/* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License, Version 1.0 only * (the "License"). You may not use this file except in compliance * with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2004 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. */ #pragma ident "%Z%%M% %I% %E% SMI" / / In-line functions for i86 kernels. / / / return current thread pointer / / NOTE: the "0x10" should be replaced by the computed value of the / offset of "cpu_thread" from the beginning of the struct cpu. / Including "assym.h" does not work, however, since that stuff / is PSM-specific and is only visible to the 'unix' build anyway. / Same with current cpu pointer, where "0xc" should be replaced / by the computed value of the offset of "cpu_self". / Ugh -- what a disaster. / .inline threadp,0 movl %gs:0x10, %eax .end / / return current cpu pointer / .inline curcpup,0 movl %gs:0xc, %eax .end / / return caller / .inline caller,0 movl 4(%ebp), %eax .end / / return value of cr3 register / .inline getcr3,0 movl %cr3, %eax .end / / reload cr3 register with its current value / .inline reload_cr3,0 movl %cr3, %eax movl %eax, %cr3 .end /* * Put a new value into cr3 (page table base register * void setcr3(void *value) */ .inline setcr3,4 movl (%esp), %eax movl %eax, %cr3 .end / / invalidate and flush cache. / .inline cache_bug,0 wbinvd .end / / convert ipl to spl. This is the identity function for i86 / .inline ipltospl,0 movl (%esp), %eax .end / / enable interrupts / .inline sti,0 sti .end / / disable interrupts / .inline cli,0 cli .end / / find the low order bit in a word / .inline lowbit,4 movl $-1, %eax bsfl (%esp), %eax incl %eax .end / / find the high order bit in a word / .inline highbit,4 movl $-1, %eax bsrl (%esp), %eax incl %eax .end / / disable interrupts and return value describing if interrupts were enabled / .inline clear_int_flag,0 pushfl cli popl %eax .end .inline intr_clear,0 pushfl cli popl %eax .end / / restore interrupt enable flag to value returned from 'clear_int_flag' above / .inline restore_int_flag,4 pushl (%esp) popfl .end .inline intr_restore,4 pushl (%esp) popfl .end / / in and out / .inline inb,4 movl (%esp), %edx xorl %eax, %eax inb (%dx) .end .inline inw,4 movl (%esp), %edx xorl %eax, %eax inw (%dx) .end .inline inl,4 movl (%esp), %edx xorl %eax, %eax inl (%dx) .end .inline outb,8 movl (%esp), %edx movl 4(%esp), %eax outb (%dx) .end .inline outw,8 movl (%esp), %edx movl 4(%esp), %eax outw (%dx) .end .inline outl,8 movl (%esp), %edx movl 4(%esp), %eax outl (%dx) .end / / Networking byte order functions (too bad, Intel has the wrong byte order) / .inline htonl,4 movl (%esp), %eax bswap %eax .end .inline ntohl,4 movl (%esp), %eax bswap %eax .end .inline htons,4 movl (%esp), %eax bswap %eax shrl $16, %eax .end .inline ntohs,4 movl (%esp), %eax bswap %eax shrl $16, %eax .end /* * multiply two long numbers and yield a u_lonlong_t result * Provided to manipulate hrtime_t values. */ .inline mul32, 8 movl 4(%esp), %eax movl (%esp), %ecx mull %ecx .end /* * Unlock hres_lock and increment the count value. (See clock.h) */ .inline unlock_hres_lock, 0 lock incl hres_lock .end .inline atomic_orb,8 movl (%esp), %eax movl 4(%esp), %edx lock orb %dl,(%eax) .end .inline atomic_andb,8 movl (%esp), %eax movl 4(%esp), %edx lock andb %dl,(%eax) .end /* * atomic inc/dec operations. * void atomic_inc16(uint16_t *addr) { ++*addr; } * void atomic_dec16(uint16_t *addr) { --*addr; } */ .inline atomic_inc16,4 movl (%esp), %eax lock incw (%eax) .end .inline atomic_dec16,4 movl (%esp), %eax lock decw (%eax) .end /* * Invalidate TLB translation to 1 page. * void mmu_tlbflush_entry(void *addr) */ .inline mmu_tlbflush_entry,4 movl (%esp), %eax invlpg (%eax) .end /* * Read Time Stamp Counter * uint64_t tsc_read(); * * usage: * uint64_t cycles = tsc_read(); * * PPro & PII take no less than 34 cycles to execute rdtsc + stores. * Pentium takes about 16 cycles. */ .inline tsc_read, 0 rdtsc / %edx:%eax = RDTSC .end /* * void tsc_clear(register) * Clear the local Time Stamp Counter via write-MSR instruction. * Note that while this is a 64-bit write, the top 32-bits are * ignored, so it isn't massively useful to write anything other * than zero. */ .inline tsc_reset, 4 movl (%esp), %ecx xorl %eax, %eax movl %eax, %edx wrmsr ret .end /* * Call the pause instruction. To the Pentium 4 Xeon processor, it acts as * a hint that the code sequence is a busy spin-wait loop. Without a pause * instruction in these loops, the P4 Xeon processor may suffer a severe * penalty when exiting the loop because the processor detects a possible * memory violation. Inserting the pause instruction significantly reduces * the likelihood of a memory order violation, improving performance. * The pause instruction is a NOP on all other IA-32 processors. */ .inline ht_pause, 0 rep / our compiler doesn't support "pause" yet, nop / so we're using "F3 90" opcode directly .end /* * Call the halt instruction. This will put the CPU to sleep until * it is again awoken via an interrupt. * This function should be called with interrupts already disabled * for the CPU. * Note that "sti" will only enable interrupts at the end of the * subsequent instruction...in this case: "hlt". */ .inline i86_halt,0 sti hlt .end