kernel-ark/include/linux/cpumask.h
Ingo Molnar a263898f62 CPU hotplug: fix cpu_is_offline() on !CONFIG_HOTPLUG_CPU
make randconfig bootup testing found that the cpufreq code
crashes on bootup, if the powernow-k8 driver is enabled and
if maxcpus=1 passed on the boot line to a !CONFIG_HOTPLUG_CPU
kernel.

First lockdep found out that there's an inconsistent unlock
sequence:

 =====================================
 [ BUG: bad unlock balance detected! ]
 -------------------------------------
 swapper/1 is trying to release lock (&per_cpu(cpu_policy_rwsem, cpu)) at:
 [<ffffffff806ffd8e>] unlock_policy_rwsem_write+0x3c/0x42
 but there are no more locks to release!

Call Trace:
 [<ffffffff806ffd8e>] unlock_policy_rwsem_write+0x3c/0x42
 [<ffffffff80251c29>] print_unlock_inbalance_bug+0x104/0x12c
 [<ffffffff80252f3a>] mark_held_locks+0x56/0x94
 [<ffffffff806ffd8e>] unlock_policy_rwsem_write+0x3c/0x42
 [<ffffffff807008b6>] cpufreq_add_dev+0x2a8/0x5c4
 ...

then shortly afterwards the cpufreq code crashed on an assert:

 ------------[ cut here ]------------
 kernel BUG at drivers/cpufreq/cpufreq.c:1068!
 invalid opcode: 0000 [1] SMP
 [...]
 Call Trace:
  [<ffffffff805145d6>] sysdev_driver_unregister+0x5b/0x91
  [<ffffffff806ff520>] cpufreq_register_driver+0x15d/0x1a2
  [<ffffffff80cc0596>] powernowk8_init+0x86/0x94
 [...]
 ---[ end trace 1e9219be2b4431de ]---

the bug was caused by maxcpus=1 bootup, which brought up the
secondary core as !cpu_online() but !cpu_is_offline() either,
which on on !CONFIG_HOTPLUG_CPU is always 0 (include/linux/cpu.h):

  /* CPUs don't go offline once they're online w/o CONFIG_HOTPLUG_CPU */
  static inline int cpu_is_offline(int cpu) { return 0; }

but the cpufreq code uses cpu_online() and cpu_is_offline() in
a mixed way - the low-level drivers use cpu_online(), while
the cpufreq core uses cpu_is_offline(). This opened up the
possibility to add the non-initialized sysdev device of the
secondary core:

 cpufreq-core: trying to register driver powernow-k8
 cpufreq-core: adding CPU 0
 powernow-k8: BIOS error - no PSB or ACPI _PSS objects
 cpufreq-core: initialization failed
 cpufreq-core: adding CPU 1
 cpufreq-core: initialization failed

which then blew up. The fix is to make cpu_is_offline() always
the negation of cpu_online(). With that fix applied the kernel
boots up fine without crashing:

 Calling initcall 0xffffffff80cc0510: powernowk8_init+0x0/0x94()
 powernow-k8: Found 1 AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ processors (1 cpu cores) (version 2.20.00)
 powernow-k8: BIOS error - no PSB or ACPI _PSS objects
 initcall 0xffffffff80cc0510: powernowk8_init+0x0/0x94() returned -19.
 initcall 0xffffffff80cc0510 ran for 19 msecs: powernowk8_init+0x0/0x94()
 Calling initcall 0xffffffff80cc328f: init_lapic_nmi_sysfs+0x0/0x39()

We could fix this by making CPU enumeration aware of max_cpus, but that
would be more fragile IMO, and the cpu_online(cpu) != cpu_is_offline(cpu)
possibility was quite confusing and a continuous source of bugs too.

Most distributions have kernels with CPU hotplug enabled, so this bug
remained hidden for a long time.

Bug forensics:

The broken cpu_is_offline() API variant was introduced via:

 commit a59d2e4e6977e7b94e003c96a41f07e96cddc340
 Author: Rusty Russell <rusty@rustcorp.com.au>
 Date:   Mon Mar 8 06:06:03 2004 -0800

     [PATCH] minor cleanups for hotplug CPUs

( this predates linux-2.6.git, this commit is available from Thomas's
  historic git tree. )

Then 1.5 years later the cpufreq code made use of it:

 commit c32b6b8e52
 Author: Ashok Raj <ashok.raj@intel.com>
 Date:   Sun Oct 30 14:59:54 2005 -0800

