License cleanup: add SPDX GPL-2.0 license identifier to files with no license
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.
By default all files without license information are under the default
license of the kernel, which is GPL version 2.
Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier. The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.
This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.
How this work was done:
Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
- file had no licensing information it it.
- file was a */uapi/* one with no licensing information in it,
- file was a */uapi/* one with existing licensing information,
Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.
The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne. Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.
The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed. Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.
Criteria used to select files for SPDX license identifier tagging was:
- Files considered eligible had to be source code files.
- Make and config files were included as candidates if they contained >5
lines of source
- File already had some variant of a license header in it (even if <5
lines).
All documentation files were explicitly excluded.
The following heuristics were used to determine which SPDX license
identifiers to apply.
- when both scanners couldn't find any license traces, file was
considered to have no license information in it, and the top level
COPYING file license applied.
For non */uapi/* files that summary was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 11139
and resulted in the first patch in this series.
If that file was a */uapi/* path one, it was "GPL-2.0 WITH
Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 WITH Linux-syscall-note 930
and resulted in the second patch in this series.
- if a file had some form of licensing information in it, and was one
of the */uapi/* ones, it was denoted with the Linux-syscall-note if
any GPL family license was found in the file or had no licensing in
it (per prior point). Results summary:
SPDX license identifier # files
---------------------------------------------------|------
GPL-2.0 WITH Linux-syscall-note 270
GPL-2.0+ WITH Linux-syscall-note 169
((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21
((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17
LGPL-2.1+ WITH Linux-syscall-note 15
GPL-1.0+ WITH Linux-syscall-note 14
((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5
LGPL-2.0+ WITH Linux-syscall-note 4
LGPL-2.1 WITH Linux-syscall-note 3
((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3
((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1
and that resulted in the third patch in this series.
- when the two scanners agreed on the detected license(s), that became
the concluded license(s).
- when there was disagreement between the two scanners (one detected a
license but the other didn't, or they both detected different
licenses) a manual inspection of the file occurred.
- In most cases a manual inspection of the information in the file
resulted in a clear resolution of the license that should apply (and
which scanner probably needed to revisit its heuristics).
- When it was not immediately clear, the license identifier was
confirmed with lawyers working with the Linux Foundation.
- If there was any question as to the appropriate license identifier,
the file was flagged for further research and to be revisited later
in time.
In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.
Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights. The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.
Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.
In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.
Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
- a full scancode scan run, collecting the matched texts, detected
license ids and scores
- reviewing anything where there was a license detected (about 500+
files) to ensure that the applied SPDX license was correct
- reviewing anything where there was no detection but the patch license
was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
SPDX license was correct
This produced a worksheet with 20 files needing minor correction. This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.
These .csv files were then reviewed by Greg. Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected. This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.) Finally Greg ran the script using the .csv files to
generate the patches.
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 07:07:57 -07:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
2005-07-27 11:44:57 -07:00
|
|
|
/*
|
|
|
|
* Out of line spinlock code.
