[Qemu-devel] [PULL 00/22] target-arm queue
[PULL 00/16] target-arm queue
1
target-arm queue for softfreeze:
1
Patches for rc1: nothing major, just some minor bugfixes and
2
This has all the big stuff I want to get in for softfreeze;
2
code cleanups.
3
there may be one or two smaller patches I pick up later in
4
the week.
5
3
6
thanks
7
-- PMM
4
-- PMM
8
5
9
The following changes since commit 0984a157c1c053394adbf64ed7de97f1aebe6a2d:
6
The following changes since commit f7e1914adad8885a5d4c70239ab90d901ed97e9f:
10
7
11
Merge remote-tracking branch 'remotes/jasowang/tags/net-pull-request' into staging (2019-03-05 09:33:20 +0000)
8
Merge remote-tracking branch 'remotes/alistair/tags/pull-riscv-to-apply-20201109' into staging (2020-11-10 09:24:56 +0000)
12
9
13
are available in the Git repository at:
10
are available in the Git repository at:
14
11
15
https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20190305
12
https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20201110
16
13
17
for you to fetch changes up to 566528f823d1a2e9eb2d7b2ed839547cb31bfc34:
14
for you to fetch changes up to b6c56c8a9a4064ea783f352f43c5df6231a110fa:
18
15
19
hw/arm/stellaris: Implement watchdog timer (2019-03-05 15:55:09 +0000)
16
target/arm/translate-neon.c: Handle VTBL UNDEF case before VFP access check (2020-11-10 11:03:48 +0000)
20
17
21
----------------------------------------------------------------
18
----------------------------------------------------------------
22
target-arm queue:
19
target-arm queue:
23
* Fix PC test for LDM (exception return)
20
* hw/arm/Kconfig: ARM_V7M depends on PTIMER
24
* Implement ARMv8.0-SB
21
* Minor coding style fixes
25
* Implement ARMv8.0-PredInv
22
* docs: add some notes on the sbsa-ref machine
26
* Implement ARMv8.4-CondM
23
* hw/arm/virt: Remove dependency on Cortex-A15 MPCore peripherals
27
* Implement ARMv8.5-CondM
24
* target/arm: Fix neon VTBL/VTBX for len > 1
28
* Implement ARMv8.5-FRINT
25
* hw/arm/armsse: Correct expansion MPC interrupt lines
29
* hw/arm/stellaris: Implement watchdog timer
26
* hw/misc/stm32f2xx_syscfg: Remove extraneous IRQ
30
* virt: support more than 255GB of RAM
27
* hw/arm/nseries: Remove invalid/unnecessary n8x0_uart_setup()
28
* hw/arm/musicpal: Don't connect two qemu_irqs directly to the same input
29
* hw/arm/musicpal: Only use qdev_get_gpio_in() when necessary
30
* hw/arm/nseries: Check return value from load_image_targphys()
31
* tests/qtest/npcm7xx_rng-test: count runs properly
32
* target/arm/translate-neon.c: Handle VTBL UNDEF case before VFP access check
31
33
32
----------------------------------------------------------------
34
----------------------------------------------------------------
33
Eric Auger (9):
35
Alex Bennée (1):
34
hw/arm/virt: Rename highmem IO regions
36
docs: add some notes on the sbsa-ref machine
35
hw/arm/virt: Split the memory map description
36
hw/boards: Add a MachineState parameter to kvm_type callback
37
kvm: add kvm_arm_get_max_vm_ipa_size
38
vl: Set machine ram_size, maxram_size and ram_slots earlier
39
hw/arm/virt: Dynamic memory map depending on RAM requirements
40
hw/arm/virt: Implement kvm_type function for 4.0 machine
41
hw/arm/virt: Check the VCPU PA range in TCG mode
42
hw/arm/virt: Bump the 255GB initial RAM limit
43
37
44
Michel Heily (1):
38
AlexChen (1):
45
hw/arm/stellaris: Implement watchdog timer
39
ssi: Fix bad printf format specifiers
46
40
47
Richard Henderson (11):
41
Andrew Jones (1):
48
target/arm: Fix PC test for LDM (exception return)
42
hw/arm/Kconfig: ARM_V7M depends on PTIMER
49
target/arm: Split out arm_sctlr
50
target/arm: Implement ARMv8.0-SB
51
target/arm: Implement ARMv8.0-PredInv
52
target/arm: Split helper_msr_i_pstate into 3
53
target/arm: Add set/clear_pstate_bits, share gen_ss_advance
54
target/arm: Rearrange disas_data_proc_reg
55
target/arm: Implement ARMv8.4-CondM
56
target/arm: Implement ARMv8.5-CondM
57
target/arm: Restructure handle_fp_1src_{single, double}
58
target/arm: Implement ARMv8.5-FRINT
59
43
60
Shameer Kolothum (1):
44
Havard Skinnemoen (1):
61
hw/arm/boot: introduce fdt_add_memory_node helper
45
tests/qtest/npcm7xx_rng-test: count runs properly
62
46
63
include/hw/arm/virt.h | 16 +-
47
Peter Maydell (2):
64
include/hw/boards.h | 5 +-
48
hw/arm/nseries: Check return value from load_image_targphys()
65
include/hw/watchdog/cmsdk-apb-watchdog.h | 8 +
49
target/arm/translate-neon.c: Handle VTBL UNDEF case before VFP access check
66
target/arm/cpu.h | 64 ++++-
67
target/arm/helper-a64.h | 3 +
68
target/arm/helper.h | 8 +-
69
target/arm/internals.h | 15 +
70
target/arm/kvm_arm.h | 13 +
71
target/arm/translate.h | 34 +++
72
accel/kvm/kvm-all.c | 2 +-
73
hw/arm/boot.c | 54 ++--
74
hw/arm/stellaris.c | 22 +-
75
hw/arm/virt-acpi-build.c | 10 +-
76
hw/arm/virt.c | 196 ++++++++++---
77
hw/ppc/mac_newworld.c | 3 +-
78
hw/ppc/mac_oldworld.c | 2 +-
79
hw/ppc/spapr.c | 2 +-
80
hw/watchdog/cmsdk-apb-watchdog.c | 74 ++++-
81
linux-user/elfload.c | 2 +
82
target/arm/cpu.c | 2 +
83
target/arm/cpu64.c | 6 +
84
target/arm/helper-a64.c | 30 ++
85
target/arm/helper.c | 63 +++-
86
target/arm/kvm.c | 10 +
87
target/arm/op_helper.c | 47 ---
88
target/arm/translate-a64.c | 478 +++++++++++++++++++++++--------
89
target/arm/translate.c | 35 ++-
90
target/arm/vfp_helper.c | 96 +++++++
91
vl.c | 6 +-
92
29 files changed, 1032 insertions(+), 274 deletions(-)
93
50
51
Philippe Mathieu-Daudé (6):
52
hw/arm/virt: Remove dependency on Cortex-A15 MPCore peripherals
53
hw/arm/armsse: Correct expansion MPC interrupt lines
54
hw/misc/stm32f2xx_syscfg: Remove extraneous IRQ
55
hw/arm/nseries: Remove invalid/unnecessary n8x0_uart_setup()
56
hw/arm/musicpal: Don't connect two qemu_irqs directly to the same input
57
hw/arm/musicpal: Only use qdev_get_gpio_in() when necessary
58
59
Richard Henderson (1):
60
target/arm: Fix neon VTBL/VTBX for len > 1
61
62
Xinhao Zhang (3):
63
target/arm: add spaces around operator
64
target/arm: Don't use '#' flag of printf format
65
target/arm: add space before the open parenthesis '('
66
67
docs/system/arm/sbsa.rst | 32 ++++++++++++++++++++++
68
docs/system/target-arm.rst | 1 +
69
include/hw/misc/stm32f2xx_syscfg.h | 2 --
70
target/arm/helper.h | 2 +-
71
hw/arm/armsse.c | 3 +-
72
hw/arm/musicpal.c | 40 +++++++++++++++++----------
73
hw/arm/nseries.c | 26 ++++++++----------
74
hw/arm/stm32f205_soc.c | 1 -
75
hw/misc/stm32f2xx_syscfg.c | 2 --
76
hw/ssi/imx_spi.c | 2 +-
77
hw/ssi/xilinx_spi.c | 2 +-
78
target/arm/arch_dump.c | 8 +++---
79
target/arm/arm-semi.c | 8 +++---
80
target/arm/helper.c | 2 +-
81
target/arm/op_helper.c | 23 +++++++++-------
82
target/arm/translate-a64.c | 4 +--
83
target/arm/translate.c | 2 +-
84
tests/qtest/npcm7xx_rng-test.c | 2 +-
85
hw/arm/Kconfig | 3 +-
86
target/arm/translate-neon.c.inc | 56 ++++++++++++++------------------------
87
20 files changed, 123 insertions(+), 98 deletions(-)
88
create mode 100644 docs/system/arm/sbsa.rst
89
diff view generated by jsdifflib
[Qemu-devel] [PULL 22/22] hw/arm/stellaris: Implement watchdog timer
[PULL 01/16] hw/arm/Kconfig: ARM_V7M depends on PTIMER
1
From: Michel Heily <michelheily@gmail.com>
1
From: Andrew Jones <drjones@redhat.com>
2
2
3
Implement the watchdog timer for the stellaris boards.
3
commit 32bd322a0134 ("hw/timer/armv7m_systick: Rewrite to use ptimers")
4
This device is a close variant of the CMSDK APB watchdog
4
changed armv7m_systick to build on ptimers. Make sure we have ptimers
5
device, so we can model it by subclassing that device and
5
in the build when building armv7m_systick.
6
tweaking the behaviour of some of its registers.
7
6
8
Signed-off-by: Michel Heily <michelheily@gmail.com>
7
Signed-off-by: Andrew Jones <drjones@redhat.com>
9
Reviewed-by: Peter Maydell <petser.maydell@linaro.org>
8
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
10
[PMM: rewrote commit message, fixed a few checkpatch nits,
9
Message-id: 20201104103343.30392-1-drjones@redhat.com
11
added comment giving the URL of the spec for the Stellaris
12
variant of the watchdog device]
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
---
11
---
15
include/hw/watchdog/cmsdk-apb-watchdog.h | 8 +++
12
hw/arm/Kconfig | 1 +
16
hw/arm/stellaris.c | 22 ++++++-
13
1 file changed, 1 insertion(+)
17
hw/watchdog/cmsdk-apb-watchdog.c | 74 +++++++++++++++++++++++-
18
3 files changed, 100 insertions(+), 4 deletions(-)
19
14
20
diff --git a/include/hw/watchdog/cmsdk-apb-watchdog.h b/include/hw/watchdog/cmsdk-apb-watchdog.h
15
diff --git a/hw/arm/Kconfig b/hw/arm/Kconfig
21
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
22
--- a/include/hw/watchdog/cmsdk-apb-watchdog.h
17
--- a/hw/arm/Kconfig
23
+++ b/include/hw/watchdog/cmsdk-apb-watchdog.h
18
+++ b/hw/arm/Kconfig
24
@@ -XXX,XX +XXX,XX @@
19
@@ -XXX,XX +XXX,XX @@ config ZYNQ
25
#define CMSDK_APB_WATCHDOG(obj) OBJECT_CHECK(CMSDKAPBWatchdog, (obj), \
20
26
TYPE_CMSDK_APB_WATCHDOG)
21
config ARM_V7M
27
22
bool
28
+/*
23
+ select PTIMER
29
+ * This shares the same struct (and cast macro) as the base
24
30
+ * cmsdk-apb-watchdog device.
25
config ALLWINNER_A10
31
+ */
26
bool
32
+#define TYPE_LUMINARY_WATCHDOG "luminary-watchdog"
33
+
34
typedef struct CMSDKAPBWatchdog {
35
/*< private >*/
36
SysBusDevice parent_obj;
37
@@ -XXX,XX +XXX,XX @@ typedef struct CMSDKAPBWatchdog {
38
MemoryRegion iomem;
39
qemu_irq wdogint;
40
uint32_t wdogclk_frq;
41
+ bool is_luminary;
42
struct ptimer_state *timer;
43
44
uint32_t control;
45
@@ -XXX,XX +XXX,XX @@ typedef struct CMSDKAPBWatchdog {
46
uint32_t itcr;
47
uint32_t itop;
48
uint32_t resetstatus;
49
+ const uint32_t *id;
50
} CMSDKAPBWatchdog;
51
52
#endif
53
diff --git a/hw/arm/stellaris.c b/hw/arm/stellaris.c
54
index XXXXXXX..XXXXXXX 100644
55
--- a/hw/arm/stellaris.c
56
+++ b/hw/arm/stellaris.c
57
@@ -XXX,XX +XXX,XX @@
58
#include "sysemu/sysemu.h"
59
#include "hw/arm/armv7m.h"
60
#include "hw/char/pl011.h"
61
+#include "hw/watchdog/cmsdk-apb-watchdog.h"
62
#include "hw/misc/unimp.h"
63
#include "cpu.h"
64
65
@@ -XXX,XX +XXX,XX @@ static void stellaris_init(MachineState *ms, stellaris_board_info *board)
66
* Stellaris LM3S6965 Microcontroller Data Sheet (rev I)
67
* http://www.ti.com/lit/ds/symlink/lm3s6965.pdf
68
*
69
- * 40000000 wdtimer (unimplemented)
70
+ * 40000000 wdtimer
71
* 40002000 i2c (unimplemented)
72
* 40004000 GPIO
73
* 40005000 GPIO
74
@@ -XXX,XX +XXX,XX @@ static void stellaris_init(MachineState *ms, stellaris_board_info *board)
75
stellaris_sys_init(0x400fe000, qdev_get_gpio_in(nvic, 28),
76
board, nd_table[0].macaddr.a);
77
78
+
79
+ if (board->dc1 & (1 << 3)) { /* watchdog present */
80
+ dev = qdev_create(NULL, TYPE_LUMINARY_WATCHDOG);
81
+
82
+ /* system_clock_scale is valid now */
83
+ uint32_t mainclk = NANOSECONDS_PER_SECOND / system_clock_scale;
84
+ qdev_prop_set_uint32(dev, "wdogclk-frq", mainclk);
85
+
86
+ qdev_init_nofail(dev);
87
+ sysbus_mmio_map(SYS_BUS_DEVICE(dev),
88
+ 0,
89
+ 0x40000000u);
90
+ sysbus_connect_irq(SYS_BUS_DEVICE(dev),
91
+ 0,
92
+ qdev_get_gpio_in(nvic, 18));
93
+ }
94
+
95
+
96
for (i = 0; i < 7; i++) {
97
if (board->dc4 & (1 << i)) {
98
gpio_dev[i] = sysbus_create_simple("pl061_luminary", gpio_addr[i],
99
@@ -XXX,XX +XXX,XX @@ static void stellaris_init(MachineState *ms, stellaris_board_info *board)
100
/* Add dummy regions for the devices we don't implement yet,
101
* so guest accesses don't cause unlogged crashes.
102
*/
103
- create_unimplemented_device("wdtimer", 0x40000000, 0x1000);
104
create_unimplemented_device("i2c-0", 0x40002000, 0x1000);
105
create_unimplemented_device("i2c-2", 0x40021000, 0x1000);
106
create_unimplemented_device("PWM", 0x40028000, 0x1000);
107
diff --git a/hw/watchdog/cmsdk-apb-watchdog.c b/hw/watchdog/cmsdk-apb-watchdog.c
108
index XXXXXXX..XXXXXXX 100644
109
--- a/hw/watchdog/cmsdk-apb-watchdog.c
110
+++ b/hw/watchdog/cmsdk-apb-watchdog.c
111
@@ -XXX,XX +XXX,XX @@
112
* System Design Kit (CMSDK) and documented in the Cortex-M System
113
* Design Kit Technical Reference Manual (ARM DDI0479C):
114
* https://developer.arm.com/products/system-design/system-design-kits/cortex-m-system-design-kit
115
+ *
116
+ * We also support the variant of this device found in the TI
117
+ * Stellaris/Luminary boards and documented in:
118
+ * http://www.ti.com/lit/ds/symlink/lm3s6965.pdf
119
*/
120
121
#include "qemu/osdep.h"
122
@@ -XXX,XX +XXX,XX @@ REG32(WDOGINTCLR, 0xc)
123
REG32(WDOGRIS, 0x10)
124
FIELD(WDOGRIS, INT, 0, 1)
125
REG32(WDOGMIS, 0x14)
126
+REG32(WDOGTEST, 0x418) /* only in Stellaris/Luminary version of the device */
127
REG32(WDOGLOCK, 0xc00)
128
#define WDOG_UNLOCK_VALUE 0x1ACCE551
129
REG32(WDOGITCR, 0xf00)
130
@@ -XXX,XX +XXX,XX @@ REG32(CID2, 0xff8)
131
REG32(CID3, 0xffc)
132
133
/* PID/CID values */
134
-static const int watchdog_id[] = {
135
+static const uint32_t cmsdk_apb_watchdog_id[] = {
136
0x04, 0x00, 0x00, 0x00, /* PID4..PID7 */
137
0x24, 0xb8, 0x1b, 0x00, /* PID0..PID3 */
138
0x0d, 0xf0, 0x05, 0xb1, /* CID0..CID3 */
139
};
140
141
+static const uint32_t luminary_watchdog_id[] = {
142
+ 0x00, 0x00, 0x00, 0x00, /* PID4..PID7 */
143
+ 0x05, 0x18, 0x18, 0x01, /* PID0..PID3 */
144
+ 0x0d, 0xf0, 0x05, 0xb1, /* CID0..CID3 */
145
+};
146
+
147
static bool cmsdk_apb_watchdog_intstatus(CMSDKAPBWatchdog *s)
148
{
149
/* Return masked interrupt status */
150
@@ -XXX,XX +XXX,XX @@ static void cmsdk_apb_watchdog_update(CMSDKAPBWatchdog *s)
151
bool wdogres;
152
153
if (s->itcr) {
154
+ /*
155
+ * Not checking that !s->is_luminary since s->itcr can't be written
156
+ * when s->is_luminary in the first place.
157
+ */
158
wdogint = s->itop & R_WDOGITOP_WDOGINT_MASK;
159
wdogres = s->itop & R_WDOGITOP_WDOGRES_MASK;
160
} else {
161
@@ -XXX,XX +XXX,XX @@ static uint64_t cmsdk_apb_watchdog_read(void *opaque, hwaddr offset,
162
r = s->lock;
163
break;
164
case A_WDOGITCR:
165
+ if (s->is_luminary) {
166
+ goto bad_offset;
167
+ }
168
r = s->itcr;
169
break;
170
case A_PID4 ... A_CID3:
171
- r = watchdog_id[(offset - A_PID4) / 4];
172
+ r = s->id[(offset - A_PID4) / 4];
173
break;
174
case A_WDOGINTCLR:
175
case A_WDOGITOP:
176
+ if (s->is_luminary) {
177
+ goto bad_offset;
178
+ }
179
qemu_log_mask(LOG_GUEST_ERROR,
180
"CMSDK APB watchdog read: read of WO offset %x\n",
181
(int)offset);
182
r = 0;
183
break;
184
+ case A_WDOGTEST:
185
+ if (!s->is_luminary) {
186
+ goto bad_offset;
187
+ }
188
+ qemu_log_mask(LOG_UNIMP,
189
+ "Luminary watchdog read: stall not implemented\n");
190
+ r = 0;
191
+ break;
192
default:
193
+bad_offset:
194
qemu_log_mask(LOG_GUEST_ERROR,
195
"CMSDK APB watchdog read: bad offset %x\n", (int)offset);
196
r = 0;
197
@@ -XXX,XX +XXX,XX @@ static void cmsdk_apb_watchdog_write(void *opaque, hwaddr offset,
198
ptimer_run(s->timer, 0);
199
break;
200
case A_WDOGCONTROL:
201
+ if (s->is_luminary && 0 != (R_WDOGCONTROL_INTEN_MASK & s->control)) {
202
+ /*
203
+ * The Luminary version of this device ignores writes to
204
+ * this register after the guest has enabled interrupts
205
+ * (so they can only be disabled again via reset).
206
+ */
207
+ break;
208
+ }
209
s->control = value & R_WDOGCONTROL_VALID_MASK;
210
cmsdk_apb_watchdog_update(s);
211
break;
212
@@ -XXX,XX +XXX,XX @@ static void cmsdk_apb_watchdog_write(void *opaque, hwaddr offset,
213
s->lock = (value != WDOG_UNLOCK_VALUE);
214
break;
215
case A_WDOGITCR:
216
+ if (s->is_luminary) {
217
+ goto bad_offset;
218
+ }
219
s->itcr = value & R_WDOGITCR_VALID_MASK;
220
cmsdk_apb_watchdog_update(s);
221
break;
222
case A_WDOGITOP:
223
+ if (s->is_luminary) {
224
+ goto bad_offset;
225
+ }
226
s->itop = value & R_WDOGITOP_VALID_MASK;
227
cmsdk_apb_watchdog_update(s);
228
break;
229
@@ -XXX,XX +XXX,XX @@ static void cmsdk_apb_watchdog_write(void *opaque, hwaddr offset,
230
"CMSDK APB watchdog write: write to RO offset 0x%x\n",
231
(int)offset);
232
break;
233
+ case A_WDOGTEST:
234
+ if (!s->is_luminary) {
235
+ goto bad_offset;
236
+ }
237
+ qemu_log_mask(LOG_UNIMP,
238
+ "Luminary watchdog write: stall not implemented\n");
239
+ break;
240
default:
241
+bad_offset:
242
qemu_log_mask(LOG_GUEST_ERROR,
243
"CMSDK APB watchdog write: bad offset 0x%x\n",
244
(int)offset);
245
@@ -XXX,XX +XXX,XX @@ static void cmsdk_apb_watchdog_init(Object *obj)
246
s, "cmsdk-apb-watchdog", 0x1000);
247
sysbus_init_mmio(sbd, &s->iomem);
248
sysbus_init_irq(sbd, &s->wdogint);
249
+
250
+ s->is_luminary = false;
251
+ s->id = cmsdk_apb_watchdog_id;
252
}
253
254
static void cmsdk_apb_watchdog_realize(DeviceState *dev, Error **errp)
255
@@ -XXX,XX +XXX,XX @@ static const TypeInfo cmsdk_apb_watchdog_info = {
256
.class_init = cmsdk_apb_watchdog_class_init,
257
};
258
259
+static void luminary_watchdog_init(Object *obj)
260
+{
261
+ CMSDKAPBWatchdog *s = CMSDK_APB_WATCHDOG(obj);
262
+
263
+ s->is_luminary = true;
264
+ s->id = luminary_watchdog_id;
265
+}
266
+
267
+static const TypeInfo luminary_watchdog_info = {
268
+ .name = TYPE_LUMINARY_WATCHDOG,
269
+ .parent = TYPE_CMSDK_APB_WATCHDOG,
270
+ .instance_init = luminary_watchdog_init
271
+};
272
+
273
static void cmsdk_apb_watchdog_register_types(void)
274
{
275
type_register_static(&cmsdk_apb_watchdog_info);
276
+ type_register_static(&luminary_watchdog_info);
277
}
278
279
type_init(cmsdk_apb_watchdog_register_types);
280
--
27
--
281
2.20.1
28
2.20.1
282
29
283
30
diff view generated by jsdifflib
[Qemu-devel] [PULL 07/22] target/arm: Rearrange disas_data_proc_reg
[PULL 02/16] ssi: Fix bad printf format specifiers
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: AlexChen <alex.chen@huawei.com>
2
2
3
This decoding more closely matches the ARMv8.4 Table C4-6,
3
We should use printf format specifier "%u" instead of "%d" for
4
Encoding table for Data Processing - Register Group.
4
argument of type "unsigned int".
5
5
6
In particular, op2 == 0 is now more than just Add/sub (with carry).