     [PATCH] create and destroy cpufreq sysfs entries based on cpu notifiers

 +       if (cpu_is_offline(cpu))
 +               return 0;

which is a correct use of the subtly broken new API. v2.6.15 then
shipped with this bug included.

then it took two more years for random-kernel qa to hit it.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-01-06 12:39:42 -08:00

416 lines
14 KiB
C

#ifndef __LINUX_CPUMASK_H
#define __LINUX_CPUMASK_H
/*
* Cpumasks provide a bitmap suitable for representing the
* set of CPU's in a system, one bit position per CPU number.
*
* See detailed comments in the file linux/bitmap.h describing the
* data type on which these cpumasks are based.
*
* For details of cpumask_scnprintf() and cpumask_parse_user(),
* see bitmap_scnprintf() and bitmap_parse_user() in lib/bitmap.c.
* For details of cpulist_scnprintf() and cpulist_parse(), see
* bitmap_scnlistprintf() and bitmap_parselist(), also in bitmap.c.
* For details of cpu_remap(), see bitmap_bitremap in lib/bitmap.c
* For details of cpus_remap(), see bitmap_remap in lib/bitmap.c.
*
* The available cpumask operations are:
*
* void cpu_set(cpu, mask) turn on bit 'cpu' in mask
* void cpu_clear(cpu, mask) turn off bit 'cpu' in mask
* void cpus_setall(mask) set all bits
* void cpus_clear(mask) clear all bits
* int cpu_isset(cpu, mask) true iff bit 'cpu' set in mask
* int cpu_test_and_set(cpu, mask) test and set bit 'cpu' in mask
*
* void cpus_and(dst, src1, src2) dst = src1 & src2 [intersection]
* void cpus_or(dst, src1, src2) dst = src1 | src2 [union]
* void cpus_xor(dst, src1, src2) dst = src1 ^ src2
* void cpus_andnot(dst, src1, src2) dst = src1 & ~src2
* void cpus_complement(dst, src) dst = ~src
*
* int cpus_equal(mask1, mask2) Does mask1 == mask2?
* int cpus_intersects(mask1, mask2) Do mask1 and mask2 intersect?
* int cpus_subset(mask1, mask2) Is mask1 a subset of mask2?
* int cpus_empty(mask) Is mask empty (no bits sets)?
* int cpus_full(mask) Is mask full (all bits sets)?
* int cpus_weight(mask) Hamming weigh - number of set bits
*
* void cpus_shift_right(dst, src, n) Shift right
* void cpus_shift_left(dst, src, n) Shift left
*
* int first_cpu(mask) Number lowest set bit, or NR_CPUS
* int next_cpu(cpu, mask) Next cpu past 'cpu', or NR_CPUS
*
* cpumask_t cpumask_of_cpu(cpu) Return cpumask with bit 'cpu' set
* CPU_MASK_ALL Initializer - all bits set
* CPU_MASK_NONE Initializer - no bits set
* unsigned long *cpus_addr(mask) Array of unsigned long's in mask
*
* int cpumask_scnprintf(buf, len, mask) Format cpumask for printing
* int cpumask_parse_user(ubuf, ulen, mask) Parse ascii string as cpumask
* int cpulist_scnprintf(buf, len, mask) Format cpumask as list for printing
* int cpulist_parse(buf, map) Parse ascii string as cpulist
* int cpu_remap(oldbit, old, new) newbit = map(old, new)(oldbit)
* int cpus_remap(dst, src, old, new) *dst = map(old, new)(src)
*
* for_each_cpu_mask(cpu, mask) for-loop cpu over mask
*
* int num_online_cpus() Number of online CPUs
* int num_possible_cpus() Number of all possible CPUs
* int num_present_cpus() Number of present CPUs
*
* int cpu_online(cpu) Is some cpu online?
* int cpu_possible(cpu) Is some cpu possible?
* int cpu_present(cpu) Is some cpu present (can schedule)?
*
* int any_online_cpu(mask) First online cpu in mask
*
* for_each_possible_cpu(cpu) for-loop cpu over cpu_possible_map
* for_each_online_cpu(cpu) for-loop cpu over cpu_online_map
* for_each_present_cpu(cpu) for-loop cpu over cpu_present_map
*
* Subtlety:
* 1) The 'type-checked' form of cpu_isset() causes gcc (3.3.2, anyway)
