blob: 66ef8f35613d2190b4e2f5e42084a4736ff3dc69 [file] [log] [blame]
Ashok Rajc8094062006-01-08 01:03:17 -08001 CPU hotplug Support in Linux(tm) Kernel
2
3 Maintainers:
4 CPU Hotplug Core:
Rusty Russelld2800802010-08-11 23:04:08 -06005 Rusty Russell <rusty@rustcorp.com.au>
Ashok Rajc8094062006-01-08 01:03:17 -08006 Srivatsa Vaddagiri <vatsa@in.ibm.com>
7 i386:
8 Zwane Mwaikambo <zwane@arm.linux.org.uk>
9 ppc64:
10 Nathan Lynch <nathanl@austin.ibm.com>
11 Joel Schopp <jschopp@austin.ibm.com>
12 ia64/x86_64:
13 Ashok Raj <ashok.raj@intel.com>
Heiko Carstens255acee2006-02-17 13:52:46 -080014 s390:
15 Heiko Carstens <heiko.carstens@de.ibm.com>
Ashok Rajc8094062006-01-08 01:03:17 -080016
17Authors: Ashok Raj <ashok.raj@intel.com>
18Lots of feedback: Nathan Lynch <nathanl@austin.ibm.com>,
19 Joel Schopp <jschopp@austin.ibm.com>
20
21Introduction
22
23Modern advances in system architectures have introduced advanced error
24reporting and correction capabilities in processors. CPU architectures permit
25partitioning support, where compute resources of a single CPU could be made
26available to virtual machine environments. There are couple OEMS that
27support NUMA hardware which are hot pluggable as well, where physical
28node insertion and removal require support for CPU hotplug.
29
30Such advances require CPUs available to a kernel to be removed either for
31provisioning reasons, or for RAS purposes to keep an offending CPU off
32system execution path. Hence the need for CPU hotplug support in the
33Linux kernel.
34
35A more novel use of CPU-hotplug support is its use today in suspend
36resume support for SMP. Dual-core and HT support makes even
37a laptop run SMP kernels which didn't support these methods. SMP support
38for suspend/resume is a work in progress.
39
40General Stuff about CPU Hotplug
41--------------------------------
42
43Command Line Switches
44---------------------
45maxcpus=n Restrict boot time cpus to n. Say if you have 4 cpus, using
46 maxcpus=2 will only boot 2. You can choose to bring the
47 other cpus later online, read FAQ's for more info.
48
Satoru Takeuchica926e82006-10-19 23:29:06 -070049additional_cpus=n (*) Use this to limit hotpluggable cpus. This option sets
Rusty Russell5f054e32012-03-29 15:38:31 +103050 cpu_possible_mask = cpu_present_mask + additional_cpus
Ashok Raj8f8b1132006-02-16 14:01:48 -080051
Gautham R Shenoy3aa565f2009-10-29 19:22:53 +000052cede_offline={"off","on"} Use this option to disable/enable putting offlined
53 processors to an extended H_CEDE state on
54 supported pseries platforms.
55 If nothing is specified,
56 cede_offline is set to "on".
57
Heiko Carstens6303dbf2006-02-20 18:27:58 -080058(*) Option valid only for following architectures
Mike Travis3b11ce72008-12-17 15:21:39 -080059- ia64
Heiko Carstens6303dbf2006-02-20 18:27:58 -080060
Mike Travis3b11ce72008-12-17 15:21:39 -080061ia64 uses the number of disabled local apics in ACPI tables MADT to
62determine the number of potentially hot-pluggable cpus. The implementation
63should only rely on this to count the # of cpus, but *MUST* not rely
64on the apicid values in those tables for disabled apics. In the event
65BIOS doesn't mark such hot-pluggable cpus as disabled entries, one could
66use this parameter "additional_cpus=x" to represent those cpus in the
Rusty Russell5f054e32012-03-29 15:38:31 +103067cpu_possible_mask.
Ashok Raj8f8b1132006-02-16 14:01:48 -080068
Mike Travis3b11ce72008-12-17 15:21:39 -080069possible_cpus=n [s390,x86_64] use this to set hotpluggable cpus.
Heiko Carstens37a33022006-02-17 13:52:47 -080070 This option sets possible_cpus bits in
Rusty Russell5f054e32012-03-29 15:38:31 +103071 cpu_possible_mask. Thus keeping the numbers of bits set
Heiko Carstens37a33022006-02-17 13:52:47 -080072 constant even if the machine gets rebooted.