|
|
|
|
*
|
2012-07-20 02:15:04 -07:00
|
|
|
* Copyright IBM Corp. 2004, 2006
|
2005-07-27 11:44:57 -07:00
|
|
|
* Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com)
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/types.h>
|
2017-02-09 13:20:25 -07:00
|
|
|
#include <linux/export.h>
|
2005-07-27 11:44:57 -07:00
|
|
|
#include <linux/spinlock.h>
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
#include <linux/jiffies.h>
|
2005-07-27 11:44:57 -07:00
|
|
|
#include <linux/init.h>
|
2012-03-11 08:59:26 -07:00
|
|
|
#include <linux/smp.h>
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
#include <linux/percpu.h>
|
2023-06-22 01:46:32 -07:00
|
|
|
#include <linux/io.h>
|
2017-10-12 04:01:47 -07:00
|
|
|
#include <asm/alternative.h>
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2015-01-14 09:52:33 -07:00
|
|
|
int spin_retry = -1;
|
|
|
|
|
|
|
|
static int __init spin_retry_init(void)
|
|
|
|
{
|
|
|
|
if (spin_retry < 0)
|
2017-03-24 09:00:45 -07:00
|
|
|
spin_retry = 1000;
|
2015-01-14 09:52:33 -07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
early_initcall(spin_retry_init);
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2021-10-21 06:14:43 -07:00
|
|
|
/*
|
2005-07-27 11:44:57 -07:00
|
|
|
* spin_retry= parameter
|
|
|
|
*/
|
|
|
|
static int __init spin_retry_setup(char *str)
|
|
|
|
{
|
|
|
|
spin_retry = simple_strtoul(str, &str, 0);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("spin_retry=", spin_retry_setup);
|
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
struct spin_wait {
|
|
|
|
struct spin_wait *next, *prev;
|
|
|
|
int node_id;
|
|
|
|
} __aligned(32);
|
|
|
|
|
|
|
|
static DEFINE_PER_CPU_ALIGNED(struct spin_wait, spin_wait[4]);
|
|
|
|
|
|
|
|
#define _Q_LOCK_CPU_OFFSET 0
|
|
|
|
#define _Q_LOCK_STEAL_OFFSET 16
|
|
|
|
#define _Q_TAIL_IDX_OFFSET 18
|
|
|
|
#define _Q_TAIL_CPU_OFFSET 20
|
|
|
|
|
|
|
|
#define _Q_LOCK_CPU_MASK 0x0000ffff
|
|
|
|
#define _Q_LOCK_STEAL_ADD 0x00010000
|
|
|
|
#define _Q_LOCK_STEAL_MASK 0x00030000
|
|
|
|
#define _Q_TAIL_IDX_MASK 0x000c0000
|
|
|
|
#define _Q_TAIL_CPU_MASK 0xfff00000
|
|
|
|
|
|
|
|
#define _Q_LOCK_MASK (_Q_LOCK_CPU_MASK | _Q_LOCK_STEAL_MASK)
|
|
|
|
#define _Q_TAIL_MASK (_Q_TAIL_IDX_MASK | _Q_TAIL_CPU_MASK)
|
|
|
|
|
|
|
|
void arch_spin_lock_setup(int cpu)
|
|
|
|
{
|
|
|
|
struct spin_wait *node;
|
|
|
|
int ix;
|
|
|
|
|
|
|
|
node = per_cpu_ptr(&spin_wait[0], cpu);
|
|
|
|
for (ix = 0; ix < 4; ix++, node++) {
|
|
|
|
memset(node, 0, sizeof(*node));
|
|
|
|
node->node_id = ((cpu + 1) << _Q_TAIL_CPU_OFFSET) +
|
|
|
|
(ix << _Q_TAIL_IDX_OFFSET);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-04-19 05:54:05 -07:00
|
|
|
static inline int arch_load_niai4(int *lock)
|
|
|
|
{
|
|
|
|
int owner;
|
|
|
|
|
2019-10-18 04:23:16 -07:00
|
|
|
asm_inline volatile(
|
2024-07-16 04:50:53 -07:00
|
|
|
ALTERNATIVE("nop", ".insn rre,0xb2fa0000,4,0", ALT_FACILITY(49)) /* NIAI 4 */
|
2017-04-19 05:54:05 -07:00
|
|
|
" l %0,%1\n"
|
|
|
|
: "=d" (owner) : "Q" (*lock) : "memory");
|
2017-11-14 05:47:00 -07:00
|
|
|
return owner;
|
2017-04-19 05:54:05 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline int arch_cmpxchg_niai8(int *lock, int old, int new)
|
|
|
|
{
|
|
|
|
int expected = old;