* to generate slightly worse code. Note for example the additional
* 40 lines of assembly code compiling the "for each possible cpu"
* loops buried in the disk_stat_read() macros calls when compiling
* drivers/block/genhd.c (arch i386, CONFIG_SMP=y). So use a simple
* one-line #define for cpu_isset(), instead of wrapping an inline
* inside a macro, the way we do the other calls.
*/
#include <linux/kernel.h>
#include <linux/threads.h>
#include <linux/bitmap.h>
typedef struct { DECLARE_BITMAP(bits, NR_CPUS); } cpumask_t;
extern cpumask_t _unused_cpumask_arg_;
#define cpu_set(cpu, dst) __cpu_set((cpu), &(dst))
static inline void __cpu_set(int cpu, volatile cpumask_t *dstp)
{
set_bit(cpu, dstp->bits);
}
#define cpu_clear(cpu, dst) __cpu_clear((cpu), &(dst))
static inline void __cpu_clear(int cpu, volatile cpumask_t *dstp)
{
clear_bit(cpu, dstp->bits);
}
#define cpus_setall(dst) __cpus_setall(&(dst), NR_CPUS)
static inline void __cpus_setall(cpumask_t *dstp, int nbits)
{
bitmap_fill(dstp->bits, nbits);
}
#define cpus_clear(dst) __cpus_clear(&(dst), NR_CPUS)
static inline void __cpus_clear(cpumask_t *dstp, int nbits)
{
bitmap_zero(dstp->bits, nbits);
}
/* No static inline type checking - see Subtlety (1) above. */
#define cpu_isset(cpu, cpumask) test_bit((cpu), (cpumask).bits)
#define cpu_test_and_set(cpu, cpumask) __cpu_test_and_set((cpu), &(cpumask))
static inline int __cpu_test_and_set(int cpu, cpumask_t *addr)
{
return test_and_set_bit(cpu, addr->bits);
}
#define cpus_and(dst, src1, src2) __cpus_and(&(dst), &(src1), &(src2), NR_CPUS)
static inline void __cpus_and(cpumask_t *dstp, const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
bitmap_and(dstp->bits, src1p->bits, src2p->bits, nbits);
}
#define cpus_or(dst, src1, src2) __cpus_or(&(dst), &(src1), &(src2), NR_CPUS)
static inline void __cpus_or(cpumask_t *dstp, const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
bitmap_or(dstp->bits, src1p->bits, src2p->bits, nbits);
}
#define cpus_xor(dst, src1, src2) __cpus_xor(&(dst), &(src1), &(src2), NR_CPUS)
static inline void __cpus_xor(cpumask_t *dstp, const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
bitmap_xor(dstp->bits, src1p->bits, src2p->bits, nbits);
}
#define cpus_andnot(dst, src1, src2) \
__cpus_andnot(&(dst), &(src1), &(src2), NR_CPUS)
static inline void __cpus_andnot(cpumask_t *dstp, const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
bitmap_andnot(dstp->bits, src1p->bits, src2p->bits, nbits);
}
#define cpus_complement(dst, src) __cpus_complement(&(dst), &(src), NR_CPUS)
static inline void __cpus_complement(cpumask_t *dstp,
const cpumask_t *srcp, int nbits)
{
bitmap_complement(dstp->bits, srcp->bits, nbits);
}
#define cpus_equal(src1, src2) __cpus_equal(&(src1), &(src2), NR_CPUS)
static inline int __cpus_equal(const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
return bitmap_equal(src1p->bits, src2p->bits, nbits);
}
#define cpus_intersects(src1, src2) __cpus_intersects(&(src1), &(src2), NR_CPUS)
static inline int __cpus_intersects(const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
return bitmap_intersects(src1p->bits, src2p->bits, nbits);
}
#define cpus_subset(src1, src2) __cpus_subset(&(src1), &(src2), NR_CPUS)
static inline int __cpus_subset(const cpumask_t *src1p,
const cpumask_t *src2p, int nbits)
{
return bitmap_subset(src1p->bits, src2p->bits, nbits);
}
#define cpus_empty(src) __cpus_empty(&(src), NR_CPUS)
static inline int __cpus_empty(const cpumask_t *srcp, int nbits)
{
return bitmap_empty(srcp->bits, nbits);
}
#define cpus_full(cpumask) __cpus_full(&(cpumask), NR_CPUS)
