linux/arch/mips/kernel
Ying Huang 966a967116 smp: Avoid using two cache lines for struct call_single_data
struct call_single_data is used in IPIs to transfer information between
CPUs.  Its size is bigger than sizeof(unsigned long) and less than
cache line size.  Currently it is not allocated with any explicit alignment
requirements.  This makes it possible for allocated call_single_data to
cross two cache lines, which results in double the number of the cache lines
that need to be transferred among CPUs.

This can be fixed by requiring call_single_data to be aligned with the
size of call_single_data. Currently the size of call_single_data is the
power of 2.  If we add new fields to call_single_data, we may need to
add padding to make sure the size of new definition is the power of 2
as well.

Fortunately, this is enforced by GCC, which will report bad sizes.

To set alignment requirements of call_single_data to the size of
call_single_data, a struct definition and a typedef is used.

To test the effect of the patch, I used the vm-scalability multiple
thread swap test case (swap-w-seq-mt).  The test will create multiple
threads and each thread will eat memory until all RAM and part of swap
is used, so that huge number of IPIs are triggered when unmapping
memory.  In the test, the throughput of memory writing improves ~5%
compared with misaligned call_single_data, because of faster IPIs.

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Huang, Ying <ying.huang@intel.com>
[ Add call_single_data_t and align with size of call_single_data. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/87bmnqd6lz.fsf@yhuang-mobile.sh.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-08-29 15:14:38 +02:00
..
.gitignore
8250-platform.c
asm-offsets.c MIPS: IRQ Stack: Unwind IRQ stack onto task stack 2017-03-22 11:53:57 +01:00
binfmt_elfn32.c fs/binfmt: Convert obsolete cputime type to nsecs 2017-02-01 09:13:51 +01:00
binfmt_elfo32.c fs/binfmt: Convert obsolete cputime type to nsecs 2017-02-01 09:13:51 +01:00
bmips_5xxx_init.S MIPS: BMIPS: Make whitespacely correct. 2016-05-09 12:00:01 +02:00
bmips_vec.S MIPS: BMIPS: Add Whirlwind (BMIPS5200) initialization code 2016-05-09 12:00:01 +02:00
branch.c MIPS: Use pr_debug' for messages from __compute_return_epc_for_insn' 2017-06-29 02:42:27 +02:00
cacheinfo.c MIPS: Fix cacheinfo overflow 2017-02-13 18:57:34 +00:00
cevt-bcm1480.c MIPS: clockevent drivers: Set ->min_delta_ticks and ->max_delta_ticks 2017-04-14 13:11:16 -07:00
cevt-ds1287.c MIPS: clockevent drivers: Set ->min_delta_ticks and ->max_delta_ticks 2017-04-14 13:11:16 -07:00
cevt-gt641xx.c MIPS: clockevent drivers: Set ->min_delta_ticks and ->max_delta_ticks 2017-04-14 13:11:16 -07:00
cevt-r4k.c MIPS: cevt-r4k: Fix out-of-bounds array access 2017-04-10 13:31:12 +02:00
cevt-sb1250.c MIPS: clockevent drivers: Set ->min_delta_ticks and ->max_delta_ticks 2017-04-14 13:11:16 -07:00
cevt-txx9.c MIPS: clockevent drivers: Set ->min_delta_ticks and ->max_delta_ticks 2017-04-14 13:11:16 -07:00
cmpxchg.c MIPS: cmpxchg: Implement 1 byte & 2 byte cmpxchg() 2017-06-29 02:42:25 +02:00
cps-vec-ns16550.S MIPS: CPS: Early debug using an ns16550-compatible UART 2015-11-11 08:34:25 +01:00
cps-vec.S MIPS: CPS: Handle spurious VP starts more gracefully 2017-06-29 02:42:28 +02:00
cpu-bugs64.c MIPS: Fix printk continuations in cpu-bugs64.c 2017-01-24 18:30:34 +01:00
cpu-probe.c MIPS: MIPS16e2: Identify ASE presence 2017-07-05 14:06:44 +02:00
crash_dump.c Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
crash.c sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task_stack.h> 2017-03-02 08:42:36 +01:00