|
|
|
|
|
2019-10-18 04:23:16 -07:00
|
|
|
asm_inline volatile(
|
2024-07-16 04:50:53 -07:00
|
|
|
ALTERNATIVE("nop", ".insn rre,0xb2fa0000,8,0", ALT_FACILITY(49)) /* NIAI 8 */
|
2017-04-19 05:54:05 -07:00
|
|
|
" cs %0,%3,%1\n"
|
|
|
|
: "=d" (old), "=Q" (*lock)
|
|
|
|
: "0" (old), "d" (new), "Q" (*lock)
|
|
|
|
: "cc", "memory");
|
|
|
|
return expected == old;
|
|
|
|
}
|
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
static inline struct spin_wait *arch_spin_decode_tail(int lock)
|
2005-07-27 11:44:57 -07:00
|
|
|
{
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
int ix, cpu;
|
|
|
|
|
|
|
|
ix = (lock & _Q_TAIL_IDX_MASK) >> _Q_TAIL_IDX_OFFSET;
|
|
|
|
cpu = (lock & _Q_TAIL_CPU_MASK) >> _Q_TAIL_CPU_OFFSET;
|
|
|
|
return per_cpu_ptr(&spin_wait[ix], cpu - 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int arch_spin_yield_target(int lock, struct spin_wait *node)
|
|
|
|
{
|
|
|
|
if (lock & _Q_LOCK_CPU_MASK)
|
|
|
|
return lock & _Q_LOCK_CPU_MASK;
|
|
|
|
if (node == NULL || node->prev == NULL)
|
|
|
|
return 0; /* 0 -> no target cpu */
|
|
|
|
while (node->prev)
|
|
|
|
node = node->prev;
|
|
|
|
return node->node_id >> _Q_TAIL_CPU_OFFSET;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void arch_spin_lock_queued(arch_spinlock_t *lp)
|
|
|
|
{
|
|
|
|
struct spin_wait *node, *next;
|
|
|
|
int lockval, ix, node_id, tail_id, old, new, owner, count;
|
|
|
|
|
2024-06-10 04:45:25 -07:00
|
|
|
ix = get_lowcore()->spinlock_index++;
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
barrier();
|
|
|
|
lockval = SPINLOCK_LOCKVAL; /* cpu + 1 */
|
|
|
|
node = this_cpu_ptr(&spin_wait[ix]);
|
|
|
|
node->prev = node->next = NULL;
|
|
|
|
node_id = node->node_id;
|
|
|
|
|
|
|
|
/* Enqueue the node for this CPU in the spinlock wait queue */
|
|
|
|
while (1) {
|
|
|
|
old = READ_ONCE(lp->lock);
|
|
|
|
if ((old & _Q_LOCK_CPU_MASK) == 0 &&
|
|
|
|
(old & _Q_LOCK_STEAL_MASK) != _Q_LOCK_STEAL_MASK) {
|
|
|
|
/*
|
|
|
|
* The lock is free but there may be waiters.
|
|
|
|
* With no waiters simply take the lock, if there
|
|
|
|
* are waiters try to steal the lock. The lock may
|
|
|
|
* be stolen three times before the next queued
|
|
|
|
* waiter will get the lock.
|
|
|
|
*/
|
|
|
|
new = (old ? (old + _Q_LOCK_STEAL_ADD) : 0) | lockval;
|
|
|
|
if (__atomic_cmpxchg_bool(&lp->lock, old, new))
|
|
|
|
/* Got the lock */
|
|
|
|
goto out;
|
|
|
|
/* lock passing in progress */
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
/* Make the node of this CPU the new tail. */
|
|
|
|
new = node_id | (old & _Q_LOCK_MASK);
|
|
|
|
if (__atomic_cmpxchg_bool(&lp->lock, old, new))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
/* Set the 'next' pointer of the tail node in the queue */
|
|
|
|
tail_id = old & _Q_TAIL_MASK;
|
|
|
|
if (tail_id != 0) {
|
|
|
|
node->prev = arch_spin_decode_tail(tail_id);
|
|
|
|
WRITE_ONCE(node->prev->next, node);
|
|
|
|
}
|
2017-04-19 05:54:05 -07:00
|
|
|
|
|
|
|
/* Pass the virtual CPU to the lock holder if it is not running */
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
owner = arch_spin_yield_target(old, node);
|
2016-12-04 06:36:04 -07:00
|
|
|
if (owner && arch_vcpu_is_preempted(owner - 1))
|
|
|
|
smp_yield_cpu(owner - 1);