static inline int __cpus_full(const cpumask_t *srcp, int nbits)
{
return bitmap_full(srcp->bits, nbits);
}
#define cpus_weight(cpumask) __cpus_weight(&(cpumask), NR_CPUS)
static inline int __cpus_weight(const cpumask_t *srcp, int nbits)
{
return bitmap_weight(srcp->bits, nbits);
}
#define cpus_shift_right(dst, src, n) \
__cpus_shift_right(&(dst), &(src), (n), NR_CPUS)
static inline void __cpus_shift_right(cpumask_t *dstp,
const cpumask_t *srcp, int n, int nbits)
{
bitmap_shift_right(dstp->bits, srcp->bits, n, nbits);
}
#define cpus_shift_left(dst, src, n) \
__cpus_shift_left(&(dst), &(src), (n), NR_CPUS)
static inline void __cpus_shift_left(cpumask_t *dstp,
const cpumask_t *srcp, int n, int nbits)
{
bitmap_shift_left(dstp->bits, srcp->bits, n, nbits);
}
#ifdef CONFIG_SMP
int __first_cpu(const cpumask_t *srcp);
#define first_cpu(src) __first_cpu(&(src))
int __next_cpu(int n, const cpumask_t *srcp);
#define next_cpu(n, src) __next_cpu((n), &(src))
#else
#define first_cpu(src) 0
#define next_cpu(n, src) 1
#endif
#define cpumask_of_cpu(cpu) \
({ \
typeof(_unused_cpumask_arg_) m; \
if (sizeof(m) == sizeof(unsigned long)) { \
m.bits[0] = 1UL<<(cpu); \
} else { \
cpus_clear(m); \
cpu_set((cpu), m); \
} \
m; \
})
#define CPU_MASK_LAST_WORD BITMAP_LAST_WORD_MASK(NR_CPUS)
#if NR_CPUS <= BITS_PER_LONG
#define CPU_MASK_ALL \
(cpumask_t) { { \
[BITS_TO_LONGS(NR_CPUS)-1] = CPU_MASK_LAST_WORD \
} }
#else
#define CPU_MASK_ALL \
(cpumask_t) { { \
[0 ... BITS_TO_LONGS(NR_CPUS)-2] = ~0UL, \
[BITS_TO_LONGS(NR_CPUS)-1] = CPU_MASK_LAST_WORD \
} }
#endif
#define CPU_MASK_NONE \
(cpumask_t) { { \
[0 ... BITS_TO_LONGS(NR_CPUS)-1] = 0UL \
} }
#define CPU_MASK_CPU0 \
(cpumask_t) { { \
[0] = 1UL \
} }
#define cpus_addr(src) ((src).bits)
#define cpumask_scnprintf(buf, len, src) \
__cpumask_scnprintf((buf), (len), &(src), NR_CPUS)
static inline int __cpumask_scnprintf(char *buf, int len,
const cpumask_t *srcp, int nbits)
{
return bitmap_scnprintf(buf, len, srcp->bits, nbits);
}
#define cpumask_parse_user(ubuf, ulen, dst) \
__cpumask_parse_user((ubuf), (ulen), &(dst), NR_CPUS)
static inline int __cpumask_parse_user(const char __user *buf, int len,
cpumask_t *dstp, int nbits)
{
return bitmap_parse_user(buf, len, dstp->bits, nbits);
}
#define cpulist_scnprintf(buf, len, src) \
__cpulist_scnprintf((buf), (len), &(src), NR_CPUS)
static inline int __cpulist_scnprintf(char *buf, int len,
const cpumask_t *srcp, int nbits)
{
return bitmap_scnlistprintf(buf, len, srcp->bits, nbits);
}
#define cpulist_parse(buf, dst) __cpulist_parse((buf), &(dst), NR_CPUS)
static inline int __cpulist_parse(const char *buf, cpumask_t *dstp, int nbits)
{
return bitmap_parselist(buf, dstp->bits, nbits);
}
#define cpu_remap(oldbit, old, new) \
__cpu_remap((oldbit), &(old), &(new), NR_CPUS)
static inline int __cpu_remap(int oldbit,
const cpumask_t *oldp, const cpumask_t *newp, int nbits)
{
return bitmap_bitremap(oldbit, oldp->bits, newp->bits, nbits);
}
#define cpus_remap(dst, src, old, new) \
__cpus_remap(&(dst), &(src), &(old), &(new), NR_CPUS)
static inline void __cpus_remap(cpumask_t *dstp, const cpumask_t *srcp,
const cpumask_t *oldp, const cpumask_t *newp, int nbits)
{
bitmap_remap(dstp->bits, srcp->bits, oldp->bits, newp->bits, nbits);
}
#if NR_CPUS > 1
#define for_each_cpu_mask(cpu, mask) \
for ((cpu) = first_cpu(mask); \
(cpu) < NR_CPUS; \
(cpu) = next_cpu((cpu), (mask)))
#else /* NR_CPUS == 1 */
#define for_each_cpu_mask(cpu, mask) \
for ((cpu) = 0; (cpu) < 1; (cpu)++, (void)mask)
#endif /* NR_CPUS */
/*
* The following particular system cpumasks and operations manage
