LinuxLists.cc - [PATCH v4 00/17] Convert cpu_up/down to device

2020-03-23 13:52:04

Subject: [PATCH v4 00/17] Convert cpu_up/down to device_online/offline

=============
Changes in v4
=============

* Split arm and arm64 patches so that the change to use reboot_cpu goes
into its own separate patch (Russell)
* Collected new Acked-by
* Rebased on top of v5.6-rc6
* Trimmed the CC list on the cover letter as lists were rejecting it

git clone git://linux-arm.org/linux-qy.git -b cpu-hp-cleanup-v4

Older post can be found here
----------------------------

https://lore.kernel.org/lkml/[email protected]/

=============
Test Coverage
=============

All tests ran with LOCKDEP enabled.

Platform: Juno-r2: arm64
------------------------

* Overnight rcutorture
* Overnight locktorture
* kexec -f Image --command="$(cat /proc/cmdline) reboot=s[0-5]"
* Hibernate to disk (using suspend option)
* Userspace hotplug via sysfs
* PSCI firemware checker

Notes:

* Couldn't convince Juno to hibernate using [reboot] or [shutdown]
options.

Platform: qemu (8 vCPUs) and VM (2 vCPUs): x86_64
-------------------------------------------------

* Overnight rcutorture
* Overnight locktorture
* Userspace hotplug via sysfs
* echo mmiotrace > /sys/kernel/debug/tracing/current_tracer &&
echo nop > /sys/kernel/debug/tracing/current_tracer
* Ran with CONFIG_DEBUG_HOTPLUG_CPU0 and CONFIG_BOOTPARAM_HOTPLUG_CPU0

Notes:

* qemu failed to bring cpu0 after offlining. Same behavior observed on
vanilla v5.6-rc6. Worked fine on the VM.

* mmiotrace successfully brought down all cpus when enabled,
then back online again when disabled. Including when cpu0 was
offline.

* My xen shenanigans are too 'humble' too create environment to test
the change in xen yet..

=====================
Original Cover Letter
=====================

Using cpu_up/down directly to bring cpus online/offline loses synchronization
with sysfs and could suffer from a race similar to what is described in
commit a6717c01ddc2 ("powerpc/rtas: use device model APIs and serialization
during LPM").

cpu_up/down seem to be more of a internal implementation detail for the cpu
subsystem to use to boot up cpus, perform suspend/resume and low level hotplug
operations. Users outside of the cpu subsystem would be better using the device
core API to bring a cpu online/offline which is the interface used to hotplug
memory and other system devices.

Several users have already migrated to use the device core API, this series
converts the remaining users and hides cpu_up/down from internal users at the
end.

I noticed this problem while working on a hack to disable offlining
a particular CPU but noticed that setting the offline_disabled attribute in the
device struct isn't enough because users can easily bypass the device core.
While my hack isn't a valid use case but it did highlight the inconsistency in
the way cpus are being onlined/offlined and this attempt hopefully improves on
this.

The first patch introduces new API to {add,remove}_cpu() using device_{online,
offline}() with correct locks held and export it.

The following 10 patches fix arch users.

The remaining 6 patches fix generic code users. Particularly creating a new
special exported API for the device core to use instead of cpu_up/down.

The last patch removes cpu_up/down from cpu.h and unexport the functions.

In some cases where the use of cpu_up/down seemed legitimate, I encapsulated
the logic in a higher level - special purposed function; and converted the code
to use that instead.

CC: Thomas Gleixner <[email protected]>
CC: Tony Luck <[email protected]>
CC: Fenghua Yu <[email protected]>
CC: Russell King <[email protected]>
CC: Catalin Marinas <[email protected]>
CC: Michael Ellerman <[email protected]>
CC: "David S. Miller" <[email protected]>
CC: Helge Deller <[email protected]>
CC: Juergen Gross <[email protected]>
CC: Mark Rutland <[email protected]>
CC: Lorenzo Pieralisi <[email protected]>
CC: "Paul E. McKenney" <[email protected]>
CC: Greg Kroah-Hartman <[email protected]>
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]

Qais Yousef (17):
cpu: Add new {add,remove}_cpu() functions
smp: Create a new function to shutdown nonboot cpus
ia64: Replace cpu_down with smp_shutdown_nonboot_cpus()
arm: Don't use disable_nonboot_cpus()
arm: Use reboot_cpu instead of hardcoding it to 0
arm64: Don't use disable_nonboot_cpus()
arm64: Use reboot_cpu instead of hardconding it to 0
arm64: hibernate.c: Create a new function to handle cpu_up(sleep_cpu)
x86: Replace cpu_up/down with add/remove_cpu
powerpc: Replace cpu_up/down with add/remove_cpu
sparc: Replace cpu_up/down with add/remove_cpu
parisc: Replace cpu_up/down with add/remove_cpu
driver: xen: Replace cpu_up/down with device_online/offline
firmware: psci: Replace cpu_up/down with add/remove_cpu
torture: Replace cpu_up/down with add/remove_cpu
smp: Create a new function to bringup nonboot cpus online
cpu: Hide cpu_up/down