|
2005-07-27 11:44:57 -07:00
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
/* Spin on the CPU local node->prev pointer */
|
|
|
|
if (tail_id != 0) {
|
|
|
|
count = spin_retry;
|
|
|
|
while (READ_ONCE(node->prev) != NULL) {
|
|
|
|
if (count-- >= 0)
|
|
|
|
continue;
|
|
|
|
count = spin_retry;
|
|
|
|
/* Query running state of lock holder again. */
|
|
|
|
owner = arch_spin_yield_target(old, node);
|
|
|
|
if (owner && arch_vcpu_is_preempted(owner - 1))
|
|
|
|
smp_yield_cpu(owner - 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Spin on the lock value in the spinlock_t */
|
2017-04-19 05:54:05 -07:00
|
|
|
count = spin_retry;
|
2005-07-27 11:44:57 -07:00
|
|
|
while (1) {
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
old = READ_ONCE(lp->lock);
|
|
|
|
owner = old & _Q_LOCK_CPU_MASK;
|
2014-05-16 06:11:12 -07:00
|
|
|
if (!owner) {
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
tail_id = old & _Q_TAIL_MASK;
|
|
|
|
new = ((tail_id != node_id) ? tail_id : 0) | lockval;
|
|
|
|
if (__atomic_cmpxchg_bool(&lp->lock, old, new))
|
|
|
|
/* Got the lock */
|
|
|
|
break;
|
2014-05-16 06:11:12 -07:00
|
|
|
continue;
|
2005-07-27 11:44:57 -07:00
|
|
|
}
|
2017-04-19 05:54:05 -07:00
|
|
|
if (count-- >= 0)
|
2014-05-16 06:11:12 -07:00
|
|
|
continue;
|
|
|
|
count = spin_retry;
|
2016-12-04 06:36:04 -07:00
|
|
|
if (!MACHINE_IS_LPAR || arch_vcpu_is_preempted(owner - 1))
|
|
|
|
smp_yield_cpu(owner - 1);
|
2005-07-27 11:44:57 -07:00
|
|
|
}
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
|
|
|
|
/* Pass lock_spin job to next CPU in the queue */
|
|
|
|
if (node_id && tail_id != node_id) {
|
|
|
|
/* Wait until the next CPU has set up the 'next' pointer */
|
|
|
|
while ((next = READ_ONCE(node->next)) == NULL)
|
|
|
|
;
|
|
|
|
next->prev = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
out:
|
2024-06-10 04:45:25 -07:00
|
|
|
get_lowcore()->spinlock_index--;
|
2005-07-27 11:44:57 -07:00
|
|
|
}
|
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
static inline void arch_spin_lock_classic(arch_spinlock_t *lp)
|
2008-01-26 06:11:28 -07:00
|
|
|
{
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
int lockval, old, new, owner, count;
|
2008-01-26 06:11:28 -07:00
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
lockval = SPINLOCK_LOCKVAL; /* cpu + 1 */
|
2017-04-19 05:54:05 -07:00
|
|
|
|
|
|
|
/* Pass the virtual CPU to the lock holder if it is not running */
|
2017-11-13 13:38:26 -07:00
|
|
|
owner = arch_spin_yield_target(READ_ONCE(lp->lock), NULL);
|
2016-12-04 06:36:04 -07:00
|
|
|
if (owner && arch_vcpu_is_preempted(owner - 1))
|
|
|
|
smp_yield_cpu(owner - 1);
|
2017-04-19 05:54:05 -07:00
|
|
|
|
|
|
|
count = spin_retry;
|
2008-01-26 06:11:28 -07:00
|
|
|
while (1) {
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
old = arch_load_niai4(&lp->lock);
|
|
|
|
owner = old & _Q_LOCK_CPU_MASK;
|
2014-05-16 06:11:12 -07:00
|
|
|
/* Try to get the lock if it is free. */
|
|
|
|
if (!owner) {
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
new = (old & _Q_TAIL_MASK) | lockval;
|
2017-11-14 05:47:00 -07:00
|
|
|
if (arch_cmpxchg_niai8(&lp->lock, old, new)) {
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
/* Got the lock */
|
2017-11-14 05:47:00 -07:00
