0xec/0x100
>>>>>        ret_from_fork+0x10/0x20
>>>>>       ---[ end trace 0000000000000000 ]---
>>>>>
>>>>> Cc: stable@vger.kernel.org
>>>>> Fixes: 56eb0c0ed345 ("ACPI: CPPC: Fix remaining
>>>>> for_each_possible_cpu() to use online CPUs")
>>>>> Signed-off-by: Jinjie Ruan <ruanjinjie@huawei.com>
>>>>> ---
>>>>>     drivers/acpi/cppc_acpi.c | 4 ++--
>>>>>     1 file changed, 2 insertions(+), 2 deletions(-)
>>>>>
>>>>> diff --git a/drivers/acpi/cppc_acpi.c b/drivers/acpi/cppc_acpi.c
>>>>> index f0e513e9ed5d..9ae29f2c6db8 100644
>>>>> --- a/drivers/acpi/cppc_acpi.c
>>>>> +++ b/drivers/acpi/cppc_acpi.c
>>>>> @@ -362,7 +362,7 @@ static int send_pcc_cmd(int pcc_ss_id, u16 cmd)
>>>>>     end:
>>>>>         if (cmd == CMD_WRITE) {
>>>>>             if (unlikely(ret)) {
>>>>> -            for_each_online_cpu(i) {
>>>>> +            for_each_possible_cpu(i) {
>>>>>                     struct cpc_desc *desc = per_cpu(cpc_desc_ptr, i);
>>>>>                       if (!desc)
>>>>> @@ -524,7 +524,7 @@ int acpi_get_psd_map(unsigned int cpu, struct
>>>>> cppc_cpudata *cpu_data)
>>>>>         else if (pdomain->coord_type == DOMAIN_COORD_TYPE_SW_ANY)
>>>>>             cpu_data->shared_type = CPUFREQ_SHARED_TYPE_ANY;
>>>>>     -    for_each_online_cpu(i) {
>>>>> +    for_each_possible_cpu(i) {
>>>>>             if (i == cpu)
>>>>>                 continue;
>>>>>     
>>>>
>>>>
>>>
>>
>>
>