arch/arm/kernel/reboot.c | 4 +-
arch/arm64/kernel/hibernate.c | 13 +--
arch/arm64/kernel/process.c | 4 +-
arch/ia64/kernel/process.c | 8 +-
arch/parisc/kernel/processor.c | 2 +-
arch/powerpc/kexec/core_64.c | 2 +-
arch/sparc/kernel/ds.c | 4 +-
arch/x86/kernel/topology.c | 22 ++---
arch/x86/mm/mmio-mod.c | 4 +-
arch/x86/xen/smp.c | 2 +-
drivers/base/cpu.c | 4 +-
drivers/firmware/psci/psci_checker.c | 4 +-
drivers/xen/cpu_hotplug.c | 2 +-
include/linux/cpu.h | 10 +-
kernel/cpu.c | 134 ++++++++++++++++++++++++++-
kernel/smp.c | 9 +-
kernel/torture.c | 9 +-
17 files changed, 172 insertions(+), 65 deletions(-)

--
2.17.1

2020-03-23 13:52:07

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 02/17] smp: Create a new function to shutdown nonboot cpus

This function will be used later in machine_shutdown() for some archs.

disable_nonboot_cpus() is not safe to use when doing machine_down(),
because it relies on freeze_secondary_cpus() which in turn is
a suspend/resume related freeze and could abort if the logic detects any
pending activities that can prevent finishing the offlining process.

Signed-off-by: Qais Yousef <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Josh Poimboeuf <[email protected]>
CC: "Peter Zijlstra (Intel)" <[email protected]>
CC: Jiri Kosina <[email protected]>
CC: Nicholas Piggin <[email protected]>
CC: Daniel Lezcano <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: Eiichi Tsukata <[email protected]>
CC: Zhenzhong Duan <[email protected]>
CC: Nadav Amit <[email protected]>
CC: Greg Kroah-Hartman <[email protected]>
CC: "Rafael J. Wysocki" <[email protected]>
CC: Tony Luck <[email protected]>
CC: Fenghua Yu <[email protected]>
CC: Russell King <[email protected]>
CC: Catalin Marinas <[email protected]>
CC: Will Deacon <[email protected]>
CC: [email protected]
CC: [email protected]
CC: [email protected]
---
include/linux/cpu.h | 2 ++
kernel/cpu.c | 42 ++++++++++++++++++++++++++++++++++++++++++
2 files changed, 44 insertions(+)

diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index cf8cf38dca43..64a246e9c8db 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -120,6 +120,7 @@ extern void cpu_hotplug_enable(void);
void clear_tasks_mm_cpumask(int cpu);
int cpu_down(unsigned int cpu);
int remove_cpu(unsigned int cpu);
+extern void smp_shutdown_nonboot_cpus(unsigned int primary_cpu);

#else /* CONFIG_HOTPLUG_CPU */

@@ -131,6 +132,7 @@ static inline int cpus_read_trylock(void) { return true; }
static inline void lockdep_assert_cpus_held(void) { }
static inline void cpu_hotplug_disable(void) { }
static inline void cpu_hotplug_enable(void) { }
+static inline void smp_shutdown_nonboot_cpus(unsigned int primary_cpu) { }
#endif /* !CONFIG_HOTPLUG_CPU */

/* Wrappers which go away once all code is converted */
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 069802f7010f..03c727195b65 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1069,6 +1069,48 @@ int remove_cpu(unsigned int cpu)
}
EXPORT_SYMBOL_GPL(remove_cpu);

+void smp_shutdown_nonboot_cpus(unsigned int primary_cpu)
+{
+ unsigned int cpu;
+ int error;
+
+ cpu_maps_update_begin();
+
+ /*
+ * Make certain the cpu I'm about to reboot on is online.
+ *
+ * This is inline to what migrate_to_reboot_cpu() already do.
+ */
+ if (!cpu_online(primary_cpu))
+ primary_cpu = cpumask_first(cpu_online_mask);
+
+ for_each_online_cpu(cpu) {
+ if (cpu == primary_cpu)
+ continue;
+
+ error = cpu_down_maps_locked(cpu, CPUHP_OFFLINE);
+ if (error) {
+ pr_err("Failed to offline CPU%d - error=%d",
+ cpu, error);
+ break;
+ }
+ }
+
+ /*
+ * Ensure all but the reboot CPU are offline.
+ */
+ BUG_ON(num_online_cpus() > 1);
+
+ /*
+ * Make sure the CPUs won't be enabled by someone else after this
+ * point. Kexec will reboot to a new kernel shortly resetting
+ * everything along the way.
+ */
+ cpu_hotplug_disabled++;
+
+ cpu_maps_update_done();
+}
+
#else
#define takedown_cpu NULL
#endif /*CONFIG_HOTPLUG_CPU*/
--
2.17.1

2020-03-23 13:52:18

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 05/17] arm: Use reboot_cpu instead of hardcoding it to 0

Use `reboot_cpu` variable instead of hardcoding 0 as the reboot cpu in
machine_shutdown().

