[PULL 00/26] target-arm queue
[PULL v2 00/12] target-arm queue
1
Hi; here's the latest arm pullreq...
1
v2: minor tweak to fix format string issue on Windows hosts...
2
2
3
-- PMM
4
3
5
The following changes since commit 03a3a62fbd0aa5227e978eef3c67d3978aec9e5f:
4
The following changes since commit 6eeea6725a70e6fcb5abba0764496bdab07ddfb3:
6
5
7
Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging (2023-09-07 10:29:06 -0400)
6
Merge remote-tracking branch 'remotes/huth-gitlab/tags/pull-request-2020-10-06' into staging (2020-10-06 21:13:34 +0100)
8
7
9
are available in the Git repository at:
8
are available in the Git repository at:
10
9
11
https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20230908
10
https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20201008-1
12
11
13
for you to fetch changes up to c8f2eb5d414b788420b938f2ffdde891aa6c3ae8:
12
for you to fetch changes up to d1b6b7017572e8d82f26eb827a1dba0e8cf3cae6:
14
13
15
arm/kvm: Enable support for KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE (2023-09-08 16:41:36 +0100)
14
target/arm: Make '-cpu max' have a 48-bit PA (2020-10-08 21:40:01 +0100)
16
15
17
----------------------------------------------------------------
16
----------------------------------------------------------------
18
target-arm queue:
17
target-arm queue:
19
* New CPU type: cortex-a710
18
* hw/ssi/npcm7xx_fiu: Fix handling of unsigned integer
20
* Implement new architectural features:
19
* hw/arm/fsl-imx25: Fix a typo
21
- FEAT_PACQARMA3
20
* hw/arm/sbsa-ref : Fix SMMUv3 Initialisation
22
- FEAT_EPAC
21
* hw/arm/sbsa-ref : allocate IRQs for SMMUv3
23
- FEAT_Pauth2
22
* hw/char/bcm2835_aux: Allow less than 32-bit accesses
24
- FEAT_FPAC
23
* hw/arm/virt: Implement kvm-steal-time
25
- FEAT_FPACCOMBINE
24
* target/arm: Make '-cpu max' have a 48-bit PA
26
- FEAT_TIDCP1
27
* Xilinx Versal: Model the CFU/CFI
28
* Implement RMR_ELx registers
29
* Implement handling of HCR_EL2.TIDCP trap bit
30
* arm/kvm: Enable support for KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE
31
* hw/intc/arm_gicv3_its: Avoid maybe-uninitialized error in get_vte()
32
* target/arm: Do not use gen_mte_checkN in trans_STGP
33
* arm64: Restore trapless ptimer access
34
25
35
----------------------------------------------------------------
26
----------------------------------------------------------------
36
Aaron Lindsay (6):
27
Andrew Jones (6):
37
target/arm: Add ID_AA64ISAR2_EL1
28
linux headers: sync to 5.9-rc7
38
target/arm: Add feature detection for FEAT_Pauth2 and extensions
29
target/arm/kvm: Make uncalled stubs explicitly unreachable
39
target/arm: Implement FEAT_EPAC
30
hw/arm/virt: Move post cpu realize check into its own function
40
target/arm: Implement FEAT_Pauth2
31
hw/arm/virt: Move kvm pmu setup to virt_cpu_post_init
41
target/arm: Inform helpers whether a PAC instruction is 'combined'
32
tests/qtest: Restore aarch64 arm-cpu-features test
42
target/arm: Implement FEAT_FPAC and FEAT_FPACCOMBINE
33
hw/arm/virt: Implement kvm-steal-time
43
34
44
Colton Lewis (1):
35
Graeme Gregory (2):
45
arm64: Restore trapless ptimer access
36
hw/arm/sbsa-ref : Fix SMMUv3 Initialisation
