Thread (11 messages) flat view 11 messages, 3 authors, 2016-08-09
STALE3684d

[PATCH] arm64: Support hard limit of cpu count by nr_cpus

From: Suzuki.Poulose@arm.com (Suzuki K Poulose)
Date: 2016-08-05 10:19:12
Subsystem: arm64 port (aarch64 architecture), the rest · Maintainers: Catalin Marinas, Will Deacon, Linus Torvalds

On 05/08/16 10:58, Kefeng Wang wrote:
quoted hunk ↗ jump to hunk

On 2016/8/5 16:32, Will Deacon wrote:
quoted
On Fri, Aug 05, 2016 at 02:03:55PM +0800, Kefeng Wang wrote:
quoted
Enable the hard limit of cpu count by nr_cpus on arm64.
The code is borrowed from MIPS.

Reported-by: Shiyuan Hu <redacted>
Signed-off-by: Kefeng Wang <redacted>
---
 arch/arm64/kernel/setup.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
diff --git a/arch/arm64/kernel/setup.c b/arch/arm64/kernel/setup.c
index 536dce2..597b777 100644
--- a/arch/arm64/kernel/setup.c
+++ b/arch/arm64/kernel/setup.c
@@ -224,6 +224,21 @@ static void __init request_standard_resources(void)

 u64 __cpu_logical_map[NR_CPUS] = { [0 ... NR_CPUS-1] = INVALID_HWID };

+static void __init prefill_possible_map(void)
+{
+	int i, possible = num_possible_cpus();
+
+	if (possible > nr_cpu_ids)
+		possible = nr_cpu_ids;
+
+	for (i = 0; i < possible; i++)
+		set_cpu_possible(i, true);
+	for (; i < NR_CPUS; i++)
+		set_cpu_possible(i, false);
+
+	nr_cpu_ids = possible;
+}
Shouldn't we just avoid marking those CPUs as possible in smp_cpu_setup,
rather than rewriting things later on?
How about this one,

From 0965eebecbc5aea0473df0cb45d32b2fccf72a9a Mon Sep 17 00:00:00 2001
From: Kefeng Wang <redacted>
Date: Fri, 5 Aug 2016 17:44:33 +0800
Subject: [PATCH v2] arm64: Support hard limit of cpu count by nr_cpus

Enable the hard limit of cpu count by set boot options nr_cpus=x
on arm64.

Signed-off-by: Kefeng Wang <redacted>
---
 arch/arm64/kernel/smp.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 76a6d92..0ac3ff5 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -496,7 +496,10 @@ static int __init smp_cpu_setup(int cpu)
 	if (cpu_ops[cpu]->cpu_init(cpu))
 		return -ENODEV;

-	set_cpu_possible(cpu, true);
+	if (cpu < nr_cpu_ids)
+		set_cpu_possible(cpu, true);
+	else
+		return -EINVAL;

 	return 0;
 }
Or is this even better ?
diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 490db85..dbd6a5e 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -659,9 +659,9 @@ void __init smp_init_cpus(void)
                 acpi_table_parse_madt(ACPI_MADT_TYPE_GENERIC_INTERRUPT,
                                       acpi_parse_gic_cpu_interface, 0);
  
-       if (cpu_count > NR_CPUS)
+       if (cpu_count > nr_cpu_ids)
                 pr_warn("no. of cores (%d) greater than configured maximum of %d - clipping\n",
-                       cpu_count, NR_CPUS);
+                       cpu_count, nr_cpu_ids);
  
         if (!bootcpu_valid) {
                 pr_err("missing boot CPU MPIDR, not enabling secondaries\n");
@@ -675,7 +675,7 @@ void __init smp_init_cpus(void)
          * with entries in cpu_logical_map while initializing the cpus.
          * If the cpu set-up fails, invalidate the cpu_logical_map entry.
          */
-       for (i = 1; i < NR_CPUS; i++) {
+       for (i = 1; i < nr_cpu_ids; i++) {
                 if (cpu_logical_map(i) != INVALID_HWID) {
                         if (smp_cpu_setup(i))
                                 cpu_logical_map(i) = INVALID_HWID;

Cheers
Suzuki
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help