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
|
--- a/vmmon/Makefile
+++ b/vmmon/Makefile
@@ -43,7 +43,11 @@
endif
+ifdef KVERSION
+VM_UNAME = $(KVERSION)
+else
VM_UNAME = $(shell uname -r)
+endif
# Header directory for the running kernel
ifdef LINUXINCLUDE
@@ -137,7 +141,6 @@
endif
# Add Spectre options when available
-CC_OPTS += $(call vm_check_gcc,-mindirect-branch=thunk -mindirect-branch-register,)
include $(SRCROOT)/Makefile.kernel
--- a/vmmon/linux/hostif.c
+++ b/vmmon/linux/hostif.c
@@ -47,6 +47,8 @@
#include <asm/asm.h>
#include <asm/io.h>
#include <asm/page.h>
+#include <asm/tlbflush.h>
+#include <asm/irq_vectors.h>
#include <asm/uaccess.h>
#include <linux/capability.h>
#include <linux/kthread.h>
@@ -54,6 +56,7 @@
#include <linux/hrtimer.h>
#include <linux/signal.h>
#include <linux/taskstats_kern.h> // For linux/sched/signal.h without version check
+#include <linux/eventfd.h>
#include "vmware.h"
#include "x86apic.h"
@@ -614,6 +617,15 @@
MutexUnlock(&fastClockMutex, callerID);
}
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 8, 0)
+static int crosspage_set_exec(pte_t *pte, unsigned long addr, void *data)
+{
+ struct page *p = data;
+
+ set_pte(pte, mk_pte(p, VM_PAGE_KERNEL_EXEC));
+ return 0;
+}
+#endif
/*
*----------------------------------------------------------------------
@@ -634,7 +646,29 @@
static void *
MapCrossPage(struct page *p) // IN:
{
+#if COMPAT_LINUX_VERSION_CHECK_LT(5, 8, 0)
return vmap(&p, 1, VM_MAP, VM_PAGE_KERNEL_EXEC);
+#else
+ void *addr;
+
+ addr = vmap(&p, 1, VM_MAP, VM_PAGE_KERNEL_EXEC);
+ if (!addr)
+ return NULL;
+
+ /* Starting with 5.8, vmap() always sets the NX bit, but the cross
+ * page needs to be executable. */
+ if (apply_to_page_range(current->mm, (unsigned long)addr, PAGE_SIZE,
+ crosspage_set_exec, p)) {
+ vunmap(addr);
+ return NULL;
+ }
+
+ preempt_disable();
+ __flush_tlb_all();
+ preempt_enable();
+
+ return addr;
+#endif
}
@@ -1739,7 +1773,10 @@
lockedPages += global_page_state(NR_PAGETABLE);
#endif
/* NR_SLAB_* moved from zone to node in 4.13. */
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 13, 0)
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(5, 9, 0)
+ lockedPages += global_node_page_state_pages(NR_SLAB_UNRECLAIMABLE_B);
+#elif LINUX_VERSION_CODE >= KERNEL_VERSION(4, 13, 0)
lockedPages += global_node_page_state(NR_SLAB_UNRECLAIMABLE);
#else
lockedPages += global_page_state(NR_SLAB_UNRECLAIMABLE);
@@ -2389,16 +2426,22 @@
static Bool
isVAReadable(VA r) // IN:
{
- mm_segment_t old_fs;
uint32 dummy;
int ret;
+#ifdef HAVE_GET_KERNEL_NOFAULT
+ ret = get_kernel_nofault(dummy, (void *)r);
+#else
+ {
+ mm_segment_t old_fs;
+
old_fs = get_fs();
set_fs(KERNEL_DS);
r = APICR_TO_ADDR(r, APICR_VERSION);
ret = HostIF_CopyFromUser(&dummy, r, sizeof dummy);
set_fs(old_fs);
-
+ }
+#endif
return ret == 0;
}
#endif
@@ -2553,7 +2596,6 @@
uint64 *args) // IN:
{
struct file *file;
- mm_segment_t old_fs;
int res;
int waitFD = args[0];
int timeoutms = args[2];
@@ -2566,9 +2610,6 @@
return MX_WAITERROR;
}
- old_fs = get_fs();
- set_fs(KERNEL_DS);
-
{
struct poll_wqueues table;
unsigned int mask;
@@ -2590,9 +2631,11 @@
* the code to happily deal with a pipe or an eventfd. We only care about
* reading no bytes (EAGAIN - non blocking fd) or sizeof(uint64).
*/
-
- res = file->f_op->read(file, (char *) &value, sizeof value, &file->f_pos);
-
+#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 14, 0)
+ res = kernel_read(file, file->f_pos, (char *)&value, sizeof value);
+#else
+ res = kernel_read(file, &value, sizeof value, &file->f_pos);
+#endif
if (res == sizeof value) {
res = MX_WAITNORMAL;
} else {
@@ -2601,7 +2644,6 @@
}
}
- set_fs(old_fs);
fput(file);
/*
@@ -2688,8 +2730,8 @@
int
HostIF_SemaphoreSignal(uint64 *args) // IN:
{
+ struct eventfd_ctx *eventfd;
struct file *file;
- mm_segment_t old_fs;
int res;
int signalFD = args[1];
uint64 value = 1; // make an eventfd happy should it be there
@@ -2699,22 +2741,32 @@
return MX_WAITERROR;
}
- old_fs = get_fs();
- set_fs(KERNEL_DS);
+ /*
+ * If it's eventfd, use specific eventfd interface as kernel writes
+ * to eventfd may not be allowed in kernel 5.10 and later.
+ */
+ eventfd = eventfd_ctx_fileget(file);
+ if (!IS_ERR(eventfd)) {
+ eventfd_signal(eventfd, 1);
+ fput(file);
+ return MX_WAITNORMAL;
+ }
/*
* Always write sizeof(uint64) bytes. This works fine for eventfd and
* pipes. The data written is formatted to make an eventfd happy should
* it be present.
*/
-
- res = file->f_op->write(file, (char *) &value, sizeof value, &file->f_pos);
+#if LINUX_VERSION_CODE < KERNEL_VERSION(4, 14, 0)
+ res = kernel_write(file, (char *)&value, sizeof value, file->f_pos);
+#else
+ res = kernel_write(file, &value, sizeof value, &file->f_pos);
+#endif
if (res == sizeof value) {
res = MX_WAITNORMAL;
}
- set_fs(old_fs);
fput(file);
/*
@@ -3261,12 +3313,9 @@
HostIFFastClockThread(void *unused) // IN:
{
int res;
- mm_segment_t oldFS;
unsigned int rate = 0;
unsigned int prevRate = 0;
- oldFS = get_fs();
- set_fs(KERNEL_DS);
allow_signal(SIGKILL);
while ((rate = linuxState.fastClockRate) > MIN_RATE) {
@@ -3289,8 +3338,6 @@
}
out:
- set_fs(oldFS);
-
/*
* Do not exit thread until we are told to do so.
*/
|