37
hw/arm/sbsa-ref : allocate IRQs for SMMUv3
46
38
47
Francisco Iglesias (8):
39
Peter Maydell (1):
48
hw/misc: Introduce the Xilinx CFI interface
40
target/arm: Make '-cpu max' have a 48-bit PA
49
hw/misc: Introduce a model of Xilinx Versal's CFU_APB
50
hw/misc/xlnx-versal-cfu: Introduce a model of Xilinx Versal CFU_FDRO
51
hw/misc/xlnx-versal-cfu: Introduce a model of Xilinx Versal's CFU_SFR
52
hw/misc: Introduce a model of Xilinx Versal's CFRAME_REG
53
hw/misc: Introduce a model of Xilinx Versal's CFRAME_BCAST_REG
54
hw/arm/xlnx-versal: Connect the CFU_APB, CFU_FDRO and CFU_SFR
55
hw/arm/versal: Connect the CFRAME_REG and CFRAME_BCAST_REG
56
41
57
Philippe Mathieu-Daudé (1):
42
Philippe Mathieu-Daudé (3):
58
hw/intc/arm_gicv3_its: Avoid maybe-uninitialized error in get_vte()
43
hw/ssi/npcm7xx_fiu: Fix handling of unsigned integer
44
hw/arm/fsl-imx25: Fix a typo
45
hw/char/bcm2835_aux: Allow less than 32-bit accesses
59
46
60
Richard Henderson (9):
47
docs/system/arm/cpu-features.rst | 11 ++++
61
tests/tcg/aarch64: Adjust pauth tests for FEAT_FPAC
48
include/hw/arm/fsl-imx25.h | 2 +-
62
target/arm: Don't change pauth features when changing algorithm
49
include/hw/arm/virt.h | 5 ++
63
target/arm: Implement FEAT_PACQARMA3
50
linux-headers/linux/kvm.h | 6 ++-
64
target/arm: Do not use gen_mte_checkN in trans_STGP
51
target/arm/cpu.h | 4 ++
65
target/arm: Implement RMR_ELx
52
target/arm/kvm_arm.h | 94 ++++++++++++++++++++++++++-------
66
target/arm: Implement cortex-a710
53
hw/arm/sbsa-ref.c | 3 +-
67
target/arm: Implement HCR_EL2.TIDCP
54
hw/arm/virt.c | 111 ++++++++++++++++++++++++++++-----------
68
target/arm: Implement FEAT_TIDCP1
55
hw/char/bcm2835_aux.c | 4 +-
69
target/arm: Enable SCTLR_EL1.TIDCP for user-only
56
hw/ssi/npcm7xx_fiu.c | 12 ++---
57
target/arm/cpu.c | 8 +++
58
target/arm/cpu64.c | 4 ++
59
target/arm/kvm.c | 16 ++++++
60
target/arm/kvm64.c | 64 ++++++++++++++++++++--
61
target/arm/monitor.c | 2 +-
62
tests/qtest/arm-cpu-features.c | 25 +++++++--
63
hw/ssi/trace-events | 2 +-
64
tests/qtest/meson.build | 3 +-
65
18 files changed, 304 insertions(+), 72 deletions(-)
70
66
71
Shameer Kolothum (1):
72
arm/kvm: Enable support for KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE
73
74
MAINTAINERS | 10 +
75
docs/system/arm/cpu-features.rst | 21 +-
76
docs/system/arm/emulation.rst | 8 +
77
docs/system/arm/virt.rst | 1 +
78
include/hw/arm/xlnx-versal.h | 85 +++
79
include/hw/misc/xlnx-cfi-if.h | 59 +++
80
include/hw/misc/xlnx-versal-cframe-reg.h | 303 +++++++++++
81
include/hw/misc/xlnx-versal-cfu.h | 258 ++++++++++
82
include/sysemu/kvm_int.h | 1 +
83
target/arm/cpu.h | 54 +-
84
target/arm/helper.h | 2 +
85
target/arm/syndrome.h | 7 +
86
target/arm/tcg/helper-a64.h | 4 +
87
tests/tcg/aarch64/pauth.h | 23 +
88
accel/kvm/kvm-all.c | 1 +
89
hw/arm/virt.c | 1 +
90
hw/arm/xlnx-versal.c | 155 +++++-
91
hw/intc/arm_gicv3_its.c | 15 +-
92
hw/misc/xlnx-cfi-if.c | 34 ++
93
hw/misc/xlnx-versal-cframe-reg.c | 858 +++++++++++++++++++++++++++++++
94
hw/misc/xlnx-versal-cfu.c | 563 ++++++++++++++++++++
95
target/arm/arm-qmp-cmds.c | 2 +-
96
target/arm/cpu.c | 4 +
97
