linux-kernel - Re: [PATCH v3] PM: EM: Fix potential division-by-zero error in em_compute

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: <tencent_8E3A87C6D6A193F757BA846F0C41887CC405@qq.com>
Date: Wed, 16 Apr 2025 10:56:44 +0800
From: Yaxiong Tian <iambestgod@...com>
To: "Rafael J. Wysocki" <rafael@...nel.org>
Cc: lukasz.luba@....com, linux-pm@...r.kernel.org,
 linux-kernel@...r.kernel.org, Yaxiong Tian <tianyaxiong@...inos.cn>
Subject: Re: [PATCH v3] PM: EM: Fix potential division-by-zero error in
 em_compute_costs()



在 2025/4/16 01:17, Rafael J. Wysocki 写道:
> On Mon, Apr 14, 2025 at 11:09 AM Yaxiong Tian <iambestgod@...com> wrote:
>>
>> From: Yaxiong Tian <tianyaxiong@...inos.cn>
>>
>> When the device is of a non-CPU type, table[i].performance won't be
>> initialized in the previous em_init_performance(), resulting in division
>> by zero when calculating costs in em_compute_costs().
>>
>> Since the 'cost' algorithm is only used for EAS energy efficiency
>> calculations and is currently not utilized by other device drivers, we
>> should add the _is_cpu_device(dev) check to prevent this division-by-zero
>> issue.
>>
>> Fixes: <1b600da51073> ("PM: EM: Optimize em_cpu_energy() and remove division")
> 
> Please look at the Fixes: tags in the kernel git history.  They don't
> look like the one above.
> 
Yes, there's an extra '<>' here.

>> Signed-off-by: Yaxiong Tian <tianyaxiong@...inos.cn>
>> ---
>>   kernel/power/energy_model.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
>> index d9b7e2b38c7a..fc972cc1fc12 100644
>> --- a/kernel/power/energy_model.c
>> +++ b/kernel/power/energy_model.c
>> @@ -235,7 +235,7 @@ static int em_compute_costs(struct device *dev, struct em_perf_state *table,
>>
>>          /* Compute the cost of each performance state. */
>>          for (i = nr_states - 1; i >= 0; i--) {
>> -               unsigned long power_res, cost;
>> +               unsigned long power_res, cost = 0;
>>
>>                  if ((flags & EM_PERF_DOMAIN_ARTIFICIAL) && cb->get_cost) {
>>                          ret = cb->get_cost(dev, table[i].frequency, &cost);
>> @@ -244,7 +244,7 @@ static int em_compute_costs(struct device *dev, struct em_perf_state *table,
>>                                          cost, ret);
>>                                  return -EINVAL;
>>                          }
>> -               } else {
>> +               } else if (_is_cpu_device(dev)) {
> 
> Can't you just check this upfront at the beginning of the function and
> make it bail out if dev is not a CPU device?
> 
Sure, But the current implementation applies em_compute_costs() to both 
non-CPU devices and CPU devices. After carefully reviewing the latest code,
I've found this issue has expanded in scope.

There are currently three call paths for invoking em_compute_costs():

1) Registering performance domains (for both non-CPU and CPU devices)
em_dev_register_perf_domain() → em_create_pd() →
em_create_perf_table() → em_compute_costs()

2)EM update paths (CPU devices only)

Periodic 1000ms update check via em_update_work work item:
em_check_capacity_update() → em_adjust_new_capacity() → 
em_recalc_and_update() → em_compute_costs()

Exynos-chip initialization:
em_dev_update_chip_binning() → em_recalc_and_update() → em_compute_costs()

3) Device cost computation (non-CPU devices only - currently unused)
em_dev_compute_costs() → em_compute_costs()

Note: In em_dev_compute_costs(), when calling em_compute_costs(),
neither the callback (cb) nor flags are set.In fact, it either does
nothing at all or performs incorrect operations.

Therefore, should we mandate that non-CPU devices must provide a
get_cost callback?

So Should we add a check at the beginning of the em_compute_costs() to:

	if (!_is_cpu_device(dev) && !cb->get_cost) {
		dev_dbg(dev, "EM: No get_cost provided, cost unset.\n");
		return 0;
	}
And Modify em_dev_compute_costs() to require callers to provide the cb
callback function,Also need to update its corresponding comments.


>>                          /* increase resolution of 'cost' precision */
>>                          power_res = table[i].power * 10;
>>                          cost = power_res / table[i].performance;
>> --