1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
|
// Works on all IA-64 platforms: Linux, HP-UX, Win64i...
// On Win64i compile with ias.exe.
.text
.global OPENSSL_cpuid_setup#
.proc OPENSSL_cpuid_setup#
OPENSSL_cpuid_setup:
{ .mib; br.ret.sptk.many b0 };;
.endp OPENSSL_cpuid_setup#
.global OPENSSL_rdtsc#
.proc OPENSSL_rdtsc#
OPENSSL_rdtsc:
{ .mib; mov r8=ar.itc
br.ret.sptk.many b0 };;
.endp OPENSSL_rdtsc#
.global OPENSSL_atomic_add#
.proc OPENSSL_atomic_add#
.align 32
OPENSSL_atomic_add:
{ .mii; ld4 r2=[r32]
nop.i 0
nop.i 0 };;
.Lspin:
{ .mii; mov ar.ccv=r2
add r8=r2,r33
mov r3=r2 };;
{ .mmi; mf
cmpxchg4.acq r2=[r32],r8,ar.ccv
nop.i 0 };;
{ .mib; cmp.ne p6,p0=r2,r3
nop.i 0
(p6) br.dpnt .Lspin };;
{ .mib; nop.m 0
sxt4 r8=r8
br.ret.sptk.many b0 };;
.endp OPENSSL_atomic_add#
// Returns a structure comprising pointer to the top of stack of
// the caller and pointer beyond backing storage for the current
// register frame. The latter is required, because it might be
// insufficient to wipe backing storage for the current frame
// (as this procedure does), one might have to go further, toward
// higher addresses to reach for whole "retroactively" saved
// context...
.global OPENSSL_wipe_cpu#
.proc OPENSSL_wipe_cpu#
.align 32
OPENSSL_wipe_cpu:
.prologue
.fframe 0
.save ar.pfs,r2
.save ar.lc,r3
{ .mib; alloc r2=ar.pfs,0,96,0,96
mov r3=ar.lc
brp.loop.imp .L_wipe_top,.L_wipe_end-16
};;
{ .mii; mov r9=ar.bsp
mov r8=pr
mov ar.lc=96 };;
.body
{ .mii; add r9=96*8-8,r9
mov ar.ec=1 };;
// One can sweep double as fast, but then we can't quarantee
// that backing storage is wiped...
.L_wipe_top:
{ .mfi; st8 [r9]=r0,-8
mov f127=f0
mov r127=r0 }
{ .mfb; nop.m 0
nop.f 0
br.ctop.sptk .L_wipe_top };;
.L_wipe_end:
{ .mfi; mov r11=r0
mov f6=f0
mov r14=r0 }
{ .mfi; mov r15=r0
mov f7=f0
mov r16=r0 }
{ .mfi; mov r17=r0
mov f8=f0
mov r18=r0 }
{ .mfi; mov r19=r0
mov f9=f0
mov r20=r0 }
{ .mfi; mov r21=r0
mov f10=f0
mov r22=r0 }
{ .mfi; mov r23=r0
mov f11=f0
mov r24=r0 }
{ .mfi; mov r25=r0
mov f12=f0
mov r26=r0 }
{ .mfi; mov r27=r0
mov f13=f0
mov r28=r0 }
{ .mfi; mov r29=r0
mov f14=f0
mov r30=r0 }
{ .mfi; mov r31=r0
mov f15=f0
nop.i 0 }
{ .mfi; mov f16=f0 }
{ .mfi; mov f17=f0 }
{ .mfi; mov f18=f0 }
{ .mfi; mov f19=f0 }
{ .mfi; mov f20=f0 }
{ .mfi; mov f21=f0 }
{ .mfi; mov f22=f0 }
{ .mfi; mov f23=f0 }
{ .mfi; mov f24=f0 }
{ .mfi; mov f25=f0 }
{ .mfi; mov f26=f0 }
{ .mfi; mov f27=f0 }
{ .mfi; mov f28=f0 }
{ .mfi; mov f29=f0 }
{ .mfi; mov f30=f0 }
{ .mfi; add r9=96*8+8,r9
mov f31=f0
mov pr=r8,0x1ffff }
{ .mib; mov r8=sp
mov ar.lc=r3
br.ret.sptk b0 };;
.endp OPENSSL_wipe_cpu#
|