6
Reported-by: Euler Robot <euler.robot@huawei.com>
7
7
Signed-off-by: Alex Chen <alex.chen@huawei.com>
8
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
8
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
9
Message-id: 20190301200501.16533-7-richard.henderson@linaro.org
9
Message-id: 5FA280F5.8060902@huawei.com
10
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
---
11
---
13
target/arm/translate-a64.c | 98 ++++++++++++++++++++++----------------
12
hw/ssi/imx_spi.c | 2 +-
14
1 file changed, 57 insertions(+), 41 deletions(-)
13
hw/ssi/xilinx_spi.c | 2 +-
14
2 files changed, 2 insertions(+), 2 deletions(-)
15
15
16
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
16
diff --git a/hw/ssi/imx_spi.c b/hw/ssi/imx_spi.c
17
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/translate-a64.c
18
--- a/hw/ssi/imx_spi.c
19
+++ b/target/arm/translate-a64.c
19
+++ b/hw/ssi/imx_spi.c
20
@@ -XXX,XX +XXX,XX @@ static void disas_data_proc_3src(DisasContext *s, uint32_t insn)
20
@@ -XXX,XX +XXX,XX @@ static const char *imx_spi_reg_name(uint32_t reg)
21
}
21
case ECSPI_MSGDATA:
22
22
return "ECSPI_MSGDATA";
23
/* Add/subtract (with carry)
23
default:
24
- * 31 30 29 28 27 26 25 24 23 22 21 20 16 15 10 9 5 4 0
24
- sprintf(unknown, "%d ?", reg);
25
- * +--+--+--+------------------------+------+---------+------+-----+
25
+ sprintf(unknown, "%u ?", reg);
26
- * |sf|op| S| 1 1 0 1 0 0 0 0 | rm | opcode2 | Rn | Rd |
26
return unknown;
27
- * +--+--+--+------------------------+------+---------+------+-----+
28
- * [000000]
29
+ * 31 30 29 28 27 26 25 24 23 22 21 20 16 15 10 9 5 4 0
30
+ * +--+--+--+------------------------+------+-------------+------+-----+
31
+ * |sf|op| S| 1 1 0 1 0 0 0 0 | rm | 0 0 0 0 0 0 | Rn | Rd |
32
+ * +--+--+--+------------------------+------+-------------+------+-----+
33
*/
34
35
static void disas_adc_sbc(DisasContext *s, uint32_t insn)
36
@@ -XXX,XX +XXX,XX @@ static void disas_adc_sbc(DisasContext *s, uint32_t insn)
37
unsigned int sf, op, setflags, rm, rn, rd;
38
TCGv_i64 tcg_y, tcg_rn, tcg_rd;
39
40
- if (extract32(insn, 10, 6) != 0) {
41
- unallocated_encoding(s);
42
- return;
43
- }
44
-
45
sf = extract32(insn, 31, 1);
46
op = extract32(insn, 30, 1);
47
setflags = extract32(insn, 29, 1);
48
@@ -XXX,XX +XXX,XX @@ static void disas_data_proc_2src(DisasContext *s, uint32_t insn)
49
}
27
}
50
}
28
}
51
29
diff --git a/hw/ssi/xilinx_spi.c b/hw/ssi/xilinx_spi.c
52
-/* Data processing - register */
30
index XXXXXXX..XXXXXXX 100644
53
+/*
31
--- a/hw/ssi/xilinx_spi.c
54
+ * Data processing - register
32
+++ b/hw/ssi/xilinx_spi.c
55
+ * 31 30 29 28 25 21 20 16 10 0
33
@@ -XXX,XX +XXX,XX @@ static void xlx_spi_update_irq(XilinxSPI *s)
56
+ * +--+---+--+---+-------+-----+-------+-------+---------+
34
irq chain unless things really changed. */
57
+ * | |op0| |op1| 1 0 1 | op2 | | op3 | |
35
if (pending != s->irqline) {
58
+ * +--+---+--+---+-------+-----+-------+-------+---------+
36
s->irqline = pending;
59
+ */
37
- DB_PRINT("irq_change of state %d ISR:%x IER:%X\n",
60
static void disas_data_proc_reg(DisasContext *s, uint32_t insn)
38
+ DB_PRINT("irq_change of state %u ISR:%x IER:%X\n",
61
{
39
pending, s->regs[R_IPISR], s->regs[R_IPIER]);
62
- switch (extract32(insn, 24, 5)) {
40
qemu_set_irq(s->irq, pending);
63
- case 0x0a: /* Logical (shifted register) */
64
- disas_logic_reg(s, insn);
65
- break;
66
- case 0x0b: /* Add/subtract */
67
- if (insn & (1 << 21)) { /* (extended register) */
68
- disas_add_sub_ext_reg(s, insn);
69
+ int op0 = extract32(insn, 30, 1);
70
+ int op1 = extract32(insn, 28, 1);
71
+ int op2 = extract32(insn, 21, 4);
72
+ int op3 = extract32(insn, 10, 6);
73
+
74
+ if (!op1) {
75
+ if (op2 & 8) {
76
+ if (op2 & 1) {
77
+ /* Add/sub (extended register) */
78
+ disas_add_sub_ext_reg(s, insn);
79
+ } else {
80
+ /* Add/sub (shifted register) */
81
+ disas_add_sub_reg(s, insn);
82
+ }
83
} else {
84
- disas_add_sub_reg(s, insn);
85
+ /* Logical (shifted register) */
86
+ disas_logic_reg(s, insn);
87
}
88
- break;
89
- case 0x1b: /* Data-processing (3 source) */
90
- disas_data_proc_3src(s, insn);
91
- break;
92
- case 0x1a:
93
- switch (extract32(insn, 21, 3)) {
94
- case 0x0: /* Add/subtract (with carry) */
95
+ return;
96
+ }
97
+
98
+ switch (op2) {
99
+ case 0x0:
100
+ switch (op3) {
101
+ case 0x00: /* Add/subtract (with carry) */
102
disas_adc_sbc(s, insn);
103
break;
104
- case 0x2: /* Conditional compare */
105
- disas_cc(s, insn); /* both imm and reg forms */
106
- break;
107
- case 0x4: /* Conditional select */
108
- disas_cond_select(s, insn);
109
- break;
110
- case 0x6: /* Data-processing */
111
- if (insn & (1 << 30)) { /* (1 source) */
112
- disas_data_proc_1src(s, insn);
113
- } else { /* (2 source) */
114
- disas_data_proc_2src(s, insn);
115
- }
116
- break;
117
+
118
default:
119
- unallocated_encoding(s);
120
- break;
121
+ goto do_unallocated;
122
}
123
break;
124
+
125
+ case 0x2: /* Conditional compare */
126
+ disas_cc(s, insn); /* both imm and reg forms */
127
+ break;
128
+
129
+ case 0x4: /* Conditional select */
130
+ disas_cond_select(s, insn);
131
+ break;
132
+
133
+ case 0x6: /* Data-processing */
134
+ if (op0) { /* (1 source) */
135
+ disas_data_proc_1src(s, insn);
136
+ } else { /* (2 source) */
137
+ disas_data_proc_2src(s, insn);
138
+ }
139
+ break;
140
+ case 0x8 ... 0xf: /* (3 source) */
141
+ disas_data_proc_3src(s, insn);
142
+ break;
143
+
144
default:
145
+ do_unallocated:
146
unallocated_encoding(s);
147
break;
148
}
41
}
149
--
42
--
150
2.20.1
43
2.20.1
151
44
152
45
diff view generated by jsdifflib
[Qemu-devel] [PULL 04/22] target/arm: Implement ARMv8.0-PredInv
[PULL 03/16] target/arm: add spaces around operator
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Xinhao Zhang <zhangxinhao1@huawei.com>
2
2
3
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
3
Fix code style. Operator needs spaces both sides.
4
Message-id: 20190301200501.16533-4-richard.henderson@linaro.org
4
5
Signed-off-by: Xinhao Zhang <zhangxinhao1@huawei.com>
6
Signed-off-by: Kai Deng <dengkai1@huawei.com>
7
Message-id: 20201103114529.638233-1-zhangxinhao1@huawei.com
5
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7
---
10
---
8
target/arm/cpu.h | 13 ++++++++++-
11
target/arm/arch_dump.c | 8 ++++----
9
target/arm/cpu.c | 1 +
12
target/arm/arm-semi.c | 8 ++++----
10
target/arm/cpu64.c | 2 ++
13
target/arm/helper.c | 2 +-
11
target/arm/helper.c | 55 +++++++++++++++++++++++++++++++++++++++++++++
14
3 files changed, 9 insertions(+), 9 deletions(-)
12
4 files changed, 70 insertions(+), 1 deletion(-)
13
15
14
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
16
diff --git a/target/arm/arch_dump.c b/target/arm/arch_dump.c
15
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
16
--- a/target/arm/cpu.h
18
--- a/target/arm/arch_dump.c
17
+++ b/target/arm/cpu.h
19
+++ b/target/arm/arch_dump.c
18
@@ -XXX,XX +XXX,XX @@ void pmu_init(ARMCPU *cpu);
20
@@ -XXX,XX +XXX,XX @@ static int aarch64_write_elf64_prfpreg(WriteCoreDumpFunction f,
19
#define SCTLR_R (1U << 9) /* up to v6; RAZ in v7 */
21
20
#define SCTLR_UMA (1U << 9) /* v8 onward, AArch64 only */
22
for (i = 0; i < 32; ++i) {
21
#define SCTLR_F (1U << 10) /* up to v6 */
23
uint64_t *q = aa64_vfp_qreg(env, i);
22
-#define SCTLR_SW (1U << 10) /* v7, RES0 in v8 */
24
- note.vfp.vregs[2*i + 0] = cpu_to_dump64(s, q[0]);
23
+#define SCTLR_SW (1U << 10) /* v7 */
25
- note.vfp.vregs[2*i + 1] = cpu_to_dump64(s, q[1]);
24
+#define SCTLR_EnRCTX (1U << 10) /* in v8.0-PredInv */
26
+ note.vfp.vregs[2 * i + 0] = cpu_to_dump64(s, q[0]);
25
#define SCTLR_Z (1U << 11) /* in v7, RES1 in v8 */
27
+ note.vfp.vregs[2 * i + 1] = cpu_to_dump64(s, q[1]);
26
#define SCTLR_EOS (1U << 11) /* v8.5-ExS */
28
}
27
#define SCTLR_I (1U << 12)
29
28
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa32_sb(const ARMISARegisters *id)
30
if (s->dump_info.d_endian == ELFDATA2MSB) {
29
return FIELD_EX32(id->id_isar6, ID_ISAR6, SB) != 0;
31
@@ -XXX,XX +XXX,XX @@ static int aarch64_write_elf64_prfpreg(WriteCoreDumpFunction f,
30
}
32
*/
31
33
for (i = 0; i < 32; ++i) {
32
+static inline bool isar_feature_aa32_predinv(const ARMISARegisters *id)
34
uint64_t tmp = note.vfp.vregs[2*i];
33
+{
35
- note.vfp.vregs[2*i] = note.vfp.vregs[2*i+1];
34
+ return FIELD_EX32(id->id_isar6, ID_ISAR6, SPECRES) != 0;
36
- note.vfp.vregs[2*i+1] = tmp;
35
+}
37
+ note.vfp.vregs[2 * i] = note.vfp.vregs[2 * i + 1];
36
+
38
+ note.vfp.vregs[2 * i + 1] = tmp;
37
static inline bool isar_feature_aa32_fp16_arith(const ARMISARegisters *id)
39
}
38
{
40
}
39
/*
41
40
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_sb(const ARMISARegisters *id)
42
diff --git a/target/arm/arm-semi.c b/target/arm/arm-semi.c
41
return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, SB) != 0;
42
}
43
44
+static inline bool isar_feature_aa64_predinv(const ARMISARegisters *id)
45
+{
46
+ return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, SPECRES) != 0;
47
+}
48
+
49
static inline bool isar_feature_aa64_fp16(const ARMISARegisters *id)
50
{
51
/* We always set the AdvSIMD and FP fields identically wrt FP16. */
52
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
53
index XXXXXXX..XXXXXXX 100644
43
index XXXXXXX..XXXXXXX 100644
54
--- a/target/arm/cpu.c
44
--- a/target/arm/arm-semi.c
55
+++ b/target/arm/cpu.c
45
+++ b/target/arm/arm-semi.c
56
@@ -XXX,XX +XXX,XX @@ static void arm_max_initfn(Object *obj)
46
@@ -XXX,XX +XXX,XX @@ target_ulong do_arm_semihosting(CPUARMState *env)
57
t = FIELD_DP32(t, ID_ISAR6, DP, 1);
47
if (use_gdb_syscalls()) {
58
t = FIELD_DP32(t, ID_ISAR6, FHM, 1);
48
arm_semi_open_guestfd = guestfd;
59
t = FIELD_DP32(t, ID_ISAR6, SB, 1);
49
ret = arm_gdb_syscall(cpu, arm_semi_open_cb, "open,%s,%x,1a4", arg0,
60
+ t = FIELD_DP32(t, ID_ISAR6, SPECRES, 1);
50
- (int)arg2+1, gdb_open_modeflags[arg1]);
61
cpu->isar.id_isar6 = t;
51
+ (int)arg2 + 1, gdb_open_modeflags[arg1]);
62
52
} else {
63
t = cpu->id_mmfr4;
53
ret = set_swi_errno(env, open(s, open_modeflags[arg1], 0644));
64
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
54
if (ret == (uint32_t)-1) {
65
index XXXXXXX..XXXXXXX 100644
55
@@ -XXX,XX +XXX,XX @@ target_ulong do_arm_semihosting(CPUARMState *env)
66
--- a/target/arm/cpu64.c
56
GET_ARG(1);
67
+++ b/target/arm/cpu64.c
57
if (use_gdb_syscalls()) {
68
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
58
ret = arm_gdb_syscall(cpu, arm_semi_cb, "unlink,%s",
69
t = FIELD_DP64(t, ID_AA64ISAR1, GPA, 1);
59
- arg0, (int)arg1+1);
70
t = FIELD_DP64(t, ID_AA64ISAR1, GPI, 0);
60
+ arg0, (int)arg1 + 1);
71
t = FIELD_DP64(t, ID_AA64ISAR1, SB, 1);
61
} else {
72
+ t = FIELD_DP64(t, ID_AA64ISAR1, SPECRES, 1);
62
s = lock_user_string(arg0);
73
cpu->isar.id_aa64isar1 = t;
63
if (!s) {
74
64
@@ -XXX,XX +XXX,XX @@ target_ulong do_arm_semihosting(CPUARMState *env)
75
t = cpu->isar.id_aa64pfr0;
65
GET_ARG(3);
76
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
66
if (use_gdb_syscalls()) {
77
u = FIELD_DP32(u, ID_ISAR6, DP, 1);
67
return arm_gdb_syscall(cpu, arm_semi_cb, "rename,%s,%s",
78
u = FIELD_DP32(u, ID_ISAR6, FHM, 1);
68
- arg0, (int)arg1+1, arg2, (int)arg3+1);
79
u = FIELD_DP32(u, ID_ISAR6, SB, 1);
69
+ arg0, (int)arg1 + 1, arg2, (int)arg3 + 1);
80
+ u = FIELD_DP32(u, ID_ISAR6, SPECRES, 1);
70
} else {
81
cpu->isar.id_isar6 = u;
71
char *s2;
82
72
s = lock_user_string(arg0);
83
/*
73
@@ -XXX,XX +XXX,XX @@ target_ulong do_arm_semihosting(CPUARMState *env)
74
GET_ARG(1);
75
if (use_gdb_syscalls()) {
76
return arm_gdb_syscall(cpu, arm_semi_cb, "system,%s",
77
- arg0, (int)arg1+1);
78
+ arg0, (int)arg1 + 1);
79
} else {
80
s = lock_user_string(arg0);
81
if (!s) {
84
diff --git a/target/arm/helper.c b/target/arm/helper.c
82
diff --git a/target/arm/helper.c b/target/arm/helper.c
85
index XXXXXXX..XXXXXXX 100644
83
index XXXXXXX..XXXXXXX 100644
86
--- a/target/arm/helper.c
84
--- a/target/arm/helper.c
87
+++ b/target/arm/helper.c
85
+++ b/target/arm/helper.c
88
@@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo pauth_reginfo[] = {
86
@@ -XXX,XX +XXX,XX @@ uint32_t HELPER(usad8)(uint32_t a, uint32_t b)
89
};
87
uint32_t sum;
90
#endif
88
sum = do_usad(a, b);
91
89
sum += do_usad(a >> 8, b >> 8);
92
+static CPAccessResult access_predinv(CPUARMState *env, const ARMCPRegInfo *ri,
90
- sum += do_usad(a >> 16, b >>16);
93
+ bool isread)
91
+ sum += do_usad(a >> 16, b >> 16);
94
+{
92
sum += do_usad(a >> 24, b >> 24);
95
+ int el = arm_current_el(env);
93
return sum;
96
+
97
+ if (el == 0) {
98
+ uint64_t sctlr = arm_sctlr(env, el);
99
+ if (!(sctlr & SCTLR_EnRCTX)) {
100
+ return CP_ACCESS_TRAP;
101
+ }
102
+ } else if (el == 1) {
103
+ uint64_t hcr = arm_hcr_el2_eff(env);
104
+ if (hcr & HCR_NV) {
105
+ return CP_ACCESS_TRAP_EL2;
106
+ }
107
+ }
108
+ return CP_ACCESS_OK;
109
+}
110
+
111
+static const ARMCPRegInfo predinv_reginfo[] = {
112
+ { .name = "CFP_RCTX", .state = ARM_CP_STATE_AA64,
113
+ .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 3, .opc2 = 4,
114
+ .type = ARM_CP_NOP, .access = PL0_W, .accessfn = access_predinv },
115
+ { .name = "DVP_RCTX", .state = ARM_CP_STATE_AA64,
116
+ .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 3, .opc2 = 5,
117
+ .type = ARM_CP_NOP, .access = PL0_W, .accessfn = access_predinv },
118
+ { .name = "CPP_RCTX", .state = ARM_CP_STATE_AA64,
119
+ .opc0 = 1, .opc1 = 3, .crn = 7, .crm = 3, .opc2 = 7,
120
+ .type = ARM_CP_NOP, .access = PL0_W, .accessfn = access_predinv },
121
+ /*
122
+ * Note the AArch32 opcodes have a different OPC1.
123
+ */
124
+ { .name = "CFPRCTX", .state = ARM_CP_STATE_AA32,
125
+ .cp = 15, .opc1 = 0, .crn = 7, .crm = 3, .opc2 = 4,
126
+ .type = ARM_CP_NOP, .access = PL0_W, .accessfn = access_predinv },
127
+ { .name = "DVPRCTX", .state = ARM_CP_STATE_AA32,
128
+ .cp = 15, .opc1 = 0, .crn = 7, .crm = 3, .opc2 = 5,
129
+ .type = ARM_CP_NOP, .access = PL0_W, .accessfn = access_predinv },
130
+ { .name = "CPPRCTX", .state = ARM_CP_STATE_AA32,
131
+ .cp = 15, .opc1 = 0, .crn = 7, .crm = 3, .opc2 = 7,
132
+ .type = ARM_CP_NOP, .access = PL0_W, .accessfn = access_predinv },
133
+ REGINFO_SENTINEL
134
+};
135
+
136
void register_cp_regs_for_features(ARMCPU *cpu)
137
{
138
/* Register all the coprocessor registers based on feature bits */
139
@@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu)
140
define_arm_cp_regs(cpu, pauth_reginfo);
141
}
142
#endif
143
+
144
+ /*
145
+ * While all v8.0 cpus support aarch64, QEMU does have configurations
146
+ * that do not set ID_AA64ISAR1, e.g. user-only qemu-arm -cpu max,
147
+ * which will set ID_ISAR6.
148
+ */
149
+ if (arm_feature(&cpu->env, ARM_FEATURE_AARCH64)
150
+ ? cpu_isar_feature(aa64_predinv, cpu)
151
+ : cpu_isar_feature(aa32_predinv, cpu)) {
152
+ define_arm_cp_regs(cpu, predinv_reginfo);
153
+ }
154
}
94
}
155
156
void arm_cpu_register_gdb_regs_for_features(ARMCPU *cpu)
157
--
95
--
158
2.20.1
96
2.20.1
159
97
160
98
diff view generated by jsdifflib
[Qemu-devel] [PULL 10/22] target/arm: Restructure handle_fp_1src_{single, double}
[PULL 04/16] target/arm: Don't use '#' flag of printf format
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Xinhao Zhang <zhangxinhao1@huawei.com>
2
2
3
This will allow sharing code that adjusts rmode beyond
3
Fix code style. Don't use '#' flag of printf format ('%#') in
4
the existing users.