target/arm/cpu64.c | 86 +++-
98
target/arm/helper.c | 68 ++-
99
target/arm/hvf/hvf.c | 1 +
100
target/arm/kvm.c | 61 +++
101
target/arm/kvm64.c | 3 +
102
target/arm/tcg/cpu64.c | 215 ++++++++
103
target/arm/tcg/op_helper.c | 33 ++
104
target/arm/tcg/pauth_helper.c | 180 +++++--
105
target/arm/tcg/translate-a64.c | 74 +--
106
target/arm/tcg/translate.c | 33 ++
107
tests/qtest/arm-cpu-features.c | 12 +-
108
tests/tcg/aarch64/pauth-2.c | 54 +-
109
tests/tcg/aarch64/pauth-4.c | 18 +-
110
tests/tcg/aarch64/pauth-5.c | 10 +
111
hw/misc/meson.build | 3 +
112
qemu-options.hx | 15 +
113
tests/tcg/aarch64/Makefile.target | 6 +-
114
40 files changed, 3184 insertions(+), 157 deletions(-)
115
create mode 100644 include/hw/misc/xlnx-cfi-if.h
116
create mode 100644 include/hw/misc/xlnx-versal-cframe-reg.h
117
create mode 100644 include/hw/misc/xlnx-versal-cfu.h
118
create mode 100644 tests/tcg/aarch64/pauth.h
119
create mode 100644 hw/misc/xlnx-cfi-if.c
120
create mode 100644 hw/misc/xlnx-versal-cframe-reg.c
121
create mode 100644 hw/misc/xlnx-versal-cfu.c
122
diff view generated by jsdifflib
[PULL 01/26] tests/tcg/aarch64: Adjust pauth tests for FEAT_FPAC
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
With FEAT_FPAC, AUT* instructions that fail authentication
4
do not produce an error value but instead fault.
5
6
For pauth-2, install a signal handler and verify it gets called.
7
8
For pauth-4 and pauth-5, we are explicitly testing the error value,
9
so there's nothing to test with FEAT_FPAC, so exit early.
10
Adjust the makefile to use -cpu neoverse-v1, which has FEAT_EPAC
11
but not FEAT_FPAC.
12
13
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
14
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
15
Message-id: 20230829232335.965414-2-richard.henderson@linaro.org
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
18
tests/tcg/aarch64/pauth.h | 23 +++++++++++++
19
tests/tcg/aarch64/pauth-2.c | 54 ++++++++++++++++++++++++++-----
20
tests/tcg/aarch64/pauth-4.c | 18 ++++++++---
21
tests/tcg/aarch64/pauth-5.c | 10 ++++++
22
tests/tcg/aarch64/Makefile.target | 6 +++-
23
5 files changed, 98 insertions(+), 13 deletions(-)
24
create mode 100644 tests/tcg/aarch64/pauth.h
25
26
diff --git a/tests/tcg/aarch64/pauth.h b/tests/tcg/aarch64/pauth.h
27
new file mode 100644
28
index XXXXXXX..XXXXXXX
29
--- /dev/null
30
+++ b/tests/tcg/aarch64/pauth.h
31
@@ -XXX,XX +XXX,XX @@
32
+/*
33
+ * Helper for pauth test case
34
+ *
35
+ * Copyright (c) 2023 Linaro Ltd
36
+ * SPDX-License-Identifier: GPL-2.0-or-later
37
+ */
38
+
39
+#include <assert.h>
40
+#include <sys/auxv.h>
41
+
42
+static int get_pac_feature(void)
43
+{
44
+ unsigned long isar1, isar2;
45
+
46
+ assert(getauxval(AT_HWCAP) & HWCAP_CPUID);
47
+
48
+ asm("mrs %0, id_aa64isar1_el1" : "=r"(isar1));
49
+ asm("mrs %0, S3_0_C0_C6_2" : "=r"(isar2)); /* id_aa64isar2_el1 */
50
+
51
+ return ((isar1 >> 4) & 0xf) /* APA */
52
+ | ((isar1 >> 8) & 0xf) /* API */
53
+ | ((isar2 >> 12) & 0xf); /* APA3 */
54
+}
55
diff --git a/tests/tcg/aarch64/pauth-2.c b/tests/tcg/aarch64/pauth-2.c
56
index XXXXXXX..XXXXXXX 100644
57
--- a/tests/tcg/aarch64/pauth-2.c
58