csrc-bcm1480.c clocksource: Use a plain u64 instead of cycle_t 2016-12-25 11:04:12 +01:00
csrc-ioasic.c clocksource: Use a plain u64 instead of cycle_t 2016-12-25 11:04:12 +01:00
csrc-r4k.c clocksource: Use a plain u64 instead of cycle_t 2016-12-25 11:04:12 +01:00
csrc-sb1250.c clocksource: Use a plain u64 instead of cycle_t 2016-12-25 11:04:12 +01:00
early_printk_8250.c
early_printk.c
elf.c MIPS: Avoid BUG warning in arch_check_elf 2017-04-12 15:22:13 +02:00
entry.S MIPS: Fix IRQ tracing & lockdep when rescheduling 2017-06-30 04:40:18 +02:00
ftrace.c MIPS: ftrace: fix init functions tracing 2017-06-08 14:51:59 +02:00
genex.S MIPS: IRQ Stack: Unwind IRQ stack onto task stack 2017-03-22 11:53:57 +01:00
gpio_txx9.c MIPS: txx9: switch to gpiochip_add_data() 2016-02-19 09:51:44 +01:00
head.S MIPS: head: Reorder instructions missing a delay slot 2017-06-27 23:35:21 +02:00
i8253.c
idle.c MIPS: Loongson: Add Loongson-3A R2 basic support 2016-05-13 14:02:14 +02:00
irq_txx9.c
irq-gt641xx.c
irq-msc01.c MIPS: MSC: Prevent out-of-bounds writes to MIPS SC ioremap'd region 2014-06-26 10:48:23 +01:00
irq-rm7000.c
irq.c MIPS: Introduce irq_stack 2017-01-03 16:34:34 +01:00
jump_label.c jump_label: Reorder hotplug lock and jump_label_lock 2017-05-26 10:10:45 +02:00
kgdb.c MIPS: KGDB: Use kernel context for sleeping threads 2017-04-12 22:29:22 +02:00
kprobes.c MIPS: tracing: disable uprobe/kprobe on compact branch instructions 2016-10-06 17:37:40 +02:00
linux32.c MIPS: Return directly in 32_mmap2() 2017-01-25 02:51:11 +01:00
machine_kexec.c MIPS: kexec: add debug info about the new kexec'ed image 2017-01-03 16:34:46 +01:00
Makefile MIPS: cmpxchg: Implement 1 byte & 2 byte xchg() 2017-06-29 02:42:25 +02:00
mcount.S MIPS: Export _mcount alongside its definition 2017-01-03 16:34:49 +01:00
mips_machine.c
mips-cm.c MIPS: CM: WARN on attempt to lock invalid VP, not BUG 2017-06-29 02:42:28 +02:00
mips-cpc.c MIPS: CPC: Provide default mips_cpc_default_phys_base to ignore CPC 2016-11-04 00:17:46 +01:00
mips-mt-fpaff.c sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task.h> 2017-03-02 08:42:35 +01:00
mips-mt.c MIPS: MT: Remove SMTC support 2014-05-24 00:07:01 +02:00
mips-r2-to-r6-emul.c Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus 2017-05-12 09:56:30 -07:00
module.c Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus 2017-07-15 10:59:54 -07:00
octeon_switch.S MIPS: Fix octeon FP context switch handling 2015-10-02 19:16:06 +02:00
perf_event_mipsxx.c MIPS: perf: add I6500 handling 2017-06-28 12:22:39 +02:00
perf_event.c sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task_stack.h> 2017-03-02 08:42:36 +01:00
pm-cps.c MIPS: pm-cps: Drop manual cache-line alignment of ready_count 2017-06-30 04:38:55 +02:00
pm.c MIPS: Fix watchpoint restoration 2016-05-09 12:00:02 +02:00
probes-common.h MIPS: tracing: disable uprobe/kprobe on compact branch instructions 2016-10-06 17:37:40 +02:00
proc.c MIPS: Fix MIPS I ISA /proc/cpuinfo reporting 2017-07-11 14:13:50 +02:00
process.c kthread: fix boot hang (regression) on MIPS/OpenRISC 2017-05-29 09:40:54 -07:00
prom.c MIPS: Use early_init_fdt_reserve_self to protect DTB location 2017-01-03 16:34:45 +01:00
ptrace32.c sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task_stack.h> 2017-03-02 08:42:36 +01:00
ptrace.c MIPS: Traced negative syscalls should return -ENOSYS 2017-07-11 14:13:06 +02:00
r4k_fpu.S MIPS: Use copy_s.fmt rather than copy_u.fmt 2016-05-09 12:00:04 +02:00
r4k_switch.S MIPS: Remove dead define of ST_OFF 2017-04-12 13:52:21 +02:00
r2300_fpu.S MIPS: Correct MIPS I FP sigcontext layout 2016-11-04 01:39:23 +01:00