4
format strings, use '0x' prefix instead
5
5
6
Tested-by: Laurent Desnogues <laurent.desnogues@gmail.com>
6
Signed-off-by: Xinhao Zhang <zhangxinhao1@huawei.com>
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Signed-off-by: Kai Deng <dengkai1@huawei.com>
8
Message-id: 20190301200501.16533-10-richard.henderson@linaro.org
8
Message-id: 20201103114529.638233-2-zhangxinhao1@huawei.com
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
11
---
12
target/arm/translate-a64.c | 90 +++++++++++++++++++++-----------------
12
target/arm/translate-a64.c | 4 ++--
13
1 file changed, 49 insertions(+), 41 deletions(-)
13
1 file changed, 2 insertions(+), 2 deletions(-)
14
14
15
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
15
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
16
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/translate-a64.c
17
--- a/target/arm/translate-a64.c
18
+++ b/target/arm/translate-a64.c
18
+++ b/target/arm/translate-a64.c
19
@@ -XXX,XX +XXX,XX @@ static void handle_fp_1src_half(DisasContext *s, int opcode, int rd, int rn)
19
@@ -XXX,XX +XXX,XX @@ static void disas_simd_three_reg_same_fp16(DisasContext *s, uint32_t insn)
20
/* Floating-point data-processing (1 source) - single precision */
20
gen_helper_advsimd_acgt_f16(tcg_res, tcg_op1, tcg_op2, fpst);
21
static void handle_fp_1src_single(DisasContext *s, int opcode, int rd, int rn)
21
break;
22
{
22
default:
23
+ void (*gen_fpst)(TCGv_i32, TCGv_i32, TCGv_ptr);
23
- fprintf(stderr, "%s: insn %#04x, fpop %#2x @ %#" PRIx64 "\n",
24
+ TCGv_i32 tcg_op, tcg_res;
24
+ fprintf(stderr, "%s: insn 0x%04x, fpop 0x%2x @ 0x%" PRIx64 "\n",
25
TCGv_ptr fpst;
25
__func__, insn, fpopcode, s->pc_curr);
26
- TCGv_i32 tcg_op;
26
g_assert_not_reached();
27
- TCGv_i32 tcg_res;
27
}
28
+ int rmode = -1;
28
@@ -XXX,XX +XXX,XX @@ static void disas_simd_two_reg_misc_fp16(DisasContext *s, uint32_t insn)
29
29
case 0x7f: /* FSQRT (vector) */
30
- fpst = get_fpstatus_ptr(false);
31
tcg_op = read_fp_sreg(s, rn);
32
tcg_res = tcg_temp_new_i32();
33
34
switch (opcode) {
35
case 0x0: /* FMOV */
36
tcg_gen_mov_i32(tcg_res, tcg_op);
37
- break;
38
+ goto done;
39
case 0x1: /* FABS */
40
gen_helper_vfp_abss(tcg_res, tcg_op);
41
- break;
42
+ goto done;
43
case 0x2: /* FNEG */
44
gen_helper_vfp_negs(tcg_res, tcg_op);
45
- break;
46
+ goto done;
47
case 0x3: /* FSQRT */
48
gen_helper_vfp_sqrts(tcg_res, tcg_op, cpu_env);
49
- break;
50
+ goto done;
51
case 0x8: /* FRINTN */
52
case 0x9: /* FRINTP */
53
case 0xa: /* FRINTM */
54
case 0xb: /* FRINTZ */
55
case 0xc: /* FRINTA */
56
- {
57
- TCGv_i32 tcg_rmode = tcg_const_i32(arm_rmode_to_sf(opcode & 7));
58
-
59
- gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
60
- gen_helper_rints(tcg_res, tcg_op, fpst);
61
-
62
- gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
63
- tcg_temp_free_i32(tcg_rmode);
64
+ rmode = arm_rmode_to_sf(opcode & 7);
65
+ gen_fpst = gen_helper_rints;
66
break;
67
- }
68
case 0xe: /* FRINTX */
69
- gen_helper_rints_exact(tcg_res, tcg_op, fpst);
70
+ gen_fpst = gen_helper_rints_exact;
71
break;
72
case 0xf: /* FRINTI */
73
- gen_helper_rints(tcg_res, tcg_op, fpst);
74
+ gen_fpst = gen_helper_rints;
75
break;
30
break;
76
default:
31
default:
77
- abort();
32
- fprintf(stderr, "%s: insn %#04x fpop %#2x\n", __func__, insn, fpop);
78
+ g_assert_not_reached();
33
+ fprintf(stderr, "%s: insn 0x%04x fpop 0x%2x\n", __func__, insn, fpop);
34
g_assert_not_reached();
79
}
35
}
80
36
81
- write_fp_sreg(s, rd, tcg_res);
82
-
83
+ fpst = get_fpstatus_ptr(false);
84
+ if (rmode >= 0) {
85
+ TCGv_i32 tcg_rmode = tcg_const_i32(rmode);
86
+ gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
87
+ gen_fpst(tcg_res, tcg_op, fpst);
88
+ gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
89
+ tcg_temp_free_i32(tcg_rmode);
90
+ } else {
91
+ gen_fpst(tcg_res, tcg_op, fpst);
92
+ }
93
tcg_temp_free_ptr(fpst);
94
+
95
+ done:
96
+ write_fp_sreg(s, rd, tcg_res);
97
tcg_temp_free_i32(tcg_op);
98
tcg_temp_free_i32(tcg_res);
99
}
100
@@ -XXX,XX +XXX,XX @@ static void handle_fp_1src_single(DisasContext *s, int opcode, int rd, int rn)
101
/* Floating-point data-processing (1 source) - double precision */
102
static void handle_fp_1src_double(DisasContext *s, int opcode, int rd, int rn)
103
{
104
+ void (*gen_fpst)(TCGv_i64, TCGv_i64, TCGv_ptr);
105
+ TCGv_i64 tcg_op, tcg_res;
106
TCGv_ptr fpst;
107
- TCGv_i64 tcg_op;
108
- TCGv_i64 tcg_res;
109
+ int rmode = -1;
110
111
switch (opcode) {
112
case 0x0: /* FMOV */
113
@@ -XXX,XX +XXX,XX @@ static void handle_fp_1src_double(DisasContext *s, int opcode, int rd, int rn)
114
return;
115
}
116
117
- fpst = get_fpstatus_ptr(false);
118
tcg_op = read_fp_dreg(s, rn);
119
tcg_res = tcg_temp_new_i64();
120
121
switch (opcode) {
122
case 0x1: /* FABS */
123
gen_helper_vfp_absd(tcg_res, tcg_op);
124
- break;
125
+ goto done;
126
case 0x2: /* FNEG */
127
gen_helper_vfp_negd(tcg_res, tcg_op);
128
- break;
129
+ goto done;
130
case 0x3: /* FSQRT */
131
gen_helper_vfp_sqrtd(tcg_res, tcg_op, cpu_env);
132
- break;
133
+ goto done;
134
case 0x8: /* FRINTN */
135
case 0x9: /* FRINTP */
136
case 0xa: /* FRINTM */
137
case 0xb: /* FRINTZ */
138
case 0xc: /* FRINTA */
139
- {
140
- TCGv_i32 tcg_rmode = tcg_const_i32(arm_rmode_to_sf(opcode & 7));
141
-
142
- gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
143
- gen_helper_rintd(tcg_res, tcg_op, fpst);
144
-
145
- gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
146
- tcg_temp_free_i32(tcg_rmode);
147
+ rmode = arm_rmode_to_sf(opcode & 7);
148
+ gen_fpst = gen_helper_rintd;
149
break;
150
- }
151
case 0xe: /* FRINTX */
152
- gen_helper_rintd_exact(tcg_res, tcg_op, fpst);
153
+ gen_fpst = gen_helper_rintd_exact;
154
break;
155
case 0xf: /* FRINTI */
156
- gen_helper_rintd(tcg_res, tcg_op, fpst);
157
+ gen_fpst = gen_helper_rintd;
158
break;
159
default:
160
- abort();
161
+ g_assert_not_reached();
162
}
163
164
- write_fp_dreg(s, rd, tcg_res);
165
-
166
+ fpst = get_fpstatus_ptr(false);
167
+ if (rmode >= 0) {
168
+ TCGv_i32 tcg_rmode = tcg_const_i32(rmode);
169
+ gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
170
+ gen_fpst(tcg_res, tcg_op, fpst);
171
+ gen_helper_set_rmode(tcg_rmode, tcg_rmode, fpst);
172
+ tcg_temp_free_i32(tcg_rmode);
173
+ } else {
174
+ gen_fpst(tcg_res, tcg_op, fpst);
175
+ }
176
tcg_temp_free_ptr(fpst);
177
+
178
+ done:
179
+ write_fp_dreg(s, rd, tcg_res);
180
tcg_temp_free_i64(tcg_op);
181
tcg_temp_free_i64(tcg_res);
182
}
183
--
37
--
184
2.20.1
38
2.20.1
185
39
186
40
diff view generated by jsdifflib
[Qemu-devel] [PULL 01/22] target/arm: Fix PC test for LDM (exception return)
[PULL 05/16] target/arm: add space before the open parenthesis '('
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Xinhao Zhang <zhangxinhao1@huawei.com>
2
2
3
Found by inspection: Rn is the base register against which the
3
Fix code style. Space required before the open parenthesis '('.
4
load began; I is the register within the mask being processed.
5
The exception return should of course be processed from the loaded PC.
6
4
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
5
Signed-off-by: Xinhao Zhang <zhangxinhao1@huawei.com>
8
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
6
Signed-off-by: Kai Deng <dengkai1@huawei.com>
9
Message-id: 20190301202921.21209-1-richard.henderson@linaro.org
7
Message-id: 20201103114529.638233-3-zhangxinhao1@huawei.com
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
10
---
12
target/arm/translate.c | 2 +-
11
target/arm/translate.c | 2 +-
13
1 file changed, 1 insertion(+), 1 deletion(-)
12
1 file changed, 1 insertion(+), 1 deletion(-)
14
13
15
diff --git a/target/arm/translate.c b/target/arm/translate.c
14
diff --git a/target/arm/translate.c b/target/arm/translate.c
16
index XXXXXXX..XXXXXXX 100644
15
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/translate.c
16
--- a/target/arm/translate.c
18
+++ b/target/arm/translate.c
17
+++ b/target/arm/translate.c
19
@@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn)
18
@@ -XXX,XX +XXX,XX @@ static void arm_tr_tb_stop(DisasContextBase *dcbase, CPUState *cpu)
20
} else if (i == rn) {
19
- Hardware watchpoints.
21
loaded_var = tmp;
20
Hardware breakpoints have already been handled and skip this code.
22
loaded_base = 1;
21
*/
23
- } else if (rn == 15 && exc_return) {
22
- switch(dc->base.is_jmp) {
24
+ } else if (i == 15 && exc_return) {
23
+ switch (dc->base.is_jmp) {
25
store_pc_exc_ret(s, tmp);
24
case DISAS_NEXT:
26
} else {
25
case DISAS_TOO_MANY:
27
store_reg_from_load(s, i, tmp);
26
gen_goto_tb(dc, 1, dc->base.pc_next);
28
--
27
--
29
2.20.1
28
2.20.1
30
29
31
30
diff view generated by jsdifflib
[Qemu-devel] [PULL 21/22] hw/arm/virt: Bump the 255GB initial RAM limit
[PULL 06/16] docs: add some notes on the sbsa-ref machine
1
From: Eric Auger <eric.auger@redhat.com>
1
From: Alex Bennée <alex.bennee@linaro.org>
2
2
3
Now we have the extended memory map (high IO regions beyond the
3
We should at least document what this machine is about.
4
scalable RAM) and dynamic IPA range support at KVM/ARM level
5
we can bump the legacy 255GB initial RAM limit. The actual maximum
6
RAM size now depends on the physical CPU and host kernel, in
7
accelerated mode. In TCG mode, it depends on the VCPU
8
AA64MMFR0.PARANGE.
9
4
10
Signed-off-by: Eric Auger <eric.auger@redhat.com>
5
Reviewed-by: Graeme Gregory <graeme@nuviainc.com>
11
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
6
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
12
Message-id: 20190304101339.25970-11-eric.auger@redhat.com
7
Message-id: 20201104165254.24822-1-alex.bennee@linaro.org
8
Cc: Leif Lindholm <leif@nuviainc.com>
9
Cc: Shashi Mallela <shashi.mallela@linaro.org>
10
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
11
[PMM: fixed filename mismatch]
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
---
13
---
15
hw/arm/virt.c | 21 +--------------------
14
docs/system/arm/sbsa.rst | 32 ++++++++++++++++++++++++++++++++
16
1 file changed, 1 insertion(+), 20 deletions(-)
15
docs/system/target-arm.rst | 1 +
16
2 files changed, 33 insertions(+)
17
create mode 100644 docs/system/arm/sbsa.rst
17
18
18
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
19
diff --git a/docs/system/arm/sbsa.rst b/docs/system/arm/sbsa.rst
20
new file mode 100644
21
index XXXXXXX..XXXXXXX
22
--- /dev/null
23
+++ b/docs/system/arm/sbsa.rst
24
@@ -XXX,XX +XXX,XX @@
25
+Arm Server Base System Architecture Reference board (``sbsa-ref``)
26
+==================================================================
27
+
28
+While the `virt` board is a generic board platform that doesn't match
29
+any real hardware the `sbsa-ref` board intends to look like real
30
+hardware. The `Server Base System Architecture
31
+<https://developer.arm.com/documentation/den0029/latest>` defines a
32
+minimum base line of hardware support and importantly how the firmware
33
+reports that to any operating system. It is a static system that
34
+reports a very minimal DT to the firmware for non-discoverable
35
+information about components affected by the qemu command line (i.e.
36
+cpus and memory). As a result it must have a firmware specifically
37
+built to expect a certain hardware layout (as you would in a real
38
+machine).
39
+
40
+It is intended to be a machine for developing firmware and testing
41
+standards compliance with operating systems.
42
+
43
+Supported devices
44
+"""""""""""""""""
45
+
46
+The sbsa-ref board supports:
47
+
48
+ - A configurable number of AArch64 CPUs
49
+ - GIC version 3
50
+ - System bus AHCI controller
51
+ - System bus EHCI controller
52
+ - CDROM and hard disc on AHCI bus
53
+ - E1000E ethernet card on PCIe bus
54
+ - VGA display adaptor on PCIe bus
55
+ - A generic SBSA watchdog device
56
+
57
diff --git a/docs/system/target-arm.rst b/docs/system/target-arm.rst
19
index XXXXXXX..XXXXXXX 100644
58
index XXXXXXX..XXXXXXX 100644
20
--- a/hw/arm/virt.c
59
--- a/docs/system/target-arm.rst
21
+++ b/hw/arm/virt.c
60
+++ b/docs/system/target-arm.rst
22
@@ -XXX,XX +XXX,XX @@
61
@@ -XXX,XX +XXX,XX @@ undocumented; you can get a complete list by running
23
62
arm/mps2
24
#define PLATFORM_BUS_NUM_IRQS 64
63
arm/musca
25
64
arm/realview
26
-/* RAM limit in GB. Since VIRT_MEM starts at the 1GB mark, this means
65
+ arm/sbsa
27
- * RAM can go up to the 256GB mark, leaving 256GB of the physical
66
arm/versatile
28
- * address space unallocated and free for future use between 256G and 512G.
67
arm/vexpress
29
- * If we need to provide more RAM to VMs in the future then we need to:
68
arm/aspeed
30
- * * allocate a second bank of RAM starting at 2TB and working up
31
- * * fix the DT and ACPI table generation code in QEMU to correctly
32
- * report two split lumps of RAM to the guest
33
- * * fix KVM in the host kernel to allow guests with >40 bit address spaces
34
- * (We don't want to fill all the way up to 512GB with RAM because
35
- * we might want it for non-RAM purposes later. Conversely it seems
36
- * reasonable to assume that anybody configuring a VM with a quarter
37
- * of a terabyte of RAM will be doing it on a host with more than a
38
- * terabyte of physical address space.)
39
- */
40
+/* Legacy RAM limit in GB (< version 4.0) */
41
#define LEGACY_RAMLIMIT_GB 255
42
#define LEGACY_RAMLIMIT_BYTES (LEGACY_RAMLIMIT_GB * GiB)
43
44
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
45
46
vms->smp_cpus = smp_cpus;
47
48
- if (machine->ram_size > vms->memmap[VIRT_MEM].size) {
49
- error_report("mach-virt: cannot model more than %dGB RAM",
50
- LEGACY_RAMLIMIT_GB);
51
- exit(1);
52
- }
53
-
54
if (vms->virt && kvm_enabled()) {
55
error_report("mach-virt: KVM does not support providing "
56
"Virtualization extensions to the guest CPU");
57
--
69
--
58
2.20.1
70
2.20.1
59
71
60
72
diff view generated by jsdifflib
[Qemu-devel] [PULL 20/22] hw/arm/virt: Check the VCPU PA range in TCG mode
[PULL 07/16] hw/arm/virt: Remove dependency on Cortex-A15 MPCore peripherals
1
From: Eric Auger <eric.auger@redhat.com>
1
From: Philippe Mathieu-Daudé <philmd@redhat.com>
2
2
3
We are about to allow the memory map to grow beyond 1TB and
3
When using a Cortex-A15, the Virt machine does not use any
4
potentially overshoot the VCPU AA64MMFR0.PARANGE.
4
MPCore peripherals. Remove the dependency.
5
5
6
In aarch64 mode and when highmem is set, let's check the VCPU
6
Fixes: 7951c7b7c05 ("hw/arm: Express dependencies of the virt machine with Kconfig")
7
PA range is sufficient to address the highest GPA of the memory
7
Reported-by: Miroslav Rezanina <mrezanin@redhat.com>
8
map.
8
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
9
9
Message-id: 20201107114852.271922-1-philmd@redhat.com
10
Signed-off-by: Eric Auger <eric.auger@redhat.com>
10
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
11
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
12
Message-id: 20190304101339.25970-10-eric.auger@redhat.com
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
---
12
---
15
hw/arm/virt.c | 17 +++++++++++++++++
13
hw/arm/Kconfig | 1 -
16
1 file changed, 17 insertions(+)
14
1 file changed, 1 deletion(-)
17
15
18
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
16
diff --git a/hw/arm/Kconfig b/hw/arm/Kconfig
19
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
20
--- a/hw/arm/virt.c
18
--- a/hw/arm/Kconfig
21
+++ b/hw/arm/virt.c
19
+++ b/hw/arm/Kconfig
22
@@ -XXX,XX +XXX,XX @@
20
@@ -XXX,XX +XXX,XX @@ config ARM_VIRT
23
#include "standard-headers/linux/input.h"
21
imply VFIO_PLATFORM
24
#include "hw/arm/smmuv3.h"
22
imply VFIO_XGMAC
25
#include "hw/acpi/acpi.h"
23
imply TPM_TIS_SYSBUS
26
+#include "target/arm/internals.h"
24
- select A15MPCORE
27
25
select ACPI
28
#define DEFINE_VIRT_MACHINE_LATEST(major, minor, latest) \
26
select ARM_SMMUV3
29
static void virt_##major##_##minor##_class_init(ObjectClass *oc, \
27
select GPIO_KEY
30
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
31
fdt_add_timer_nodes(vms);
32
fdt_add_cpu_nodes(vms);
33
34
+ if (!kvm_enabled()) {
35
+ ARMCPU *cpu = ARM_CPU(first_cpu);
36
+ bool aarch64 = object_property_get_bool(OBJECT(cpu), "aarch64", NULL);
37
+
38
+ if (aarch64 && vms->highmem) {
39
+ int requested_pa_size, pamax = arm_pamax(cpu);
40
+
41
+ requested_pa_size = 64 - clz64(vms->highest_gpa);
42
+ if (pamax < requested_pa_size) {
43
+ error_report("VCPU supports less PA bits (%d) than requested "
44
+ "by the memory map (%d)", pamax, requested_pa_size);
45
+ exit(1);
46
+ }
47
+ }
48
+ }
49
+
50
memory_region_allocate_system_memory(ram, NULL, "mach-virt.ram",
51
machine->ram_size);
52
memory_region_add_subregion(sysmem, vms->memmap[VIRT_MEM].base, ram);
53
--
28
--
54
2.20.1
29
2.20.1
55
30
56
31
diff view generated by jsdifflib
[Qemu-devel] [PULL 05/22] target/arm: Split helper_msr_i_pstate into 3
[PULL 08/16] target/arm: Fix neon VTBL/VTBX for len > 1
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
The EL0+UMA check is unique to DAIF. While SPSel had avoided the
3
The helper function did not get updated when we reorganized
4
check by nature of already checking EL >= 1, the other post v8.0
4
the vector register file for SVE. Since then, the neon dregs
5
extensions to MSR (imm) allow EL0 and do not require UMA. Avoid
5
are non-sequential and cannot be simply indexed.
6
the unconditional write to pc and use raise_exception_ra to unwind.
7
6
7
At the same time, make the helper function operate on 64-bit
8
quantities so that we do not have to call it twice.
9
10
Fixes: c39c2b9043e
11
Reported-by: Ard Biesheuvel <ardb@kernel.org>
8
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
12
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
9
Message-id: 20190301200501.16533-5-richard.henderson@linaro.org
13
[PMM: use aa32_vfp_dreg() rather than opencoding]
14
Message-id: 20201105171126.88014-1-richard.henderson@linaro.org
10
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
---
17
---
13
target/arm/helper-a64.h | 3 +++
18
target/arm/helper.h | 2 +-
14
target/arm/helper.h | 1 -
19
target/arm/op_helper.c | 23 +++++++++--------
15
target/arm/internals.h | 15 ++++++++++++++
20
target/arm/translate-neon.c.inc | 44 +++++++++++----------------------
16
target/arm/helper-a64.c | 30 +++++++++++++++++++++++++++
21
3 files changed, 29 insertions(+), 40 deletions(-)
17
target/arm/op_helper.c | 42 --------------------------------------
18
target/arm/translate-a64.c | 41 ++++++++++++++++++++++---------------
19
6 files changed, 73 insertions(+), 59 deletions(-)
20
22
21
diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
22
index XXXXXXX..XXXXXXX 100644
23
--- a/target/arm/helper-a64.h
24
+++ b/target/arm/helper-a64.h
25
@@ -XXX,XX +XXX,XX @@
26
DEF_HELPER_FLAGS_2(udiv64, TCG_CALL_NO_RWG_SE, i64, i64, i64)
27
DEF_HELPER_FLAGS_2(sdiv64, TCG_CALL_NO_RWG_SE, s64, s64, s64)
28
DEF_HELPER_FLAGS_1(rbit64, TCG_CALL_NO_RWG_SE, i64, i64)
29
+DEF_HELPER_2(msr_i_spsel, void, env, i32)
30
+DEF_HELPER_2(msr_i_daifset, void, env, i32)
31
+DEF_HELPER_2(msr_i_daifclear, void, env, i32)
32
DEF_HELPER_3(vfp_cmph_a64, i64, f16, f16, ptr)
33
DEF_HELPER_3(vfp_cmpeh_a64, i64, f16, f16, ptr)
34
DEF_HELPER_3(vfp_cmps_a64, i64, f32, f32, ptr)
35
diff --git a/target/arm/helper.h b/target/arm/helper.h
23
diff --git a/target/arm/helper.h b/target/arm/helper.h
36
index XXXXXXX..XXXXXXX 100644
24
index XXXXXXX..XXXXXXX 100644
37
--- a/target/arm/helper.h
25
--- a/target/arm/helper.h
38
+++ b/target/arm/helper.h
26
+++ b/target/arm/helper.h
39
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(get_cp_reg, i32, env, ptr)
27
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(rsqrte_f32, TCG_CALL_NO_RWG, f32, f32, ptr)
40
DEF_HELPER_3(set_cp_reg64, void, env, ptr, i64)
28
DEF_HELPER_FLAGS_2(rsqrte_f64, TCG_CALL_NO_RWG, f64, f64, ptr)
41
DEF_HELPER_2(get_cp_reg64, i64, env, ptr)
29
DEF_HELPER_FLAGS_1(recpe_u32, TCG_CALL_NO_RWG, i32, i32)
42
30
DEF_HELPER_FLAGS_1(rsqrte_u32, TCG_CALL_NO_RWG, i32, i32)
43
-DEF_HELPER_3(msr_i_pstate, void, env, i32, i32)
31
-DEF_HELPER_FLAGS_4(neon_tbl, TCG_CALL_NO_RWG, i32, i32, i32, ptr, i32)
44
DEF_HELPER_1(clear_pstate_ss, void, env)
32
+DEF_HELPER_FLAGS_4(neon_tbl, TCG_CALL_NO_RWG, i64, env, i32, i64, i64)
45
33
46
DEF_HELPER_2(get_r13_banked, i32, env, i32)
34
DEF_HELPER_3(shl_cc, i32, env, i32, i32)
47
diff --git a/target/arm/internals.h b/target/arm/internals.h
35
DEF_HELPER_3(shr_cc, i32, env, i32, i32)
48
index XXXXXXX..XXXXXXX 100644
49
--- a/target/arm/internals.h
50
+++ b/target/arm/internals.h
51
@@ -XXX,XX +XXX,XX @@ ARMVAParameters aa64_va_parameters_both(CPUARMState *env, uint64_t va,
52
ARMVAParameters aa64_va_parameters(CPUARMState *env, uint64_t va,
53
ARMMMUIdx mmu_idx, bool data);
54
55
+static inline int exception_target_el(CPUARMState *env)
56
+{
57
+ int target_el = MAX(1, arm_current_el(env));
58
+
59
+ /*
60
+ * No such thing as secure EL1 if EL3 is aarch32,
61
+ * so update the target EL to EL3 in this case.
62
+ */
63
+ if (arm_is_secure(env) && !arm_el_is_aa64(env, 3) && target_el == 1) {
64
+ target_el = 3;
65
+ }
66
+
67
+ return target_el;
68
+}
69
+
70
#endif
71
diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c
72
index XXXXXXX..XXXXXXX 100644
73
--- a/target/arm/helper-a64.c
74
+++ b/target/arm/helper-a64.c
75
@@ -XXX,XX +XXX,XX @@ uint64_t HELPER(rbit64)(uint64_t x)
76
return revbit64(x);
77
}
78
79
+void HELPER(msr_i_spsel)(CPUARMState *env, uint32_t imm)
80
+{
81
+ update_spsel(env, imm);
82
+}
83
+
84
+static void daif_check(CPUARMState *env, uint32_t op,
85
+ uint32_t imm, uintptr_t ra)
86
+{
87
+ /* DAIF update to PSTATE. This is OK from EL0 only if UMA is set. */
88
+ if (arm_current_el(env) == 0 && !(env->cp15.sctlr_el[1] & SCTLR_UMA)) {
89
+ raise_exception_ra(env, EXCP_UDEF,
90
+ syn_aa64_sysregtrap(0, extract32(op, 0, 3),
91
+ extract32(op, 3, 3), 4,
92
+ imm, 0x1f, 0),
93
+ exception_target_el(env), ra);
94
+ }
95
+}
96
+
97
+void HELPER(msr_i_daifset)(CPUARMState *env, uint32_t imm)
98
+{
99
+ daif_check(env, 0x1e, imm, GETPC());
100
+ env->daif |= (imm << 6) & PSTATE_DAIF;
101
+}
102
+
103
+void HELPER(msr_i_daifclear)(CPUARMState *env, uint32_t imm)
104
+{
105
+ daif_check(env, 0x1f, imm, GETPC());
106
+ env->daif &= ~((imm << 6) & PSTATE_DAIF);
107
+}
108
+
109
/* Convert a softfloat float_relation_ (as returned by
110
* the float*_compare functions) to the correct ARM
111
* NZCV flag state.
