Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1421357 > unrolled thread

Re: [PATCH v6 10/11] cpuidle/powernv: Add support for POWER ISA v3 idle states

Started byBenjamin Herrenschmidt <benh@au1.ibm.com>
First post2016-06-13 23:50 +0200
Last post2016-06-14 13:40 +0200
Articles 3 — 2 participants

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v6 10/11] cpuidle/powernv: Add support for POWER ISA v3  idle states Benjamin Herrenschmidt <benh@au1.ibm.com> - 2016-06-13 23:50 +0200
    Re: [PATCH v6 10/11] cpuidle/powernv: Add support for POWER ISA v3  idle states Shreyas B Prabhu <shreyas@linux.vnet.ibm.com> - 2016-06-14 13:20 +0200
      Re: [PATCH v6 10/11] cpuidle/powernv: Add support for POWER ISA v3  idle states Benjamin Herrenschmidt <benh@au1.ibm.com> - 2016-06-14 13:40 +0200

#1421357 — Re: [PATCH v6 10/11] cpuidle/powernv: Add support for POWER ISA v3 idle states

FromBenjamin Herrenschmidt <benh@au1.ibm.com>
Date2016-06-13 23:50 +0200
SubjectRe: [PATCH v6 10/11] cpuidle/powernv: Add support for POWER ISA v3 idle states
Message-ID<rJHU5-3wH-1@gated-at.bofh.it>
On Wed, 2016-06-08 at 11:54 -0500, Shreyas B. Prabhu wrote:
> 
>  /*
>   * States for dedicated partition case.
>   */
> @@ -167,6 +183,8 @@ static int powernv_add_idle_states(void)
>  	int nr_idle_states = 1; /* Snooze */
>  	int dt_idle_states;
>  	u32 *latency_ns, *residency_ns, *flags;
> +	u64 *psscr_val = NULL;
> +	const char *names[CPUIDLE_STATE_MAX];
>  	int i, rc;
>  
>  	/* Currently we have snooze statically defined */
> @@ -199,12 +217,41 @@ static int powernv_add_idle_states(void)
>  		goto out_free_latency;
>  	}
>  
> +	rc = of_property_read_string_array(power_mgt,
> +					   "ibm,cpu-idle-state-names", names,
> +					   dt_idle_states);

Ok so from this I assume that dt_idle_states is the number of entries,
which has been checked properly to be < CPUIDLE_STATE_MAX correct ?

Beause ...

> +	if (rc < 0) {
> +		pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-names in DT\n");
> +		goto out_free_latency;
> +	}
> +
> +	/*
> +	 * If the idle states use stop instruction, probe for psscr values
> +	 * which are necessary to specify required stop level.
> +	 */
> +	if (flags[0] & (OPAL_PM_STOP_INST_FAST | OPAL_PM_STOP_INST_DEEP)) {
> +		psscr_val = kcalloc(dt_idle_states, sizeof(*psscr_val),
> +				    GFP_KERNEL);
> +		rc = of_property_read_u64_array(power_mgt,
> +						"ibm,cpu-idle-state-psscr",
> +						psscr_val, dt_idle_states);

Here, psscr val is only one u64 ... shouldn't you kmalloc sizeof(..) *
dt_idle_states ?

> +		if (rc) {
> +			pr_warn("cpuidle-powernv: missing ibm,cpu-idle-states-psscr in DT\n");
> +			goto out_free_psscr;
> +		}
> +	}
>  	residency_ns = kzalloc(sizeof(*residency_ns) * dt_idle_states, GFP_KERNEL);

Just like we do here

>  	rc = of_property_read_u32_array(power_mgt, 		"ibm,cpu-idle-state-residency-ns", residency_ns, dt_idle_states);
>  	for (i = 0; i < dt_idle_states; i++) {
> -
> +		/*
> +		 * If an idle state has exit latency beyond
> +		 * POWERNV_THRESHOLD_LATENCY_NS then don't use it
> +		 * in cpu-idle.
> +		 */
> +		if (latency_ns[i] > POWERNV_THRESHOLD_LATENCY_NS)
> +			continue;
>  		/*
>  		 * Cpuidle accepts exit_latency and target_residency in us.
>  		 * Use default target_residency values if f/w does not expose it.
> @@ -216,6 +263,16 @@ static int powernv_add_idle_states(void)
>  			powernv_states[nr_idle_states].flags = 0;
>  			powernv_states[nr_idle_states].target_residency = 100;
>  			powernv_states[nr_idle_states].enter = &nap_loop;
> +		} else if ((flags[i] & OPAL_PM_STOP_INST_FAST) &&
> +				!(flags[i] & OPAL_PM_TIMEBASE_STOP)) {
> +			strncpy(powernv_states[nr_idle_states].name,
> +				names[i], CPUIDLE_NAME_LEN);
> +			strncpy(powernv_states[nr_idle_states].desc,
> +				names[i], CPUIDLE_NAME_LEN);
> +			powernv_states[nr_idle_states].flags = 0;
> +
> +			powernv_states[nr_idle_states].enter = stop_loop;
> +			stop_psscr_table[nr_idle_states] = psscr_val[i];
>  		}
>  
>  		/*
> @@ -231,6 +288,16 @@ static int powernv_add_idle_states(void)
>  			powernv_states[nr_idle_states].flags = CPUIDLE_FLAG_TIMER_STOP;
>  			powernv_states[nr_idle_states].target_residency = 300000;
>  			powernv_states[nr_idle_states].enter = &fastsleep_loop;
> +		} else if ((flags[i] & OPAL_PM_STOP_INST_DEEP) &&
> +				(flags[i] & OPAL_PM_TIMEBASE_STOP)) {
> +			strncpy(powernv_states[nr_idle_states].name,
> +				names[i], CPUIDLE_NAME_LEN);
> +			strncpy(powernv_states[nr_idle_states].desc,
> +				names[i], CPUIDLE_NAME_LEN);
> +
> +			powernv_states[nr_idle_states].flags = CPUIDLE_FLAG_TIMER_STOP;
> +			powernv_states[nr_idle_states].enter = stop_loop;
> +			stop_psscr_table[nr_idle_states] = psscr_val[i];
>  		}
>  #endif
>  		powernv_states[nr_idle_states].exit_latency =
> @@ -245,6 +312,8 @@ static int powernv_add_idle_states(void)
>  	}
>  
>  	kfree(residency_ns);
> +out_free_psscr:
> +	kfree(psscr_val);
>  out_free_latency:
>  	kfree(latency_ns);
>  out_free_flags:

