[PATCHv9 2/2] powerpc/setup: Loosen the mapping between cpu logical id and its seq in dt

Hari Bathini hbathini at linux.ibm.com
Mon Jan 8 20:33:02 PST 2024



On 09/01/24 9:57 am, Hari Bathini wrote:
> Hi Michael,
> 

Sorry, Michael.
I am just about getting back to work and I spoke too soon.
You already seem to have posted a set with the approach you had in mind:

   https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=388350

Thanks
Hari

> I am fine with either approach. I was trying to address your concerns
> in my way. Looking for your inputs here on how to go about this now..
> 
> On 29/11/23 7:00 am, Pingfan Liu wrote:
>> Hi Hari,
>>
>>
>> On Mon, Nov 27, 2023 at 12:30 PM Hari Bathini <hbathini at linux.ibm.com> 
>> wrote:
>>>
>>> Hi Pingfan, Michael,
>>>
>>> On 17/10/23 4:03 pm, Hari Bathini wrote:
>>>>
>>>>
>>>> On 17/10/23 7:58 am, Pingfan Liu wrote:
>>>>> *** Idea ***
>>>>> For kexec -p, the boot cpu can be not the cpu0, this causes the 
>>>>> problem
>>>>> of allocating memory for paca_ptrs[]. However, in theory, there is no
>>>>> requirement to assign cpu's logical id as its present sequence in the
>>>>> device tree. But there is something like cpu_first_thread_sibling(),
>>>>> which makes assumption on the mapping inside a core. Hence partially
>>>>> loosening the mapping, i.e. unbind the mapping of core while keep the
>>>>> mapping inside a core.
>>>>>
>>>>> *** Implement ***
>>>>> At this early stage, there are plenty of memory to utilize. Hence, 
>>>>> this
>>>>> patch allocates interim memory to link the cpu info on a list, then
>>>>> reorder cpus by changing the list head. As a result, there is a rotate
>>>>> shift between the sequence number in dt and the cpu logical number.
>>>>>
>>>>> *** Result ***
>>>>> After this patch, a boot-cpu's logical id will always be mapped 
>>>>> into the
>>>>> range [0,threads_per_core).
>>>>>
>>>>> Besides this, at this phase, all threads in the boot core are 
>>>>> forced to
>>>>> be onlined. This restriction will be lifted in a later patch with
>>>>> extra effort.
>>>>>
>>>>> Signed-off-by: Pingfan Liu <piliu at redhat.com>
>>>>> Cc: Michael Ellerman <mpe at ellerman.id.au>
>>>>> Cc: Nicholas Piggin <npiggin at gmail.com>
>>>>> Cc: Christophe Leroy <christophe.leroy at csgroup.eu>
>>>>> Cc: Mahesh Salgaonkar <mahesh at linux.ibm.com>
>>>>> Cc: Wen Xiong <wenxiong at us.ibm.com>
>>>>> Cc: Baoquan He <bhe at redhat.com>
>>>>> Cc: Ming Lei <ming.lei at redhat.com>
>>>>> Cc: Sourabh Jain <sourabhjain at linux.ibm.com>
>>>>> Cc: Hari Bathini <hbathini at linux.ibm.com>
>>>>> Cc: kexec at lists.infradead.org
>>>>> To: linuxppc-dev at lists.ozlabs.org
>>>>
>>>> Thanks for working on this, Pingfan.
>>>> Looks good to me.
>>>>
>>>> Acked-by: Hari Bathini <hbathini at linux.ibm.com>
>>>>
>>>
>>> On second thoughts, probably better off with no impact for
>>> bootcpu < nr_cpu_ids case and changing only two cores logical
>>> numbering otherwise. Something like the below (Please share
>>> your thoughts):
>>>
>>
>> I am afraid that it may not be as ideal as it looks, considering the
>> following factors:
>> -1. For the case of 'bootcpu < nr_cpu_ids', crash can happen evenly
>> across any cpu in the system, which seriously undermines the
>> protection intended here (Under the most optimistic scenario, there is
>> a 50% chance of success)
>>
>> -2. For the re-ordering of logical numbering, IMHO, if there is
>> concern that re-ordering will break something, the partial re-ordering
>> can not avoid that.  We ought to spot probable hazards so as to ease