Signed-off-by: Qais Yousef <[email protected]>
CC: Russell King <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/arm/kernel/reboot.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm/kernel/reboot.c b/arch/arm/kernel/reboot.c
index 58ad1a70f770..0ce388f15422 100644
--- a/arch/arm/kernel/reboot.c
+++ b/arch/arm/kernel/reboot.c
@@ -92,7 +92,7 @@ void soft_restart(unsigned long addr)
*/
void machine_shutdown(void)
{
- smp_shutdown_nonboot_cpus(0);
+ smp_shutdown_nonboot_cpus(reboot_cpu);
}

/*
--
2.17.1

2020-03-23 13:52:25

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 08/17] arm64: hibernate.c: Create a new function to handle cpu_up(sleep_cpu)

In preparation to make cpu_up/down private - move the user in arm64
hibernate.c to use a new generic function that provides what arm64
needs.

Acked-by: Catalin Marinas <[email protected]>
Signed-off-by: Qais Yousef <[email protected]>
CC: Catalin Marinas <[email protected]>
CC: Will Deacon <[email protected]>
CC: Steve Capper <[email protected]>
CC: Richard Fontana <[email protected]>
CC: James Morse <[email protected]>
CC: Mark Rutland <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Josh Poimboeuf <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: "Peter Zijlstra (Intel)" <[email protected]>
CC: Nicholas Piggin <[email protected]>
CC: Daniel Lezcano <[email protected]>
CC: Jiri Kosina <[email protected]>
CC: Pavankumar Kondeti <[email protected]>
CC: Zhenzhong Duan <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/arm64/kernel/hibernate.c | 13 +++++--------
include/linux/cpu.h | 1 +
kernel/cpu.c | 24 ++++++++++++++++++++++++
3 files changed, 30 insertions(+), 8 deletions(-)

diff --git a/arch/arm64/kernel/hibernate.c b/arch/arm64/kernel/hibernate.c
index 590963c9c609..5b73e92c99e3 100644
--- a/arch/arm64/kernel/hibernate.c
+++ b/arch/arm64/kernel/hibernate.c
@@ -166,14 +166,11 @@ int arch_hibernation_header_restore(void *addr)
sleep_cpu = -EINVAL;
return -EINVAL;
}
- if (!cpu_online(sleep_cpu)) {
- pr_info("Hibernated on a CPU that is offline! Bringing CPU up.\n");
- ret = cpu_up(sleep_cpu);
- if (ret) {
- pr_err("Failed to bring hibernate-CPU up!\n");
- sleep_cpu = -EINVAL;
- return ret;
- }
+
+ ret = bringup_hibernate_cpu(sleep_cpu);
+ if (ret) {
+ sleep_cpu = -EINVAL;
+ return ret;
}

resume_hdr = *hdr;
diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 64a246e9c8db..aebe8186cb07 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -93,6 +93,7 @@ int add_cpu(unsigned int cpu);
void notify_cpu_starting(unsigned int cpu);
extern void cpu_maps_update_begin(void);
extern void cpu_maps_update_done(void);
+extern int bringup_hibernate_cpu(unsigned int sleep_cpu);

#else /* CONFIG_SMP */
#define cpuhp_tasks_frozen 0
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 03c727195b65..bf39c5bfb6d9 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1275,6 +1275,30 @@ int add_cpu(unsigned int cpu)
}
EXPORT_SYMBOL_GPL(add_cpu);

+/**
+ * bringup_hibernate_cpu - Bring up the CPU that we hibernated on
+ * @sleep_cpu: The cpu we hibernated on and should be brought up.
+ *
+ * On some archs like arm64, we can hibernate on any CPU, but on wake up the
+ * CPU we hibernated on might be offline as a side effect of using maxcpus= for
+ * example.
+ */
+int bringup_hibernate_cpu(unsigned int sleep_cpu)
+{
+ int ret;
+
+ if (!cpu_online(sleep_cpu)) {
+ pr_info("Hibernated on a CPU that is offline! Bringing CPU up.\n");
+ ret = cpu_up(sleep_cpu);
+ if (ret) {
+ pr_err("Failed to bring hibernate-CPU up!\n");
+ return ret;
+ }
+ }
+
+ return 0;
+}
+
#ifdef CONFIG_PM_SLEEP_SMP
static cpumask_var_t frozen_cpus;

--
2.17.1

2020-03-23 13:52:32

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 11/17] sparc: Replace cpu_up/down with add/remove_cpu

2020-03-23 13:52:37

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 13/17] driver: xen: Replace cpu_up/down with device_online/offline

2020-03-23 13:52:38

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 14/17] firmware: psci: Replace cpu_up/down with add/remove_cpu

2020-03-23 13:52:49

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 17/17] cpu: Hide cpu_up/down

Provide a special exported function for the device core to bring a cpu
up/down and hide the real cpu_up/down as they are treated as private
functions. cpu_up/down are lower level API and users outside the cpu
subsystem should use add/remove_cpu() which will take care of extra
housekeeping work like keeping sysfs in sync.

