s390/preempt: move preempt_count to the lowcore
Convert s390 to use a field in the struct lowcore for the CPU
preemption count. It is a bit cheaper to access a lowcore field
compared to a thread_info variable and it removes the depencency
on a task related structure.
bloat-o-meter on the vmlinux image for the default configuration
(CONFIG_PREEMPT_NONE=y) reports a small reduction in text size:
add/remove: 0/0 grow/shrink: 18/578 up/down: 228/-5448 (-5220)
A larger improvement is achieved with the default configuration
but with CONFIG_PREEMPT=y and CONFIG_DEBUG_PREEMPT=n:
add/remove: 2/6 grow/shrink: 59/4477 up/down: 1618/-228762 (-227144)
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
diff --git a/arch/s390/include/asm/lowcore.h b/arch/s390/include/asm/lowcore.h
index 7b93b78..1df1e62 100644
--- a/arch/s390/include/asm/lowcore.h
+++ b/arch/s390/include/asm/lowcore.h
@@ -126,7 +126,8 @@ struct lowcore {
__u64 percpu_offset; /* 0x0378 */
__u64 vdso_per_cpu_data; /* 0x0380 */
__u64 machine_flags; /* 0x0388 */
- __u8 pad_0x0390[0x0398-0x0390]; /* 0x0390 */
+ __u32 preempt_count; /* 0x0390 */
+ __u8 pad_0x0394[0x0398-0x0394]; /* 0x0394 */
__u64 gmap; /* 0x0398 */
__u32 spinlock_lockval; /* 0x03a0 */
__u32 fpu_flags; /* 0x03a4 */
diff --git a/arch/s390/include/asm/preempt.h b/arch/s390/include/asm/preempt.h
new file mode 100644
index 0000000..b0776b2
--- /dev/null
+++ b/arch/s390/include/asm/preempt.h
@@ -0,0 +1,137 @@
+#ifndef __ASM_PREEMPT_H
+#define __ASM_PREEMPT_H
+
+#include <asm/current.h>
+#include <linux/thread_info.h>
+#include <asm/atomic_ops.h>
+
+#ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
+
+#define PREEMPT_ENABLED (0 + PREEMPT_NEED_RESCHED)
+
+static inline int preempt_count(void)
+{
+ return READ_ONCE(S390_lowcore.preempt_count) & ~PREEMPT_NEED_RESCHED;
+}
+
+static inline void preempt_count_set(int pc)
+{
+ int old, new;
+
+ do {
+ old = READ_ONCE(S390_lowcore.preempt_count);
+ new = (old & PREEMPT_NEED_RESCHED) |
+ (pc & ~PREEMPT_NEED_RESCHED);
+ } while (__atomic_cmpxchg(&S390_lowcore.preempt_count,
+ old, new) != old);
+}
+
+#define init_task_preempt_count(p) do { } while (0)
+
+#define init_idle_preempt_count(p, cpu) do { \
+ S390_lowcore.preempt_count = PREEMPT_ENABLED; \
+} while (0)
+
+static inline void set_preempt_need_resched(void)
+{
+ __atomic_and(~PREEMPT_NEED_RESCHED, &S390_lowcore.preempt_count);
+}
+
+static inline void clear_preempt_need_resched(void)
+{
+ __atomic_or(PREEMPT_NEED_RESCHED, &S390_lowcore.preempt_count);
+}
+
+static inline bool test_preempt_need_resched(void)
+{
+ return !(READ_ONCE(S390_lowcore.preempt_count) & PREEMPT_NEED_RESCHED);
+}
+
+static inline void __preempt_count_add(int val)
+{
+ if (__builtin_constant_p(val) && (val >= -128) && (val <= 127))
+ __atomic_add_const(val, &S390_lowcore.preempt_count);
+ else
+ __atomic_add(val, &S390_lowcore.preempt_count);
+}
+
+static inline void __preempt_count_sub(int val)
+{
+ __preempt_count_add(-val);
+}
+
+static inline bool __preempt_count_dec_and_test(void)
+{
+ return __atomic_add(-1, &S390_lowcore.preempt_count) == 1;
+}
+
+static inline bool should_resched(int preempt_offset)
+{
+ return unlikely(READ_ONCE(S390_lowcore.preempt_count) ==
+ preempt_offset);
+}
+
+#else /* CONFIG_HAVE_MARCH_Z196_FEATURES */
+
+#define PREEMPT_ENABLED (0)
+
+static inline int preempt_count(void)
+{
+ return READ_ONCE(S390_lowcore.preempt_count);
+}
+
+static inline void preempt_count_set(int pc)
+{
+ S390_lowcore.preempt_count = pc;
+}
+
+#define init_task_preempt_count(p) do { } while (0)
+
+#define init_idle_preempt_count(p, cpu) do { \
+ S390_lowcore.preempt_count = PREEMPT_ENABLED; \
+} while (0)
+
+static inline void set_preempt_need_resched(void)
+{
+}
+
+static inline void clear_preempt_need_resched(void)
+{
+}
+
+static inline bool test_preempt_need_resched(void)
+{
+ return false;
+}
+
+static inline void __preempt_count_add(int val)
+{
+ S390_lowcore.preempt_count += val;
+}
+
+static inline void __preempt_count_sub(int val)
+{
+ S390_lowcore.preempt_count -= val;