112
diff --git a/target/arm/op_helper.c b/target/arm/op_helper.c
36
diff --git a/target/arm/op_helper.c b/target/arm/op_helper.c
113
index XXXXXXX..XXXXXXX 100644
37
index XXXXXXX..XXXXXXX 100644
114
--- a/target/arm/op_helper.c
38
--- a/target/arm/op_helper.c
115
+++ b/target/arm/op_helper.c
39
+++ b/target/arm/op_helper.c
116
@@ -XXX,XX +XXX,XX @@ void raise_exception_ra(CPUARMState *env, uint32_t excp, uint32_t syndrome,
40
@@ -XXX,XX +XXX,XX @@ void raise_exception_ra(CPUARMState *env, uint32_t excp, uint32_t syndrome,
117
cpu_loop_exit_restore(cs, ra);
41
cpu_loop_exit_restore(cs, ra);
118
}
42
}
119
43
120
-static int exception_target_el(CPUARMState *env)
44
-uint32_t HELPER(neon_tbl)(uint32_t ireg, uint32_t def, void *vn,
121
-{
45
- uint32_t maxindex)
122
- int target_el = MAX(1, arm_current_el(env));
46
+uint64_t HELPER(neon_tbl)(CPUARMState *env, uint32_t desc,
123
-
47
+ uint64_t ireg, uint64_t def)
124
- /* No such thing as secure EL1 if EL3 is aarch32, so update the target EL
48
{
125
- * to EL3 in this case.
49
- uint32_t val, shift;
126
- */
50
- uint64_t *table = vn;
127
- if (arm_is_secure(env) && !arm_el_is_aa64(env, 3) && target_el == 1) {
51
+ uint64_t tmp, val = 0;
128
- target_el = 3;
52
+ uint32_t maxindex = ((desc & 3) + 1) * 8;
53
+ uint32_t base_reg = desc >> 2;
54
+ uint32_t shift, index, reg;
55
56
- val = 0;
57
- for (shift = 0; shift < 32; shift += 8) {
58
- uint32_t index = (ireg >> shift) & 0xff;
59
+ for (shift = 0; shift < 64; shift += 8) {
60
+ index = (ireg >> shift) & 0xff;
61
if (index < maxindex) {
62
- uint32_t tmp = (table[index >> 3] >> ((index & 7) << 3)) & 0xff;
63
- val |= tmp << shift;
64
+ reg = base_reg + (index >> 3);
65
+ tmp = *aa32_vfp_dreg(env, reg);
66
+ tmp = ((tmp >> ((index & 7) << 3)) & 0xff) << shift;
67
} else {
68
- val |= def & (0xff << shift);
69
+ tmp = def & (0xffull << shift);
70
}
71
+ val |= tmp;
72
}
73
return val;
74
}
75
diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc
76
index XXXXXXX..XXXXXXX 100644
77
--- a/target/arm/translate-neon.c.inc
78
+++ b/target/arm/translate-neon.c.inc
79
@@ -XXX,XX +XXX,XX @@ static bool trans_VEXT(DisasContext *s, arg_VEXT *a)
80
81
static bool trans_VTBL(DisasContext *s, arg_VTBL *a)
82
{
83
- int n;
84
- TCGv_i32 tmp, tmp2, tmp3, tmp4;
85
- TCGv_ptr ptr1;
86
+ TCGv_i64 val, def;
87
+ TCGv_i32 desc;
88
89
if (!arm_dc_feature(s, ARM_FEATURE_NEON)) {
90
return false;
91
@@ -XXX,XX +XXX,XX @@ static bool trans_VTBL(DisasContext *s, arg_VTBL *a)
92
return true;
93
}
94
95
- n = a->len + 1;
96
- if ((a->vn + n) > 32) {
97
+ if ((a->vn + a->len + 1) > 32) {
98
/*
99
* This is UNPREDICTABLE; we choose to UNDEF to avoid the
100
* helper function running off the end of the register file.
101
*/
102
return false;
103
}
104
- n <<= 3;
105
- tmp = tcg_temp_new_i32();
106
- if (a->op) {
107
- read_neon_element32(tmp, a->vd, 0, MO_32);
108
- } else {
109
- tcg_gen_movi_i32(tmp, 0);
129
- }
110
- }
130
-
111
- tmp2 = tcg_temp_new_i32();
131
- return target_el;
112
- read_neon_element32(tmp2, a->vm, 0, MO_32);
132
-}
113
- ptr1 = vfp_reg_ptr(true, a->vn);
133
-
114
- tmp4 = tcg_const_i32(n);
134
uint32_t HELPER(neon_tbl)(uint32_t ireg, uint32_t def, void *vn,
115
- gen_helper_neon_tbl(tmp2, tmp2, tmp, ptr1, tmp4);
135
uint32_t maxindex)
116
136
{
117
+ desc = tcg_const_i32((a->vn << 2) | a->len);
137
@@ -XXX,XX +XXX,XX @@ uint64_t HELPER(get_cp_reg64)(CPUARMState *env, void *rip)
118
+ def = tcg_temp_new_i64();
138
return res;
119
if (a->op) {
120
- read_neon_element32(tmp, a->vd, 1, MO_32);
121
+ read_neon_element64(def, a->vd, 0, MO_64);
122
} else {
123
- tcg_gen_movi_i32(tmp, 0);
124
+ tcg_gen_movi_i64(def, 0);
125
}
126
- tmp3 = tcg_temp_new_i32();
127
- read_neon_element32(tmp3, a->vm, 1, MO_32);
128
- gen_helper_neon_tbl(tmp3, tmp3, tmp, ptr1, tmp4);
129
- tcg_temp_free_i32(tmp);
130
- tcg_temp_free_i32(tmp4);
131
- tcg_temp_free_ptr(ptr1);
132
+ val = tcg_temp_new_i64();
133
+ read_neon_element64(val, a->vm, 0, MO_64);
134
135
- write_neon_element32(tmp2, a->vd, 0, MO_32);
136
- write_neon_element32(tmp3, a->vd, 1, MO_32);
137
- tcg_temp_free_i32(tmp2);
138
- tcg_temp_free_i32(tmp3);
139
+ gen_helper_neon_tbl(val, cpu_env, desc, val, def);
140
+ write_neon_element64(val, a->vd, 0, MO_64);
141
+
142
+ tcg_temp_free_i64(def);
143
+ tcg_temp_free_i64(val);
144
+ tcg_temp_free_i32(desc);
145
return true;
139
}
146
}
140
147
141
-void HELPER(msr_i_pstate)(CPUARMState *env, uint32_t op, uint32_t imm)
142
-{
143
- /* MSR_i to update PSTATE. This is OK from EL0 only if UMA is set.
144
- * Note that SPSel is never OK from EL0; we rely on handle_msr_i()
145
- * to catch that case at translate time.
146
- */
147
- if (arm_current_el(env) == 0 && !(env->cp15.sctlr_el[1] & SCTLR_UMA)) {
148
- uint32_t syndrome = syn_aa64_sysregtrap(0, extract32(op, 0, 3),
149
- extract32(op, 3, 3), 4,
150
- imm, 0x1f, 0);
151
- raise_exception(env, EXCP_UDEF, syndrome, exception_target_el(env));
152
- }
153
-
154
- switch (op) {
155
- case 0x05: /* SPSel */
156
- update_spsel(env, imm);
157
- break;
158
- case 0x1e: /* DAIFSet */
159
- env->daif |= (imm << 6) & PSTATE_DAIF;
160
- break;
161
- case 0x1f: /* DAIFClear */
162
- env->daif &= ~((imm << 6) & PSTATE_DAIF);
163
- break;
164
- default:
165
- g_assert_not_reached();
166
- }
167
-}
168
-
169
void HELPER(clear_pstate_ss)(CPUARMState *env)
170
{
171
env->pstate &= ~PSTATE_SS;
172
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
173
index XXXXXXX..XXXXXXX 100644
174
--- a/target/arm/translate-a64.c
175
+++ b/target/arm/translate-a64.c
176
@@ -XXX,XX +XXX,XX @@ static void handle_sync(DisasContext *s, uint32_t insn,
177
static void handle_msr_i(DisasContext *s, uint32_t insn,
178
unsigned int op1, unsigned int op2, unsigned int crm)
179
{
180
+ TCGv_i32 t1;
181
int op = op1 << 3 | op2;
182
+
183
+ /* End the TB by default, chaining is ok. */
184
+ s->base.is_jmp = DISAS_TOO_MANY;
185
+
186
switch (op) {
187
case 0x05: /* SPSel */
188
if (s->current_el == 0) {
189
- unallocated_encoding(s);
190
- return;
191
+ goto do_unallocated;
192
}
193
- /* fall through */
194
- case 0x1e: /* DAIFSet */
195
- case 0x1f: /* DAIFClear */
196
- {
197
- TCGv_i32 tcg_imm = tcg_const_i32(crm);
198
- TCGv_i32 tcg_op = tcg_const_i32(op);
199
- gen_a64_set_pc_im(s->pc - 4);
200
- gen_helper_msr_i_pstate(cpu_env, tcg_op, tcg_imm);
201
- tcg_temp_free_i32(tcg_imm);
202
- tcg_temp_free_i32(tcg_op);
203
- /* For DAIFClear, exit the cpu loop to re-evaluate pending IRQs. */
204
- gen_a64_set_pc_im(s->pc);
205
- s->base.is_jmp = (op == 0x1f ? DISAS_EXIT : DISAS_JUMP);
206
+ t1 = tcg_const_i32(crm & PSTATE_SP);
207
+ gen_helper_msr_i_spsel(cpu_env, t1);
208
+ tcg_temp_free_i32(t1);
209
break;
210
- }
211
+
212
+ case 0x1e: /* DAIFSet */
213
+ t1 = tcg_const_i32(crm);
214
+ gen_helper_msr_i_daifset(cpu_env, t1);
215
+ tcg_temp_free_i32(t1);
216
+ break;
217
+
218
+ case 0x1f: /* DAIFClear */
219
+ t1 = tcg_const_i32(crm);
220
+ gen_helper_msr_i_daifclear(cpu_env, t1);
221
+ tcg_temp_free_i32(t1);
222
+ /* For DAIFClear, exit the cpu loop to re-evaluate pending IRQs. */
223
+ s->base.is_jmp = DISAS_UPDATE;
224
+ break;
225
+
226
default:
227
+ do_unallocated:
228
unallocated_encoding(s);
229
return;
230
}
231
--
148
--
232
2.20.1
149
2.20.1
233
150
234
151
diff view generated by jsdifflib
[Qemu-devel] [PULL 02/22] target/arm: Split out arm_sctlr
[PULL 09/16] hw/arm/armsse: Correct expansion MPC interrupt lines
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Philippe Mathieu-Daudé <f4bug@amsat.org>
2
2
3
Minimize the number of places that will need updating when
3
We can use one MPC per SRAM bank, but we currently only wire the
4
the virtual host extensions are added.
4
IRQ from the first expansion MPC to the IRQ splitter. Fix that.
5
5
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Fixes: bb75e16d5e6 ("hw/arm/iotkit: Wire up MPC interrupt lines")
7
Message-id: 20190301200501.16533-2-richard.henderson@linaro.org
7
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
Message-id: 20201107193403.436146-2-f4bug@amsat.org
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
---
11
target/arm/cpu.h | 26 ++++++++++++++++----------
12
hw/arm/armsse.c | 3 ++-
12
target/arm/helper.c | 8 ++------
13
1 file changed, 2 insertions(+), 1 deletion(-)
13
2 files changed, 18 insertions(+), 16 deletions(-)
14
14
15
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
15
diff --git a/hw/arm/armsse.c b/hw/arm/armsse.c
16
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/cpu.h
17
--- a/hw/arm/armsse.c
18
+++ b/target/arm/cpu.h
18
+++ b/hw/arm/armsse.c
19
@@ -XXX,XX +XXX,XX @@ static inline bool arm_sctlr_b(CPUARMState *env)
19
@@ -XXX,XX +XXX,XX @@ static void armsse_realize(DeviceState *dev, Error **errp)
20
(env->cp15.sctlr_el[1] & SCTLR_B) != 0;
20
qdev_get_gpio_in(dev_splitter, 0));
21
}
21
qdev_connect_gpio_out(dev_splitter, 0,
22
22
qdev_get_gpio_in_named(dev_secctl,
23
+static inline uint64_t arm_sctlr(CPUARMState *env, int el)
23
- "mpc_status", 0));
24
+{
24
+ "mpc_status",
25
+ if (el == 0) {
25
+ i - IOTS_NUM_EXP_MPC));
26
+ /* FIXME: ARMv8.1-VHE S2 translation regime. */
27
+ return env->cp15.sctlr_el[1];
28
+ } else {
29
+ return env->cp15.sctlr_el[el];
30
+ }
31
+}
32
+
33
+
34
/* Return true if the processor is in big-endian mode. */
35
static inline bool arm_cpu_data_is_big_endian(CPUARMState *env)
36
{
37
- int cur_el;
38
-
39
/* In 32bit endianness is determined by looking at CPSR's E bit */
40
if (!is_a64(env)) {
41
return
42
@@ -XXX,XX +XXX,XX @@ static inline bool arm_cpu_data_is_big_endian(CPUARMState *env)
43
arm_sctlr_b(env) ||
44
#endif
45
((env->uncached_cpsr & CPSR_E) ? 1 : 0);
46
+ } else {
47
+ int cur_el = arm_current_el(env);
48
+ uint64_t sctlr = arm_sctlr(env, cur_el);
49
+
50
+ return (sctlr & (cur_el ? SCTLR_EE : SCTLR_E0E)) != 0;
51
}
52
-
53
- cur_el = arm_current_el(env);
54
-
55
- if (cur_el == 0) {
56
- return (env->cp15.sctlr_el[1] & SCTLR_E0E) != 0;
57
- }
58
-
59
- return (env->cp15.sctlr_el[cur_el] & SCTLR_EE) != 0;
60
}
61
62
#include "exec/cpu-all.h"
63
diff --git a/target/arm/helper.c b/target/arm/helper.c
64
index XXXXXXX..XXXXXXX 100644
65
--- a/target/arm/helper.c
66
+++ b/target/arm/helper.c
67
@@ -XXX,XX +XXX,XX @@ void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc,
68
flags = FIELD_DP32(flags, TBFLAG_A64, ZCR_LEN, zcr_len);
69
}
26
}
70
27
71
- if (current_el == 0) {
28
qdev_connect_gpio_out(dev_splitter, 1,
72
- /* FIXME: ARMv8.1-VHE S2 translation regime. */
73
- sctlr = env->cp15.sctlr_el[1];
74
- } else {
75
- sctlr = env->cp15.sctlr_el[current_el];
76
- }
77
+ sctlr = arm_sctlr(env, current_el);
78
+
79
if (cpu_isar_feature(aa64_pauth, cpu)) {
80
/*
81
* In order to save space in flags, we record only whether
82
--
29
--
83
2.20.1
30
2.20.1
84
31
85
32
diff view generated by jsdifflib
[Qemu-devel] [PULL 14/22] hw/arm/virt: Split the memory map description
[PULL 10/16] hw/misc/stm32f2xx_syscfg: Remove extraneous IRQ
1
From: Eric Auger <eric.auger@redhat.com>
1
From: Philippe Mathieu-Daudé <f4bug@amsat.org>
2
2
3
In the prospect to introduce an extended memory map supporting more
3
The system configuration controller (SYSCFG) doesn't have
4
RAM, let's split the memory map array into two parts:
4
any output IRQ (and the INTC input #71 belongs to the UART6).
5
Remove the invalid code.
5
6
6
- the former a15memmap, renamed base_memmap, contains regions below
7
Fixes: db635521a02 ("stm32f205: Add the stm32f205 SoC")
7
and including the RAM. MemMapEntries initialized in this array
8
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
have a static size and base address.
9
Message-id: 20201107193403.436146-3-f4bug@amsat.org
9
- extended_memmap, only initialized with entries located after the
10
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
RAM. MemMapEntries initialized in this array only get their size
11
initialized. Their base address is dynamically computed depending
12
on the the top of the RAM, with same alignment as their size.
13
14
Eventually base_memmap entries are copied into the extended_memmap
15
array. Using two separate arrays however clarifies which entries
16
are statically allocated and those which are dynamically allocated.
17
18
This new split will allow to grow the RAM size without changing the
19
description of the high IO entries.
20
21
We introduce a new virt_set_memmap() helper function which
22
"freezes" the memory map. We call it in machvirt_init as
23
memory attributes of the machine are not yet set when
24
virt_instance_init() gets called.
25
26
The memory map is unchanged (the top of the initial RAM still is
27
256GiB). Then come the high IO regions with same layout as before.
28
29
Signed-off-by: Eric Auger <eric.auger@redhat.com>
30
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
31
Message-id: 20190304101339.25970-4-eric.auger@redhat.com
32
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
33
---
12
---
34
include/hw/arm/virt.h | 13 +++++++----
13
include/hw/misc/stm32f2xx_syscfg.h | 2 --
35
hw/arm/virt.c | 50 +++++++++++++++++++++++++++++++++++++------
14
hw/arm/stm32f205_soc.c | 1 -
36
2 files changed, 53 insertions(+), 10 deletions(-)
15
hw/misc/stm32f2xx_syscfg.c | 2 --
16
3 files changed, 5 deletions(-)
37
17
38
diff --git a/include/hw/arm/virt.h b/include/hw/arm/virt.h
18
diff --git a/include/hw/misc/stm32f2xx_syscfg.h b/include/hw/misc/stm32f2xx_syscfg.h
39
index XXXXXXX..XXXXXXX 100644
19
index XXXXXXX..XXXXXXX 100644
40
--- a/include/hw/arm/virt.h
20
--- a/include/hw/misc/stm32f2xx_syscfg.h
41
+++ b/include/hw/arm/virt.h
21
+++ b/include/hw/misc/stm32f2xx_syscfg.h
42
@@ -XXX,XX +XXX,XX @@ enum {
22
@@ -XXX,XX +XXX,XX @@ struct STM32F2XXSyscfgState {
43
VIRT_GIC_VCPU,
23
uint32_t syscfg_exticr3;
44
VIRT_GIC_ITS,
24
uint32_t syscfg_exticr4;
45
VIRT_GIC_REDIST,
25
uint32_t syscfg_cmpcr;
46
- VIRT_HIGH_GIC_REDIST2,
26
-
47
VIRT_SMMU,
27
- qemu_irq irq;
48
VIRT_UART,
49
VIRT_MMIO,
50
@@ -XXX,XX +XXX,XX @@ enum {
51
VIRT_PCIE_MMIO,
52
VIRT_PCIE_PIO,
53
VIRT_PCIE_ECAM,
54
- VIRT_HIGH_PCIE_ECAM,
55
VIRT_PLATFORM_BUS,
56
- VIRT_HIGH_PCIE_MMIO,
57
VIRT_GPIO,
58
VIRT_SECURE_UART,
59
VIRT_SECURE_MEM,
60
+ VIRT_LOWMEMMAP_LAST,
61
+};
62
+
63
+/* indices of IO regions located after the RAM */
64
+enum {
65
+ VIRT_HIGH_GIC_REDIST2 = VIRT_LOWMEMMAP_LAST,
66
+ VIRT_HIGH_PCIE_ECAM,
67
+ VIRT_HIGH_PCIE_MMIO,
68
};
28
};
69
29
70
typedef enum VirtIOMMUType {
30
#endif /* HW_STM32F2XX_SYSCFG_H */
71
@@ -XXX,XX +XXX,XX @@ typedef struct {
31
diff --git a/hw/arm/stm32f205_soc.c b/hw/arm/stm32f205_soc.c
72
int32_t gic_version;
73
VirtIOMMUType iommu;
74
struct arm_boot_info bootinfo;
75
- const MemMapEntry *memmap;
76
+ MemMapEntry *memmap;
77
const int *irqmap;
78
int smp_cpus;
79
void *fdt;
80
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
81
index XXXXXXX..XXXXXXX 100644
32
index XXXXXXX..XXXXXXX 100644
82
--- a/hw/arm/virt.c
33
--- a/hw/arm/stm32f205_soc.c
83
+++ b/hw/arm/virt.c
34
+++ b/hw/arm/stm32f205_soc.c
84
@@ -XXX,XX +XXX,XX @@
35
@@ -XXX,XX +XXX,XX @@ static void stm32f205_soc_realize(DeviceState *dev_soc, Error **errp)
85
*/
36
}
86
37
busdev = SYS_BUS_DEVICE(dev);
87
#include "qemu/osdep.h"
38
sysbus_mmio_map(busdev, 0, 0x40013800);
88
+#include "qemu/units.h"
39
- sysbus_connect_irq(busdev, 0, qdev_get_gpio_in(armv7m, 71));
89
#include "qapi/error.h"
40
90
#include "hw/sysbus.h"
41
/* Attach UART (uses USART registers) and USART controllers */
91
#include "hw/arm/arm.h"
42
for (i = 0; i < STM_NUM_USARTS; i++) {
92
@@ -XXX,XX +XXX,XX @@
43
diff --git a/hw/misc/stm32f2xx_syscfg.c b/hw/misc/stm32f2xx_syscfg.c
93
* Note that devices should generally be placed at multiples of 0x10000,
44
index XXXXXXX..XXXXXXX 100644
94
* to accommodate guests using 64K pages.
45
--- a/hw/misc/stm32f2xx_syscfg.c
95
*/
46
+++ b/hw/misc/stm32f2xx_syscfg.c
96
-static const MemMapEntry a15memmap[] = {
47
@@ -XXX,XX +XXX,XX @@ static void stm32f2xx_syscfg_init(Object *obj)
97
+static const MemMapEntry base_memmap[] = {
98
/* Space up to 0x8000000 is reserved for a boot ROM */
99
[VIRT_FLASH] = { 0, 0x08000000 },
100
[VIRT_CPUPERIPHS] = { 0x08000000, 0x00020000 },
101
@@ -XXX,XX +XXX,XX @@ static const MemMapEntry a15memmap[] = {
102
[VIRT_PCIE_PIO] = { 0x3eff0000, 0x00010000 },
103
[VIRT_PCIE_ECAM] = { 0x3f000000, 0x01000000 },
104
[VIRT_MEM] = { 0x40000000, RAMLIMIT_BYTES },
105
+};
106
+
107
+/*
108
+ * Highmem IO Regions: This memory map is floating, located after the RAM.
109
+ * Each MemMapEntry base (GPA) will be dynamically computed, depending on the
110
+ * top of the RAM, so that its base get the same alignment as the size,
111
+ * ie. a 512GiB entry will be aligned on a 512GiB boundary. If there is
112
+ * less than 256GiB of RAM, the floating area starts at the 256GiB mark.
113
+ * Note the extended_memmap is sized so that it eventually also includes the
114
+ * base_memmap entries (VIRT_HIGH_GIC_REDIST2 index is greater than the last
115
+ * index of base_memmap).