* possible, present and online cpus. Each of them is a fixed size
* bitmap of size NR_CPUS.
*
* #ifdef CONFIG_HOTPLUG_CPU
* cpu_possible_map - has bit 'cpu' set iff cpu is populatable
* cpu_present_map - has bit 'cpu' set iff cpu is populated
* cpu_online_map - has bit 'cpu' set iff cpu available to scheduler
* #else
* cpu_possible_map - has bit 'cpu' set iff cpu is populated
* cpu_present_map - copy of cpu_possible_map
* cpu_online_map - has bit 'cpu' set iff cpu available to scheduler
* #endif
*
* In either case, NR_CPUS is fixed at compile time, as the static
* size of these bitmaps. The cpu_possible_map is fixed at boot
* time, as the set of CPU id's that it is possible might ever
* be plugged in at anytime during the life of that system boot.
* The cpu_present_map is dynamic(*), representing which CPUs
* are currently plugged in. And cpu_online_map is the dynamic
* subset of cpu_present_map, indicating those CPUs available
* for scheduling.
*
* If HOTPLUG is enabled, then cpu_possible_map is forced to have
* all NR_CPUS bits set, otherwise it is just the set of CPUs that
* ACPI reports present at boot.
*
* If HOTPLUG is enabled, then cpu_present_map varies dynamically,
* depending on what ACPI reports as currently plugged in, otherwise
* cpu_present_map is just a copy of cpu_possible_map.
*
* (*) Well, cpu_present_map is dynamic in the hotplug case. If not
* hotplug, it's a copy of cpu_possible_map, hence fixed at boot.
*
* Subtleties:
* 1) UP arch's (NR_CPUS == 1, CONFIG_SMP not defined) hardcode
* assumption that their single CPU is online. The UP
* cpu_{online,possible,present}_maps are placebos. Changing them
* will have no useful affect on the following num_*_cpus()
* and cpu_*() macros in the UP case. This ugliness is a UP
* optimization - don't waste any instructions or memory references
* asking if you're online or how many CPUs there are if there is
* only one CPU.
* 2) Most SMP arch's #define some of these maps to be some
* other map specific to that arch. Therefore, the following
* must be #define macros, not inlines. To see why, examine
* the assembly code produced by the following. Note that
* set1() writes phys_x_map, but set2() writes x_map:
* int x_map, phys_x_map;
* #define set1(a) x_map = a
* inline void set2(int a) { x_map = a; }
* #define x_map phys_x_map
* main(){ set1(3); set2(5); }
*/
extern cpumask_t cpu_possible_map;
extern cpumask_t cpu_online_map;
extern cpumask_t cpu_present_map;
#if NR_CPUS > 1
#define num_online_cpus() cpus_weight(cpu_online_map)
#define num_possible_cpus() cpus_weight(cpu_possible_map)
#define num_present_cpus() cpus_weight(cpu_present_map)
#define cpu_online(cpu) cpu_isset((cpu), cpu_online_map)
#define cpu_possible(cpu) cpu_isset((cpu), cpu_possible_map)
#define cpu_present(cpu) cpu_isset((cpu), cpu_present_map)
#else
#define num_online_cpus() 1
#define num_possible_cpus() 1
#define num_present_cpus() 1
#define cpu_online(cpu) ((cpu) == 0)
#define cpu_possible(cpu) ((cpu) == 0)
#define cpu_present(cpu) ((cpu) == 0)
#endif
#define cpu_is_offline(cpu) unlikely(!cpu_online(cpu))
#ifdef CONFIG_SMP
extern int nr_cpu_ids;
#define any_online_cpu(mask) __any_online_cpu(&(mask))
int __any_online_cpu(const cpumask_t *mask);
#else
#define nr_cpu_ids 1
#define any_online_cpu(mask) 0
#endif
#define for_each_possible_cpu(cpu) for_each_cpu_mask((cpu), cpu_possible_map)
#define for_each_online_cpu(cpu) for_each_cpu_mask((cpu), cpu_online_map)
#define for_each_present_cpu(cpu) for_each_cpu_mask((cpu), cpu_present_map)
#endif /* __LINUX_CPUMASK_H */