+++ b/tests/tcg/aarch64/pauth-2.c
59
@@ -XXX,XX +XXX,XX @@
60
#include <stdint.h>
61
+#include <signal.h>
62
+#include <stdlib.h>
63
#include <assert.h>
64
+#include "pauth.h"
65
+
66
+
67
+static void sigill(int sig, siginfo_t *info, void *vuc)
68
+{
69
+ ucontext_t *uc = vuc;
70
+ uint64_t test;
71
+
72
+ /* There is only one insn below that is allowed to fault. */
73
+ asm volatile("adr %0, auth2_insn" : "=r"(test));
74
+ assert(test == uc->uc_mcontext.pc);
75
+ exit(0);
76
+}
77
+
78
+static int pac_feature;
79
80
void do_test(uint64_t value)
81
{
82
@@ -XXX,XX +XXX,XX @@ void do_test(uint64_t value)
83
* An invalid salt usually fails authorization, but again there
84
* is a chance of choosing another salt that works.
85
* Iterate until we find another salt which does fail.
86
+ *
87
+ * With FEAT_FPAC, this will SIGILL instead of producing a result.
88
*/
89
for (salt2 = salt1 + 1; ; salt2++) {
90
- asm volatile("autda %0, %2" : "=r"(decode) : "0"(encode), "r"(salt2));
91
+ asm volatile("auth2_insn: autda %0, %2"
92
+ : "=r"(decode) : "0"(encode), "r"(salt2));
93
if (decode != value) {
94
break;
95
}
96
}
97
98
+ assert(pac_feature < 4); /* No FEAT_FPAC */
99
+
100
/* The VA bits, bit 55, and the TBI bits, should be unchanged. */
101
assert(((decode ^ value) & 0xff80ffffffffffffull) == 0);
102
103
/*
104
- * Bits [54:53] are an error indicator based on the key used;
105
- * the DA key above is keynumber 0, so error == 0b01. Otherwise
106
- * bit 55 of the original is sign-extended into the rest of the auth.
107
+ * Without FEAT_Pauth2, bits [54:53] are an error indicator based on
108
+ * the key used; the DA key above is keynumber 0, so error == 0b01.
109
+ * Otherwise, bit 55 of the original is sign-extended into the rest
110
+ * of the auth.
111
*/
112
- if ((value >> 55) & 1) {
113
- assert(((decode >> 48) & 0xff) == 0b10111111);
114
- } else {
115
- assert(((decode >> 48) & 0xff) == 0b00100000);
116
+ if (pac_feature < 3) {
117
+ if ((value >> 55) & 1) {
118
+ assert(((decode >> 48) & 0xff) == 0b10111111);
119
+ } else {
120
+ assert(((decode >> 48) & 0xff) == 0b00100000);
121
+ }
122
}
123
}
124
125
int main()
126
{
127
+ static const struct sigaction sa = {
128
+ .sa_sigaction = sigill,
129
+ .sa_flags = SA_SIGINFO
130
+ };
131
+
132
+ pac_feature = get_pac_feature();
133
+ assert(pac_feature != 0);
134
+
135
+ if (pac_feature >= 4) {
136
+ /* FEAT_FPAC */
137
+ sigaction(SIGILL, &sa, NULL);
138
+ }
139
+
140
do_test(0);
141
do_test(0xda004acedeadbeefull);
142
return 0;
143
diff --git a/tests/tcg/aarch64/pauth-4.c b/tests/tcg/aarch64/pauth-4.c
144
index XXXXXXX..XXXXXXX 100644
145
--- a/tests/tcg/aarch64/pauth-4.c
146
+++ b/tests/tcg/aarch64/pauth-4.c
147
@@ -XXX,XX +XXX,XX @@
148
#include <assert.h>
149
#include <stdio.h>
150
#include <stdlib.h>
151
+#include "pauth.h"
152
153
#define TESTS 1000
154
155
int main()
156
{
157
+ char base[TESTS];
158
int i, count = 0;
159
float perc;
160
- void *base = malloc(TESTS);
161
+ int pac_feature = get_pac_feature();
162
+
163
+ /*
164
+ * Exit if no PAuth or FEAT_FPAC, which will SIGILL on AUTIA failure
165
+ * rather than return an error for us to check below.