>> worries.
>>
>>
>> Thanks,
>>
>> Pingfan
>>
>>> diff --git a/arch/powerpc/kernel/prom.c b/arch/powerpc/kernel/prom.c
>>> index ec82f5bda908..78a8312aa8c4 100644
>>> --- a/arch/powerpc/kernel/prom.c
>>> +++ b/arch/powerpc/kernel/prom.c
>>> @@ -76,7 +76,9 @@ u64 ppc64_rma_size;
>>>    unsigned int boot_cpu_node_count __ro_after_init;
>>>    #endif
>>>    static phys_addr_t first_memblock_size;
>>> +#ifdef CONFIG_SMP
>>>    static int __initdata boot_cpu_count;
>>> +#endif
>>>
>>>    static int __init early_parse_mem(char *p)
>>>    {
>>> @@ -357,6 +359,25 @@ static int __init early_init_dt_scan_cpus(unsigned
>>> long node,
>>>                          fdt_boot_cpuid_phys(initial_boot_params)) {
>>>                          found = boot_cpu_count;
>>>                          found_thread = i;
>>> +                       /*
>>> +                        * Map boot-cpu logical id into the range
>>> +                        * of [0, thread_per_core) if it can't be
>>> +                        * accommodated within nr_cpu_ids.
>>> +                        */
>>> +                       if (i != boot_cpu_count && boot_cpu_count >= 
>>> nr_cpu_ids) {
>>> +                               boot_cpuid = i;
>>> +                               DBG("Logical CPU number for boot CPU 
>>> changed from %d to %d\n",
>>> +                                   boot_cpu_count, i);
>>> +                       } else {
>>> +                               boot_cpuid = boot_cpu_count;
>>> +                       }
>>> +
>>> +                       /* Ensure boot thread is acconted for in 
>>> nr_cpu_ids */
>>> +                       if (boot_cpuid >= nr_cpu_ids) {
>>> +                               set_nr_cpu_ids(boot_cpuid + 1);
>>> +                               DBG("Adjusted nr_cpu_ids to %u, to 
>>> include boot CPU.\n",
>>> +                                   nr_cpu_ids);
>>> +                       }
>>>                  }
>>>    #ifdef CONFIG_SMP
>>>                  /* logical cpu id is always 0 on UP kernels */
>>> @@ -368,9 +389,8 @@ static int __init early_init_dt_scan_cpus(unsigned
>>> long node,
>>>          if (found < 0)
>>>                  return 0;
>>>
>>> -       DBG("boot cpu: logical %d physical %d\n", found,
>>> +       DBG("boot cpu: logical %d physical %d\n", boot_cpuid,
>>>              be32_to_cpu(intserv[found_thread]));
>>> -       boot_cpuid = found;
>>>
>>>          boot_cpu_hwid = be32_to_cpu(intserv[found_thread]);
>>>
>>> diff --git a/arch/powerpc/kernel/setup-common.c
>>> b/arch/powerpc/kernel/setup-common.c
>>> index b7b733474b60..f7179525c774 100644
>>> --- a/arch/powerpc/kernel/setup-common.c
>>> +++ b/arch/powerpc/kernel/setup-common.c
>>> @@ -409,6 +409,12 @@ static void __init cpu_init_thread_core_maps(int 
>>> tpc)
>>>
>>>    u32 *cpu_to_phys_id = NULL;
>>>
>>> +struct interrupt_server_node {
>>> +       bool    avail;
>>> +       int     len;
>>> +       __be32 intserv[];
>>> +};
>>> +
>>>    /**
>>>     * setup_cpu_maps - initialize the following cpu maps:
>>>     *                  cpu_possible_mask
>>> @@ -429,9 +435,13 @@ u32 *cpu_to_phys_id = NULL;
>>>     */
>>>    void __init smp_setup_cpu_maps(void)
>>>    {
>>> +       struct interrupt_server_node *core0_node = NULL, *bt_node = 
>>> NULL;
>>> +       int orig_boot_cpu = -1, orig_boot_thread = -1;
>>> +       bool found_boot_cpu = false;
>>>          struct device_node *dn;
>>> -       int cpu = 0;
>>>          int nthreads = 1;
>>> +       int cpu = 0;
>>> +       int j, len;
>>>
>>>          DBG("smp_setup_cpu_maps()\n");