116
+ */
117
+static MemMapEntry extended_memmap[] = {
118
/* Additional 64 MB redist region (can contain up to 512 redistributors) */
119
- [VIRT_HIGH_GIC_REDIST2] = { 0x4000000000ULL, 0x4000000 },
120
- [VIRT_HIGH_PCIE_ECAM] = { 0x4010000000ULL, 0x10000000 },
121
- /* Second PCIe window, 512GB wide at the 512GB boundary */
122
- [VIRT_HIGH_PCIE_MMIO] = { 0x8000000000ULL, 0x8000000000ULL },
123
+ [VIRT_HIGH_GIC_REDIST2] = { 0x0, 64 * MiB },
124
+ [VIRT_HIGH_PCIE_ECAM] = { 0x0, 256 * MiB },
125
+ /* Second PCIe window */
126
+ [VIRT_HIGH_PCIE_MMIO] = { 0x0, 512 * GiB },
127
};
128
129
static const int a15irqmap[] = {
130
@@ -XXX,XX +XXX,XX @@ static uint64_t virt_cpu_mp_affinity(VirtMachineState *vms, int idx)
131
return arm_cpu_mp_affinity(idx, clustersz);
132
}
133
134
+static void virt_set_memmap(VirtMachineState *vms)
135
+{
136
+ hwaddr base;
137
+ int i;
138
+
139
+ vms->memmap = extended_memmap;
140
+
141
+ for (i = 0; i < ARRAY_SIZE(base_memmap); i++) {
142
+ vms->memmap[i] = base_memmap[i];
143
+ }
144
+
145
+ base = 256 * GiB; /* Top of the legacy initial RAM region */
146
+
147
+ for (i = VIRT_LOWMEMMAP_LAST; i < ARRAY_SIZE(extended_memmap); i++) {
148
+ hwaddr size = extended_memmap[i].size;
149
+
150
+ base = ROUND_UP(base, size);
151
+ vms->memmap[i].base = base;
152
+ vms->memmap[i].size = size;
153
+ base += size;
154
+ }
155
+}
156
+
157
static void machvirt_init(MachineState *machine)
158
{
48
{
159
VirtMachineState *vms = VIRT_MACHINE(machine);
49
STM32F2XXSyscfgState *s = STM32F2XX_SYSCFG(obj);
160
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
50
161
bool firmware_loaded = bios_name || drive_get(IF_PFLASH, 0, 0);
51
- sysbus_init_irq(SYS_BUS_DEVICE(obj), &s->irq);
162
bool aarch64 = true;
52
-
163
53
memory_region_init_io(&s->mmio, obj, &stm32f2xx_syscfg_ops, s,
164
+ virt_set_memmap(vms);
54
TYPE_STM32F2XX_SYSCFG, 0x400);
165
+
55
sysbus_init_mmio(SYS_BUS_DEVICE(obj), &s->mmio);
166
/* We can probe only here because during property set
167
* KVM is not available yet
168
*/
169
@@ -XXX,XX +XXX,XX @@ static void virt_instance_init(Object *obj)
170
"Valid values are none and smmuv3",
171
NULL);
172
173
- vms->memmap = a15memmap;
174
vms->irqmap = a15irqmap;
175
}
176
177
--
56
--
178
2.20.1
57
2.20.1
179
58
180
59
diff view generated by jsdifflib
[Qemu-devel] [PULL 06/22] target/arm: Add set/clear_pstate_bits, share gen_ss_advance
[PULL 11/16] hw/arm/nseries: Remove invalid/unnecessary n8x0_uart_setup()
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Philippe Mathieu-Daudé <f4bug@amsat.org>
2
2
3
We do not need an out-of-line helper for manipulating bits in pstate.
3
omap2420_mpu_init() introduced in commit 827df9f3c5f ("Add basic
4
While changing things, share the implementation of gen_ss_advance.
4
OMAP2 chip support") takes care of creating the 3 UARTs.
5
5
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Then commit 58a26b477e9 ("Emulate a serial bluetooth HCI with H4+
7
Message-id: 20190301200501.16533-6-richard.henderson@linaro.org
7
extensions and attach to n8x0's UART") added n8x0_uart_setup()
8
which create the UART and connects it to an IRQ output,
9
overwritting the existing peripheral and its IRQ connection.
10
This is incorrect.
11
12
Fortunately we don't need to fix this, because commit 6da68df7f9b
13
("hw/arm/nseries: Replace the bluetooth chardev with a "null"
14
chardev") removed the use of this peripheral. We can simply
15
remove the code.
16
17
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
18
Message-id: 20201107193403.436146-4-f4bug@amsat.org
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
19
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
20
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
21
---
11
target/arm/helper.h | 2 --
22
hw/arm/nseries.c | 11 -----------
12
target/arm/translate.h | 34 ++++++++++++++++++++++++++++++++++
23
1 file changed, 11 deletions(-)
13
target/arm/op_helper.c | 5 -----
14
target/arm/translate-a64.c | 11 -----------
15
target/arm/translate.c | 11 -----------
16
5 files changed, 34 insertions(+), 29 deletions(-)
17
24
18
diff --git a/target/arm/helper.h b/target/arm/helper.h
25
diff --git a/hw/arm/nseries.c b/hw/arm/nseries.c
19
index XXXXXXX..XXXXXXX 100644
26
index XXXXXXX..XXXXXXX 100644
20
--- a/target/arm/helper.h
27
--- a/hw/arm/nseries.c
21
+++ b/target/arm/helper.h
28
+++ b/hw/arm/nseries.c
22
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_2(get_cp_reg, i32, env, ptr)
29
@@ -XXX,XX +XXX,XX @@ static void n8x0_cbus_setup(struct n800_s *s)
23
DEF_HELPER_3(set_cp_reg64, void, env, ptr, i64)
30
cbus_attach(cbus, s->tahvo = tahvo_init(tahvo_irq, 1));
24
DEF_HELPER_2(get_cp_reg64, i64, env, ptr)
25
26
-DEF_HELPER_1(clear_pstate_ss, void, env)
27
-
28
DEF_HELPER_2(get_r13_banked, i32, env, i32)
29
DEF_HELPER_3(set_r13_banked, void, env, i32, i32)
30
31
diff --git a/target/arm/translate.h b/target/arm/translate.h
32
index XXXXXXX..XXXXXXX 100644
33
--- a/target/arm/translate.h
34
+++ b/target/arm/translate.h
35
@@ -XXX,XX +XXX,XX @@ static inline TCGv_i32 get_ahp_flag(void)
36
return ret;
37
}
31
}
38
32
39
+/* Set bits within PSTATE. */
33
-static void n8x0_uart_setup(struct n800_s *s)
40
+static inline void set_pstate_bits(uint32_t bits)
41
+{
42
+ TCGv_i32 p = tcg_temp_new_i32();
43
+
44
+ tcg_debug_assert(!(bits & CACHED_PSTATE_BITS));
45
+
46
+ tcg_gen_ld_i32(p, cpu_env, offsetof(CPUARMState, pstate));
47
+ tcg_gen_ori_i32(p, p, bits);
48
+ tcg_gen_st_i32(p, cpu_env, offsetof(CPUARMState, pstate));
49
+ tcg_temp_free_i32(p);
50
+}
51
+
52
+/* Clear bits within PSTATE. */
53
+static inline void clear_pstate_bits(uint32_t bits)
54
+{
55
+ TCGv_i32 p = tcg_temp_new_i32();
56
+
57
+ tcg_debug_assert(!(bits & CACHED_PSTATE_BITS));
58
+
59
+ tcg_gen_ld_i32(p, cpu_env, offsetof(CPUARMState, pstate));
60
+ tcg_gen_andi_i32(p, p, ~bits);
61
+ tcg_gen_st_i32(p, cpu_env, offsetof(CPUARMState, pstate));
62
+ tcg_temp_free_i32(p);
63
+}
64
+
65
+/* If the singlestep state is Active-not-pending, advance to Active-pending. */
66
+static inline void gen_ss_advance(DisasContext *s)
67
+{
68
+ if (s->ss_active) {
69
+ s->pstate_ss = 0;
70
+ clear_pstate_bits(PSTATE_SS);
71
+ }
72
+}
73
74
/* Vector operations shared between ARM and AArch64. */
75
extern const GVecGen3 bsl_op;
76
diff --git a/target/arm/op_helper.c b/target/arm/op_helper.c
77
index XXXXXXX..XXXXXXX 100644
78
--- a/target/arm/op_helper.c
79
+++ b/target/arm/op_helper.c
80
@@ -XXX,XX +XXX,XX @@ uint64_t HELPER(get_cp_reg64)(CPUARMState *env, void *rip)
81
return res;
82
}
83
84
-void HELPER(clear_pstate_ss)(CPUARMState *env)
85
-{
34
-{
86
- env->pstate &= ~PSTATE_SS;
35
- Chardev *radio = qemu_chr_new("bt-dummy-uart", "null", NULL);
36
- /*
37
- * Note: We used to connect N8X0_BT_RESET_GPIO and N8X0_BT_WKUP_GPIO
38
- * here, but this code has been removed with the bluetooth backend.
39
- */
40
- omap_uart_attach(s->mpu->uart[BT_UART], radio);
87
-}
41
-}
88
-
42
-
89
void HELPER(pre_hvc)(CPUARMState *env)
43
static void n8x0_usb_setup(struct n800_s *s)
90
{
44
{
91
ARMCPU *cpu = arm_env_get_cpu(env);
45
SysBusDevice *dev;
92
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
46
@@ -XXX,XX +XXX,XX @@ static void n8x0_init(MachineState *machine,
93
index XXXXXXX..XXXXXXX 100644
47
n8x0_spi_setup(s);
94
--- a/target/arm/translate-a64.c
48
n8x0_dss_setup(s);
95
+++ b/target/arm/translate-a64.c
49
n8x0_cbus_setup(s);
96
@@ -XXX,XX +XXX,XX @@ static void gen_exception_bkpt_insn(DisasContext *s, int offset,
50
- n8x0_uart_setup(s);
97
s->base.is_jmp = DISAS_NORETURN;
51
if (machine_usb(machine)) {
98
}
52
n8x0_usb_setup(s);
99
53
}
100
-static void gen_ss_advance(DisasContext *s)
101
-{
102
- /* If the singlestep state is Active-not-pending, advance to
103
- * Active-pending.
104
- */
105
- if (s->ss_active) {
106
- s->pstate_ss = 0;
107
- gen_helper_clear_pstate_ss(cpu_env);
108
- }
109
-}
110
-
111
static void gen_step_complete_exception(DisasContext *s)
112
{
113
/* We just completed step of an insn. Move from Active-not-pending
114
diff --git a/target/arm/translate.c b/target/arm/translate.c
115
index XXXXXXX..XXXXXXX 100644
116
--- a/target/arm/translate.c
117
+++ b/target/arm/translate.c
118
@@ -XXX,XX +XXX,XX @@ static void gen_exception(int excp, uint32_t syndrome, uint32_t target_el)
119
tcg_temp_free_i32(tcg_excp);
120
}
121
122
-static void gen_ss_advance(DisasContext *s)
123
-{
124
- /* If the singlestep state is Active-not-pending, advance to
125
- * Active-pending.
126
- */
127
- if (s->ss_active) {
128
- s->pstate_ss = 0;
129
- gen_helper_clear_pstate_ss(cpu_env);
130
- }
131
-}
132
-
133
static void gen_step_complete_exception(DisasContext *s)
134
{
135
/* We just completed step of an insn. Move from Active-not-pending
136
--
54
--
137
2.20.1
55
2.20.1
138
56
139
57
diff view generated by jsdifflib
[Qemu-devel] [PULL 11/22] target/arm: Implement ARMv8.5-FRINT
[PULL 12/16] hw/arm/musicpal: Don't connect two qemu_irqs directly to the same input
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Philippe Mathieu-Daudé <f4bug@amsat.org>
2
2
3
Tested-by: Laurent Desnogues <laurent.desnogues@gmail.com>
3
The MusicPal board code connects both of the IRQ outputs of the UART
4
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
4
to the same INTC qemu_irq. Connecting two qemu_irqs outputs directly
5
Message-id: 20190301200501.16533-11-richard.henderson@linaro.org
5
to the same input is not valid as it produces subtly wrong behaviour
6
(for instance if both the IRQ lines are high, and then one goes
7
low, the INTC input will see this as a high-to-low transition
8
even though the second IRQ line should still be holding it high).
9
10
This kind of wiring needs an explicitly created OR gate; add one.
11
12
Inspired-by: Peter Maydell <peter.maydell@linaro.org>
13
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
14
Message-id: 20201107193403.436146-5-f4bug@amsat.org
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
---
17
---
9
target/arm/cpu.h | 5 ++
18
hw/arm/musicpal.c | 17 +++++++++++++----
10
target/arm/helper.h | 5 ++
19
hw/arm/Kconfig | 1 +
11
target/arm/cpu64.c | 1 +
20
2 files changed, 14 insertions(+), 4 deletions(-)
12
target/arm/translate-a64.c | 71 ++++++++++++++++++++++++++--
13
target/arm/vfp_helper.c | 96 ++++++++++++++++++++++++++++++++++++++
14
5 files changed, 173 insertions(+), 5 deletions(-)
15
21
16
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
22
diff --git a/hw/arm/musicpal.c b/hw/arm/musicpal.c
17
index XXXXXXX..XXXXXXX 100644
23
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/cpu.h
24
--- a/hw/arm/musicpal.c
19
+++ b/target/arm/cpu.h
25
+++ b/hw/arm/musicpal.c
20
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_predinv(const ARMISARegisters *id)
26
@@ -XXX,XX +XXX,XX @@
21
return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, SPECRES) != 0;
27
#include "ui/console.h"
22
}
28
#include "hw/i2c/i2c.h"
23
29
#include "hw/irq.h"
24
+static inline bool isar_feature_aa64_frint(const ARMISARegisters *id)
30
+#include "hw/or-irq.h"
25
+{
31
#include "hw/audio/wm8750.h"
26
+ return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, FRINTTS) != 0;
32
#include "sysemu/block-backend.h"
27
+}
33
#include "sysemu/runstate.h"
34
@@ -XXX,XX +XXX,XX @@
35
#define MP_TIMER4_IRQ 7
36
#define MP_EHCI_IRQ 8
37
#define MP_ETH_IRQ 9
38
-#define MP_UART1_IRQ 11
39
-#define MP_UART2_IRQ 11
40
+#define MP_UART_SHARED_IRQ 11
41
#define MP_GPIO_IRQ 12
42
#define MP_RTC_IRQ 28
43
#define MP_AUDIO_IRQ 30
44
@@ -XXX,XX +XXX,XX @@ static void musicpal_init(MachineState *machine)
45
ARMCPU *cpu;
46
qemu_irq pic[32];
47
DeviceState *dev;
48
+ DeviceState *uart_orgate;
49
DeviceState *i2c_dev;
50
DeviceState *lcd_dev;
51
DeviceState *key_dev;
52
@@ -XXX,XX +XXX,XX @@ static void musicpal_init(MachineState *machine)
53
pic[MP_TIMER2_IRQ], pic[MP_TIMER3_IRQ],
54
pic[MP_TIMER4_IRQ], NULL);
55
56
- serial_mm_init(address_space_mem, MP_UART1_BASE, 2, pic[MP_UART1_IRQ],
57
+ /* Logically OR both UART IRQs together */
58
+ uart_orgate = DEVICE(object_new(TYPE_OR_IRQ));
59
+ object_property_set_int(OBJECT(uart_orgate), "num-lines", 2, &error_fatal);
60
+ qdev_realize_and_unref(uart_orgate, NULL, &error_fatal);
61
+ qdev_connect_gpio_out(DEVICE(uart_orgate), 0, pic[MP_UART_SHARED_IRQ]);
28
+
62
+
29
static inline bool isar_feature_aa64_fp16(const ARMISARegisters *id)
63
+ serial_mm_init(address_space_mem, MP_UART1_BASE, 2,
30
{
64
+ qdev_get_gpio_in(uart_orgate, 0),
31
/* We always set the AdvSIMD and FP fields identically wrt FP16. */
65
1825000, serial_hd(0), DEVICE_NATIVE_ENDIAN);
32
diff --git a/target/arm/helper.h b/target/arm/helper.h
66
- serial_mm_init(address_space_mem, MP_UART2_BASE, 2, pic[MP_UART2_IRQ],
67
+ serial_mm_init(address_space_mem, MP_UART2_BASE, 2,
68
+ qdev_get_gpio_in(uart_orgate, 1),
69
1825000, serial_hd(1), DEVICE_NATIVE_ENDIAN);
70
71
/* Register flash */
72
diff --git a/hw/arm/Kconfig b/hw/arm/Kconfig
33
index XXXXXXX..XXXXXXX 100644
73
index XXXXXXX..XXXXXXX 100644
34
--- a/target/arm/helper.h
74
--- a/hw/arm/Kconfig
35
+++ b/target/arm/helper.h
75
+++ b/hw/arm/Kconfig
36
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(gvec_fmlal_idx_a32, TCG_CALL_NO_RWG,
76
@@ -XXX,XX +XXX,XX @@ config MUSCA
37
DEF_HELPER_FLAGS_5(gvec_fmlal_idx_a64, TCG_CALL_NO_RWG,
77
38
void, ptr, ptr, ptr, ptr, i32)
78
config MUSICPAL
39
79
bool
40
+DEF_HELPER_FLAGS_2(frint32_s, TCG_CALL_NO_RWG, f32, f32, ptr)
80
+ select OR_IRQ
41
+DEF_HELPER_FLAGS_2(frint64_s, TCG_CALL_NO_RWG, f32, f32, ptr)
81
select BITBANG_I2C
42
+DEF_HELPER_FLAGS_2(frint32_d, TCG_CALL_NO_RWG, f64, f64, ptr)
82
select MARVELL_88W8618
43
+DEF_HELPER_FLAGS_2(frint64_d, TCG_CALL_NO_RWG, f64, f64, ptr)
83
select PTIMER
44
+
45
#ifdef TARGET_AARCH64
46
#include "helper-a64.h"
47
#include "helper-sve.h"
48
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
49
index XXXXXXX..XXXXXXX 100644
50
--- a/target/arm/cpu64.c
51
+++ b/target/arm/cpu64.c
52
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
53
t = FIELD_DP64(t, ID_AA64ISAR1, GPI, 0);
54
t = FIELD_DP64(t, ID_AA64ISAR1, SB, 1);
55
t = FIELD_DP64(t, ID_AA64ISAR1, SPECRES, 1);
56
+ t = FIELD_DP64(t, ID_AA64ISAR1, FRINTTS, 1);
57
cpu->isar.id_aa64isar1 = t;
58
59
t = cpu->isar.id_aa64pfr0;
60
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
61
index XXXXXXX..XXXXXXX 100644
62
--- a/target/arm/translate-a64.c
63
+++ b/target/arm/translate-a64.c
64
@@ -XXX,XX +XXX,XX @@ static void handle_fp_1src_single(DisasContext *s, int opcode, int rd, int rn)
65
case 0xf: /* FRINTI */
66
gen_fpst = gen_helper_rints;
67
break;
68
+ case 0x10: /* FRINT32Z */
69
+ rmode = float_round_to_zero;
70
+ gen_fpst = gen_helper_frint32_s;
71
+ break;
72
+ case 0x11: /* FRINT32X */
73
+ gen_fpst = gen_helper_frint32_s;
74
+ break;
75
+ case 0x12: /* FRINT64Z */
76
+ rmode = float_round_to_zero;
77
+ gen_fpst = gen_helper_frint64_s;
78
+ break;
79
+ case 0x13: /* FRINT64X */
80
+ gen_fpst = gen_helper_frint64_s;
81
+ break;
82
default:
83
g_assert_not_reached();
84
}
85
@@ -XXX,XX +XXX,XX @@ static void handle_fp_1src_double(DisasContext *s, int opcode, int rd, int rn)
86
case 0xf: /* FRINTI */
87
gen_fpst = gen_helper_rintd;
88
break;
89
+ case 0x10: /* FRINT32Z */
90
+ rmode = float_round_to_zero;
91
+ gen_fpst = gen_helper_frint32_d;
92
+ break;
93
+ case 0x11: /* FRINT32X */
94
+ gen_fpst = gen_helper_frint32_d;
95
+ break;
96
+ case 0x12: /* FRINT64Z */
97
+ rmode = float_round_to_zero;
98
+ gen_fpst = gen_helper_frint64_d;
99
+ break;
100
+ case 0x13: /* FRINT64X */
101
+ gen_fpst = gen_helper_frint64_d;
102
+ break;
103
default:
104
g_assert_not_reached();
105
}
106
@@ -XXX,XX +XXX,XX @@ static void disas_fp_1src(DisasContext *s, uint32_t insn)
107
handle_fp_fcvt(s, opcode, rd, rn, dtype, type);
108
break;
109
}
110
+
111
+ case 0x10 ... 0x13: /* FRINT{32,64}{X,Z} */
112
+ if (type > 1 || !dc_isar_feature(aa64_frint, s)) {
113
+ unallocated_encoding(s);
114
+ return;
115
+ }
116
+ /* fall through */
117
case 0x0 ... 0x3:
118
case 0x8 ... 0xc:
119
case 0xe ... 0xf:
120
@@ -XXX,XX +XXX,XX @@ static void disas_fp_1src(DisasContext *s, uint32_t insn)
121
if (!fp_access_check(s)) {
122
return;
123
}
124
-
125
handle_fp_1src_single(s, opcode, rd, rn);
126
break;
127
case 1:
128
if (!fp_access_check(s)) {
129
return;
130
}
131
-
132
handle_fp_1src_double(s, opcode, rd, rn);
133
break;
134
case 3:
135
@@ -XXX,XX +XXX,XX @@ static void disas_fp_1src(DisasContext *s, uint32_t insn)
136
if (!fp_access_check(s)) {
137
return;
138
}
139
-
140
handle_fp_1src_half(s, opcode, rd, rn);
141
break;
142
default:
143
unallocated_encoding(s);
144
}
145
break;
146
+
147
default:
148
unallocated_encoding(s);
149
break;
150
@@ -XXX,XX +XXX,XX @@ static void handle_2misc_64(DisasContext *s, int opcode, bool u,
151
case 0x59: /* FRINTX */
152
gen_helper_rintd_exact(tcg_rd, tcg_rn, tcg_fpstatus);
153
break;
154
+ case 0x1e: /* FRINT32Z */
155
+ case 0x5e: /* FRINT32X */
156
+ gen_helper_frint32_d(tcg_rd, tcg_rn, tcg_fpstatus);
157
+ break;
158
+ case 0x1f: /* FRINT64Z */
159
+ case 0x5f: /* FRINT64X */
160
+ gen_helper_frint64_d(tcg_rd, tcg_rn, tcg_fpstatus);
161
+ break;
162
default:
163
g_assert_not_reached();
164
}
165
@@ -XXX,XX +XXX,XX @@ static void disas_simd_two_reg_misc(DisasContext *s, uint32_t insn)
166
}
167
break;
168
case 0xc ... 0xf:
169
- case 0x16 ... 0x1d:
170
- case 0x1f:
171
+ case 0x16 ... 0x1f:
172
{
173
/* Floating point: U, size[1] and opcode indicate operation;
174
* size[0] indicates single or double precision.