r2300_switch.S MIPS: Export _save_fp & _save_msa alongside their definitions 2017-01-03 16:34:49 +01:00
r6000_fpu.S MIPS: Fix ISA I/II FP signal context offsets 2016-11-04 01:38:52 +01:00
relocate_kernel.S MIPS: Replace add and sub instructions in relocate_kernel.S with addiu 2015-08-03 15:26:30 +02:00
relocate.c MIPS: KASLR: Add missing header files 2017-04-12 17:02:04 +02:00
reset.c MIPS: Provide fallback reboot/poweroff/halt implementations 2015-04-01 17:21:58 +02:00
rtlx-cmp.c MIPS: APRP: Fix an issue when device_create() fails. 2014-08-01 17:30:35 +02:00
rtlx-mt.c MIPS: APRP: Fix an issue when device_create() fails. 2014-08-01 17:30:35 +02:00
rtlx.c sched/headers: Prepare to move signal wakeup & sigpending methods from <linux/sched.h> into <linux/sched/signal.h> 2017-03-02 08:42:32 +01:00
scall32-o32.S MIPS: Save static registers before sysmips 2017-06-28 12:22:40 +02:00
scall64-64.S MIPS: Save static registers before sysmips 2017-06-28 12:22:40 +02:00
scall64-n32.S MIPS: Save static registers before sysmips 2017-06-28 12:22:40 +02:00
scall64-o32.S MIPS: Save static registers before sysmips 2017-06-28 12:22:40 +02:00
segment.c MIPS: Print segment physical address when EU=1 2016-07-28 11:44:30 +02:00
setup.c MIPS: cmdline: Add support for 'memmap' parameter 2017-06-29 02:42:23 +02:00
signal32.c Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
signal_n32.c Replace <asm/uaccess.h> with <linux/uaccess.h> globally 2016-12-24 11:46:01 -08:00
signal_o32.c sched/headers: Prepare for new header dependencies before moving code to <linux/sched/signal.h> 2017-03-02 08:42:29 +01:00
signal-common.h MIPS: Save MSA extended context around signals 2015-09-03 12:07:59 +02:00
signal.c Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus 2016-08-06 09:13:11 -04:00
smp-bmips.c MIPS: Add missing include files 2017-03-08 10:38:06 +01:00
smp-cmp.c MIPS: Make smp CMP, CPS and MT use the new generic IPI functions 2016-02-25 10:56:58 +01:00
smp-cps.c MIPS: CPS: Handle cores not powering down more gracefully 2017-06-29 02:42:28 +02:00
smp-mt.c MIPS: smp-mt: Use CPU interrupt controller IPI IRQ domain support 2017-04-12 23:13:13 +02:00
smp-up.c MIPS: SMP: Remove plat_smp_ops cpus_done method. 2014-05-27 11:06:42 +02:00
smp.c smp: Avoid using two cache lines for struct call_single_data 2017-08-29 15:14:38 +02:00
spinlock_test.c MIPS: Declare mips_debugfs_dir in a header 2015-10-26 09:49:42 +01:00
spram.c MIPS: Add P6600 cases to CPU switch statements 2016-05-13 14:01:52 +02:00
stacktrace.c sched/headers: Prepare for new header dependencies before moving code to <linux/sched/task_stack.h> 2017-03-02 08:42:36 +01:00
sync-r4k.c MIPS: sync-r4k: Fix KERN_CONT fallout 2017-02-13 18:58:39 +00:00
syscall.c MIPS: Branch straight to ll in mips_atomic_set() 2017-06-28 12:22:40 +02:00
sysrq.c MIPS: Refactor dumping of TLB registers for r3k/r4k 2015-09-03 12:07:45 +02:00
time.c KVM: MIPS: Implement VZ support 2017-03-28 14:53:54 +01:00
topology.c
traps.c lib/extable.c: use bsearch() library function in search_extable() 2017-07-10 16:32:35 -07:00
unaligned.c MIPS: MIPS16e2: Subdecode extended LWSP/SWSP instructions 2017-07-05 14:07:20 +02:00
uprobes.c MIPS: uprobes: Remove __weak attribute from arch_uprobe_copy_ixol. 2017-01-03 16:34:48 +01:00
vdso.c userfaultfd: non-cooperative: add event for memory unmaps 2017-02-24 17:46:55 -08:00
vmlinux.lds.S debug: Fix __bug_table[] in arch linker scripts 2017-04-03 10:22:40 +02:00
vpe-cmp.c
vpe-mt.c MIPS: MT: Remove SMTC support 2014-05-24 00:07:01 +02:00
vpe.c module: use a structure to encapsulate layout. 2015-12-04 22:46:25 +01:00
watch.c MIPS: Add and use watch register field definitions 2016-05-13 14:02:13 +02:00