>>>
>>> @@ -442,9 +452,9 @@ void __init smp_setup_cpu_maps(void)
>>>                        __func__, nr_cpu_ids * sizeof(u32), 
>>> __alignof__(u32));
>>>
>>>          for_each_node_by_type(dn, "cpu") {
>>> +               bool avail, skip = false;
>>>                  const __be32 *intserv;
>>>                  __be32 cpu_be;
>>> -               int j, len;
>>>
>>>                  DBG("  * %pOF...\n", dn);
>>>
>>> @@ -466,29 +476,121 @@ void __init smp_setup_cpu_maps(void)
>>>
>>>                  nthreads = len / sizeof(int);
>>>
>>> -               for (j = 0; j < nthreads && cpu < nr_cpu_ids; j++) {
>>> -                       bool avail;
>>> +               avail = of_device_is_available(dn);
>>> +               if (!avail)
>>> +                       avail = !of_property_match_string(dn,
>>> +                                       "enable-method", "spin-table");
>>> +
>>> +               for (j = 0; (cpu == 0 || !found_boot_cpu) && j < 
>>> nthreads; j++) {
>>> +                       if (be32_to_cpu(intserv[j]) == boot_cpu_hwid) {
>>> +                               found_boot_cpu = true;
>>> +                               if (cpu == 0)
>>> +                                       break;
>>> +
>>> +                               /* Original logical CPU number of 
>>> thread0 in boot core */
>>> +                               orig_boot_cpu = cpu;
>>> +                               orig_boot_thread = j;
>>> +                               bt_node = 
>>> memblock_alloc(sizeof(struct interrupt_server_node) + len,
>>> +                                               __alignof__(u32));
>>> +                               if (!bt_node)
>>> +                                       panic("%s: Failed to allocate 
>>> %zu bytes align=0x%zx\n",
>>> +                                               __func__,
>>> +                                               sizeof(struct 
>>> interrupt_server_node) + len,
>>> +                                               __alignof__(u32));
>>> +                               bt_node->len = len;
>>> +                               memcpy(bt_node->intserv, intserv, len);
>>> +                               bt_node->avail = avail;
>>> +                               skip = true;
>>> +                               break;
>>> +                       }
>>> +               }
>>>
>>> +               /*
>>> +                * Boot CPU not on core0.
>>> +                * Hold off adding core0 until boot core is found as 
>>> core0
>>> +                * may have to be replaced with boot core if boot 
>>> core can
>>> +                * not be accommodated within nr_cpu_ids with its 
>>> original
>>> +                * logical CPU numbering.
>>> +                */
>>> +               if (cpu == 0 && !found_boot_cpu) {
>>> +                       core0_node = memblock_alloc(sizeof(struct 
>>> interrupt_server_node) + len,
>>> +                                       __alignof__(u32));
>>> +                       if (!core0_node)
>>> +                               panic("%s: Failed to allocate %zu 
>>> bytes align=0x%zx\n",
>>> +                                       __func__,
>>> +                                       sizeof(struct 
>>> interrupt_server_node) + len,
>>> +                                       __alignof__(u32));
>>> +                       core0_node->len = len;
>>> +                       memcpy(core0_node->intserv, intserv, len);
>>> +                       core0_node->avail = avail;
>>> +                       skip = true;