[toc] | [next] | [standalone]

#1421777

FromShreyas B Prabhu <shreyas@linux.vnet.ibm.com>
Date2016-06-14 13:20 +0200
Message-ID<rJUxX-46H-11@gated-at.bofh.it>
In reply to#1421357

On 06/14/2016 03:18 AM, Benjamin Herrenschmidt wrote:
> On Wed, 2016-06-08 at 11:54 -0500, Shreyas B. Prabhu wrote:
>>
>>  /*
>>   * States for dedicated partition case.
>>   */
>> @@ -167,6 +183,8 @@ static int powernv_add_idle_states(void)
>>  	int nr_idle_states = 1; /* Snooze */
>>  	int dt_idle_states;
>>  	u32 *latency_ns, *residency_ns, *flags;
>> +	u64 *psscr_val = NULL;
>> +	const char *names[CPUIDLE_STATE_MAX];
>>  	int i, rc;
>>  
>>  	/* Currently we have snooze statically defined */
>> @@ -199,12 +217,41 @@ static int powernv_add_idle_states(void)
>>  		goto out_free_latency;
>>  	}
>>  
>> +	rc = of_property_read_string_array(power_mgt,
>> +					   "ibm,cpu-idle-state-names", names,
>> +					   dt_idle_states);
> 
> Ok so from this I assume that dt_idle_states is the number of entries,
> which has been checked properly to be < CPUIDLE_STATE_MAX correct ?
> 
> Beause ...
>

While dt_idle_states should not be > CPUIDLE_STATE_MAX, if that were the
case we will end up corrupting memory while updating powernv_states[].
I'll add a WARN_ON for such a case and
handle adding idle states to powernv_states accordingly. Thanks for
pointing this out.

>> +	if (rc < 0) {
>> +		pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-names in DT\n");
>> +		goto out_free_latency;
>> +	}
>> +
>> +	/*
>> +	 * If the idle states use stop instruction, probe for psscr values
>> +	 * which are necessary to specify required stop level.
>> +	 */
>> +	if (flags[0] & (OPAL_PM_STOP_INST_FAST | OPAL_PM_STOP_INST_DEEP)) {
>> +		psscr_val = kcalloc(dt_idle_states, sizeof(*psscr_val),
>> +				    GFP_KERNEL);
>> +		rc = of_property_read_u64_array(power_mgt,
>> +						"ibm,cpu-idle-state-psscr",
>> +						psscr_val, dt_idle_states);
> 
> Here, psscr val is only one u64 ... shouldn't you kmalloc sizeof(..) *
> dt_idle_states ?

I'm using kcalloc here since checkpatch script suggested kcalloc over
kzalloc for allocating memory for arrays.
I'll also include a patch to use kcalloc throughout the file for
uniformity in next version. I was originally planning to post that
cleanup separately.

Thanks,
Shreyas

[toc] | [prev] | [next] | [standalone]

#1421799

FromBenjamin Herrenschmidt <benh@au1.ibm.com>
Date2016-06-14 13:40 +0200
Message-ID<rJURj-4dg-5@gated-at.bofh.it>
In reply to#1421777
	
 <1465404871-5406-11-git-send-email-shreyas@linux.vnet.ibm.com>
	
	
 <1465854492.3022.30.camel@au1.ibm.com>
	 <575FE64C.9080107@linux.vnet.ibm.com>
Organization: IBM Australia
Content-Type: text/plain; charset="UTF-8"
X-Mailer: Evolution 3.20.3 (3.20.3-1.fc24) 
Mime-Version: 1.0
Content-Transfer-Encoding: 8bit

On Tue, 2016-06-14 at 16:41 +0530, Shreyas B Prabhu wrote:
> >> +            psscr_val = kcalloc(dt_idle_states, sizeof(*psscr_val),
> >> +                                GFP_KERNEL);
> >> +            rc = of_property_read_u64_array(power_mgt,
> >> +                                            "ibm,cpu-idle-state-psscr",
> >> +                                            psscr_val, dt_idle_states);
> > 
> > Here, psscr val is only one u64 ... shouldn't you kmalloc sizeof(..) *
> > dt_idle_states ?
> 
> I'm using kcalloc here since checkpatch script suggested kcalloc over
> kzalloc for allocating memory for arrays.
> I'll also include a patch to use kcalloc throughout the file for
> uniformity in next version. I was originally planning to post that
> cleanup separately.

Ah ok, I missed the use of kcalloc (I didn't even know its existence),
my brain just read kmalloc :-)

Still, I find it inconsistent that you allocate here while you use the
stack for the names. Any reason for that ?

Cheers,
Ben.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web