Heiko Carstens37a33022006-02-17 13:52:47 -080073
Ashok Rajc8094062006-01-08 01:03:17 -080074CPU maps and such
75-----------------
76[More on cpumaps and primitive to manipulate, please check
77include/linux/cpumask.h that has more descriptive text.]
78
Rusty Russell5f054e32012-03-29 15:38:31 +103079cpu_possible_mask: Bitmap of possible CPUs that can ever be available in the
Ashok Rajc8094062006-01-08 01:03:17 -080080system. This is used to allocate some boot time memory for per_cpu variables
81that aren't designed to grow/shrink as CPUs are made available or removed.
82Once set during boot time discovery phase, the map is static, i.e no bits
83are added or removed anytime. Trimming it accurately for your system needs
84upfront can save some boot time memory. See below for how we use heuristics
85in x86_64 case to keep this under check.
86
Rusty Russell5f054e32012-03-29 15:38:31 +103087cpu_online_mask: Bitmap of all CPUs currently online. Its set in __cpu_up()
Ashok Rajc8094062006-01-08 01:03:17 -080088after a cpu is available for kernel scheduling and ready to receive
89interrupts from devices. Its cleared when a cpu is brought down using
90__cpu_disable(), before which all OS services including interrupts are
91migrated to another target CPU.
92
Rusty Russell5f054e32012-03-29 15:38:31 +103093cpu_present_mask: Bitmap of CPUs currently present in the system. Not all
Ashok Rajc8094062006-01-08 01:03:17 -080094of them may be online. When physical hotplug is processed by the relevant
95subsystem (e.g ACPI) can change and new bit either be added or removed
96from the map depending on the event is hot-add/hot-remove. There are currently
97no locking rules as of now. Typical usage is to init topology during boot,
98at which time hotplug is disabled.
99
100You really dont need to manipulate any of the system cpu maps. They should
101be read-only for most use. When setting up per-cpu resources almost always use
Rusty Russell5f054e32012-03-29 15:38:31 +1030102cpu_possible_mask/for_each_possible_cpu() to iterate.
Ashok Rajc8094062006-01-08 01:03:17 -0800103
104Never use anything other than cpumask_t to represent bitmap of CPUs.
105
Satoru Takeuchica926e82006-10-19 23:29:06 -0700106 #include <linux/cpumask.h>
Ashok Rajc8094062006-01-08 01:03:17 -0800107
Rusty Russell5f054e32012-03-29 15:38:31 +1030108 for_each_possible_cpu - Iterate over cpu_possible_mask
109 for_each_online_cpu - Iterate over cpu_online_mask
110 for_each_present_cpu - Iterate over cpu_present_mask
Satoru Takeuchica926e82006-10-19 23:29:06 -0700111 for_each_cpu_mask(x,mask) - Iterate over some random collection of cpu mask.
Ashok Rajc8094062006-01-08 01:03:17 -0800112
Satoru Takeuchica926e82006-10-19 23:29:06 -0700113 #include <linux/cpu.h>
Gautham R Shenoy86ef5c92008-01-25 21:08:02 +0100114 get_online_cpus() and put_online_cpus():
Ashok Rajc8094062006-01-08 01:03:17 -0800115
Gautham R Shenoy86ef5c92008-01-25 21:08:02 +0100116The above calls are used to inhibit cpu hotplug operations. While the
Rusty Russell5f054e32012-03-29 15:38:31 +1030117cpu_hotplug.refcount is non zero, the cpu_online_mask will not change.
Gautham R Shenoy86ef5c92008-01-25 21:08:02 +0100118If you merely need to avoid cpus going away, you could also use
119preempt_disable() and preempt_enable() for those sections.
120Just remember the critical section cannot call any
Ashok Rajc8094062006-01-08 01:03:17 -0800121function that can sleep or schedule this process away. The preempt_disable()
122will work as long as stop_machine_run() is used to take a cpu down.
123
124CPU Hotplug - Frequently Asked Questions.
125
Satoru Takeuchica926e82006-10-19 23:29:06 -0700126Q: How to enable my kernel to support CPU hotplug?
Ashok Rajc8094062006-01-08 01:03:17 -0800127A: When doing make defconfig, Enable CPU hotplug support
128
129 "Processor type and Features" -> Support for Hotpluggable CPUs
130
131Make sure that you have CONFIG_HOTPLUG, and CONFIG_SMP turned on as well.