>>> +               }
>>> +
>>> +               if (skip) {
>>> +                       /* Assumes same number of threads for all 
>>> cores */
>>> +                       cpu += nthreads;
>>> +                       continue;
>>> +               }
>>> +
>>> +               for (j = 0; j < nthreads && cpu < nr_cpu_ids; j++) {
>>>                          DBG("    thread %d -> cpu %d (hard id %d)\n",
>>>                              j, cpu, be32_to_cpu(intserv[j]));
>>>
>>> -                       avail = of_device_is_available(dn);
>>> -                       if (!avail)
>>> -                               avail = !of_property_match_string(dn,
>>> -                                               "enable-method", 
>>> "spin-table");
>>> -
>>>                          set_cpu_present(cpu, avail);
>>>                          set_cpu_possible(cpu, true);
>>>                          cpu_to_phys_id[cpu] = be32_to_cpu(intserv[j]);
>>>                          cpu++;
>>>                  }
>>>
>>> -               if (cpu >= nr_cpu_ids) {
>>> +               if (found_boot_cpu && cpu >= nr_cpu_ids) {
>>>                          of_node_put(dn);
>>>                          break;
>>>                  }
>>>          }
>>>
>>> +       /*
>>> +        * Boot CPU not on core0.
>>> +        *
>>> +        * If nr_cpu_ids does not accommodate the original logical CPU
>>> numbering for
>>> +        * boot CPU core, use logical CPU numbers 0 to nthreads for 
>>> boot CPU core.
>>> +        * Note that boot cpu is already assigned with logical CPU 
>>> number
>>> somewhere
>>> +        * between 0 to nthreads (depending on the boot thread within 
>>> the core) in
>>> +        * early_init_dt_scan_cpus() for this case.
>>> +        *
>>> +        * Otherwise, stick with the original logical CPU numbering.
>>> +        */
>>> +       if (bt_node) {
>>> +               int core0_cpu;
>>> +
>>> +               if (orig_boot_cpu + orig_boot_thread >= nr_cpu_ids) {
>>> +                       cpu = 0;
>>> +                       core0_cpu = orig_boot_cpu;
>>> +               } else {
>>> +                       cpu = orig_boot_cpu;
>>> +                       core0_cpu = 0;
>>> +               }
>>> +
>>> +               for (j = 0; j < nthreads && core0_cpu < nr_cpu_ids; 
>>> j++) {
>>> +                       DBG("    thread %d -> cpu %d (hard id %d)\n",
>>> +                           j, core0_cpu, 
>>> be32_to_cpu(core0_node->intserv[j]));
>>> +
>>> +                       set_cpu_present(core0_cpu, core0_node->avail);
>>> +                       set_cpu_possible(core0_cpu, true);
>>> +                       cpu_to_phys_id[core0_cpu] = 
>>> be32_to_cpu(core0_node->intserv[j]);
>>> +                       core0_cpu++;
>>> +               }
>>> +
>>> +               for (j = 0; j < nthreads && cpu < nr_cpu_ids; j++) {
>>> +                       DBG("    thread %d -> cpu %d (hard id %d)\n",
>>> +                           j, cpu, be32_to_cpu(bt_node->intserv[j]));
>>> +
>>> +                       set_cpu_present(cpu, bt_node->avail);
>>> +                       set_cpu_possible(cpu, true);
>>> +                       cpu_to_phys_id[cpu] = 
>>> be32_to_cpu(bt_node->intserv[j]);
>>> +                       cpu++;
>>> +               }
>>> +       }
>>> +
>>>          /* If no SMT supported, nthreads is forced to 1 */
>>>          if (!cpu_has_feature(CPU_FTR_SMT)) {
>>>                  DBG("  SMT disabled ! nthreads forced to 1\n");
>>>
>>
> 
> Thanks
> Hari



More information about the kexec mailing list