175
@@ -XXX,XX +XXX,XX @@ static void disas_simd_two_reg_misc(DisasContext *s, uint32_t insn)
176
}
177
need_fpstatus = true;
178
break;
179
+ case 0x1e: /* FRINT32Z */
180
+ case 0x1f: /* FRINT64Z */
181
+ need_rmode = true;
182
+ rmode = FPROUNDING_ZERO;
183
+ /* fall through */
184
+ case 0x5e: /* FRINT32X */
185
+ case 0x5f: /* FRINT64X */
186
+ need_fpstatus = true;
187
+ if ((size == 3 && !is_q) || !dc_isar_feature(aa64_frint, s)) {
188
+ unallocated_encoding(s);
189
+ return;
190
+ }
191
+ break;
192
default:
193
unallocated_encoding(s);
194
return;
195
@@ -XXX,XX +XXX,XX @@ static void disas_simd_two_reg_misc(DisasContext *s, uint32_t insn)
196
case 0x7c: /* URSQRTE */
197
gen_helper_rsqrte_u32(tcg_res, tcg_op, tcg_fpstatus);
198
break;
199
+ case 0x1e: /* FRINT32Z */
200
+ case 0x5e: /* FRINT32X */
201
+ gen_helper_frint32_s(tcg_res, tcg_op, tcg_fpstatus);
202
+ break;
203
+ case 0x1f: /* FRINT64Z */
204
+ case 0x5f: /* FRINT64X */
205
+ gen_helper_frint64_s(tcg_res, tcg_op, tcg_fpstatus);
206
+ break;
207
default:
208
g_assert_not_reached();
209
}
210
diff --git a/target/arm/vfp_helper.c b/target/arm/vfp_helper.c
211
index XXXXXXX..XXXXXXX 100644
212
--- a/target/arm/vfp_helper.c
213
+++ b/target/arm/vfp_helper.c
214
@@ -XXX,XX +XXX,XX @@ uint32_t HELPER(vjcvt)(float64 value, CPUARMState *env)
215
216
return result;
217
}
218
+
219
+/* Round a float32 to an integer that fits in int32_t or int64_t. */
220
+static float32 frint_s(float32 f, float_status *fpst, int intsize)
221
+{
222
+ int old_flags = get_float_exception_flags(fpst);
223
+ uint32_t exp = extract32(f, 23, 8);
224
+
225
+ if (unlikely(exp == 0xff)) {
226
+ /* NaN or Inf. */
227
+ goto overflow;
228
+ }
229
+
230
+ /* Round and re-extract the exponent. */
231
+ f = float32_round_to_int(f, fpst);
232
+ exp = extract32(f, 23, 8);
233
+
234
+ /* Validate the range of the result. */
235
+ if (exp < 126 + intsize) {
236
+ /* abs(F) <= INT{N}_MAX */
237
+ return f;
238
+ }
239
+ if (exp == 126 + intsize) {
240
+ uint32_t sign = extract32(f, 31, 1);
241
+ uint32_t frac = extract32(f, 0, 23);
242
+ if (sign && frac == 0) {
243
+ /* F == INT{N}_MIN */
244
+ return f;
245
+ }
246
+ }
247
+
248
+ overflow:
249
+ /*
250
+ * Raise Invalid and return INT{N}_MIN as a float. Revert any
251
+ * inexact exception float32_round_to_int may have raised.
252
+ */
253
+ set_float_exception_flags(old_flags | float_flag_invalid, fpst);
254
+ return (0x100u + 126u + intsize) << 23;
255
+}
256
+
257
+float32 HELPER(frint32_s)(float32 f, void *fpst)
258
+{
259
+ return frint_s(f, fpst, 32);
260
+}
261
+
262
+float32 HELPER(frint64_s)(float32 f, void *fpst)
263
+{
264
+ return frint_s(f, fpst, 64);
265
+}
266
+
267
+/* Round a float64 to an integer that fits in int32_t or int64_t. */
268
+static float64 frint_d(float64 f, float_status *fpst, int intsize)
269
+{
270
+ int old_flags = get_float_exception_flags(fpst);
271
+ uint32_t exp = extract64(f, 52, 11);
272
+
273
+ if (unlikely(exp == 0x7ff)) {
274
+ /* NaN or Inf. */
275
+ goto overflow;
276
+ }
277
+
278
+ /* Round and re-extract the exponent. */
279
+ f = float64_round_to_int(f, fpst);
280
+ exp = extract64(f, 52, 11);
281
+
282
+ /* Validate the range of the result. */
283
+ if (exp < 1022 + intsize) {
284
+ /* abs(F) <= INT{N}_MAX */
285
+ return f;
286
+ }
287
+ if (exp == 1022 + intsize) {
288
+ uint64_t sign = extract64(f, 63, 1);
289
+ uint64_t frac = extract64(f, 0, 52);
290
+ if (sign && frac == 0) {
291
+ /* F == INT{N}_MIN */
292
+ return f;
293
+ }
294
+ }
295
+
296
+ overflow:
297
+ /*
298
+ * Raise Invalid and return INT{N}_MIN as a float. Revert any
299
+ * inexact exception float64_round_to_int may have raised.
300
+ */
301
+ set_float_exception_flags(old_flags | float_flag_invalid, fpst);
302
+ return (uint64_t)(0x800 + 1022 + intsize) << 52;
303
+}
304
+
305
+float64 HELPER(frint32_d)(float64 f, void *fpst)
306
+{
307
+ return frint_d(f, fpst, 32);
308
+}
309
+
310
+float64 HELPER(frint64_d)(float64 f, void *fpst)
311
+{
312
+ return frint_d(f, fpst, 64);
313
+}
314
--
84
--
315
2.20.1
85
2.20.1
316
86
317
87
diff view generated by jsdifflib
[Qemu-devel] [PULL 03/22] target/arm: Implement ARMv8.0-SB
[PULL 13/16] hw/arm/musicpal: Only use qdev_get_gpio_in() when necessary
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Philippe Mathieu-Daudé <f4bug@amsat.org>
2
2
3
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
3
We don't need to fill the full pic[] array if we only use
4
Message-id: 20190301200501.16533-3-richard.henderson@linaro.org
4
few of the interrupt lines. Directly call qdev_get_gpio_in()
5
when necessary.
6
7
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
Message-id: 20201107193403.436146-6-f4bug@amsat.org
5
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7
---
11
---
8
target/arm/cpu.h | 10 ++++++++++
12
hw/arm/musicpal.c | 25 +++++++++++++------------
9
linux-user/elfload.c | 1 +
13
1 file changed, 13 insertions(+), 12 deletions(-)
10
target/arm/cpu.c | 1 +
11
target/arm/cpu64.c | 2 ++
12
target/arm/translate-a64.c | 14 ++++++++++++++
13
target/arm/translate.c | 22 ++++++++++++++++++++++
14
6 files changed, 50 insertions(+)
15
14
16
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
15
diff --git a/hw/arm/musicpal.c b/hw/arm/musicpal.c
17
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/cpu.h
17
--- a/hw/arm/musicpal.c
19
+++ b/target/arm/cpu.h
18
+++ b/hw/arm/musicpal.c
20
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa32_fhm(const ARMISARegisters *id)
19
@@ -XXX,XX +XXX,XX @@ static struct arm_boot_info musicpal_binfo = {
21
return FIELD_EX32(id->id_isar6, ID_ISAR6, FHM) != 0;
20
static void musicpal_init(MachineState *machine)
22
}
23
24
+static inline bool isar_feature_aa32_sb(const ARMISARegisters *id)
25
+{
26
+ return FIELD_EX32(id->id_isar6, ID_ISAR6, SB) != 0;
27
+}
28
+
29
static inline bool isar_feature_aa32_fp16_arith(const ARMISARegisters *id)
30
{
21
{
31
/*
22
ARMCPU *cpu;
32
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_pauth(const ARMISARegisters *id)
23
- qemu_irq pic[32];
33
FIELD_DP64(0, ID_AA64ISAR1, GPI, 0xf))) != 0;
24
DeviceState *dev;
34
}
25
+ DeviceState *pic;
35
26
DeviceState *uart_orgate;
36
+static inline bool isar_feature_aa64_sb(const ARMISARegisters *id)
27
DeviceState *i2c_dev;
37
+{
28
DeviceState *lcd_dev;
38
+ return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, SB) != 0;
29
@@ -XXX,XX +XXX,XX @@ static void musicpal_init(MachineState *machine)
39
+}
30
&error_fatal);
40
+
31
memory_region_add_subregion(address_space_mem, MP_SRAM_BASE, sram);
41
static inline bool isar_feature_aa64_fp16(const ARMISARegisters *id)
32
42
{
33
- dev = sysbus_create_simple(TYPE_MV88W8618_PIC, MP_PIC_BASE,
43
/* We always set the AdvSIMD and FP fields identically wrt FP16. */
34
+ pic = sysbus_create_simple(TYPE_MV88W8618_PIC, MP_PIC_BASE,
44
diff --git a/linux-user/elfload.c b/linux-user/elfload.c
35
qdev_get_gpio_in(DEVICE(cpu), ARM_CPU_IRQ));
45
index XXXXXXX..XXXXXXX 100644
36
- for (i = 0; i < 32; i++) {
46
--- a/linux-user/elfload.c
37
- pic[i] = qdev_get_gpio_in(dev, i);
47
+++ b/linux-user/elfload.c
38
- }
48
@@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap(void)
39
- sysbus_create_varargs(TYPE_MV88W8618_PIT, MP_PIT_BASE, pic[MP_TIMER1_IRQ],
49
GET_FEATURE_ID(aa64_pauth, ARM_HWCAP_A64_PACA | ARM_HWCAP_A64_PACG);
40
- pic[MP_TIMER2_IRQ], pic[MP_TIMER3_IRQ],
50
GET_FEATURE_ID(aa64_fhm, ARM_HWCAP_A64_ASIMDFHM);
41
- pic[MP_TIMER4_IRQ], NULL);
51
GET_FEATURE_ID(aa64_jscvt, ARM_HWCAP_A64_JSCVT);
42
+ sysbus_create_varargs(TYPE_MV88W8618_PIT, MP_PIT_BASE,
52
+ GET_FEATURE_ID(aa64_sb, ARM_HWCAP_A64_SB);
43
+ qdev_get_gpio_in(pic, MP_TIMER1_IRQ),
53
44
+ qdev_get_gpio_in(pic, MP_TIMER2_IRQ),
54
#undef GET_FEATURE_ID
45
+ qdev_get_gpio_in(pic, MP_TIMER3_IRQ),
55
46
+ qdev_get_gpio_in(pic, MP_TIMER4_IRQ), NULL);
56
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
47
57
index XXXXXXX..XXXXXXX 100644
48
/* Logically OR both UART IRQs together */
58
--- a/target/arm/cpu.c
49
uart_orgate = DEVICE(object_new(TYPE_OR_IRQ));
59
+++ b/target/arm/cpu.c
50
object_property_set_int(OBJECT(uart_orgate), "num-lines", 2, &error_fatal);
60
@@ -XXX,XX +XXX,XX @@ static void arm_max_initfn(Object *obj)
51
qdev_realize_and_unref(uart_orgate, NULL, &error_fatal);
61
t = FIELD_DP32(t, ID_ISAR6, JSCVT, 1);
52
- qdev_connect_gpio_out(DEVICE(uart_orgate), 0, pic[MP_UART_SHARED_IRQ]);
62
t = FIELD_DP32(t, ID_ISAR6, DP, 1);
53
+ qdev_connect_gpio_out(DEVICE(uart_orgate), 0,
63
t = FIELD_DP32(t, ID_ISAR6, FHM, 1);
54
+ qdev_get_gpio_in(pic, MP_UART_SHARED_IRQ));
64
+ t = FIELD_DP32(t, ID_ISAR6, SB, 1);
55
65
cpu->isar.id_isar6 = t;
56
serial_mm_init(address_space_mem, MP_UART1_BASE, 2,
66
57
qdev_get_gpio_in(uart_orgate, 0),
67
t = cpu->id_mmfr4;
58
@@ -XXX,XX +XXX,XX @@ static void musicpal_init(MachineState *machine)
68
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
59
OBJECT(get_system_memory()), &error_fatal);
69
index XXXXXXX..XXXXXXX 100644
60
sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
70
--- a/target/arm/cpu64.c
61
sysbus_mmio_map(SYS_BUS_DEVICE(dev), 0, MP_ETH_BASE);
71
+++ b/target/arm/cpu64.c
62
- sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, pic[MP_ETH_IRQ]);
72
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
63
+ sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0,
73
t = FIELD_DP64(t, ID_AA64ISAR1, API, 0);
64
+ qdev_get_gpio_in(pic, MP_ETH_IRQ));
74
t = FIELD_DP64(t, ID_AA64ISAR1, GPA, 1);
65
75
t = FIELD_DP64(t, ID_AA64ISAR1, GPI, 0);
66
sysbus_create_simple("mv88w8618_wlan", MP_WLAN_BASE, NULL);
76
+ t = FIELD_DP64(t, ID_AA64ISAR1, SB, 1);
67
77
cpu->isar.id_aa64isar1 = t;
68
sysbus_create_simple(TYPE_MUSICPAL_MISC, MP_MISC_BASE, NULL);
78
69
79
t = cpu->isar.id_aa64pfr0;
70
dev = sysbus_create_simple(TYPE_MUSICPAL_GPIO, MP_GPIO_BASE,
80
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
71
- pic[MP_GPIO_IRQ]);
81
u = FIELD_DP32(u, ID_ISAR6, JSCVT, 1);
72
+ qdev_get_gpio_in(pic, MP_GPIO_IRQ));
82
u = FIELD_DP32(u, ID_ISAR6, DP, 1);
73
i2c_dev = sysbus_create_simple("gpio_i2c", -1, NULL);
83
u = FIELD_DP32(u, ID_ISAR6, FHM, 1);
74
i2c = (I2CBus *)qdev_get_child_bus(i2c_dev, "i2c");
84
+ u = FIELD_DP32(u, ID_ISAR6, SB, 1);
75
85
cpu->isar.id_isar6 = u;
76
@@ -XXX,XX +XXX,XX @@ static void musicpal_init(MachineState *machine)
86
77
NULL);
87
/*
78
sysbus_realize_and_unref(s, &error_fatal);
88
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
79
sysbus_mmio_map(s, 0, MP_AUDIO_BASE);
89
index XXXXXXX..XXXXXXX 100644
80
- sysbus_connect_irq(s, 0, pic[MP_AUDIO_IRQ]);
90
--- a/target/arm/translate-a64.c
81
+ sysbus_connect_irq(s, 0, qdev_get_gpio_in(pic, MP_AUDIO_IRQ));
91
+++ b/target/arm/translate-a64.c
82
92
@@ -XXX,XX +XXX,XX @@ static void handle_sync(DisasContext *s, uint32_t insn,
83
musicpal_binfo.ram_size = MP_RAM_DEFAULT_SIZE;
93
reset_btype(s);
84
arm_load_kernel(cpu, machine, &musicpal_binfo);
94
gen_goto_tb(s, 0, s->pc);
95
return;
96
+
97
+ case 7: /* SB */
98
+ if (crm != 0 || !dc_isar_feature(aa64_sb, s)) {
99
+ goto do_unallocated;
100
+ }
101
+ /*
102
+ * TODO: There is no speculation barrier opcode for TCG;
103
+ * MB and end the TB instead.
104
+ */
105
+ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC);
106
+ gen_goto_tb(s, 0, s->pc);
107
+ return;
108
+
109
default:
110
+ do_unallocated:
111
unallocated_encoding(s);
112
return;
113
}
114
diff --git a/target/arm/translate.c b/target/arm/translate.c
115
index XXXXXXX..XXXXXXX 100644
116
--- a/target/arm/translate.c
117
+++ b/target/arm/translate.c
118
@@ -XXX,XX +XXX,XX @@ static void disas_arm_insn(DisasContext *s, unsigned int insn)
119
*/
120
gen_goto_tb(s, 0, s->pc & ~1);
121
return;
122
+ case 7: /* sb */
123
+ if ((insn & 0xf) || !dc_isar_feature(aa32_sb, s)) {
124
+ goto illegal_op;
125
+ }
126
+ /*
127
+ * TODO: There is no speculation barrier opcode
128
+ * for TCG; MB and end the TB instead.
129
+ */
130
+ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC);
131
+ gen_goto_tb(s, 0, s->pc & ~1);
132
+ return;
133
default:
134
goto illegal_op;
135
}
136
@@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn)
137
*/
138
gen_goto_tb(s, 0, s->pc & ~1);
139
break;
140
+ case 7: /* sb */
141
+ if ((insn & 0xf) || !dc_isar_feature(aa32_sb, s)) {
142
+ goto illegal_op;
143
+ }
144
+ /*
145
+ * TODO: There is no speculation barrier opcode
146
+ * for TCG; MB and end the TB instead.
147
+ */
148
+ tcg_gen_mb(TCG_MO_ALL | TCG_BAR_SC);
149
+ gen_goto_tb(s, 0, s->pc & ~1);
150
+ break;
151
default:
152
goto illegal_op;
153
}
154
--
85
--
155
2.20.1
86
2.20.1
156
87
157
88
diff view generated by jsdifflib
[Qemu-devel] [PULL 08/22] target/arm: Implement ARMv8.4-CondM
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
Tested-by: Laurent Desnogues <laurent.desnogues@gmail.com>
4
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
5
Message-id: 20190301200501.16533-8-richard.henderson@linaro.org
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
[PMM: fixed up block comment style]
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
---
10
target/arm/cpu.h | 5 ++
11
linux-user/elfload.c | 1 +
12
target/arm/cpu64.c | 1 +
13
target/arm/translate-a64.c | 99 +++++++++++++++++++++++++++++++++++++-
14
4 files changed, 105 insertions(+), 1 deletion(-)
15
16
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
17
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/cpu.h
19
+++ b/target/arm/cpu.h
20
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_fhm(const ARMISARegisters *id)
21
return FIELD_EX64(id->id_aa64isar0, ID_AA64ISAR0, FHM) != 0;
22
}
23
24
+static inline bool isar_feature_aa64_condm_4(const ARMISARegisters *id)
25
+{
26
+ return FIELD_EX64(id->id_aa64isar0, ID_AA64ISAR0, TS) != 0;
27
+}
28
+
29
static inline bool isar_feature_aa64_jscvt(const ARMISARegisters *id)
30
{
31
return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, JSCVT) != 0;
32
diff --git a/linux-user/elfload.c b/linux-user/elfload.c
33
index XXXXXXX..XXXXXXX 100644
34
--- a/linux-user/elfload.c
35
+++ b/linux-user/elfload.c
36
@@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap(void)
37
GET_FEATURE_ID(aa64_fhm, ARM_HWCAP_A64_ASIMDFHM);
38
GET_FEATURE_ID(aa64_jscvt, ARM_HWCAP_A64_JSCVT);
39
GET_FEATURE_ID(aa64_sb, ARM_HWCAP_A64_SB);
40
+ GET_FEATURE_ID(aa64_condm_4, ARM_HWCAP_A64_FLAGM);
41
42
#undef GET_FEATURE_ID
43
44
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
45
index XXXXXXX..XXXXXXX 100644
46
--- a/target/arm/cpu64.c
47
+++ b/target/arm/cpu64.c
48
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
49
t = FIELD_DP64(t, ID_AA64ISAR0, SM4, 1);
50
t = FIELD_DP64(t, ID_AA64ISAR0, DP, 1);
51
t = FIELD_DP64(t, ID_AA64ISAR0, FHM, 1);
52
+ t = FIELD_DP64(t, ID_AA64ISAR0, TS, 1);
53
cpu->isar.id_aa64isar0 = t;
54
55
t = cpu->isar.id_aa64isar1;
56
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
57
index XXXXXXX..XXXXXXX 100644
58
--- a/target/arm/translate-a64.c
59
+++ b/target/arm/translate-a64.c
60
@@ -XXX,XX +XXX,XX @@ static void handle_msr_i(DisasContext *s, uint32_t insn,
61
s->base.is_jmp = DISAS_TOO_MANY;
62
63
switch (op) {
64
+ case 0x00: /* CFINV */
65
+ if (crm != 0 || !dc_isar_feature(aa64_condm_4, s)) {
66
+ goto do_unallocated;
67
+ }
68
+ tcg_gen_xori_i32(cpu_CF, cpu_CF, 1);
69
+ s->base.is_jmp = DISAS_NEXT;
70
+ break;
71
+
72
case 0x05: /* SPSel */
73
if (s->current_el == 0) {
74
goto do_unallocated;
75
@@ -XXX,XX +XXX,XX @@ static void gen_get_nzcv(TCGv_i64 tcg_rt)
76
}
77
78
static void gen_set_nzcv(TCGv_i64 tcg_rt)
79
-
80
{
81
TCGv_i32 nzcv = tcg_temp_new_i32();
82
83
@@ -XXX,XX +XXX,XX @@ static void disas_adc_sbc(DisasContext *s, uint32_t insn)
84
}
85
}
86
87
+/*
88
+ * Rotate right into flags
89
+ * 31 30 29 21 15 10 5 4 0
90
+ * +--+--+--+-----------------+--------+-----------+------+--+------+
91
+ * |sf|op| S| 1 1 0 1 0 0 0 0 | imm6 | 0 0 0 0 1 | Rn |o2| mask |
92
+ * +--+--+--+-----------------+--------+-----------+------+--+------+
93
+ */
94
+static void disas_rotate_right_into_flags(DisasContext *s, uint32_t insn)
95
+{
96
+ int mask = extract32(insn, 0, 4);
97
+ int o2 = extract32(insn, 4, 1);
98
+ int rn = extract32(insn, 5, 5);
99
+ int imm6 = extract32(insn, 15, 6);
100
+ int sf_op_s = extract32(insn, 29, 3);
101
+ TCGv_i64 tcg_rn;
102
+ TCGv_i32 nzcv;
103
+
104
+ if (sf_op_s != 5 || o2 != 0 || !dc_isar_feature(aa64_condm_4, s)) {
105
+ unallocated_encoding(s);
106
+ return;
107
+ }
108
+
109
+ tcg_rn = read_cpu_reg(s, rn, 1);
110
+ tcg_gen_rotri_i64(tcg_rn, tcg_rn, imm6);
111
+
112
+ nzcv = tcg_temp_new_i32();
113
+ tcg_gen_extrl_i64_i32(nzcv, tcg_rn);
114
+
115
+ if (mask & 8) { /* N */
116
+ tcg_gen_shli_i32(cpu_NF, nzcv, 31 - 3);
117
+ }
118
+ if (mask & 4) { /* Z */
119
+ tcg_gen_not_i32(cpu_ZF, nzcv);
120
+ tcg_gen_andi_i32(cpu_ZF, cpu_ZF, 4);
121
+ }
122
+ if (mask & 2) { /* C */
123
+ tcg_gen_extract_i32(cpu_CF, nzcv, 1, 1);
124
+ }
125
+ if (mask & 1) { /* V */
126
+ tcg_gen_shli_i32(cpu_VF, nzcv, 31 - 0);
127
+ }
128
+
129
+ tcg_temp_free_i32(nzcv);
130
+}
131
+
132
+/*
133
+ * Evaluate into flags
134
+ * 31 30 29 21 15 14 10 5 4 0
135
+ * +--+--+--+-----------------+---------+----+---------+------+--+------+
136
+ * |sf|op| S| 1 1 0 1 0 0 0 0 | opcode2 | sz | 0 0 1 0 | Rn |o3| mask |
137
+ * +--+--+--+-----------------+---------+----+---------+------+--+------+
138
+ */
139
+static void disas_evaluate_into_flags(DisasContext *s, uint32_t insn)
140
+{
141
+ int o3_mask = extract32(insn, 0, 5);
142
+ int rn = extract32(insn, 5, 5);
143
+ int o2 = extract32(insn, 15, 6);
144
+ int sz = extract32(insn, 14, 1);
145
+ int sf_op_s = extract32(insn, 29, 3);
146
+ TCGv_i32 tmp;
147
+ int shift;
148
+
149
+ if (sf_op_s != 1 || o2 != 0 || o3_mask != 0xd ||
150
+ !dc_isar_feature(aa64_condm_4, s)) {
151
+ unallocated_encoding(s);
152
+ return;
153
+ }
154
+ shift = sz ? 16 : 24; /* SETF16 or SETF8 */
155
+
156
+ tmp = tcg_temp_new_i32();
157
+ tcg_gen_extrl_i64_i32(tmp, cpu_reg(s, rn));
158
+ tcg_gen_shli_i32(cpu_NF, tmp, shift);
159
+ tcg_gen_shli_i32(cpu_VF, tmp, shift - 1);
160
+ tcg_gen_mov_i32(cpu_ZF, cpu_NF);
161
+ tcg_gen_xor_i32(cpu_VF, cpu_VF, cpu_NF);
162
+ tcg_temp_free_i32(tmp);
163
+}
164
+
165
/* Conditional compare (immediate / register)
166
* 31 30 29 28 27 26 25 24 23 22 21 20 16 15 12 11 10 9 5 4 3 0
167
* +--+--+--+------------------------+--------+------+----+--+------+--+-----+
168
@@ -XXX,XX +XXX,XX @@ static void disas_data_proc_reg(DisasContext *s, uint32_t insn)
169
disas_adc_sbc(s, insn);
170
break;
171
172
+ case 0x01: /* Rotate right into flags */
173
+ case 0x21:
174
+ disas_rotate_right_into_flags(s, insn);
175
+ break;
176
+
177
+ case 0x02: /* Evaluate into flags */
178
+ case 0x12:
179
+ case 0x22:
180
+ case 0x32:
181
+ disas_evaluate_into_flags(s, insn);
182
+ break;
183
+
184
default:
185
goto do_unallocated;
186
}
187
--
188
2.20.1
189
190
diff view generated by jsdifflib
[Qemu-devel] [PULL 19/22] hw/arm/virt: Implement kvm_type function for 4.0 machine
[PULL 14/16] hw/arm/nseries: Check return value from load_image_targphys()
1
From: Eric Auger <eric.auger@redhat.com>
1
The nseries machines have a codepath that allows them to load a
2
secondary bootloader. This code wasn't checking that the
3
load_image_targphys() succeeded. Check the return value and report
4
the error to the user.
