Re: [RFC PATCH 1/3] sched/topology: Split out SD_* flags declaration to its own file

From: Valentin Schneider
Date: Mon Mar 23 2020 - 13:10:09 EST



Hi Morten,

Just as a heads-up, I think those changes would better fit 2/3, or be
in their own patch. 1/3 is just a straight up code move, with no changes
to the existing comments.

On Mon, Mar 23 2020, Morten Rasmussen wrote:
> On Wed, Mar 11, 2020 at 06:33:18PM +0000, Valentin Schneider wrote:
>> diff --git a/include/linux/sched/sd_flags.h b/include/linux/sched/sd_flags.h
>> new file mode 100644
>> index 000000000000..685bbe736945
>> --- /dev/null
>> +++ b/include/linux/sched/sd_flags.h
>> @@ -0,0 +1,33 @@
>> +// SPDX-License-Identifier: GPL-2.0
>> +/*
>> + * sched-domains (multiprocessor balancing) flag declarations.
>> + */
>> +
>> +/* Balance when about to become idle */
>> +SD_FLAG(SD_BALANCE_NEWIDLE, 0)
>> +/* Balance on exec */
>> +SD_FLAG(SD_BALANCE_EXEC, 1)
>> +/* Balance on fork, clone */
>> +SD_FLAG(SD_BALANCE_FORK, 2)
>> +/* Balance on wakeup */
>> +SD_FLAG(SD_BALANCE_WAKE, 3)
>> +/* Wake task to waking CPU */
>> +SD_FLAG(SD_WAKE_AFFINE, 4)
>
> Isn't it more like: "Consider waking task on waking CPU"?
>
> IIRC, with this flag set the wake-up can happen either near prev_cpu or
> this_cpu.
>

Right, it's not a hard guarantee.

>> +/* Domain members have different CPU capacities */
>> +SD_FLAG(SD_ASYM_CPUCAPACITY, 5)
>> +/* Domain members share CPU capacity */
>> +SD_FLAG(SD_SHARE_CPUCAPACITY, 6)
>
> Perhaps add +" (SMT)" to the comment to help the uninitiated
> understanding it a bit easier?
>

Sounds good.

>> +/* Domain members share power domain */
>> +SD_FLAG(SD_SHARE_POWERDOMAIN, 7)
>
> This flag is set only by 32-bit arm and has never had any effect. I
> think it was the beginning of something years ago that hasn't
> progressed. Perhaps we can remove it now?
>

Right, I don't think I've seen anything recent that uses that flag.

>> +/* Domain members share CPU pkg resources */
>> +SD_FLAG(SD_SHARE_PKG_RESOURCES, 8)
>
> +" (e.g. caches)" ?
>

Agreed! I actually already have that one in 2/3.

>> +/* Only a single load balancing instance */
>> +SD_FLAG(SD_SERIALIZE, 9)
>> +/* Place busy groups earlier in the domain */
>> +SD_FLAG(SD_ASYM_PACKING, 10)
>
> Place busy _tasks_ earlier in the domain?
>

Ack.

> It is a bit unclear what 'earlier' means here but since the packing
> ordering can actually be defined by the architecture, we can't be much
> more specific I guess.
>

This probably dates back to when ASYM_PACKING was really just for
bubbling tasks up to the first CPU of each core, and hasn't been changed
when the asym_priority thing was introduced. I can add a pointer to that.

> Morten