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 for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <SL2PR06MB30828CF9FF2879AFC9DC53D2BD0C9@SL2PR06MB3082.apcprd06.prod.outlook.com>
Date:   Fri, 11 Mar 2022 02:03:46 +0000
From:   王擎 <wangqing@...o.com>
To:     Vincent Guittot <vincent.guittot@...aro.org>
CC:     Catalin Marinas <Catalin.Marinas@....com>,
        Will Deacon <will@...nel.org>,
        Sudeep Holla <sudeep.holla@....com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "Rafael J. Wysocki" <rafael@...nel.org>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Juri Lelli <juri.lelli@...hat.com>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Ben Segall <bsegall@...gle.com>, Mel Gorman <mgorman@...e.de>,
        Daniel Bristot de Oliveira <bristot@...hat.com>,
        "linux-arm-kernel@...ts.infradead.org" 
        <linux-arm-kernel@...ts.infradead.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH] sched: topology: make cache topology separate from cpu
 topology

 
>On Thu, 10 Mar 2022 at 13:59, Qing Wang <wangqing@...o.com> wrote:
>>
>> From: Wang Qing <wangqing@...o.com>
>>
>> Some architectures(e.g. ARM64), caches are implemented below:
>> cluster:                     ****** cluster 0 *****      ****** cluster 1 *****
>> core:                         0      1          2      3          4      5           6      7
(add cache level 1)        c0    c1        c2    c3         c4    c5         c6    c7
>> cache(Leveln):         **cache0**  **cache1**  **cache2**  **cache3**
(add cache level 3)        *************share level 3 cache ***************
>> sd_llc_id(current):     0      0          0      0          4      4           4      4
>> sd_llc_id(should be): 0      0          2      2          4      4           6      6
>>
Here, n always be 2 in ARM64, but others are also possible.
core[0,1] form a complex(ARMV9),  which share L2 cache, core[2,3] is the same.

>> Caches and cpus have different topology, this causes cpus_share_cache()
>> return the wrong value, which will affect the CPU load balance.
>>
>What does your current scheduler topology  look like?
>
>For CPU 0 to 3, do you have the below ?
>DIE [0     -     3] [4-7]
>MC  [0] [1] [2] [3]

The current scheduler topology consistent with CPU topology:
DIE  [0-7]           
MC  [0-3] [4-7]  (SD_SHARE_PKG_RESOURCES)
Most Android phones have this topology.
>
>But you would like something like below for cpu 0-1 instead ?
>DIE [0     -     3] [4-7]
>CLS [0 - 1] [2 - 3]
>MC  [0] [1]
>
>with SD_SHARE_PKG_RESOURCES only set to MC level ?

We don't change the current scheduler topology, but the  
cache topology should be separated like below:
[0-7]                          (shared level 3 cache )
[0-1] [2-3][4-5][6-7]   (shared level 2 cache )
>
>>
>> Cache topology should be separated with CPU topology, it can be obtained
>> from "next-level-cache" in DTS preferentially.
>>
>> Signed-off-by: Wang Qing <wangqing@...o.com>
>> ---
>>  arch/arm64/kernel/smp.c       |  1 +
>>  drivers/base/arch_topology.c  | 23 +++++++++++++++++++++++
>>  include/linux/arch_topology.h |  3 +++
>>  kernel/sched/topology.c       | 33 +++++++++++++++++++++++++++++++--
>>  4 files changed, 58 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
>> index 27df5c1..94cf649
>> --- a/arch/arm64/kernel/smp.c
>> +++ b/arch/arm64/kernel/smp.c
>> @@ -723,6 +723,7 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
>>         unsigned int this_cpu;
>>
>>         init_cpu_topology();
>> +       init_cpu_cache_topology();
>>
>>         this_cpu = smp_processor_id();
>>         store_cpu_topology(this_cpu);
>> diff --git a/drivers/base/arch_topology.c b/drivers/base/arch_topology.c
>> index 9761541..613213f
>> --- a/drivers/base/arch_topology.c
>> +++ b/drivers/base/arch_topology.c
>> @@ -613,6 +613,7 @@ static int __init parse_dt_topology(void)
>>   */
>>  struct cpu_topology cpu_topology[NR_CPUS];
>>  EXPORT_SYMBOL_GPL(cpu_topology);
>> +struct device_node *cache_topology[NR_CPUS][MAX_CPU_CACHE_LEVEL];
>
>AFAICT, arch_topology.c is only used by arm/arm64 and riscv so this is
>not initialized for other archs

I see, will be fixed in V2.