|
|
|
return;
|
|
|
|
}
|
2016-04-13 02:05:20 -07:00
|
|
|
continue;
|
2014-05-16 06:11:12 -07:00
|
|
|
}
|
2017-04-19 05:54:05 -07:00
|
|
|
if (count-- >= 0)
|
2014-05-16 06:11:12 -07:00
|
|
|
continue;
|
|
|
|
count = spin_retry;
|
2016-12-04 06:36:04 -07:00
|
|
|
if (!MACHINE_IS_LPAR || arch_vcpu_is_preempted(owner - 1))
|
|
|
|
smp_yield_cpu(owner - 1);
|
2008-01-26 06:11:28 -07:00
|
|
|
}
|
|
|
|
}
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
|
|
|
|
void arch_spin_lock_wait(arch_spinlock_t *lp)
|
|
|
|
{
|
|
|
|
if (test_cpu_flag(CIF_DEDICATED_CPU))
|
|
|
|
arch_spin_lock_queued(lp);
|
|
|
|
else
|
|
|
|
arch_spin_lock_classic(lp);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(arch_spin_lock_wait);
|
2008-01-26 06:11:28 -07:00
|
|
|
|
2009-12-02 12:01:25 -07:00
|
|
|
int arch_spin_trylock_retry(arch_spinlock_t *lp)
|
2005-07-27 11:44:57 -07:00
|
|
|
{
|
2016-11-28 07:50:48 -07:00
|
|
|
int cpu = SPINLOCK_LOCKVAL;
|
|
|
|
int owner, count;
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2015-01-14 09:52:33 -07:00
|
|
|
for (count = spin_retry; count > 0; count--) {
|
2017-02-10 04:34:49 -07:00
|
|
|
owner = READ_ONCE(lp->lock);
|
2015-01-14 09:52:33 -07:00
|
|
|
/* Try to get the lock if it is free. */
|
|
|
|
if (!owner) {
|
2016-11-28 07:50:48 -07:00
|
|
|
if (__atomic_cmpxchg_bool(&lp->lock, 0, cpu))
|
2015-01-14 09:52:33 -07:00
|
|
|
return 1;
|
2017-03-24 09:00:45 -07:00
|
|
|
}
|
2015-01-14 09:52:33 -07:00
|
|
|
}
|
2005-07-27 11:44:57 -07:00
|
|
|
return 0;
|
|
|
|
}
|
2009-12-02 12:01:25 -07:00
|
|
|
EXPORT_SYMBOL(arch_spin_trylock_retry);
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2017-03-24 09:32:23 -07:00
|
|
|
void arch_read_lock_wait(arch_rwlock_t *rw)
|
2005-07-27 11:44:57 -07:00
|
|
|
{
|
2017-03-24 09:32:23 -07:00
|
|
|
if (unlikely(in_interrupt())) {
|
|
|
|
while (READ_ONCE(rw->cnts) & 0x10000)
|
|
|
|
barrier();
|
|
|
|
return;
|
2005-07-27 11:44:57 -07:00
|
|
|
}
|
|
|
|
|
2017-03-24 09:32:23 -07:00
|
|
|
/* Remove this reader again to allow recursive read locking */
|
|
|
|
__atomic_add_const(-1, &rw->cnts);
|
|
|
|
/* Put the reader into the wait queue */
|
|
|
|
arch_spin_lock(&rw->wait);
|
|
|
|
/* Now add this reader to the count value again */
|
|
|
|
__atomic_add_const(1, &rw->cnts);
|
|
|
|
/* Loop until the writer is done */
|
|
|
|
while (READ_ONCE(rw->cnts) & 0x10000)
|
|
|
|
barrier();
|
|
|
|
arch_spin_unlock(&rw->wait);
|
2005-07-27 11:44:57 -07:00
|
|
|
}
|
2017-03-24 09:32:23 -07:00
|
|
|
EXPORT_SYMBOL(arch_read_lock_wait);
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2017-03-24 09:32:23 -07:00
|
|
|
void arch_write_lock_wait(arch_rwlock_t *rw)
|
2005-07-27 11:44:57 -07:00
|
|
|
{
|
2016-11-28 07:50:48 -07:00
|
|
|
int old;
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2017-03-24 09:32:23 -07:00
|
|
|
/* Add this CPU to the write waiters */
|
|
|
|
__atomic_add(0x20000, &rw->cnts);
|
2014-09-22 07:34:38 -07:00
|
|
|
|
2017-03-24 09:32:23 -07:00
|
|
|
/* Put the writer into the wait queue */
|
|
|
|
arch_spin_lock(&rw->wait);
|
2014-09-22 07:34:38 -07:00
|
|
|
|
|
|
|
while (1) {
|
2017-03-24 09:32:23 -07:00
|
|
|
old = READ_ONCE(rw->cnts);
|
|
|
|
if ((old & 0x1ffff) == 0 &&
|
|
|
|