132
133You would need to enable CONFIG_HOTPLUG_CPU for SMP suspend/resume support
134as well.
135
136Q: What architectures support CPU hotplug?
137A: As of 2.6.14, the following architectures support CPU hotplug.
138
139i386 (Intel), ppc, ppc64, parisc, s390, ia64 and x86_64
140
141Q: How to test if hotplug is supported on the newly built kernel?
142A: You should now notice an entry in sysfs.
143
144Check if sysfs is mounted, using the "mount" command. You should notice
145an entry as shown below in the output.
146
Satoru Takeuchica926e82006-10-19 23:29:06 -0700147 ....
148 none on /sys type sysfs (rw)
149 ....
Ashok Rajc8094062006-01-08 01:03:17 -0800150
Satoru Takeuchica926e82006-10-19 23:29:06 -0700151If this is not mounted, do the following.
Ashok Rajc8094062006-01-08 01:03:17 -0800152
Satoru Takeuchica926e82006-10-19 23:29:06 -0700153 #mkdir /sysfs
154 #mount -t sysfs sys /sys
Ashok Rajc8094062006-01-08 01:03:17 -0800155
Satoru Takeuchica926e82006-10-19 23:29:06 -0700156Now you should see entries for all present cpu, the following is an example
Ashok Rajc8094062006-01-08 01:03:17 -0800157in a 8-way system.
158
Satoru Takeuchica926e82006-10-19 23:29:06 -0700159 #pwd
160 #/sys/devices/system/cpu
161 #ls -l
162 total 0
163 drwxr-xr-x 10 root root 0 Sep 19 07:44 .
164 drwxr-xr-x 13 root root 0 Sep 19 07:45 ..
165 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu0
166 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu1
167 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu2
168 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu3
169 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu4
170 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu5
171 drwxr-xr-x 3 root root 0 Sep 19 07:44 cpu6
172 drwxr-xr-x 3 root root 0 Sep 19 07:48 cpu7
Ashok Rajc8094062006-01-08 01:03:17 -0800173
174Under each directory you would find an "online" file which is the control
175file to logically online/offline a processor.
176
177Q: Does hot-add/hot-remove refer to physical add/remove of cpus?
178A: The usage of hot-add/remove may not be very consistently used in the code.
Satoru Takeuchica926e82006-10-19 23:29:06 -0700179CONFIG_HOTPLUG_CPU enables logical online/offline capability in the kernel.
Ashok Rajc8094062006-01-08 01:03:17 -0800180To support physical addition/removal, one would need some BIOS hooks and
181the platform should have something like an attention button in PCI hotplug.
182CONFIG_ACPI_HOTPLUG_CPU enables ACPI support for physical add/remove of CPUs.
183
184Q: How do i logically offline a CPU?
185A: Do the following.
186
Satoru Takeuchica926e82006-10-19 23:29:06 -0700187 #echo 0 > /sys/devices/system/cpu/cpuX/online
Ashok Rajc8094062006-01-08 01:03:17 -0800188
Satoru Takeuchica926e82006-10-19 23:29:06 -0700189Once the logical offline is successful, check
Ashok Rajc8094062006-01-08 01:03:17 -0800190
Satoru Takeuchica926e82006-10-19 23:29:06 -0700191 #cat /proc/interrupts
Ashok Rajc8094062006-01-08 01:03:17 -0800192
Satoru Takeuchica926e82006-10-19 23:29:06 -0700193You should now not see the CPU that you removed. Also online file will report
Ashok Rajc8094062006-01-08 01:03:17 -0800194the state as 0 when a cpu if offline and 1 when its online.
195
Satoru Takeuchica926e82006-10-19 23:29:06 -0700196 #To display the current cpu state.
197 #cat /sys/devices/system/cpu/cpuX/online
Ashok Rajc8094062006-01-08 01:03:17 -0800198
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300199Q: Why can't i remove CPU0 on some systems?
Ashok Rajc8094062006-01-08 01:03:17 -0800200A: Some architectures may have some special dependency on a certain CPU.
201
202For e.g in IA64 platforms we have ability to sent platform interrupts to the
203OS. a.k.a Corrected Platform Error Interrupts (CPEI). In current ACPI
204specifications, we didn't have a way to change the target CPU. Hence if the
205current ACPI version doesn't support such re-direction, we disable that CPU
206by making it not-removable.
207
208In such cases you will also notice that the online file is missing under cpu0.