Thanks,
Wang
>
>>
>>  const struct cpumask *cpu_coregroup_mask(int cpu)
>>  {
>> @@ -738,4 +739,26 @@ void __init init_cpu_topology(void)
>>         else if (of_have_populated_dt() && parse_dt_topology())
>>                 reset_cpu_topology();
>>  }
>> +
>> +void __init init_cpu_cache_topology(void)
>> +{
>> +       struct device_node *node_cpu, *node_cache;
>> +       int cpu, level;
>> +
>> +       for_each_possible_cpu(cpu) {
>> +               node_cpu = of_get_cpu_node(cpu, NULL);
>> +               if (!node_cpu)
>> +                       continue;
>> +
>> +               level = 0;
>> +               node_cache = node_cpu;
>> +               while (level < MAX_CPU_CACHE_LEVEL) {
>> +                       node_cache = of_parse_phandle(node_cache, "next-level-cache", 0);
>> +                       if (!node_cache)
>> +                               break;
>> +                       cache_topology[cpu][level++] = node_cache;
>> +               }
>> +               of_node_put(node_cpu);
>> +       }
>> +}
>>  #endif
>> diff --git a/include/linux/arch_topology.h b/include/linux/arch_topology.h
>> index cce6136b..d37f47d
>> --- a/include/linux/arch_topology.h
>> +++ b/include/linux/arch_topology.h
>> @@ -72,6 +72,8 @@ struct cpu_topology {
>>  };
>>
>>  #ifdef CONFIG_GENERIC_ARCH_TOPOLOGY
>> +#define MAX_CPU_CACHE_LEVEL 7
>> +extern struct device_node *cache_topology[NR_CPUS][MAX_CPU_CACHE_LEVEL];
>>  extern struct cpu_topology cpu_topology[NR_CPUS];
>>
>>  #define topology_physical_package_id(cpu)      (cpu_topology[cpu].package_id)
>> @@ -82,6 +84,7 @@ extern struct cpu_topology cpu_topology[NR_CPUS];
>>  #define topology_cluster_cpumask(cpu)  (&cpu_topology[cpu].cluster_sibling)
>>  #define topology_llc_cpumask(cpu)      (&cpu_topology[cpu].llc_sibling)
>>  void init_cpu_topology(void);
>> +void init_cpu_cache_topology(void);
>>  void store_cpu_topology(unsigned int cpuid);
>>  const struct cpumask *cpu_coregroup_mask(int cpu);
>>  const struct cpumask *cpu_clustergroup_mask(int cpu);
>> diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
>> index d201a70..10850d6
>> --- a/kernel/sched/topology.c
>> +++ b/kernel/sched/topology.c
>> @@ -650,6 +650,36 @@ DEFINE_PER_CPU(struct sched_domain __rcu *, sd_asym_packing);
>>  DEFINE_PER_CPU(struct sched_domain __rcu *, sd_asym_cpucapacity);
>>  DEFINE_STATIC_KEY_FALSE(sched_asym_cpucapacity);
>>
>> +static void set_sd_llc(int cpu, struct sched_domain *sd, int *first_cpu, int *cpu_num)
>> +{
>> +       int cache_level, cpu_id;
>> +       int first, last;
>> +       int id = cpumask_first(sched_domain_span(sd));
>> +       int size = cpumask_weight(sched_domain_span(sd));
>> +
>> +       *first_cpu = id;
>> +       *cpu_num = size;
>> +
>> +       for (cache_level = 0; cache_level < MAX_CPU_CACHE_LEVEL; cache_level++) {
>> +               if (!cache_topology[cpu][cache_level])
>> +                       break;
>> +
>> +               first = -1;
>> +               last = id;
>> +               for (cpu_id = 0; cpu_id < NR_CPUS; cpu_id++) {
>> +                       if (cache_topology[cpu][cache_level] == cache_topology[cpu_id][cache_level]) {
>> +                               if (cpu_id >= id && cpu_id < id + size) {
>> +                                       first = (first == -1)?cpu_id:first;
>> +                                       last = cpu_id;
>> +                               } else
>> +                                       return;
>> +                       }
>> +               }
>> +               *first_cpu = first;
>> +               *cpu_num = last - first + 1;
>> +       }
>> +}
>> +
>>  static void update_top_cache_domain(int cpu)
>>  {
>>         struct sched_domain_shared *sds = NULL;
>> @@ -659,8 +689,7 @@ static void update_top_cache_domain(int cpu)
>>
>>         sd = highest_flag_domain(cpu, SD_SHARE_PKG_RESOURCES);
>>         if (sd) {
>> -               id = cpumask_first(sched_domain_span(sd));
>> -               size = cpumask_weight(sched_domain_span(sd));
>> +               set_sd_llc(cpu, sd, &id, &size);
>
>In scheduler, we look for the last level of SD_SHARE_PKG_RESOURCES to
>find shared memory. It seems that cpu_coregroup_mask doesn't return
>the correct cpumask in your case as it returns a full cluster instead
>of a subset
>
>>                 sds = sd->shared;
>
>sds must  stay aligned with id and size so instead of modifying id and
>size you should returns a cpumask that reflects your topology
>
>>         }
>>
>> --
>> 2.7.4
>

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