__atomic_cmpxchg_bool(&rw->cnts, old, old | 0x10000))
|
|
|
|
/* Got the lock */
|
2014-09-22 07:34:38 -07:00
|
|
|
break;
|
2017-03-24 09:32:23 -07:00
|
|
|
barrier();
|
2014-09-22 07:34:38 -07:00
|
|
|
}
|
2005-07-27 11:44:57 -07:00
|
|
|
|
2017-03-24 09:32:23 -07:00
|
|
|
arch_spin_unlock(&rw->wait);
|
2005-07-27 11:44:57 -07:00
|
|
|
}
|
2017-03-24 09:32:23 -07:00
|
|
|
EXPORT_SYMBOL(arch_write_lock_wait);
|
2014-09-22 07:34:38 -07:00
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
void arch_spin_relax(arch_spinlock_t *lp)
|
2005-07-27 11:44:57 -07:00
|
|
|
{
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
int cpu;
|
2005-07-27 11:44:57 -07:00
|
|
|
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
cpu = READ_ONCE(lp->lock) & _Q_LOCK_CPU_MASK;
|
2014-09-19 05:29:31 -07:00
|
|
|
if (!cpu)
|
|
|
|
return;
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
if (MACHINE_IS_LPAR && !arch_vcpu_is_preempted(cpu - 1))
|
2014-09-19 05:29:31 -07:00
|
|
|
return;
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
smp_yield_cpu(cpu - 1);
|
2014-09-19 05:29:31 -07:00
|
|
|
}
|
s390/spinlock: introduce spinlock wait queueing
The queued spinlock code for s390 follows the principles of the common
code qspinlock implementation but with a few notable differences.
The format of the spinlock_t locking word differs, s390 needs to store
the logical CPU number of the lock holder in the spinlock_t to be able
to use the diagnose 9c directed yield hypervisor call.
The inline code sequences for spin_lock and spin_unlock are nice and
short. The inline portion of a spin_lock now typically looks like this:
lhi %r0,0 # 0 indicates an empty lock
l %r1,0x3a0 # CPU number + 1 from lowcore
cs %r0,%r1,<some_lock> # lock operation
jnz call_wait # on failure call wait function
locked:
...
call_wait:
la %r2,<some_lock>
brasl %r14,arch_spin_lock_wait
j locked
A spin_unlock is as simple as before:
lhi %r0,0
sth %r0,2(%r2) # unlock operation
After a CPU has queued itself it may not enable interrupts again for the
arch_spin_lock_flags() variant. The arch_spin_lock_wait_flags wait function
is removed.
To improve performance the code implements opportunistic lock stealing.
If the wait function finds a spinlock_t that indicates that the lock is
free but there are queued waiters, the CPU may steal the lock up to three
times without queueing itself. The lock stealing update the steal counter
in the lock word to prevent more than 3 steals. The counter is reset at
the time the CPU next in the queue successfully takes the lock.
While the queued spinlocks improve performance in a system with dedicated
CPUs, in a virtualized environment with continuously overcommitted CPUs
the queued spinlocks can have a negative effect on performance. This
is due to the fact that a queued CPU that is preempted by the hypervisor
will block the queue at some point even without holding the lock. With
the classic spinlock it does not matter if a CPU is preempted that waits
for the lock. Therefore use the queued spinlock code only if the system
runs with dedicated CPUs and fall back to classic spinlocks when running
with shared CPUs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
2017-03-24 09:25:02 -07:00
|
|
|
EXPORT_SYMBOL(arch_spin_relax);
|