Signed-off-by: Qais Yousef <[email protected]>
CC: Greg Kroah-Hartman <[email protected]>
CC: "Rafael J. Wysocki" <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Josh Poimboeuf <[email protected]>
CC: Nicholas Piggin <[email protected]>
CC: "Peter Zijlstra (Intel)" <[email protected]>
CC: Jiri Kosina <[email protected]>
CC: Daniel Lezcano <[email protected]>
CC: Eiichi Tsukata <[email protected]>
CC: Zhenzhong Duan <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: Pavankumar Kondeti <[email protected]>
CC: [email protected]
---
drivers/base/cpu.c | 4 ++--
include/linux/cpu.h | 4 ++--
kernel/cpu.c | 32 ++++++++++++++++++++++++++++----
3 files changed, 32 insertions(+), 8 deletions(-)

diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c
index 6265871a4af2..9a134cd362ee 100644
--- a/drivers/base/cpu.c
+++ b/drivers/base/cpu.c
@@ -55,7 +55,7 @@ static int cpu_subsys_online(struct device *dev)
if (from_nid == NUMA_NO_NODE)
return -ENODEV;

- ret = cpu_up(cpuid);
+ ret = cpu_subsys_up(dev);
/*
* When hot adding memory to memoryless node and enabling a cpu
* on the node, node number of the cpu may internally change.
@@ -69,7 +69,7 @@ static int cpu_subsys_online(struct device *dev)

static int cpu_subsys_offline(struct device *dev)
{
- return cpu_down(dev->id);
+ return cpu_subsys_down(dev);
}

void unregister_cpu(struct cpu *cpu)
diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 3d3caecef395..e1bbd8d6ecec 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -88,8 +88,8 @@ extern ssize_t arch_cpu_release(const char *, size_t);

#ifdef CONFIG_SMP
extern bool cpuhp_tasks_frozen;
-int cpu_up(unsigned int cpu);
int add_cpu(unsigned int cpu);
+int cpu_subsys_up(struct device *dev);
void notify_cpu_starting(unsigned int cpu);
extern void cpu_maps_update_begin(void);
extern void cpu_maps_update_done(void);
@@ -120,8 +120,8 @@ extern void lockdep_assert_cpus_held(void);
extern void cpu_hotplug_disable(void);
extern void cpu_hotplug_enable(void);
void clear_tasks_mm_cpumask(int cpu);
-int cpu_down(unsigned int cpu);
int remove_cpu(unsigned int cpu);
+int cpu_subsys_down(struct device *dev);
extern void smp_shutdown_nonboot_cpus(unsigned int primary_cpu);

#else /* CONFIG_HOTPLUG_CPU */
diff --git a/kernel/cpu.c b/kernel/cpu.c
index 7c6aa427a3a3..017e16dfcccd 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1051,11 +1051,23 @@ static int do_cpu_down(unsigned int cpu, enum cpuhp_state target)
return err;
}

-int cpu_down(unsigned int cpu)
+static int cpu_down(unsigned int cpu)
{
return do_cpu_down(cpu, CPUHP_OFFLINE);
}
-EXPORT_SYMBOL(cpu_down);
+
+/**
+ * cpu_subsys_down - Bring down a cpu device
+ * @dev: Pointer to the cpu device to offline
+ *
+ * This function is meant to be used by device core cpu subsystem only.
+ *
+ * Other subsystems should use remove_cpu() instead.
+ */
+int cpu_subsys_down(struct device *dev)
+{
+ return cpu_down(dev->id);
+}

int remove_cpu(unsigned int cpu)
{
@@ -1257,11 +1269,23 @@ static int do_cpu_up(unsigned int cpu, enum cpuhp_state target)
return err;
}

-int cpu_up(unsigned int cpu)
+static int cpu_up(unsigned int cpu)
{
return do_cpu_up(cpu, CPUHP_ONLINE);
}
-EXPORT_SYMBOL_GPL(cpu_up);
+
+/**
+ * cpu_subsys_up - Bring up a cpu device
+ * @dev: Pointer to the cpu device to online
+ *
+ * This function is meant to be used by device core cpu subsystem only.
+ *
+ * Other subsystems should use add_cpu() instead.
+ */
+int cpu_subsys_up(struct device *dev)
+{
+ return cpu_up(dev->id);
+}

int add_cpu(unsigned int cpu)
{
--
2.17.1

2020-03-23 13:53:00

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 01/17] cpu: Add new {add,remove}_cpu() functions

The new functions use device_{online,offline}() which are userspace
safe.

This is in preparation to move cpu_{up, down} kernel users to use
a safer interface that is not racy with userspace.

Suggested-by: "Paul E. McKenney" <[email protected]>
Signed-off-by: Qais Yousef <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: "Paul E. McKenney" <[email protected]>
CC: Helge Deller <[email protected]>
CC: Michael Ellerman <[email protected]>
CC: "David S. Miller" <[email protected]>
CC: Juergen Gross <[email protected]>
CC: Mark Rutland <[email protected]>
CC: Lorenzo Pieralisi <[email protected]>
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]
---
include/linux/cpu.h | 2 ++
kernel/cpu.c | 24 ++++++++++++++++++++++++
2 files changed, 26 insertions(+)

diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index 1ca2baf817ed..cf8cf38dca43 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -89,6 +89,7 @@ extern ssize_t arch_cpu_release(const char *, size_t);
#ifdef CONFIG_SMP
extern bool cpuhp_tasks_frozen;
int cpu_up(unsigned int cpu);
+int add_cpu(unsigned int cpu);
void notify_cpu_starting(unsigned int cpu);
extern void cpu_maps_update_begin(void);
extern void cpu_maps_update_done(void);
@@ -118,6 +119,7 @@ extern void cpu_hotplug_disable(void);
extern void cpu_hotplug_enable(void);
void clear_tasks_mm_cpumask(int cpu);
int cpu_down(unsigned int cpu);
+int remove_cpu(unsigned int cpu);