2
5
3
This patch implements the machine class kvm_type() callback.
6
While we're in the vicinity, fix the comment style of the
4
It returns the number of bits requested to implement the whole GPA
7
comment documenting what this image load is doing.
5
range including the RAM and IO regions located beyond.
6
The returned value is passed though the KVM_CREATE_VM ioctl and
7
this allows KVM to set the stage2 tables dynamically.
8
8
9
To compute the highest GPA used in the memory map, kvm_type()
9
Fixes: Coverity CID 1192904
10
must freeze the memory map by calling virt_set_memmap().
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
12
Message-id: 20201103114918.11807-1-peter.maydell@linaro.org
13
---
14
hw/arm/nseries.c | 15 +++++++++++----
15
1 file changed, 11 insertions(+), 4 deletions(-)
11
16
12
Signed-off-by: Eric Auger <eric.auger@redhat.com>
17
diff --git a/hw/arm/nseries.c b/hw/arm/nseries.c
13
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
14
Message-id: 20190304101339.25970-9-eric.auger@redhat.com
15
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
---
17
hw/arm/virt.c | 39 ++++++++++++++++++++++++++++++++++++++-
18
1 file changed, 38 insertions(+), 1 deletion(-)
19
20
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
21
index XXXXXXX..XXXXXXX 100644
18
index XXXXXXX..XXXXXXX 100644
22
--- a/hw/arm/virt.c
19
--- a/hw/arm/nseries.c
23
+++ b/hw/arm/virt.c
20
+++ b/hw/arm/nseries.c
24
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
21
@@ -XXX,XX +XXX,XX @@ static void n8x0_init(MachineState *machine,
25
bool firmware_loaded = bios_name || drive_get(IF_PFLASH, 0, 0);
22
/* No, wait, better start at the ROM. */
26
bool aarch64 = true;
23
s->mpu->cpu->env.regs[15] = OMAP2_Q2_BASE + 0x400000;
27
24
28
- virt_set_memmap(vms);
25
- /* This is intended for loading the `secondary.bin' program from
29
+ /*
26
+ /*
30
+ * In accelerated mode, the memory map is computed earlier in kvm_type()
27
+ * This is intended for loading the `secondary.bin' program from
31
+ * to create a VM with the right number of IPA bits.
28
* Nokia images (the NOLO bootloader). The entry point seems
32
+ */
29
* to be at OMAP2_Q2_BASE + 0x400000.
33
+ if (!vms->memmap) {
30
*
34
+ virt_set_memmap(vms);
31
@@ -XXX,XX +XXX,XX @@ static void n8x0_init(MachineState *machine,
35
+ }
32
* for them the entry point needs to be set to OMAP2_SRAM_BASE.
36
33
*
37
/* We can probe only here because during property set
34
* The code above is for loading the `zImage' file from Nokia
38
* KVM is not available yet
35
- * images. */
39
@@ -XXX,XX +XXX,XX @@ static HotplugHandler *virt_machine_get_hotplug_handler(MachineState *machine,
36
- load_image_targphys(option_rom[0].name, OMAP2_Q2_BASE + 0x400000,
40
return NULL;
37
- machine->ram_size - 0x400000);
41
}
38
+ * images.
42
39
+ */
43
+/*
40
+ if (load_image_targphys(option_rom[0].name,
44
+ * for arm64 kvm_type [7-0] encodes the requested number of bits
41
+ OMAP2_Q2_BASE + 0x400000,
45
+ * in the IPA address space
42
+ machine->ram_size - 0x400000) < 0) {
46
+ */
43
+ error_report("Failed to load secondary bootloader %s",
47
+static int virt_kvm_type(MachineState *ms, const char *type_str)
44
+ option_rom[0].name);
48
+{
45
+ exit(EXIT_FAILURE);
49
+ VirtMachineState *vms = VIRT_MACHINE(ms);
46
+ }
50
+ int max_vm_pa_size = kvm_arm_get_max_vm_ipa_size(ms);
47
51
+ int requested_pa_size;
48
n800_setup_nolo_tags(nolo_tags);
52
+
49
cpu_physical_memory_write(OMAP2_SRAM_BASE, nolo_tags, 0x10000);
53
+ /* we freeze the memory map to compute the highest gpa */
54
+ virt_set_memmap(vms);
55
+
56
+ requested_pa_size = 64 - clz64(vms->highest_gpa);
57
+
58
+ if (requested_pa_size > max_vm_pa_size) {
59
+ error_report("-m and ,maxmem option values "
60
+ "require an IPA range (%d bits) larger than "
61
+ "the one supported by the host (%d bits)",
62
+ requested_pa_size, max_vm_pa_size);
63
+ exit(1);
64
+ }
65
+ /*
66
+ * By default we return 0 which corresponds to an implicit legacy
67
+ * 40b IPA setting. Otherwise we return the actual requested PA
68
+ * logsize
69
+ */
70
+ return requested_pa_size > 40 ? requested_pa_size : 0;
71
+}
72
+
73
static void virt_machine_class_init(ObjectClass *oc, void *data)
74
{
75
MachineClass *mc = MACHINE_CLASS(oc);
76
@@ -XXX,XX +XXX,XX @@ static void virt_machine_class_init(ObjectClass *oc, void *data)
77
mc->cpu_index_to_instance_props = virt_cpu_index_to_props;
78
mc->default_cpu_type = ARM_CPU_TYPE_NAME("cortex-a15");
79
mc->get_default_cpu_node_id = virt_get_default_cpu_node_id;
80
+ mc->kvm_type = virt_kvm_type;
81
assert(!mc->get_hotplug_handler);
82
mc->get_hotplug_handler = virt_machine_get_hotplug_handler;
83
hc->plug = virt_machine_device_plug_cb;
84
--
50
--
85
2.20.1
51
2.20.1
86
52
87
53
diff view generated by jsdifflib
[Qemu-devel] [PULL 09/22] target/arm: Implement ARMv8.5-CondM
[PULL 15/16] tests/qtest/npcm7xx_rng-test: count runs properly
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Havard Skinnemoen <hskinnemoen@google.com>
2
2
3
Tested-by: Laurent Desnogues <laurent.desnogues@gmail.com>
3
The number of runs is equal to the number of 0-1 and 1-0 transitions,
4
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
4
plus one. Currently, it's counting the number of times these transitions
5
Message-id: 20190301200501.16533-9-richard.henderson@linaro.org
5
do _not_ happen, plus one.
6
7
Source:
8
https://nvlpubs.nist.gov/nistpubs/Legacy/SP/nistspecialpublication800-22r1a.pdf
9
section 2.3.4 point (3).
10
11
Signed-off-by: Havard Skinnemoen <hskinnemoen@google.com>
12
Message-id: 20201103011457.2959989-2-hskinnemoen@google.com
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
13
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
---
15
---
9
target/arm/cpu.h | 5 ++++
16
tests/qtest/npcm7xx_rng-test.c | 2 +-
10
target/arm/cpu64.c | 2 +-
17
1 file changed, 1 insertion(+), 1 deletion(-)
11
target/arm/translate-a64.c | 58 ++++++++++++++++++++++++++++++++++++++
12
3 files changed, 64 insertions(+), 1 deletion(-)
13
18
14
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
19
diff --git a/tests/qtest/npcm7xx_rng-test.c b/tests/qtest/npcm7xx_rng-test.c
15
index XXXXXXX..XXXXXXX 100644
20
index XXXXXXX..XXXXXXX 100644
16
--- a/target/arm/cpu.h
21
--- a/tests/qtest/npcm7xx_rng-test.c
17
+++ b/target/arm/cpu.h
22
+++ b/tests/qtest/npcm7xx_rng-test.c
18
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_condm_4(const ARMISARegisters *id)
23
@@ -XXX,XX +XXX,XX @@ static double calc_runs_p(const unsigned long *buf, unsigned int nr_bits)
19
return FIELD_EX64(id->id_aa64isar0, ID_AA64ISAR0, TS) != 0;
24
pi = (double)nr_ones / nr_bits;
20
}
25
21
26
for (k = 0; k < nr_bits - 1; k++) {
22
+static inline bool isar_feature_aa64_condm_5(const ARMISARegisters *id)
27
- vn_obs += !(test_bit(k, buf) ^ test_bit(k + 1, buf));
23
+{
28
+ vn_obs += (test_bit(k, buf) ^ test_bit(k + 1, buf));
24
+ return FIELD_EX64(id->id_aa64isar0, ID_AA64ISAR0, TS) >= 2;
25
+}
26
+
27
static inline bool isar_feature_aa64_jscvt(const ARMISARegisters *id)
28
{
29
return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, JSCVT) != 0;
30
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
31
index XXXXXXX..XXXXXXX 100644
32
--- a/target/arm/cpu64.c
33
+++ b/target/arm/cpu64.c
34
@@ -XXX,XX +XXX,XX @@ static void aarch64_max_initfn(Object *obj)
35
t = FIELD_DP64(t, ID_AA64ISAR0, SM4, 1);
36
t = FIELD_DP64(t, ID_AA64ISAR0, DP, 1);
37
t = FIELD_DP64(t, ID_AA64ISAR0, FHM, 1);
38
- t = FIELD_DP64(t, ID_AA64ISAR0, TS, 1);
39
+ t = FIELD_DP64(t, ID_AA64ISAR0, TS, 2); /* v8.5-CondM */
40
cpu->isar.id_aa64isar0 = t;
41
42
t = cpu->isar.id_aa64isar1;
43
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
44
index XXXXXXX..XXXXXXX 100644
45
--- a/target/arm/translate-a64.c
46
+++ b/target/arm/translate-a64.c
47
@@ -XXX,XX +XXX,XX @@ static void handle_sync(DisasContext *s, uint32_t insn,
48
}
29
}
49
}
30
vn_obs += 1;
50
31
51
+static void gen_xaflag(void)
52
+{
53
+ TCGv_i32 z = tcg_temp_new_i32();
54
+
55
+ tcg_gen_setcondi_i32(TCG_COND_EQ, z, cpu_ZF, 0);
56
+
57
+ /*
58
+ * (!C & !Z) << 31
59
+ * (!(C | Z)) << 31
60
+ * ~((C | Z) << 31)
61
+ * ~-(C | Z)
62
+ * (C | Z) - 1
63
+ */
64
+ tcg_gen_or_i32(cpu_NF, cpu_CF, z);
65
+ tcg_gen_subi_i32(cpu_NF, cpu_NF, 1);
66
+
67
+ /* !(Z & C) */
68
+ tcg_gen_and_i32(cpu_ZF, z, cpu_CF);
69
+ tcg_gen_xori_i32(cpu_ZF, cpu_ZF, 1);
70
+
71
+ /* (!C & Z) << 31 -> -(Z & ~C) */
72
+ tcg_gen_andc_i32(cpu_VF, z, cpu_CF);
73
+ tcg_gen_neg_i32(cpu_VF, cpu_VF);
74
+
75
+ /* C | Z */
76
+ tcg_gen_or_i32(cpu_CF, cpu_CF, z);
77
+
78
+ tcg_temp_free_i32(z);
79
+}
80
+
81
+static void gen_axflag(void)
82
+{
83
+ tcg_gen_sari_i32(cpu_VF, cpu_VF, 31); /* V ? -1 : 0 */
84
+ tcg_gen_andc_i32(cpu_CF, cpu_CF, cpu_VF); /* C & !V */
85
+
86
+ /* !(Z | V) -> !(!ZF | V) -> ZF & !V -> ZF & ~VF */
87
+ tcg_gen_andc_i32(cpu_ZF, cpu_ZF, cpu_VF);
88
+
89
+ tcg_gen_movi_i32(cpu_NF, 0);
90
+ tcg_gen_movi_i32(cpu_VF, 0);
91
+}
92
+
93
/* MSR (immediate) - move immediate to processor state field */
94
static void handle_msr_i(DisasContext *s, uint32_t insn,
95
unsigned int op1, unsigned int op2, unsigned int crm)
96
@@ -XXX,XX +XXX,XX @@ static void handle_msr_i(DisasContext *s, uint32_t insn,
97
s->base.is_jmp = DISAS_NEXT;
98
break;
99
100
+ case 0x01: /* XAFlag */
101
+ if (crm != 0 || !dc_isar_feature(aa64_condm_5, s)) {
102
+ goto do_unallocated;
103
+ }
104
+ gen_xaflag();
105
+ s->base.is_jmp = DISAS_NEXT;
106
+ break;
107
+
108
+ case 0x02: /* AXFlag */
109
+ if (crm != 0 || !dc_isar_feature(aa64_condm_5, s)) {
110
+ goto do_unallocated;
111
+ }
112
+ gen_axflag();
113
+ s->base.is_jmp = DISAS_NEXT;
114
+ break;
115
+
116
case 0x05: /* SPSel */
117
if (s->current_el == 0) {
118
goto do_unallocated;
119
--
32
--
120
2.20.1
33
2.20.1
121
34
122
35
diff view generated by jsdifflib
[Qemu-devel] [PULL 12/22] hw/arm/boot: introduce fdt_add_memory_node helper
Deleted patch
1
From: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
2
1
3
We introduce an helper to create a memory node.
4
5
Signed-off-by: Eric Auger <eric.auger@redhat.com>
6
Signed-off-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
7
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
8
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
9
Message-id: 20190304101339.25970-2-eric.auger@redhat.com
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
hw/arm/boot.c | 54 ++++++++++++++++++++++++++++++++-------------------
13
1 file changed, 34 insertions(+), 20 deletions(-)
14
15
diff --git a/hw/arm/boot.c b/hw/arm/boot.c
16
index XXXXXXX..XXXXXXX 100644
17
--- a/hw/arm/boot.c
18
+++ b/hw/arm/boot.c
19
@@ -XXX,XX +XXX,XX @@ static void set_kernel_args_old(const struct arm_boot_info *info,
20
}
21
}
22
23
+static int fdt_add_memory_node(void *fdt, uint32_t acells, hwaddr mem_base,
24
+ uint32_t scells, hwaddr mem_len,
25
+ int numa_node_id)
26
+{
27
+ char *nodename;
28
+ int ret;
29
+
30
+ nodename = g_strdup_printf("/memory@%" PRIx64, mem_base);
31
+ qemu_fdt_add_subnode(fdt, nodename);
32
+ qemu_fdt_setprop_string(fdt, nodename, "device_type", "memory");
33
+ ret = qemu_fdt_setprop_sized_cells(fdt, nodename, "reg", acells, mem_base,
34
+ scells, mem_len);
35
+ if (ret < 0) {
36
+ goto out;
37
+ }
38
+
39
+ /* only set the NUMA ID if it is specified */
40
+ if (numa_node_id >= 0) {
41
+ ret = qemu_fdt_setprop_cell(fdt, nodename,
42
+ "numa-node-id", numa_node_id);
43
+ }
44
+out:
45
+ g_free(nodename);
46
+ return ret;
47
+}
48
+
49
static void fdt_add_psci_node(void *fdt)
50
{
51
uint32_t cpu_suspend_fn;
52
@@ -XXX,XX +XXX,XX @@ int arm_load_dtb(hwaddr addr, const struct arm_boot_info *binfo,
53
void *fdt = NULL;
54
int size, rc, n = 0;
55
uint32_t acells, scells;
56
- char *nodename;
57
unsigned int i;
58
hwaddr mem_base, mem_len;
59
char **node_path;
60
@@ -XXX,XX +XXX,XX @@ int arm_load_dtb(hwaddr addr, const struct arm_boot_info *binfo,
61
mem_base = binfo->loader_start;
62
for (i = 0; i < nb_numa_nodes; i++) {
63
mem_len = numa_info[i].node_mem;
64
- nodename = g_strdup_printf("/memory@%" PRIx64, mem_base);
65
- qemu_fdt_add_subnode(fdt, nodename);
66
- qemu_fdt_setprop_string(fdt, nodename, "device_type", "memory");
67
- rc = qemu_fdt_setprop_sized_cells(fdt, nodename, "reg",
68
- acells, mem_base,
69
- scells, mem_len);
70
+ rc = fdt_add_memory_node(fdt, acells, mem_base,
71
+ scells, mem_len, i);
72
if (rc < 0) {
73
- fprintf(stderr, "couldn't set %s/reg for node %d\n", nodename,
74
- i);
75
+ fprintf(stderr, "couldn't add /memory@%"PRIx64" node\n",
76
+ mem_base);
77
goto fail;
78
}
79
80
- qemu_fdt_setprop_cell(fdt, nodename, "numa-node-id", i);
81
mem_base += mem_len;
82
- g_free(nodename);
83
}
84
} else {
85
- nodename = g_strdup_printf("/memory@%" PRIx64, binfo->loader_start);
86
- qemu_fdt_add_subnode(fdt, nodename);
87
- qemu_fdt_setprop_string(fdt, nodename, "device_type", "memory");
88
-
89
- rc = qemu_fdt_setprop_sized_cells(fdt, nodename, "reg",
90
- acells, binfo->loader_start,
91
- scells, binfo->ram_size);
92
+ rc = fdt_add_memory_node(fdt, acells, binfo->loader_start,
93
+ scells, binfo->ram_size, -1);
94
if (rc < 0) {
95
- fprintf(stderr, "couldn't set %s reg\n", nodename);
96
+ fprintf(stderr, "couldn't add /memory@%"PRIx64" node\n",
97
+ binfo->loader_start);
98
goto fail;
99
}
100
- g_free(nodename);
101
}
102
103
rc = fdt_path_offset(fdt, "/chosen");
104
--
105
2.20.1
106
107
diff view generated by jsdifflib
[Qemu-devel] [PULL 13/22] hw/arm/virt: Rename highmem IO regions
Deleted patch
1
From: Eric Auger <eric.auger@redhat.com>
2
1
3
In preparation for a split of the memory map into a static
4
part and a dynamic part floating after the RAM, let's rename the
5
regions located after the RAM
6
7
Signed-off-by: Eric Auger <eric.auger@redhat.com>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
10
Message-id: 20190304101339.25970-3-eric.auger@redhat.com
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
---
13
include/hw/arm/virt.h | 8 ++++----
14
hw/arm/virt-acpi-build.c | 10 ++++++----
15
hw/arm/virt.c | 33 ++++++++++++++++++---------------
16
3 files changed, 28 insertions(+), 23 deletions(-)
17
18
diff --git a/include/hw/arm/virt.h b/include/hw/arm/virt.h
19
index XXXXXXX..XXXXXXX 100644
20
--- a/include/hw/arm/virt.h
21
+++ b/include/hw/arm/virt.h
22
@@ -XXX,XX +XXX,XX @@ enum {
23
VIRT_GIC_VCPU,
24
VIRT_GIC_ITS,
25
VIRT_GIC_REDIST,
26
- VIRT_GIC_REDIST2,
27
+ VIRT_HIGH_GIC_REDIST2,
28
VIRT_SMMU,
29
VIRT_UART,
30
VIRT_MMIO,
31
@@ -XXX,XX +XXX,XX @@ enum {
32
VIRT_PCIE_MMIO,
33
VIRT_PCIE_PIO,
34
VIRT_PCIE_ECAM,
35
- VIRT_PCIE_ECAM_HIGH,
36
+ VIRT_HIGH_PCIE_ECAM,
37
VIRT_PLATFORM_BUS,
38
- VIRT_PCIE_MMIO_HIGH,
39
+ VIRT_HIGH_PCIE_MMIO,
40
VIRT_GPIO,
41
VIRT_SECURE_UART,
42
VIRT_SECURE_MEM,
43
@@ -XXX,XX +XXX,XX @@ typedef struct {
44
int psci_conduit;
45
} VirtMachineState;
46
47
-#define VIRT_ECAM_ID(high) (high ? VIRT_PCIE_ECAM_HIGH : VIRT_PCIE_ECAM)
48
+#define VIRT_ECAM_ID(high) (high ? VIRT_HIGH_PCIE_ECAM : VIRT_PCIE_ECAM)
49
50
#define TYPE_VIRT_MACHINE MACHINE_TYPE_NAME("virt")
51
#define VIRT_MACHINE(obj) \
52
diff --git a/hw/arm/virt-acpi-build.c b/hw/arm/virt-acpi-build.c
53
index XXXXXXX..XXXXXXX 100644
54
--- a/hw/arm/virt-acpi-build.c
55
+++ b/hw/arm/virt-acpi-build.c
56
@@ -XXX,XX +XXX,XX @@ static void acpi_dsdt_add_pci(Aml *scope, const MemMapEntry *memmap,
57
size_pio));
58
59
if (use_highmem) {
60
- hwaddr base_mmio_high = memmap[VIRT_PCIE_MMIO_HIGH].base;
61
- hwaddr size_mmio_high = memmap[VIRT_PCIE_MMIO_HIGH].size;
62
+ hwaddr base_mmio_high = memmap[VIRT_HIGH_PCIE_MMIO].base;
63
+ hwaddr size_mmio_high = memmap[VIRT_HIGH_PCIE_MMIO].size;
64
65
aml_append(rbuf,
66
aml_qword_memory(AML_POS_DECODE, AML_MIN_FIXED, AML_MAX_FIXED,
67
@@ -XXX,XX +XXX,XX @@ build_madt(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
68
gicr = acpi_data_push(table_data, sizeof(*gicr));
69
gicr->type = ACPI_APIC_GENERIC_REDISTRIBUTOR;
70
gicr->length = sizeof(*gicr);
71
- gicr->base_address = cpu_to_le64(memmap[VIRT_GIC_REDIST2].base);
72
- gicr->range_length = cpu_to_le32(memmap[VIRT_GIC_REDIST2].size);
73
+ gicr->base_address =
74
+ cpu_to_le64(memmap[VIRT_HIGH_GIC_REDIST2].base);
75
+ gicr->range_length =
76
+ cpu_to_le32(memmap[VIRT_HIGH_GIC_REDIST2].size);
77
}
78
79
if (its_class_name() && !vmc->no_its) {
80
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
81
index XXXXXXX..XXXXXXX 100644
82
--- a/hw/arm/virt.c
83
+++ b/hw/arm/virt.c
84
@@ -XXX,XX +XXX,XX @@ static const MemMapEntry a15memmap[] = {
85
[VIRT_PCIE_ECAM] = { 0x3f000000, 0x01000000 },
86
[VIRT_MEM] = { 0x40000000, RAMLIMIT_BYTES },
87
/* Additional 64 MB redist region (can contain up to 512 redistributors) */
88
- [VIRT_GIC_REDIST2] = { 0x4000000000ULL, 0x4000000 },
89
- [VIRT_PCIE_ECAM_HIGH] = { 0x4010000000ULL, 0x10000000 },
90
+ [VIRT_HIGH_GIC_REDIST2] = { 0x4000000000ULL, 0x4000000 },
91
+ [VIRT_HIGH_PCIE_ECAM] = { 0x4010000000ULL, 0x10000000 },
92
/* Second PCIe window, 512GB wide at the 512GB boundary */
93
- [VIRT_PCIE_MMIO_HIGH] = { 0x8000000000ULL, 0x8000000000ULL },
94
+ [VIRT_HIGH_PCIE_MMIO] = { 0x8000000000ULL, 0x8000000000ULL },
95
};
96
97
static const int a15irqmap[] = {
98
@@ -XXX,XX +XXX,XX @@ static void fdt_add_gic_node(VirtMachineState *vms)
99
2, vms->memmap[VIRT_GIC_REDIST].size);
100
} else {
101
qemu_fdt_setprop_sized_cells(vms->fdt, nodename, "reg",
102
- 2, vms->memmap[VIRT_GIC_DIST].base,
103
- 2, vms->memmap[VIRT_GIC_DIST].size,
104
- 2, vms->memmap[VIRT_GIC_REDIST].base,
105
- 2, vms->memmap[VIRT_GIC_REDIST].size,
106
- 2, vms->memmap[VIRT_GIC_REDIST2].base,
107
- 2, vms->memmap[VIRT_GIC_REDIST2].size);
108
+ 2, vms->memmap[VIRT_GIC_DIST].base,
109
+ 2, vms->memmap[VIRT_GIC_DIST].size,
110
+ 2, vms->memmap[VIRT_GIC_REDIST].base,
111
+ 2, vms->memmap[VIRT_GIC_REDIST].size,
112
+ 2, vms->memmap[VIRT_HIGH_GIC_REDIST2].base,
113
+ 2, vms->memmap[VIRT_HIGH_GIC_REDIST2].size);
114
}
115
116
if (vms->virt) {
117
@@ -XXX,XX +XXX,XX @@ static void create_gic(VirtMachineState *vms, qemu_irq *pic)
118
119
if (nb_redist_regions == 2) {
120
uint32_t redist1_capacity =
121
- vms->memmap[VIRT_GIC_REDIST2].size / GICV3_REDIST_SIZE;
122
+ vms->memmap[VIRT_HIGH_GIC_REDIST2].size / GICV3_REDIST_SIZE;
123
124
qdev_prop_set_uint32(gicdev, "redist-region-count[1]",
125
MIN(smp_cpus - redist0_count, redist1_capacity));
126
@@ -XXX,XX +XXX,XX @@ static void create_gic(VirtMachineState *vms, qemu_irq *pic)
127
if (type == 3) {
128
sysbus_mmio_map(gicbusdev, 1, vms->memmap[VIRT_GIC_REDIST].base);
129
if (nb_redist_regions == 2) {
130
- sysbus_mmio_map(gicbusdev, 2, vms->memmap[VIRT_GIC_REDIST2].base);
131
+ sysbus_mmio_map(gicbusdev, 2,
132
+ vms->memmap[VIRT_HIGH_GIC_REDIST2].base);
133
}
134
} else {
135
sysbus_mmio_map(gicbusdev, 1, vms->memmap[VIRT_GIC_CPU].base);
136
@@ -XXX,XX +XXX,XX @@ static void create_pcie(VirtMachineState *vms, qemu_irq *pic)
137
{
138
hwaddr base_mmio = vms->memmap[VIRT_PCIE_MMIO].base;
139
hwaddr size_mmio = vms->memmap[VIRT_PCIE_MMIO].size;
140
- hwaddr base_mmio_high = vms->memmap[VIRT_PCIE_MMIO_HIGH].base;
141
- hwaddr size_mmio_high = vms->memmap[VIRT_PCIE_MMIO_HIGH].size;
142
+ hwaddr base_mmio_high = vms->memmap[VIRT_HIGH_PCIE_MMIO].base;
143
+ hwaddr size_mmio_high = vms->memmap[VIRT_HIGH_PCIE_MMIO].size;
144
hwaddr base_pio = vms->memmap[VIRT_PCIE_PIO].base;
145
hwaddr size_pio = vms->memmap[VIRT_PCIE_PIO].size;
146
hwaddr base_ecam, size_ecam;
147
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
148
* many redistributors we can fit into the memory map.