209
210Q: How do i find out if a particular CPU is not removable?
211A: Depending on the implementation, some architectures may show this by the
212absence of the "online" file. This is done if it can be determined ahead of
213time that this CPU cannot be removed.
214
215In some situations, this can be a run time check, i.e if you try to remove the
216last CPU, this will not be permitted. You can find such failures by
217investigating the return value of the "echo" command.
218
219Q: What happens when a CPU is being logically offlined?
220A: The following happen, listed in no particular order :-)
221
222- A notification is sent to in-kernel registered modules by sending an event
Rafael J. Wysocki8bb78442007-05-09 02:35:10 -0700223 CPU_DOWN_PREPARE or CPU_DOWN_PREPARE_FROZEN, depending on whether or not the
224 CPU is being offlined while tasks are frozen due to a suspend operation in
225 progress
Cliff Wickman470fd642007-10-18 23:40:46 -0700226- All processes are migrated away from this outgoing CPU to new CPUs.
227 The new CPU is chosen from each process' current cpuset, which may be
228 a subset of all online CPUs.
Ashok Rajc8094062006-01-08 01:03:17 -0800229- All interrupts targeted to this CPU is migrated to a new CPU
230- timers/bottom half/task lets are also migrated to a new CPU
231- Once all services are migrated, kernel calls an arch specific routine
232 __cpu_disable() to perform arch specific cleanup.
233- Once this is successful, an event for successful cleanup is sent by an event
Rafael J. Wysocki8bb78442007-05-09 02:35:10 -0700234 CPU_DEAD (or CPU_DEAD_FROZEN if tasks are frozen due to a suspend while the
235 CPU is being offlined).
Ashok Rajc8094062006-01-08 01:03:17 -0800236
237 "It is expected that each service cleans up when the CPU_DOWN_PREPARE
238 notifier is called, when CPU_DEAD is called its expected there is nothing
239 running on behalf of this CPU that was offlined"
240
241Q: If i have some kernel code that needs to be aware of CPU arrival and
242 departure, how to i arrange for proper notification?
243A: This is what you would need in your kernel code to receive notifications.
244
Satoru Takeuchica926e82006-10-19 23:29:06 -0700245 #include <linux/cpu.h>
246 static int __cpuinit foobar_cpu_callback(struct notifier_block *nfb,
Ashok Rajc8094062006-01-08 01:03:17 -0800247 unsigned long action, void *hcpu)
248 {
249 unsigned int cpu = (unsigned long)hcpu;
250
251 switch (action) {
252 case CPU_ONLINE:
Rafael J. Wysocki8bb78442007-05-09 02:35:10 -0700253 case CPU_ONLINE_FROZEN:
Ashok Rajc8094062006-01-08 01:03:17 -0800254 foobar_online_action(cpu);
255 break;
256 case CPU_DEAD:
Rafael J. Wysocki8bb78442007-05-09 02:35:10 -0700257 case CPU_DEAD_FROZEN:
Ashok Rajc8094062006-01-08 01:03:17 -0800258 foobar_dead_action(cpu);
259 break;
260 }
261 return NOTIFY_OK;
262 }
263
Chandra Seetharaman7c7165c2006-07-30 03:03:36 -0700264 static struct notifier_block __cpuinitdata foobar_cpu_notifer =
Ashok Rajc8094062006-01-08 01:03:17 -0800265 {
266 .notifier_call = foobar_cpu_callback,
267 };
268
Chandra Seetharaman7c7165c2006-07-30 03:03:36 -0700269You need to call register_cpu_notifier() from your init function.
270Init functions could be of two types:
2711. early init (init function called when only the boot processor is online).
2722. late init (init function called _after_ all the CPUs are online).
Ashok Rajc8094062006-01-08 01:03:17 -0800273
Chandra Seetharaman7c7165c2006-07-30 03:03:36 -0700274For the first case, you should add the following to your init function
Ashok Rajc8094062006-01-08 01:03:17 -0800275
276 register_cpu_notifier(&foobar_cpu_notifier);
277
Chandra Seetharaman7c7165c2006-07-30 03:03:36 -0700278For the second case, you should add the following to your init function
279
280 register_hotcpu_notifier(&foobar_cpu_notifier);
281
Ashok Rajc8094062006-01-08 01:03:17 -0800282You can fail PREPARE notifiers if something doesn't work to prepare resources.
283This will stop the activity and send a following CANCELED event back.