#else /* CONFIG_HOTPLUG_CPU */

diff --git a/kernel/cpu.c b/kernel/cpu.c
index 9c706af713fb..069802f7010f 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1057,6 +1057,18 @@ int cpu_down(unsigned int cpu)
}
EXPORT_SYMBOL(cpu_down);

+int remove_cpu(unsigned int cpu)
+{
+ int ret;
+
+ lock_device_hotplug();
+ ret = device_offline(get_cpu_device(cpu));
+ unlock_device_hotplug();
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(remove_cpu);
+
#else
#define takedown_cpu NULL
#endif /*CONFIG_HOTPLUG_CPU*/
@@ -1209,6 +1221,18 @@ int cpu_up(unsigned int cpu)
}
EXPORT_SYMBOL_GPL(cpu_up);

+int add_cpu(unsigned int cpu)
+{
+ int ret;
+
+ lock_device_hotplug();
+ ret = device_online(get_cpu_device(cpu));
+ unlock_device_hotplug();
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(add_cpu);
+
#ifdef CONFIG_PM_SLEEP_SMP
static cpumask_var_t frozen_cpus;

--
2.17.1

2020-03-23 13:53:04

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 12/17] parisc: Replace cpu_up/down with add/remove_cpu

2020-03-23 13:53:07

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 07/17] arm64: Use reboot_cpu instead of hardconding it to 0

Use `reboot_cpu` variable instead of hardcoding 0 as the reboot cpu in
machine_shutdown().

Acked-by: Catalin Marinas <[email protected]>
Signed-off-by: Qais Yousef <[email protected]>
CC: Catalin Marinas <[email protected]>
CC: Will Deacon <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/arm64/kernel/process.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
index 1b9f7b749d75..3e5a6ad66cbe 100644
--- a/arch/arm64/kernel/process.c
+++ b/arch/arm64/kernel/process.c
@@ -145,7 +145,7 @@ void arch_cpu_idle_dead(void)
*/
void machine_shutdown(void)
{
- smp_shutdown_nonboot_cpus(0);
+ smp_shutdown_nonboot_cpus(reboot_cpu);
}

/*
--
2.17.1

2020-03-23 13:53:09

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 04/17] arm: Don't use disable_nonboot_cpus()

disable_nonboot_cpus() is not safe to use when doing machine_down(),
because it relies on freeze_secondary_cpus() which in turn is
a suspend/resume related freeze and could abort if the logic detects any
pending activities that can prevent finishing the offlining process.

Beside disable_nonboot_cpus() is dependent on CONFIG_PM_SLEEP_SMP which
is an othogonal config to rely on to ensure this function works
correctly.

Signed-off-by: Qais Yousef <[email protected]>
CC: Russell King <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/arm/kernel/reboot.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm/kernel/reboot.c b/arch/arm/kernel/reboot.c
index bb18ed0539f4..58ad1a70f770 100644
--- a/arch/arm/kernel/reboot.c
+++ b/arch/arm/kernel/reboot.c
@@ -88,11 +88,11 @@ void soft_restart(unsigned long addr)
* to execute e.g. a RAM-based pin loop is not sufficient. This allows the
* kexec'd kernel to use any and all RAM as it sees fit, without having to
* avoid any code or data used by any SW CPU pin loop. The CPU hotplug
- * functionality embodied in disable_nonboot_cpus() to achieve this.
+ * functionality embodied in smp_shutdown_nonboot_cpus() to achieve this.
*/
void machine_shutdown(void)
{
- disable_nonboot_cpus();
+ smp_shutdown_nonboot_cpus(0);
}

/*
--
2.17.1

2020-03-23 13:53:10

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 09/17] x86: Replace cpu_up/down with add/remove_cpu

The core device API performs extra housekeeping bits that are missing
from directly calling cpu_up/down.

See commit a6717c01ddc2 ("powerpc/rtas: use device model APIs and
serialization during LPM") for an example description of what might go
wrong.

This also prepares to make cpu_up/down a private interface for anything
but the cpu subsystem.

Signed-off-by: Qais Yousef <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: Borislav Petkov <[email protected]>
CC: "H. Peter Anvin" <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/x86/kernel/topology.c | 22 ++++++----------------
arch/x86/mm/mmio-mod.c | 4 ++--
arch/x86/xen/smp.c | 2 +-
3 files changed, 9 insertions(+), 19 deletions(-)

diff --git a/arch/x86/kernel/topology.c b/arch/x86/kernel/topology.c
index be5bc2e47c71..b8810ebbc8ae 100644
--- a/arch/x86/kernel/topology.c
+++ b/arch/x86/kernel/topology.c
@@ -59,39 +59,29 @@ __setup("cpu0_hotplug", enable_cpu0_hotplug);
*/
int _debug_hotplug_cpu(int cpu, int action)
{
- struct device *dev = get_cpu_device(cpu);
int ret;

if (!cpu_is_hotpluggable(cpu))
return -EINVAL;