149
*/
150
if (vms->gic_version == 3) {
151
- virt_max_cpus = vms->memmap[VIRT_GIC_REDIST].size / GICV3_REDIST_SIZE;
152
- virt_max_cpus += vms->memmap[VIRT_GIC_REDIST2].size / GICV3_REDIST_SIZE;
153
+ virt_max_cpus =
154
+ vms->memmap[VIRT_GIC_REDIST].size / GICV3_REDIST_SIZE;
155
+ virt_max_cpus +=
156
+ vms->memmap[VIRT_HIGH_GIC_REDIST2].size / GICV3_REDIST_SIZE;
157
} else {
158
virt_max_cpus = GIC_NCPU;
159
}
160
--
161
2.20.1
162
163
diff view generated by jsdifflib
[Qemu-devel] [PULL 15/22] hw/boards: Add a MachineState parameter to kvm_type callback
Deleted patch
1
From: Eric Auger <eric.auger@redhat.com>
2
1
3
On ARM, the kvm_type will be resolved by querying the KVMState.
4
Let's add the MachineState handle to the callback so that we
5
can retrieve the KVMState handle. in kvm_init, when the callback
6
is called, the kvm_state variable is not yet set.
7
8
Signed-off-by: Eric Auger <eric.auger@redhat.com>
9
Acked-by: David Gibson <david@gibson.dropbear.id.au>
10
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
11
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
12
Message-id: 20190304101339.25970-5-eric.auger@redhat.com
13
[ppc parts]
14
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
18
include/hw/boards.h | 5 ++++-
19
accel/kvm/kvm-all.c | 2 +-
20
hw/ppc/mac_newworld.c | 3 +--
21
hw/ppc/mac_oldworld.c | 2 +-
22
hw/ppc/spapr.c | 2 +-
23
5 files changed, 8 insertions(+), 6 deletions(-)
24
25
diff --git a/include/hw/boards.h b/include/hw/boards.h
26
index XXXXXXX..XXXXXXX 100644
27
--- a/include/hw/boards.h
28
+++ b/include/hw/boards.h
29
@@ -XXX,XX +XXX,XX @@ typedef struct {
30
* should instead use "unimplemented-device" for all memory ranges where
31
* the guest will attempt to probe for a device that QEMU doesn't
32
* implement and a stub device is required.
33
+ * @kvm_type:
34
+ * Return the type of KVM corresponding to the kvm-type string option or
35
+ * computed based on other criteria such as the host kernel capabilities.
36
*/
37
struct MachineClass {
38
/*< private >*/
39
@@ -XXX,XX +XXX,XX @@ struct MachineClass {
40
void (*init)(MachineState *state);
41
void (*reset)(void);
42
void (*hot_add_cpu)(const int64_t id, Error **errp);
43
- int (*kvm_type)(const char *arg);
44
+ int (*kvm_type)(MachineState *machine, const char *arg);
45
46
BlockInterfaceType block_default_type;
47
int units_per_default_bus;
48
diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
49
index XXXXXXX..XXXXXXX 100644
50
--- a/accel/kvm/kvm-all.c
51
+++ b/accel/kvm/kvm-all.c
52
@@ -XXX,XX +XXX,XX @@ static int kvm_init(MachineState *ms)
53
54
kvm_type = qemu_opt_get(qemu_get_machine_opts(), "kvm-type");
55
if (mc->kvm_type) {
56
- type = mc->kvm_type(kvm_type);
57
+ type = mc->kvm_type(ms, kvm_type);
58
} else if (kvm_type) {
59
ret = -EINVAL;
60
fprintf(stderr, "Invalid argument kvm-type=%s\n", kvm_type);
61
diff --git a/hw/ppc/mac_newworld.c b/hw/ppc/mac_newworld.c
62
index XXXXXXX..XXXXXXX 100644
63
--- a/hw/ppc/mac_newworld.c
64
+++ b/hw/ppc/mac_newworld.c
65
@@ -XXX,XX +XXX,XX @@ static char *core99_fw_dev_path(FWPathProvider *p, BusState *bus,
66
67
return NULL;
68
}
69
-
70
-static int core99_kvm_type(const char *arg)
71
+static int core99_kvm_type(MachineState *machine, const char *arg)
72
{
73
/* Always force PR KVM */
74
return 2;
75
diff --git a/hw/ppc/mac_oldworld.c b/hw/ppc/mac_oldworld.c
76
index XXXXXXX..XXXXXXX 100644
77
--- a/hw/ppc/mac_oldworld.c
78
+++ b/hw/ppc/mac_oldworld.c
79
@@ -XXX,XX +XXX,XX @@ static char *heathrow_fw_dev_path(FWPathProvider *p, BusState *bus,
80
return NULL;
81
}
82
83
-static int heathrow_kvm_type(const char *arg)
84
+static int heathrow_kvm_type(MachineState *machine, const char *arg)
85
{
86
/* Always force PR KVM */
87
return 2;
88
diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
89
index XXXXXXX..XXXXXXX 100644
90
--- a/hw/ppc/spapr.c
91
+++ b/hw/ppc/spapr.c
92
@@ -XXX,XX +XXX,XX @@ static void spapr_machine_init(MachineState *machine)
93
}
94
}
95
96
-static int spapr_kvm_type(const char *vm_type)
97
+static int spapr_kvm_type(MachineState *machine, const char *vm_type)
98
{
99
if (!vm_type) {
100
return 0;
101
--
102
2.20.1
103
104
diff view generated by jsdifflib
[Qemu-devel] [PULL 16/22] kvm: add kvm_arm_get_max_vm_ipa_size
Deleted patch
1
From: Eric Auger <eric.auger@redhat.com>
2
1
3
Add the kvm_arm_get_max_vm_ipa_size() helper that returns the
4
number of bits in the IPA address space supported by KVM.
5
6
This capability needs to be known to create the VM with a
7
specific IPA max size (kvm_type passed along KVM_CREATE_VM ioctl.
8
9
Signed-off-by: Eric Auger <eric.auger@redhat.com>
10
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
11
Message-id: 20190304101339.25970-6-eric.auger@redhat.com
12
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
13
---
14
target/arm/kvm_arm.h | 13 +++++++++++++
15
target/arm/kvm.c | 10 ++++++++++
16
2 files changed, 23 insertions(+)
17
18
diff --git a/target/arm/kvm_arm.h b/target/arm/kvm_arm.h
19
index XXXXXXX..XXXXXXX 100644
20
--- a/target/arm/kvm_arm.h
21
+++ b/target/arm/kvm_arm.h
22
@@ -XXX,XX +XXX,XX @@ bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf);
23
*/
24
void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu);
25
26
+/**
27
+ * kvm_arm_get_max_vm_ipa_size - Returns the number of bits in the
28
+ * IPA address space supported by KVM
29
+ *
30
+ * @ms: Machine state handle
31
+ */
32
+int kvm_arm_get_max_vm_ipa_size(MachineState *ms);
33
+
34
/**
35
* kvm_arm_sync_mpstate_to_kvm
36
* @cpu: ARMCPU
37
@@ -XXX,XX +XXX,XX @@ static inline void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu)
38
cpu->host_cpu_probe_failed = true;
39
}
40
41
+static inline int kvm_arm_get_max_vm_ipa_size(MachineState *ms)
42
+{
43
+ return -ENOENT;
44
+}
45
+
46
static inline int kvm_arm_vgic_probe(void)
47
{
48
return 0;
49
diff --git a/target/arm/kvm.c b/target/arm/kvm.c
50
index XXXXXXX..XXXXXXX 100644
51
--- a/target/arm/kvm.c
52
+++ b/target/arm/kvm.c
53
@@ -XXX,XX +XXX,XX @@
54
#include "qemu/error-report.h"
55
#include "sysemu/sysemu.h"
56
#include "sysemu/kvm.h"
57
+#include "sysemu/kvm_int.h"
58
#include "kvm_arm.h"
59
#include "cpu.h"
60
#include "trace.h"
61
@@ -XXX,XX +XXX,XX @@ void kvm_arm_set_cpu_features_from_host(ARMCPU *cpu)
62
env->features = arm_host_cpu_features.features;
63
}
64
65
+int kvm_arm_get_max_vm_ipa_size(MachineState *ms)
66
+{
67
+ KVMState *s = KVM_STATE(ms->accelerator);
68
+ int ret;
69
+
70
+ ret = kvm_check_extension(s, KVM_CAP_ARM_VM_IPA_SIZE);
71
+ return ret > 0 ? ret : 40;
72
+}
73
+
74
int kvm_arch_init(MachineState *ms, KVMState *s)
75
{
76
/* For ARM interrupt delivery is always asynchronous,
77
--
78
2.20.1
79
80
diff view generated by jsdifflib
[Qemu-devel] [PULL 17/22] vl: Set machine ram_size, maxram_size and ram_slots earlier
Deleted patch
1
From: Eric Auger <eric.auger@redhat.com>
2
1
3
The machine RAM attributes will need to be analyzed during the
4
configure_accelerator() process. especially kvm_type() arm64
5
machine callback will use them to know how many IPA/GPA bits are
6
needed to model the whole RAM range. So let's assign those machine
7
state fields before calling configure_accelerator.
8
9
Signed-off-by: Eric Auger <eric.auger@redhat.com>
10
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
11
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
12
Message-id: 20190304101339.25970-7-eric.auger@redhat.com
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
---
15
vl.c | 6 +++---
16
1 file changed, 3 insertions(+), 3 deletions(-)
17
18
diff --git a/vl.c b/vl.c
19
index XXXXXXX..XXXXXXX 100644
20
--- a/vl.c
21
+++ b/vl.c
22
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp)
23
machine_opts = qemu_get_machine_opts();
24
qemu_opt_foreach(machine_opts, machine_set_property, current_machine,
25
&error_fatal);
26
+ current_machine->ram_size = ram_size;
27
+ current_machine->maxram_size = maxram_size;
28
+ current_machine->ram_slots = ram_slots;
29
30
configure_accelerator(current_machine, argv[0]);
31
32
@@ -XXX,XX +XXX,XX @@ int main(int argc, char **argv, char **envp)
33
replay_checkpoint(CHECKPOINT_INIT);
34
qdev_machine_init();
35
36
- current_machine->ram_size = ram_size;
37
- current_machine->maxram_size = maxram_size;
38
- current_machine->ram_slots = ram_slots;
39
current_machine->boot_order = boot_order;
40
41
/* parse features once if machine provides default cpu_type */
42
--
43
2.20.1
44
45
diff view generated by jsdifflib
[Qemu-devel] [PULL 18/22] hw/arm/virt: Dynamic memory map depending on RAM requirements
[PULL 16/16] target/arm/translate-neon.c: Handle VTBL UNDEF case before VFP access check
1
From: Eric Auger <eric.auger@redhat.com>
1
Checks for UNDEF cases should go before the "is VFP enabled?" access
2
check, except in special cases. Move a stray UNDEF check in the VTBL
3
trans function up above the access check.
2
4
3
Up to now the memory map has been static and the high IO region
5
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
4
base has always been 256GiB.
6
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
7
Message-id: 20201109145324.2859-1-peter.maydell@linaro.org
8
---
9
target/arm/translate-neon.c.inc | 8 ++++----
10
1 file changed, 4 insertions(+), 4 deletions(-)
5
11
6
This patch modifies the virt_set_memmap() function, which freezes
12
diff --git a/target/arm/translate-neon.c.inc b/target/arm/translate-neon.c.inc
7
the memory map, so that the high IO range base becomes floating,
8
located after the initial RAM and the device memory.
9
10
The function computes
11
- the base of the device memory,
12
- the size of the device memory,
13
- the high IO region base
14
- the highest GPA used in the memory map.
15
16
Entries of the high IO region are assigned a base address. The
17
device memory is initialized.
18
19
The highest GPA used in the memory map will be used at VM creation
20
to choose the requested IPA size.
21
22
Setting all the existing highmem IO regions beyond the RAM
23
allows to have a single contiguous RAM region (initial RAM and
24
possible hotpluggable device memory). That way we do not need
25
to do invasive changes in the EDK2 FW to support a dynamic
26
RAM base.
27
28
Still the user cannot request an initial RAM size greater than 255GB.
29
30
Signed-off-by: Eric Auger <eric.auger@redhat.com>
31
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
32
Message-id: 20190304101339.25970-8-eric.auger@redhat.com
33
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
34
---
35
include/hw/arm/virt.h | 1 +
36
hw/arm/virt.c | 52 ++++++++++++++++++++++++++++++++++++++-----
37
2 files changed, 47 insertions(+), 6 deletions(-)
38
39
diff --git a/include/hw/arm/virt.h b/include/hw/arm/virt.h
40
index XXXXXXX..XXXXXXX 100644
13
index XXXXXXX..XXXXXXX 100644
41
--- a/include/hw/arm/virt.h
14
--- a/target/arm/translate-neon.c.inc
42
+++ b/include/hw/arm/virt.h
15
+++ b/target/arm/translate-neon.c.inc
43
@@ -XXX,XX +XXX,XX @@ typedef struct {
16
@@ -XXX,XX +XXX,XX @@ static bool trans_VTBL(DisasContext *s, arg_VTBL *a)
44
uint32_t msi_phandle;
17
return false;
45
uint32_t iommu_phandle;
46
int psci_conduit;
47
+ hwaddr highest_gpa;
48
} VirtMachineState;
49
50
#define VIRT_ECAM_ID(high) (high ? VIRT_HIGH_PCIE_ECAM : VIRT_PCIE_ECAM)
51
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
52
index XXXXXXX..XXXXXXX 100644
53
--- a/hw/arm/virt.c
54
+++ b/hw/arm/virt.c
55
@@ -XXX,XX +XXX,XX @@
56
#include "qapi/visitor.h"
57
#include "standard-headers/linux/input.h"
58
#include "hw/arm/smmuv3.h"
59
+#include "hw/acpi/acpi.h"
60
61
#define DEFINE_VIRT_MACHINE_LATEST(major, minor, latest) \
62
static void virt_##major##_##minor##_class_init(ObjectClass *oc, \
63
@@ -XXX,XX +XXX,XX @@
64
* of a terabyte of RAM will be doing it on a host with more than a
65
* terabyte of physical address space.)
66
*/
67
-#define RAMLIMIT_GB 255
68
-#define RAMLIMIT_BYTES (RAMLIMIT_GB * 1024ULL * 1024 * 1024)
69
+#define LEGACY_RAMLIMIT_GB 255
70
+#define LEGACY_RAMLIMIT_BYTES (LEGACY_RAMLIMIT_GB * GiB)
71
72
/* Addresses and sizes of our components.
73
* 0..128MB is space for a flash device so we can run bootrom code such as UEFI.
74
@@ -XXX,XX +XXX,XX @@ static const MemMapEntry base_memmap[] = {
75
[VIRT_PCIE_MMIO] = { 0x10000000, 0x2eff0000 },
76
[VIRT_PCIE_PIO] = { 0x3eff0000, 0x00010000 },
77
[VIRT_PCIE_ECAM] = { 0x3f000000, 0x01000000 },
78
- [VIRT_MEM] = { 0x40000000, RAMLIMIT_BYTES },
79
+ /* Actual RAM size depends on initial RAM and device memory settings */
80
+ [VIRT_MEM] = { GiB, LEGACY_RAMLIMIT_BYTES },
81
};
82
83
/*
84
@@ -XXX,XX +XXX,XX @@ static uint64_t virt_cpu_mp_affinity(VirtMachineState *vms, int idx)
85
86
static void virt_set_memmap(VirtMachineState *vms)
87
{
88
- hwaddr base;
89
+ MachineState *ms = MACHINE(vms);
90
+ hwaddr base, device_memory_base, device_memory_size;
91
int i;
92
93
vms->memmap = extended_memmap;
94
@@ -XXX,XX +XXX,XX @@ static void virt_set_memmap(VirtMachineState *vms)
95
vms->memmap[i] = base_memmap[i];
96
}
18
}
97
19
98
- base = 256 * GiB; /* Top of the legacy initial RAM region */
20
- if (!vfp_access_check(s)) {
99
+ if (ms->ram_slots > ACPI_MAX_RAM_SLOTS) {
21
- return true;
100
+ error_report("unsupported number of memory slots: %"PRIu64,
22
- }
101
+ ms->ram_slots);
23
-
102
+ exit(EXIT_FAILURE);
24
if ((a->vn + a->len + 1) > 32) {
25
/*
26
* This is UNPREDICTABLE; we choose to UNDEF to avoid the
27
@@ -XXX,XX +XXX,XX @@ static bool trans_VTBL(DisasContext *s, arg_VTBL *a)
28
return false;
29
}
30
31
+ if (!vfp_access_check(s)) {
32
+ return true;
103
+ }
33
+ }
104
+
34
+
105
+ /*
35
desc = tcg_const_i32((a->vn << 2) | a->len);
106
+ * We compute the base of the high IO region depending on the
36
def = tcg_temp_new_i64();
107
+ * amount of initial and device memory. The device memory start/size
37
if (a->op) {
108
+ * is aligned on 1GiB. We never put the high IO region below 256GiB
109
+ * so that if maxram_size is < 255GiB we keep the legacy memory map.
110
+ * The device region size assumes 1GiB page max alignment per slot.
111
+ */
112
+ device_memory_base =
113
+ ROUND_UP(vms->memmap[VIRT_MEM].base + ms->ram_size, GiB);
114
+ device_memory_size = ms->maxram_size - ms->ram_size + ms->ram_slots * GiB;
115
+
116
+ /* Base address of the high IO region */
117
+ base = device_memory_base + ROUND_UP(device_memory_size, GiB);
118
+ if (base < device_memory_base) {
119
+ error_report("maxmem/slots too huge");
120
+ exit(EXIT_FAILURE);
121
+ }
122
+ if (base < vms->memmap[VIRT_MEM].base + LEGACY_RAMLIMIT_BYTES) {
123
+ base = vms->memmap[VIRT_MEM].base + LEGACY_RAMLIMIT_BYTES;
124
+ }
125
126
for (i = VIRT_LOWMEMMAP_LAST; i < ARRAY_SIZE(extended_memmap); i++) {
127
hwaddr size = extended_memmap[i].size;
128
@@ -XXX,XX +XXX,XX @@ static void virt_set_memmap(VirtMachineState *vms)
129
vms->memmap[i].size = size;
130
base += size;
131
}
132
+ vms->highest_gpa = base - 1;
133
+ if (device_memory_size > 0) {
134
+ ms->device_memory = g_malloc0(sizeof(*ms->device_memory));
135
+ ms->device_memory->base = device_memory_base;
136
+ memory_region_init(&ms->device_memory->mr, OBJECT(vms),
137
+ "device-memory", device_memory_size);
138
+ }
139
}
140
141
static void machvirt_init(MachineState *machine)
142
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
143
vms->smp_cpus = smp_cpus;
144
145
if (machine->ram_size > vms->memmap[VIRT_MEM].size) {
146
- error_report("mach-virt: cannot model more than %dGB RAM", RAMLIMIT_GB);
147
+ error_report("mach-virt: cannot model more than %dGB RAM",
148
+ LEGACY_RAMLIMIT_GB);
149
exit(1);
150
}
151
152
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
153
memory_region_allocate_system_memory(ram, NULL, "mach-virt.ram",
154
machine->ram_size);
155
memory_region_add_subregion(sysmem, vms->memmap[VIRT_MEM].base, ram);
156
+ if (machine->device_memory) {
157
+ memory_region_add_subregion(sysmem, machine->device_memory->base,
158
+ &machine->device_memory->mr);
159
+ }
160
161
create_flash(vms, sysmem, secure_sysmem ? secure_sysmem : sysmem);
162
163
--
38
--
164
2.20.1
39
2.20.1
165
40
166
41
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.