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

Hari Bathini hbathini at linux.ibm.com
Sun Nov 26 20:30:13 PST 2023


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):

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");



More information about the kexec mailing list