- lock_device_hotplug();
-
switch (action) {
case 0:
- ret = cpu_down(cpu);
- if (!ret) {
+ ret = remove_cpu(cpu);
+ if (!ret)
pr_info("DEBUG_HOTPLUG_CPU0: CPU %u is now offline\n", cpu);
- dev->offline = true;
- kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
- } else
+ else
pr_debug("Can't offline CPU%d.\n", cpu);
break;
case 1:
- ret = cpu_up(cpu);
- if (!ret) {
- dev->offline = false;
- kobject_uevent(&dev->kobj, KOBJ_ONLINE);
- } else {
+ ret = add_cpu(cpu);
+ if (ret)
pr_debug("Can't online CPU%d.\n", cpu);
- }
+
break;
default:
ret = -EINVAL;
}

- unlock_device_hotplug();
-
return ret;
}

diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c
index 673de6063345..109325d77b3e 100644
--- a/arch/x86/mm/mmio-mod.c
+++ b/arch/x86/mm/mmio-mod.c
@@ -386,7 +386,7 @@ static void enter_uniprocessor(void)
put_online_cpus();

for_each_cpu(cpu, downed_cpus) {
- err = cpu_down(cpu);
+ err = remove_cpu(cpu);
if (!err)
pr_info("CPU%d is down.\n", cpu);
else
@@ -406,7 +406,7 @@ static void leave_uniprocessor(void)
return;
pr_notice("Re-enabling CPUs...\n");
for_each_cpu(cpu, downed_cpus) {
- err = cpu_up(cpu);
+ err = add_cpu(cpu);
if (!err)
pr_info("enabled CPU%d.\n", cpu);
else
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 7a43b2ae19f1..2097fa0ebdb5 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -132,7 +132,7 @@ void __init xen_smp_cpus_done(unsigned int max_cpus)
if (xen_vcpu_nr(cpu) < MAX_VIRT_CPUS)
continue;

- rc = cpu_down(cpu);
+ rc = remove_cpu(cpu);

if (rc == 0) {
/*
--
2.17.1

2020-03-23 13:53:12

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 06/17] arm64: Don't use disable_nonboot_cpus()

disable_nonboot_cpus() is not safe to use when doing machine_down(),
because it relies on freeze_secondary_cpus() which in turn is
a suspend/resume related freeze and could abort if the logic detects any
pending activities that can prevent finishing the offlining process.

Beside disable_nonboot_cpus() is dependent on CONFIG_PM_SLEEP_SMP which
is an othogonal config to rely on to ensure this function works
correctly.

Acked-by: Catalin Marinas <[email protected]>
Signed-off-by: Qais Yousef <[email protected]>
CC: Catalin Marinas <[email protected]>
CC: Will Deacon <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/arm64/kernel/process.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/process.c b/arch/arm64/kernel/process.c
index 00626057a384..1b9f7b749d75 100644
--- a/arch/arm64/kernel/process.c
+++ b/arch/arm64/kernel/process.c
@@ -141,11 +141,11 @@ void arch_cpu_idle_dead(void)
* to execute e.g. a RAM-based pin loop is not sufficient. This allows the
* kexec'd kernel to use any and all RAM as it sees fit, without having to
* avoid any code or data used by any SW CPU pin loop. The CPU hotplug
- * functionality embodied in disable_nonboot_cpus() to achieve this.
+ * functionality embodied in smpt_shutdown_nonboot_cpus() to achieve this.
*/
void machine_shutdown(void)
{
- disable_nonboot_cpus();
+ smp_shutdown_nonboot_cpus(0);
}

/*
--
2.17.1

2020-03-23 13:53:38

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 15/17] torture: Replace cpu_up/down with add/remove_cpu

2020-03-23 13:53:55

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 16/17] smp: Create a new function to bringup nonboot cpus online

This is the last direct user of cpu_up() before we can hide it now as
internal implementation detail of the cpu subsystem.

Signed-off-by: Qais Yousef <[email protected]>
CC: Thomas Gleixner <[email protected]>
CC: Josh Poimboeuf <[email protected]>
CC: "Peter Zijlstra (Intel)" <[email protected]>
CC: Jiri Kosina <[email protected]>
CC: Nicholas Piggin <[email protected]>
CC: Daniel Lezcano <[email protected]>
CC: Ingo Molnar <[email protected]>
CC: Eiichi Tsukata <[email protected]>
CC: Zhenzhong Duan <[email protected]>
CC: Nadav Amit <[email protected]>
CC: Greg Kroah-Hartman <[email protected]>
CC: "Rafael J. Wysocki" <[email protected]>
CC: [email protected]
---
include/linux/cpu.h | 1 +
kernel/cpu.c | 12 ++++++++++++
kernel/smp.c | 9 +--------
3 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/include/linux/cpu.h b/include/linux/cpu.h
index aebe8186cb07..3d3caecef395 100644
--- a/include/linux/cpu.h
+++ b/include/linux/cpu.h
@@ -94,6 +94,7 @@ void notify_cpu_starting(unsigned int cpu);
extern void cpu_maps_update_begin(void);
extern void cpu_maps_update_done(void);
extern int bringup_hibernate_cpu(unsigned int sleep_cpu);
+extern void smp_bringup_nonboot_cpus(unsigned int setup_max_cpus);

#else /* CONFIG_SMP */
#define cpuhp_tasks_frozen 0
diff --git a/kernel/cpu.c b/kernel/cpu.c
index bf39c5bfb6d9..7c6aa427a3a3 100644
--- a/kernel/cpu.c
+++ b/kernel/cpu.c
@@ -1299,6 +1299,18 @@ int bringup_hibernate_cpu(unsigned int sleep_cpu)
return 0;
}

