From ace9429bb58fd418f0c81d4c2835699bddf6bde6 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Thu, 11 Apr 2024 10:27:49 +0200 Subject: Adding upstream version 6.6.15. Signed-off-by: Daniel Baumann --- arch/x86/lib/delay.c | 231 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 231 insertions(+) create mode 100644 arch/x86/lib/delay.c (limited to 'arch/x86/lib/delay.c') diff --git a/arch/x86/lib/delay.c b/arch/x86/lib/delay.c new file mode 100644 index 0000000000..0e65d00e23 --- /dev/null +++ b/arch/x86/lib/delay.c @@ -0,0 +1,231 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Precise Delay Loops for i386 + * + * Copyright (C) 1993 Linus Torvalds + * Copyright (C) 1997 Martin Mares + * Copyright (C) 2008 Jiri Hladky + * + * The __delay function must _NOT_ be inlined as its execution time + * depends wildly on alignment on many x86 processors. The additional + * jump magic is needed to get the timing stable on all the CPU's + * we have to worry about. + */ + +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#ifdef CONFIG_SMP +# include +#endif + +static void delay_loop(u64 __loops); + +/* + * Calibration and selection of the delay mechanism happens only once + * during boot. + */ +static void (*delay_fn)(u64) __ro_after_init = delay_loop; +static void (*delay_halt_fn)(u64 start, u64 cycles) __ro_after_init; + +/* simple loop based delay: */ +static void delay_loop(u64 __loops) +{ + unsigned long loops = (unsigned long)__loops; + + asm volatile( + " test %0,%0 \n" + " jz 3f \n" + " jmp 1f \n" + + ".align 16 \n" + "1: jmp 2f \n" + + ".align 16 \n" + "2: dec %0 \n" + " jnz 2b \n" + "3: dec %0 \n" + + : "+a" (loops) + : + ); +} + +/* TSC based delay: */ +static void delay_tsc(u64 cycles) +{ + u64 bclock, now; + int cpu; + + preempt_disable(); + cpu = smp_processor_id(); + bclock = rdtsc_ordered(); + for (;;) { + now = rdtsc_ordered(); + if ((now - bclock) >= cycles) + break; + + /* Allow RT tasks to run */ + preempt_enable(); + rep_nop(); + preempt_disable(); + + /* + * It is possible that we moved to another CPU, and + * since TSC's are per-cpu we need to calculate + * that. The delay must guarantee that we wait "at + * least" the amount of time. Being moved to another + * CPU could make the wait longer but we just need to + * make sure we waited long enough. Rebalance the + * counter for this CPU. + */ + if (unlikely(cpu != smp_processor_id())) { + cycles -= (now - bclock); + cpu = smp_processor_id(); + bclock = rdtsc_ordered(); + } + } + preempt_enable(); +} + +/* + * On Intel the TPAUSE instruction waits until any of: + * 1) the TSC counter exceeds the value provided in EDX:EAX + * 2) global timeout in IA32_UMWAIT_CONTROL is exceeded + * 3) an external interrupt occurs + */ +static void delay_halt_tpause(u64 start, u64 cycles) +{ + u64 until = start + cycles; + u32 eax, edx; + + eax = lower_32_bits(until); + edx = upper_32_bits(until); + + /* + * Hard code the deeper (C0.2) sleep state because exit latency is + * small compared to the "microseconds" that usleep() will delay. + */ + __tpause(TPAUSE_C02_STATE, edx, eax); +} + +/* + * On some AMD platforms, MWAITX has a configurable 32-bit timer, that + * counts with TSC frequency. The input value is the number of TSC cycles + * to wait. MWAITX will also exit when the timer expires. + */ +static void delay_halt_mwaitx(u64 unused, u64 cycles) +{ + u64 delay; + + delay = min_t(u64, MWAITX_MAX_WAIT_CYCLES, cycles); + /* + * Use cpu_tss_rw as a cacheline-aligned, seldomly accessed per-cpu + * variable as the monitor target. + */ + __monitorx(raw_cpu_ptr(&cpu_tss_rw), 0, 0); + + /* + * AMD, like Intel, supports the EAX hint and EAX=0xf means, do not + * enter any deep C-state and we use it here in delay() to minimize + * wakeup latency. + */ + __mwaitx(MWAITX_DISABLE_CSTATES, delay, MWAITX_ECX_TIMER_ENABLE); +} + +/* + * Call a vendor specific function to delay for a given amount of time. Because + * these functions may return earlier than requested, check for actual elapsed + * time and call again until done. + */ +static void delay_halt(u64 __cycles) +{ + u64 start, end, cycles = __cycles; + + /* + * Timer value of 0 causes MWAITX to wait indefinitely, unless there + * is a store on the memory monitored by MONITORX. + */ + if (!cycles) + return; + + start = rdtsc_ordered(); + + for (;;) { + delay_halt_fn(start, cycles); + end = rdtsc_ordered(); + + if (cycles <= end - start) + break; + + cycles -= end - start; + start = end; + } +} + +void __init use_tsc_delay(void) +{ + if (delay_fn == delay_loop) + delay_fn = delay_tsc; +} + +void __init use_tpause_delay(void) +{ + delay_halt_fn = delay_halt_tpause; + delay_fn = delay_halt; +} + +void use_mwaitx_delay(void) +{ + delay_halt_fn = delay_halt_mwaitx; + delay_fn = delay_halt; +} + +int read_current_timer(unsigned long *timer_val) +{ + if (delay_fn == delay_tsc) { + *timer_val = rdtsc(); + return 0; + } + return -1; +} + +void __delay(unsigned long loops) +{ + delay_fn(loops); +} +EXPORT_SYMBOL(__delay); + +noinline void __const_udelay(unsigned long xloops) +{ + unsigned long lpj = this_cpu_read(cpu_info.loops_per_jiffy) ? : loops_per_jiffy; + int d0; + + xloops *= 4; + asm("mull %%edx" + :"=d" (xloops), "=&a" (d0) + :"1" (xloops), "0" (lpj * (HZ / 4))); + + __delay(++xloops); +} +EXPORT_SYMBOL(__const_udelay); + +void __udelay(unsigned long usecs) +{ + __const_udelay(usecs * 0x000010c7); /* 2**32 / 1000000 (rounded up) */ +} +EXPORT_SYMBOL(__udelay); + +void __ndelay(unsigned long nsecs) +{ + __const_udelay(nsecs * 0x00005); /* 2**32 / 1000000000 (rounded up) */ +} +EXPORT_SYMBOL(__ndelay); -- cgit v1.2.3