Re: [RFC 1/4] perf: Move some access checks later in perf_event_open

[Tvrtko Ursulin <tursulin@ursulin.net>]

On 26/06/18 18:24, Alexey Budankov wrote:
> Hi,
> 
> On 26.06.2018 18:36, Tvrtko Ursulin wrote:
>> From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>>
>> To enable per-PMU access controls in a following patch first move all call
>> sites of perf_paranoid_kernel() to after the event has been created.
>>
>> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
>> Cc: Thomas Gleixner <tglx@linutronix.de>
>> Cc: Peter Zijlstra <peterz@infradead.org>
>> Cc: Ingo Molnar <mingo@redhat.com>
>> Cc: "H. Peter Anvin" <hpa@zytor.com>
>> Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
>> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
>> Cc: Jiri Olsa <jolsa@redhat.com>
>> Cc: Namhyung Kim <namhyung@kernel.org>
>> Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
>> Cc: Andi Kleen <ak@linux.intel.com>
>> Cc: Alexey Budankov <alexey.budankov@linux.intel.com>
>> Cc: linux-kernel@vger.kernel.org
>> Cc: x86@kernel.org
>> ---
>>   kernel/events/core.c | 36 ++++++++++++++++++++++--------------
>>   1 file changed, 22 insertions(+), 14 deletions(-)
>>
>> diff --git a/kernel/events/core.c b/kernel/events/core.c
>> index f490caca9aa4..12de95b0472e 100644
>> --- a/kernel/events/core.c
>> +++ b/kernel/events/core.c
>> @@ -10189,10 +10189,6 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr,
>>   			 */
>>   			attr->branch_sample_type = mask;
>>   		}
>> -		/* privileged levels capture (kernel, hv): check permissions */
>> -		if ((mask & PERF_SAMPLE_BRANCH_PERM_PLM)
>> -		    && perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN))
>> -			return -EACCES;
>>   	}
>>   
>>   	if (attr->sample_type & PERF_SAMPLE_REGS_USER) {
>> @@ -10409,11 +10405,6 @@ SYSCALL_DEFINE5(perf_event_open,
>>   	if (err)
>>   		return err;
>>   
>> -	if (!attr.exclude_kernel) {
>> -		if (perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN))
>> -			return -EACCES;
>> -	}
>> -
>>   	if (attr.namespaces) {
>>   		if (!capable(CAP_SYS_ADMIN))
>>   			return -EACCES;
>> @@ -10427,11 +10418,6 @@ SYSCALL_DEFINE5(perf_event_open,
>>   			return -EINVAL;
>>   	}
>>   
>> -	/* Only privileged users can get physical addresses */
>> -	if ((attr.sample_type & PERF_SAMPLE_PHYS_ADDR) &&
>> -	    perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN))
>> -		return -EACCES;
>> -
>>   	/*
>>   	 * In cgroup mode, the pid argument is used to pass the fd
>>   	 * opened to the cgroup directory in cgroupfs. The cpu argument
>> @@ -10501,6 +10487,28 @@ SYSCALL_DEFINE5(perf_event_open,
>>   		goto err_cred;
>>   	}
>>   
>> +	if (!attr.exclude_kernel) {
>> +		if (perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN)) {
>> +			err = -EACCES;
> 
> I would separate this combined permissions check into a function e.g.
> static bool perf_test_pmu_paranoid(const struct pmu *pmu, int *err) to avoid
> code duplication.

My thinking was for this to be as mechanical (code movement) as 
possible, but I can consider it.

Regards,

Tvrtko

>> +			goto err_alloc;
>> +		}
>> +	}
>> +
>> +	/* Only privileged users can get physical addresses */
>> +	if ((attr.sample_type & PERF_SAMPLE_PHYS_ADDR) &&
>> +	    perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN)) {
>> +		err = -EACCES;
>> +		goto err_alloc;
>> +	}
>> +
>> +	/* privileged levels capture (kernel, hv): check permissions */
>> +	if ((attr.sample_type & PERF_SAMPLE_BRANCH_STACK) &&
>> +	    (attr.branch_sample_type & PERF_SAMPLE_BRANCH_PERM_PLM) &&
>> +	    perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN)) {
>> +		err = -EACCES;
>> +		goto err_alloc;
>> +	}
>> +
>>   	if (is_sampling_event(event)) {
>>   		if (event->pmu->capabilities & PERF_PMU_CAP_NO_INTERRUPT) {
>>   			err = -EOPNOTSUPP;
>>
>