1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  23  */
  24 /*
  25  * Copyright 2004 Sun Microsystems, Inc.  All rights reserved.
  26  * Use is subject to license terms.
  27  */
  28 
  29         .file   "__swapFLAGS.s"
  30 
  31 #include "libm.h"
  32 
  33 /*
  34  * swap exception masks
  35  *
  36  * Put the complement of bits 5-0 of the argument into FPCW bits 5-0
  37  * and MXCSR bits 12-7, return the complement of the previous FPCW
  38  * bits 5-0.
  39  */
  40         ENTRY(__swapTE)         / di <-- NOT(desired xcptn_masks)
  41         subq    $8,%rsp
  42         fstcw   (%rsp)          / push current_cw on '86 stack
  43         movq    (%rsp),%rcx     / cx <-- current_cw
  44         movw    %cx,%ax         / ax <-- current_cw
  45         orw     $0x3f,%cx       / cx <-- current_cw, but masking all xcptns
  46         andw    $0x3f,%di       / make sure bits > B5 are all zero
  47         xorw    %di,%cx         / cx <-- present_cw, with new xcptn_masks
  48         movw    %cx,(%rsp)
  49         fldcw   (%rsp)          / load new cw 
  50         stmxcsr (%rsp)
  51         movq    (%rsp),%rcx
  52         orw     $0x1f80,%cx     / cx <-- current mxcsr, but masking all xcptns
  53         shlw    $7,%di
  54         xorw    %di,%cx         / cx <-- present mxcsr, with new xcptn_masks
  55         movq    %rcx,(%rsp)
  56         ldmxcsr (%rsp)
  57         andq    $0x3f,%rax      / al[5..0] <-- former xcptn_masks
  58         xorq    $0x3f,%rax      / al[5..0] <-- NOT(former xcptn_masks)
  59         addq    $8,%rsp
  60         ret
  61         .align  16
  62         SET_SIZE(__swapTE)
  63 
  64 /*
  65  * swap exception flags
  66  *
  67  * Put bits 5-0 of the argument into FPSW bits 5-0 and MXCSR bits 5-0,
  68  * return the "or" of the previous FPSW bits 5-0 and MXCSR bits 5-0.
  69  */
  70         ENTRY(__swapEX)
  71         fstsw   %ax             / ax = sw
  72         andq    $0x3f,%rdi
  73         jnz     .L1
  74                                 / input ex=0, clear all exception
  75         fnclex  
  76         subq    $8,%rsp
  77         stmxcsr (%rsp)
  78         movq    (%rsp),%rcx
  79         orw     %cx,%ax
  80         andw    $0xffc0,%cx
  81         movq    %rcx,(%rsp)
  82         ldmxcsr (%rsp)
  83         andq    $0x3f,%rax
  84         addq    $8,%rsp
  85         ret
  86 .L1:
  87                                 / input ex !=0, use fnstenv and fldenv
  88         subq    $32,%rsp        / only needed 28
  89         fnstenv (%rsp)
  90         movw    %ax,%dx
  91         andw    $0xffc0,%dx
  92         orw     %cx,%dx
  93         movw    %dx,4(%rsp)     / replace old sw by new one
  94         fldenv  (%rsp)
  95         stmxcsr (%rsp)
  96         movq    (%rsp),%rdx
  97         orw     %dx,%ax
  98         andw    $0xffc0,%dx
  99         orw     %cx,%dx
 100         movq    %rdx,(%rsp)
 101         ldmxcsr (%rsp)
 102         andq    $0x3f,%rax
 103         addq    $32,%rsp
 104         ret
 105         .align  16
 106         SET_SIZE(__swapEX)
 107 
 108 /*
 109  * swap rounding precision
 110  *
 111  * Put bits 1-0 of the argument into FPCW bits 9-8, return the
 112  * previous FPCW bits 9-8.
 113  */
 114         ENTRY(__swapRP)
 115         subq    $8,%rsp
 116         fstcw   (%rsp)
 117         movw    (%rsp),%ax
 118         movw    %ax,%cx
 119         andw    $0xfcff,%cx
 120         andq    $0x3,%rdi
 121         shlw    $8,%di
 122         orw     %di,%cx
 123         movq    %rcx,(%rsp)
 124         fldcw   (%rsp)
 125         shrw    $8,%ax
 126         andq    $0x3,%rax
 127         addq    $8,%rsp
 128         ret
 129         .align  16
 130         SET_SIZE(__swapRP)
 131 
 132 /*
 133  * swap rounding direction
 134  *
 135  * Put bits 1-0 of the argument into FPCW bits 11-10 and MXCSR
 136  * bits 14-13, return the previous FPCW bits 11-10.
 137  */
 138         ENTRY(__swapRD)
 139         subq    $8,%rsp
 140         fstcw   (%rsp)
 141         movw    (%rsp),%ax
 142         movw    %ax,%cx
 143         andw    $0xf3ff,%cx
 144         andq    $0x3,%rdi
 145         shlw    $10,%di
 146         orw     %di,%cx
 147         movq    %rcx,(%rsp)
 148         fldcw   (%rsp)
 149         stmxcsr (%rsp)
 150         movq    (%rsp),%rcx
 151         andw    $0x9fff,%cx
 152         shlw    $3,%di
 153         orw     %di,%cx
 154         movq    %rcx,(%rsp)
 155         ldmxcsr (%rsp)
 156         shrw    $10,%ax
 157         andq    $0x3,%rax
 158         addq    $8,%rsp
 159         ret
 160         .align  16
 161         SET_SIZE(__swapRD)