284
285CPU_DEAD should not be failed, its just a goodness indication, but bad
286things will happen if a notifier in path sent a BAD notify code.
287
288Q: I don't see my action being called for all CPUs already up and running?
289A: Yes, CPU notifiers are called only when new CPUs are on-lined or offlined.
290 If you need to perform some action for each cpu already in the system, then
291
Satoru Takeuchica926e82006-10-19 23:29:06 -0700292 for_each_online_cpu(i) {
Ashok Rajc8094062006-01-08 01:03:17 -0800293 foobar_cpu_callback(&foobar_cpu_notifier, CPU_UP_PREPARE, i);
Satoru Takeuchica926e82006-10-19 23:29:06 -0700294 foobar_cpu_callback(&foobar_cpu_notifier, CPU_ONLINE, i);
295 }
Ashok Rajc8094062006-01-08 01:03:17 -0800296
297Q: If i would like to develop cpu hotplug support for a new architecture,
298 what do i need at a minimum?
299A: The following are what is required for CPU hotplug infrastructure to work
300 correctly.
301
302 - Make sure you have an entry in Kconfig to enable CONFIG_HOTPLUG_CPU
303 - __cpu_up() - Arch interface to bring up a CPU
304 - __cpu_disable() - Arch interface to shutdown a CPU, no more interrupts
305 can be handled by the kernel after the routine
306 returns. Including local APIC timers etc are
307 shutdown.
308 - __cpu_die() - This actually supposed to ensure death of the CPU.
309 Actually look at some example code in other arch
310 that implement CPU hotplug. The processor is taken
311 down from the idle() loop for that specific
312 architecture. __cpu_die() typically waits for some
313 per_cpu state to be set, to ensure the processor
314 dead routine is called to be sure positively.
315
316Q: I need to ensure that a particular cpu is not removed when there is some
317 work specific to this cpu is in progress.
Rusty Russell42514172009-12-17 11:43:29 -0600318A: There are two ways. If your code can be run in interrupt context, use
319 smp_call_function_single(), otherwise use work_on_cpu(). Note that
320 work_on_cpu() is slow, and can fail due to out of memory:
Ashok Rajc8094062006-01-08 01:03:17 -0800321
Satoru Takeuchica926e82006-10-19 23:29:06 -0700322 int my_func_on_cpu(int cpu)
323 {
Rusty Russell42514172009-12-17 11:43:29 -0600324 int err;
325 get_online_cpus();
326 if (!cpu_online(cpu))
327 err = -EINVAL;
328 else
329#if NEEDS_BLOCKING
330 err = work_on_cpu(cpu, __my_func_on_cpu, NULL);
331#else
332 smp_call_function_single(cpu, __my_func_on_cpu, &err,
333 true);
334#endif
335 put_online_cpus();
336 return err;
337 }
Ashok Rajc8094062006-01-08 01:03:17 -0800338
339Q: How do we determine how many CPUs are available for hotplug.
340A: There is no clear spec defined way from ACPI that can give us that
341 information today. Based on some input from Natalie of Unisys,
342 that the ACPI MADT (Multiple APIC Description Tables) marks those possible
343 CPUs in a system with disabled status.
344
345 Andi implemented some simple heuristics that count the number of disabled
346 CPUs in MADT as hotpluggable CPUS. In the case there are no disabled CPUS
347 we assume 1/2 the number of CPUs currently present can be hotplugged.
348
349 Caveat: Today's ACPI MADT can only provide 256 entries since the apicid field
350 in MADT is only 8 bits.
351
352User Space Notification
353
354Hotplug support for devices is common in Linux today. Its being used today to
355support automatic configuration of network, usb and pci devices. A hotplug
356event can be used to invoke an agent script to perform the configuration task.
357
358You can add /etc/hotplug/cpu.agent to handle hotplug notification user space
359scripts.
360
361 #!/bin/bash
362 # $Id: cpu.agent
363 # Kernel hotplug params include:
364 #ACTION=%s [online or offline]
365 #DEVPATH=%s
366 #
367 cd /etc/hotplug
368 . ./hotplug.functions
369
370 case $ACTION in
371 online)
372 echo `date` ":cpu.agent" add cpu >> /tmp/hotplug.txt
373 ;;
374 offline)
375 echo `date` ":cpu.agent" remove cpu >>/tmp/hotplug.txt
376 ;;
377 *)
378 debug_mesg CPU $ACTION event not supported
379 exit 1
380 ;;
381 esac