+}
+
+static inline bool __preempt_count_dec_and_test(void)
+{
+ return !--S390_lowcore.preempt_count && tif_need_resched();
+}
+
+static inline bool should_resched(int preempt_offset)
+{
+ return unlikely(preempt_count() == preempt_offset &&
+ tif_need_resched());
+}
+
+#endif /* CONFIG_HAVE_MARCH_Z196_FEATURES */
+
+#ifdef CONFIG_PREEMPT
+extern asmlinkage void preempt_schedule(void);
+#define __preempt_schedule() preempt_schedule()
+extern asmlinkage void preempt_schedule_notrace(void);
+#define __preempt_schedule_notrace() preempt_schedule_notrace()
+#endif /* CONFIG_PREEMPT */
+
+#endif /* __ASM_PREEMPT_H */
diff --git a/arch/s390/include/asm/thread_info.h b/arch/s390/include/asm/thread_info.h
index f15c039..9337941 100644
--- a/arch/s390/include/asm/thread_info.h
+++ b/arch/s390/include/asm/thread_info.h
@@ -34,7 +34,6 @@ struct thread_info {
unsigned long flags; /* low level flags */
unsigned long sys_call_table; /* System call table address */
unsigned int cpu; /* current CPU */
- int preempt_count; /* 0 => preemptable, <0 => BUG */
unsigned int system_call;
__u64 user_timer;
__u64 system_timer;
@@ -49,7 +48,6 @@ struct thread_info {
.task = &tsk, \
.flags = 0, \
.cpu = 0, \
- .preempt_count = INIT_PREEMPT_COUNT, \
}
#define init_thread_info (init_thread_union.thread_info)
diff --git a/arch/s390/kernel/asm-offsets.c b/arch/s390/kernel/asm-offsets.c
index ec16cec..6be10e490 100644
--- a/arch/s390/kernel/asm-offsets.c
+++ b/arch/s390/kernel/asm-offsets.c
@@ -43,7 +43,6 @@ int main(void)
OFFSET(__TI_flags, thread_info, flags);
OFFSET(__TI_sysc_table, thread_info, sys_call_table);
OFFSET(__TI_cpu, thread_info, cpu);
- OFFSET(__TI_precount, thread_info, preempt_count);
OFFSET(__TI_user_timer, thread_info, user_timer);
OFFSET(__TI_system_timer, thread_info, system_timer);
OFFSET(__TI_last_break, thread_info, last_break);
@@ -175,6 +174,7 @@ int main(void)
OFFSET(__LC_PERCPU_OFFSET, lowcore, percpu_offset);
OFFSET(__LC_VDSO_PER_CPU, lowcore, vdso_per_cpu_data);
OFFSET(__LC_MACHINE_FLAGS, lowcore, machine_flags);
+ OFFSET(__LC_PREEMPT_COUNT, lowcore, preempt_count);
OFFSET(__LC_GMAP, lowcore, gmap);
OFFSET(__LC_PASTE, lowcore, paste);
/* software defined ABI-relevant lowcore locations 0xe00 - 0xe20 */
diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c
index 39d03a7..9cd85ad 100644
--- a/arch/s390/kernel/early.c
+++ b/arch/s390/kernel/early.c
@@ -293,6 +293,7 @@ static noinline __init void setup_lowcore_early(void)
psw.addr = (unsigned long) s390_base_pgm_handler;
S390_lowcore.program_new_psw = psw;
s390_base_pgm_handler_fn = early_pgm_check_handler;
+ S390_lowcore.preempt_count = INIT_PREEMPT_COUNT;
}
static noinline __init void setup_facility_list(void)
diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S
index 49a3073..3f4a163 100644
--- a/arch/s390/kernel/entry.S
+++ b/arch/s390/kernel/entry.S
@@ -626,7 +626,7 @@
jo .Lio_work_user # yes -> do resched & signal
#ifdef CONFIG_PREEMPT
# check for preemptive scheduling
- icm %r0,15,__TI_precount(%r12)
+ icm %r0,15,__LC_PREEMPT_COUNT
jnz .Lio_restore # preemption is disabled
TSTMSK __TI_flags(%r12),_TIF_NEED_RESCHED
jno .Lio_restore
diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c
index 7f7ba5f2..268d6c1 100644
--- a/arch/s390/kernel/setup.c
+++ b/arch/s390/kernel/setup.c
@@ -333,6 +333,7 @@ static void __init setup_lowcore(void)
lc->thread_info = (unsigned long) &init_thread_union;
lc->lpp = LPP_MAGIC;
lc->machine_flags = S390_lowcore.machine_flags;
+ lc->preempt_count = S390_lowcore.preempt_count;
lc->stfl_fac_list = S390_lowcore.stfl_fac_list;
memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list,
MAX_FACILITY_BIT/8);
diff --git a/arch/s390/mm/fault.c b/arch/s390/mm/fault.c
index 661d9fe..d1faae5 100644
--- a/arch/s390/mm/fault.c
+++ b/arch/s390/mm/fault.c
@@ -733,6 +733,7 @@ static void pfault_interrupt(struct ext_code ext_code,
* return to userspace schedule() to block. */
__set_current_state(TASK_UNINTERRUPTIBLE);
set_tsk_need_resched(tsk);
+ set_preempt_need_resched();
}
}
out: