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
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
|
/* SPDX-License-Identifier: GPL-2.0 */
/*
* Compatibility mode system call entry point for x86-64.
*
* Copyright 2000-2002 Andi Kleen, SuSE Labs.
*/
#include <asm/asm-offsets.h>
#include <asm/current.h>
#include <asm/errno.h>
#include <asm/ia32_unistd.h>
#include <asm/thread_info.h>
#include <asm/segment.h>
#include <asm/irqflags.h>
#include <asm/asm.h>
#include <asm/smap.h>
#include <asm/nospec-branch.h>
#include <linux/linkage.h>
#include <linux/err.h>
#include "calling.h"
.section .entry.text, "ax"
/*
* 32-bit SYSENTER entry.
*
* 32-bit system calls through the vDSO's __kernel_vsyscall enter here
* on 64-bit kernels running on Intel CPUs.
*
* The SYSENTER instruction, in principle, should *only* occur in the
* vDSO. In practice, a small number of Android devices were shipped
* with a copy of Bionic that inlined a SYSENTER instruction. This
* never happened in any of Google's Bionic versions -- it only happened
* in a narrow range of Intel-provided versions.
*
* SYSENTER loads SS, RSP, CS, and RIP from previously programmed MSRs.
* IF and VM in RFLAGS are cleared (IOW: interrupts are off).
* SYSENTER does not save anything on the stack,
* and does not save old RIP (!!!), RSP, or RFLAGS.
*
* Arguments:
* eax system call number
* ebx arg1
* ecx arg2
* edx arg3
* esi arg4
* edi arg5
* ebp user stack
* 0(%ebp) arg6
*/
SYM_CODE_START(entry_SYSENTER_compat)
UNWIND_HINT_ENTRY
ENDBR
/* Interrupts are off on entry. */
swapgs
pushq %rax
SWITCH_TO_KERNEL_CR3 scratch_reg=%rax
popq %rax
movq PER_CPU_VAR(pcpu_hot + X86_top_of_stack), %rsp
/* Construct struct pt_regs on stack */
pushq $__USER_DS /* pt_regs->ss */
pushq $0 /* pt_regs->sp = 0 (placeholder) */
/*
* Push flags. This is nasty. First, interrupts are currently
* off, but we need pt_regs->flags to have IF set. Second, if TS
* was set in usermode, it's still set, and we're singlestepping
* through this code. do_SYSENTER_32() will fix up IF.
*/
pushfq /* pt_regs->flags (except IF = 0) */
pushq $__USER32_CS /* pt_regs->cs */
pushq $0 /* pt_regs->ip = 0 (placeholder) */
SYM_INNER_LABEL(entry_SYSENTER_compat_after_hwframe, SYM_L_GLOBAL)
/*
* User tracing code (ptrace or signal handlers) might assume that
* the saved RAX contains a 32-bit number when we're invoking a 32-bit
* syscall. Just in case the high bits are nonzero, zero-extend
* the syscall number. (This could almost certainly be deleted
* with no ill effects.)
*/
movl %eax, %eax
pushq %rax /* pt_regs->orig_ax */
PUSH_AND_CLEAR_REGS rax=$-ENOSYS
UNWIND_HINT_REGS
cld
IBRS_ENTER
UNTRAIN_RET
/*
* SYSENTER doesn't filter flags, so we need to clear NT and AC
* ourselves. To save a few cycles, we can check whether
* either was set instead of doing an unconditional popfq.
* This needs to happen before enabling interrupts so that
* we don't get preempted with NT set.
*
* If TF is set, we will single-step all the way to here -- do_debug
* will ignore all the traps. (Yes, this is slow, but so is
* single-stepping in general. This allows us to avoid having
* a more complicated code to handle the case where a user program
* forces us to single-step through the SYSENTER entry code.)
*
* NB.: .Lsysenter_fix_flags is a label with the code under it moved
* out-of-line as an optimization: NT is unlikely to be set in the
* majority of the cases and instead of polluting the I$ unnecessarily,
* we're keeping that code behind a branch which will predict as
* not-taken and therefore its instructions won't be fetched.
*/
testl $X86_EFLAGS_NT|X86_EFLAGS_AC|X86_EFLAGS_TF, EFLAGS(%rsp)
jnz .Lsysenter_fix_flags
.Lsysenter_flags_fixed:
movq %rsp, %rdi
call do_SYSENTER_32
jmp sysret32_from_system_call
.Lsysenter_fix_flags:
pushq $X86_EFLAGS_FIXED
popfq
jmp .Lsysenter_flags_fixed
SYM_INNER_LABEL(__end_entry_SYSENTER_compat, SYM_L_GLOBAL)
SYM_CODE_END(entry_SYSENTER_compat)
/*
* 32-bit SYSCALL entry.
*
* 32-bit system calls through the vDSO's __kernel_vsyscall enter here
* on 64-bit kernels running on AMD CPUs.
*
* The SYSCALL instruction, in principle, should *only* occur in the
* vDSO. In practice, it appears that this really is the case.
* As evidence:
*
* - The calling convention for SYSCALL has changed several times without
* anyone noticing.
*
* - Prior to the in-kernel X86_BUG_SYSRET_SS_ATTRS fixup, anything
* user task that did SYSCALL without immediately reloading SS
* would randomly crash.
*
* - Most programmers do not directly target AMD CPUs, and the 32-bit
* SYSCALL instruction does not exist on Intel CPUs. Even on AMD
* CPUs, Linux disables the SYSCALL instruction on 32-bit kernels
* because the SYSCALL instruction in legacy/native 32-bit mode (as
* opposed to compat mode) is sufficiently poorly designed as to be
* essentially unusable.
*
* 32-bit SYSCALL saves RIP to RCX, clears RFLAGS.RF, then saves
* RFLAGS to R11, then loads new SS, CS, and RIP from previously
* programmed MSRs. RFLAGS gets masked by a value from another MSR
* (so CLD and CLAC are not needed). SYSCALL does not save anything on
* the stack and does not change RSP.
*
* Note: RFLAGS saving+masking-with-MSR happens only in Long mode
* (in legacy 32-bit mode, IF, RF and VM bits are cleared and that's it).
* Don't get confused: RFLAGS saving+masking depends on Long Mode Active bit
* (EFER.LMA=1), NOT on bitness of userspace where SYSCALL executes
* or target CS descriptor's L bit (SYSCALL does not read segment descriptors).
*
* Arguments:
* eax system call number
* ecx return address
* ebx arg1
* ebp arg2 (note: not saved in the stack frame, should not be touched)
* edx arg3
* esi arg4
* edi arg5
* esp user stack
* 0(%esp) arg6
*/
SYM_CODE_START(entry_SYSCALL_compat)
UNWIND_HINT_ENTRY
ENDBR
/* Interrupts are off on entry. */
swapgs
/* Stash user ESP */
movl %esp, %r8d
/* Use %rsp as scratch reg. User ESP is stashed in r8 */
SWITCH_TO_KERNEL_CR3 scratch_reg=%rsp
/* Switch to the kernel stack */
movq PER_CPU_VAR(pcpu_hot + X86_top_of_stack), %rsp
SYM_INNER_LABEL(entry_SYSCALL_compat_safe_stack, SYM_L_GLOBAL)
ANNOTATE_NOENDBR
/* Construct struct pt_regs on stack */
pushq $__USER_DS /* pt_regs->ss */
pushq %r8 /* pt_regs->sp */
pushq %r11 /* pt_regs->flags */
pushq $__USER32_CS /* pt_regs->cs */
pushq %rcx /* pt_regs->ip */
SYM_INNER_LABEL(entry_SYSCALL_compat_after_hwframe, SYM_L_GLOBAL)
movl %eax, %eax /* discard orig_ax high bits */
pushq %rax /* pt_regs->orig_ax */
PUSH_AND_CLEAR_REGS rcx=%rbp rax=$-ENOSYS
UNWIND_HINT_REGS
IBRS_ENTER
UNTRAIN_RET
movq %rsp, %rdi
call do_fast_syscall_32
sysret32_from_system_call:
/* XEN PV guests always use IRET path */
ALTERNATIVE "testb %al, %al; jz swapgs_restore_regs_and_return_to_usermode", \
"jmp swapgs_restore_regs_and_return_to_usermode", X86_FEATURE_XENPV
/*
* Opportunistic SYSRET
*
* We are not going to return to userspace from the trampoline
* stack. So let's erase the thread stack right now.
*/
STACKLEAK_ERASE
IBRS_EXIT
movq RBX(%rsp), %rbx /* pt_regs->rbx */
movq RBP(%rsp), %rbp /* pt_regs->rbp */
movq EFLAGS(%rsp), %r11 /* pt_regs->flags (in r11) */
movq RIP(%rsp), %rcx /* pt_regs->ip (in rcx) */
addq $RAX, %rsp /* Skip r8-r15 */
popq %rax /* pt_regs->rax */
popq %rdx /* Skip pt_regs->cx */
popq %rdx /* pt_regs->dx */
popq %rsi /* pt_regs->si */
popq %rdi /* pt_regs->di */
/*
* USERGS_SYSRET32 does:
* GSBASE = user's GS base
* EIP = ECX
* RFLAGS = R11
* CS = __USER32_CS
* SS = __USER_DS
*
* ECX will not match pt_regs->cx, but we're returning to a vDSO
* trampoline that will fix up RCX, so this is okay.
*
* R12-R15 are callee-saved, so they contain whatever was in them
* when the system call started, which is already known to user
* code. We zero R8-R10 to avoid info leaks.
*/
movq RSP-ORIG_RAX(%rsp), %rsp
SYM_INNER_LABEL(entry_SYSRETL_compat_unsafe_stack, SYM_L_GLOBAL)
ANNOTATE_NOENDBR
/*
* The original userspace %rsp (RSP-ORIG_RAX(%rsp)) is stored
* on the process stack which is not mapped to userspace and
* not readable after we SWITCH_TO_USER_CR3. Delay the CR3
* switch until after after the last reference to the process
* stack.
*
* %r8/%r9 are zeroed before the sysret, thus safe to clobber.
*/
SWITCH_TO_USER_CR3_NOSTACK scratch_reg=%r8 scratch_reg2=%r9
xorl %r8d, %r8d
xorl %r9d, %r9d
xorl %r10d, %r10d
swapgs
sysretl
SYM_INNER_LABEL(entry_SYSRETL_compat_end, SYM_L_GLOBAL)
ANNOTATE_NOENDBR
int3
SYM_CODE_END(entry_SYSCALL_compat)
|