+void smp_bringup_nonboot_cpus(unsigned int setup_max_cpus)
+{
+ unsigned int cpu;
+
+ for_each_present_cpu(cpu) {
+ if (num_online_cpus() >= setup_max_cpus)
+ break;
+ if (!cpu_online(cpu))
+ cpu_up(cpu);
+ }
+}
+
#ifdef CONFIG_PM_SLEEP_SMP
static cpumask_var_t frozen_cpus;

diff --git a/kernel/smp.c b/kernel/smp.c
index d0ada39eb4d4..9e793cfd6128 100644
--- a/kernel/smp.c
+++ b/kernel/smp.c
@@ -589,20 +589,13 @@ void __init setup_nr_cpu_ids(void)
void __init smp_init(void)
{
int num_nodes, num_cpus;
- unsigned int cpu;

idle_threads_init();
cpuhp_threads_init();

pr_info("Bringing up secondary CPUs ...\n");

- /* FIXME: This should be done in userspace --RR */
- for_each_present_cpu(cpu) {
- if (num_online_cpus() >= setup_max_cpus)
- break;
- if (!cpu_online(cpu))
- cpu_up(cpu);
- }
+ smp_bringup_nonboot_cpus(setup_max_cpus);

num_nodes = num_online_nodes();
num_cpus = num_online_cpus();
--
2.17.1

2020-03-23 13:54:04

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 10/17] powerpc: Replace cpu_up/down with add/remove_cpu

2020-03-23 13:54:06

by Qais Yousef

[permalink] [raw]

Subject: [PATCH v4 03/17] ia64: Replace cpu_down with smp_shutdown_nonboot_cpus()

Use the new smp_shutdown_nonboot_cpus() instead of open coding using
cpu_down() directly.

Use reboot_cpu instead of hardcoding the boot CPU to 0.

This also prepares to make cpu_up/down a private interface for anything
but the cpu subsystem.

Signed-off-by: Qais Yousef <[email protected]>
CC: Tony Luck <[email protected]>
CC: Fenghua Yu <[email protected]>
CC: [email protected]
CC: [email protected]
---
arch/ia64/kernel/process.c | 8 +-------
1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
index 968b5f33e725..bf4c0cdb7a25 100644
--- a/arch/ia64/kernel/process.c
+++ b/arch/ia64/kernel/process.c
@@ -646,14 +646,8 @@ cpu_halt (void)

void machine_shutdown(void)
{
-#ifdef CONFIG_HOTPLUG_CPU
- int cpu;
+ smp_shutdown_nonboot_cpus(reboot_cpu);

- for_each_online_cpu(cpu) {
- if (cpu != smp_processor_id())
- cpu_down(cpu);
- }
-#endif
#ifdef CONFIG_KEXEC
kexec_disable_iosapic();
#endif
--
2.17.1

2020-03-23 15:01:42

by Paul E. McKenney

[permalink] [raw]

Subject: Re: [PATCH v4 01/17] cpu: Add new {add,remove}_cpu() functions

On Mon, Mar 23, 2020 at 01:50:54PM +0000, Qais Yousef wrote:
> The new functions use device_{online,offline}() which are userspace
> safe.
>
> This is in preparation to move cpu_{up, down} kernel users to use
> a safer interface that is not racy with userspace.
>
> Suggested-by: "Paul E. McKenney" <[email protected]>
> Signed-off-by: Qais Yousef <[email protected]>
> CC: Thomas Gleixner <[email protected]>
> CC: "Paul E. McKenney" <[email protected]>

Reviewed-by: Paul E. McKenney <[email protected]>

> CC: Helge Deller <[email protected]>
> CC: Michael Ellerman <[email protected]>
> CC: "David S. Miller" <[email protected]>
> CC: Juergen Gross <[email protected]>
> CC: Mark Rutland <[email protected]>
> CC: Lorenzo Pieralisi <[email protected]>
> CC: [email protected]
> CC: [email protected]
> CC: [email protected]
> CC: [email protected]
> CC: [email protected]
> CC: [email protected]
> CC: [email protected]
> ---
> include/linux/cpu.h | 2 ++
> kernel/cpu.c | 24 ++++++++++++++++++++++++
> 2 files changed, 26 insertions(+)
>
> diff --git a/include/linux/cpu.h b/include/linux/cpu.h
> index 1ca2baf817ed..cf8cf38dca43 100644
> --- a/include/linux/cpu.h
> +++ b/include/linux/cpu.h
> @@ -89,6 +89,7 @@ extern ssize_t arch_cpu_release(const char *, size_t);
> #ifdef CONFIG_SMP
> extern bool cpuhp_tasks_frozen;
> int cpu_up(unsigned int cpu);
> +int add_cpu(unsigned int cpu);
> void notify_cpu_starting(unsigned int cpu);
> extern void cpu_maps_update_begin(void);
> extern void cpu_maps_update_done(void);
> @@ -118,6 +119,7 @@ extern void cpu_hotplug_disable(void);
> extern void cpu_hotplug_enable(void);
> void clear_tasks_mm_cpumask(int cpu);
> int cpu_down(unsigned int cpu);
> +int remove_cpu(unsigned int cpu);
>
> #else /* CONFIG_HOTPLUG_CPU */
>
> diff --git a/kernel/cpu.c b/kernel/cpu.c
> index 9c706af713fb..069802f7010f 100644
> --- a/kernel/cpu.c
> +++ b/kernel/cpu.c
> @@ -1057,6 +1057,18 @@ int cpu_down(unsigned int cpu)
> }
> EXPORT_SYMBOL(cpu_down);
>
> +int remove_cpu(unsigned int cpu)
> +{
> + int ret;
> +
> + lock_device_hotplug();
> + ret = device_offline(get_cpu_device(cpu));
> + unlock_device_hotplug();
> +
> + return ret;
> +}
> +EXPORT_SYMBOL_GPL(remove_cpu);
> +
> #else
> #define takedown_cpu NULL
> #endif /*CONFIG_HOTPLUG_CPU*/
> @@ -1209,6 +1221,18 @@ int cpu_up(unsigned int cpu)
> }
> EXPORT_SYMBOL_GPL(cpu_up);
>
> +int add_cpu(unsigned int cpu)
> +{
> + int ret;
> +
> + lock_device_hotplug();
> + ret = device_online(get_cpu_device(cpu));
> + unlock_device_hotplug();
> +
> + return ret;
> +}
> +EXPORT_SYMBOL_GPL(add_cpu);
> +
> #ifdef CONFIG_PM_SLEEP_SMP
> static cpumask_var_t frozen_cpus;
>
> --
> 2.17.1
>

2020-03-25 12:06:59

Subject: [PATCH v4 00/17] Convert cpu_up/down to device_online/offline

Subject: [PATCH v4 02/17] smp: Create a new function to shutdown nonboot cpus

Subject: [PATCH v4 05/17] arm: Use reboot_cpu instead of hardcoding it to 0

Subject: [PATCH v4 08/17] arm64: hibernate.c: Create a new function to handle cpu_up(sleep_cpu)

Subject: [PATCH v4 11/17] sparc: Replace cpu_up/down with add/remove_cpu

Subject: [PATCH v4 13/17] driver: xen: Replace cpu_up/down with device_online/offline

Subject: [PATCH v4 14/17] firmware: psci: Replace cpu_up/down with add/remove_cpu

Subject: [PATCH v4 17/17] cpu: Hide cpu_up/down

Subject: [PATCH v4 01/17] cpu: Add new {add,remove}_cpu() functions

Subject: [PATCH v4 12/17] parisc: Replace cpu_up/down with add/remove_cpu

Subject: [PATCH v4 07/17] arm64: Use reboot_cpu instead of hardconding it to 0

Subject: [PATCH v4 04/17] arm: Don't use disable_nonboot_cpus()

Subject: [PATCH v4 09/17] x86: Replace cpu_up/down with add/remove_cpu

Subject: [PATCH v4 06/17] arm64: Don't use disable_nonboot_cpus()

Subject: [PATCH v4 15/17] torture: Replace cpu_up/down with add/remove_cpu

Subject: [PATCH v4 16/17] smp: Create a new function to bringup nonboot cpus online

Subject: [PATCH v4 10/17] powerpc: Replace cpu_up/down with add/remove_cpu

Subject: [PATCH v4 03/17] ia64: Replace cpu_down with smp_shutdown_nonboot_cpus()

Subject: Re: [PATCH v4 01/17] cpu: Add new {add,remove}_cpu() functions

Subject: [tip: smp/core] cpu/hotplug: Hide cpu_up/down()

Subject: [tip: smp/core] cpu/hotplug: Move bringup of secondary CPUs out of smp_init()

Subject: [tip: smp/core] cpu/hotplug: Create a new function to shutdown nonboot cpus

Subject: [tip: smp/core] parisc: Replace cpu_up/down() with add/remove_cpu()

Subject: [tip: smp/core] arm64: Don't use disable_nonboot_cpus()

Subject: [tip: smp/core] sparc: Replace cpu_up/down() with add/remove_cpu()

Subject: [tip: smp/core] ia64: Replace cpu_down() with smp_shutdown_nonboot_cpus()

Subject: [tip: smp/core] ARM: Use reboot_cpu instead of hardcoding it to 0

Subject: [tip: smp/core] cpu/hotplug: Add new {add,remove}_cpu() functions

Subject: [tip: smp/core] ARM: Don't use disable_nonboot_cpus()

Subject: [tip: smp/core] arm64: Use reboot_cpu instead of hardconding it to 0

Subject: [tip: smp/core] cpu/hotplug: Provide bringup_hibernate_cpu()

Subject: [tip: smp/core] powerpc: Replace cpu_up/down() with add/remove_cpu()

Subject: [tip: smp/core] xen/cpuhotplug: Replace cpu_up/down() with device_online/offline()

Subject: [tip: smp/core] x86/smp: Replace cpu_up/down() with add/remove_cpu()

Subject: [tip: smp/core] torture: Replace cpu_up/down() with add/remove_cpu()

Subject: [tip: smp/core] arm64: hibernate: Use bringup_hibernate_cpu()

Subject: [tip: smp/core] firmware: psci: Replace cpu_up/down() with add/remove_cpu()