166
+ */
167
+ if (pac_feature == 0 || pac_feature >= 4) {
168
+ return 0;
169
+ }
170
171
for (i = 0; i < TESTS; i++) {
172
uintptr_t in, x, y;
173
@@ -XXX,XX +XXX,XX @@ int main()
174
in = i + (uintptr_t) base;
175
176
asm("mov %0, %[in]\n\t"
177
- "pacia %0, sp\n\t" /* sigill if pauth not supported */
178
+ "pacia %0, sp\n\t"
179
"eor %0, %0, #4\n\t" /* corrupt single bit */
180
"mov %1, %0\n\t"
181
"autia %1, sp\n\t" /* validate corrupted pointer */
182
@@ -XXX,XX +XXX,XX @@ int main()
183
if (x != y) {
184
count++;
185
}
186
-
187
}
188
+
189
perc = (float) count / (float) TESTS;
190
- printf("Checks Passed: %0.2f%%", perc * 100.0);
191
+ printf("Checks Passed: %0.2f%%\n", perc * 100.0);
192
assert(perc > 0.95);
193
return 0;
194
}
195
diff --git a/tests/tcg/aarch64/pauth-5.c b/tests/tcg/aarch64/pauth-5.c
196
index XXXXXXX..XXXXXXX 100644
197
--- a/tests/tcg/aarch64/pauth-5.c
198
+++ b/tests/tcg/aarch64/pauth-5.c
199
@@ -XXX,XX +XXX,XX @@
200
#include <assert.h>
201
+#include "pauth.h"
202
203
static int x;
204
205
@@ -XXX,XX +XXX,XX @@ int main()
206
{
207
int *p0 = &x, *p1, *p2, *p3;
208
unsigned long salt = 0;
209
+ int pac_feature = get_pac_feature();
210
+
211
+ /*
212
+ * Exit if no PAuth or FEAT_FPAC, which will SIGILL on AUTDA failure
213
+ * rather than return an error for us to check below.
214
+ */
215
+ if (pac_feature == 0 || pac_feature >= 4) {
216
+ return 0;
217
+ }
218
219
/*
220
* With TBI enabled and a 48-bit VA, there are 7 bits of auth, and so
221
diff --git a/tests/tcg/aarch64/Makefile.target b/tests/tcg/aarch64/Makefile.target
222
index XXXXXXX..XXXXXXX 100644
223
--- a/tests/tcg/aarch64/Makefile.target
224
+++ b/tests/tcg/aarch64/Makefile.target
225
@@ -XXX,XX +XXX,XX @@ endif
226
ifneq ($(CROSS_CC_HAS_ARMV8_3),)
227
AARCH64_TESTS += pauth-1 pauth-2 pauth-4 pauth-5
228
pauth-%: CFLAGS += -march=armv8.3-a
229
-run-pauth-%: QEMU_OPTS += -cpu max
230
+run-pauth-1: QEMU_OPTS += -cpu max
231
+run-pauth-2: QEMU_OPTS += -cpu max
232
+# Choose a cpu with FEAT_Pauth but without FEAT_FPAC for pauth-[45].
233
+run-pauth-4: QEMU_OPTS += -cpu neoverse-v1
234
+run-pauth-5: QEMU_OPTS += -cpu neoverse-v1
235
endif
236
237
# BTI Tests
238
--
239
2.34.1
diff view generated by jsdifflib
[PULL 02/26] target/arm: Add ID_AA64ISAR2_EL1
Deleted patch
1
From: Aaron Lindsay <aaron@os.amperecomputing.com>
2
1
3
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
4
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
5
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
6
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
8
Message-id: 20230829232335.965414-3-richard.henderson@linaro.org
9
[PMM: drop the HVF part of the patch and just comment that
10
we need to do something when the register appears in that API]
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
---
15
target/arm/cpu.h | 1 +
16
target/arm/helper.c | 4 ++--
17
target/arm/hvf/hvf.c | 1 +
18
target/arm/kvm64.c | 2 ++
19
4 files changed, 6 insertions(+), 2 deletions(-)
20
21
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
22
index XXXXXXX..XXXXXXX 100644
23
--- a/target/arm/cpu.h
24
+++ b/target/arm/cpu.h
25
@@ -XXX,XX +XXX,XX @@ struct ArchCPU {
26
uint32_t dbgdevid1;
27
uint64_t id_aa64isar0;
28
uint64_t id_aa64isar1;
29
+ uint64_t id_aa64isar2;
30
uint64_t id_aa64pfr0;
31
uint64_t id_aa64pfr1;
32
uint64_t id_aa64mmfr0;
33
diff --git a/target/arm/helper.c b/target/arm/helper.c
34
index XXXXXXX..XXXXXXX 100644
35
--- a/target/arm/helper.c
36
+++ b/target/arm/helper.c
37
@@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu)
38
.access = PL1_R, .type = ARM_CP_CONST,
39
.accessfn = access_aa64_tid3,
40
.resetvalue = cpu->isar.id_aa64isar1 },
41
- { .name = "ID_AA64ISAR2_EL1_RESERVED", .state = ARM_CP_STATE_AA64,
42
+ { .name = "ID_AA64ISAR2_EL1", .state = ARM_CP_STATE_AA64,
43
.opc0 = 3, .opc1 = 0, .crn = 0, .crm = 6, .opc2 = 2,
44
.access = PL1_R, .type = ARM_CP_CONST,
45
.accessfn = access_aa64_tid3,
46
- .resetvalue = 0 },
47
+ .resetvalue = cpu->isar.id_aa64isar2 },
48
{ .name = "ID_AA64ISAR3_EL1_RESERVED", .state = ARM_CP_STATE_AA64,
49
.opc0 = 3, .opc1 = 0, .crn = 0, .crm = 6, .opc2 = 3,
50
.access = PL1_R, .type = ARM_CP_CONST,
51
diff --git a/target/arm/hvf/hvf.c b/target/arm/hvf/hvf.c
52
index XXXXXXX..XXXXXXX 100644
53
--- a/target/arm/hvf/hvf.c
54
+++ b/target/arm/hvf/hvf.c
55
@@ -XXX,XX +XXX,XX @@ static bool hvf_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
56
{ HV_SYS_REG_ID_AA64DFR1_EL1, &host_isar.id_aa64dfr1 },
57
{ HV_SYS_REG_ID_AA64ISAR0_EL1, &host_isar.id_aa64isar0 },
58
{ HV_SYS_REG_ID_AA64ISAR1_EL1, &host_isar.id_aa64isar1 },
59
+ /* Add ID_AA64ISAR2_EL1 here when HVF supports it */
60
{ HV_SYS_REG_ID_AA64MMFR0_EL1, &host_isar.id_aa64mmfr0 },
61
{ HV_SYS_REG_ID_AA64MMFR1_EL1, &host_isar.id_aa64mmfr1 },
62
{ HV_SYS_REG_ID_AA64MMFR2_EL1, &host_isar.id_aa64mmfr2 },
63
diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
64
index XXXXXXX..XXXXXXX 100644
65
--- a/target/arm/kvm64.c
66
+++ b/target/arm/kvm64.c
67
@@ -XXX,XX +XXX,XX @@ bool kvm_arm_get_host_cpu_features(ARMHostCPUFeatures *ahcf)
68
ARM64_SYS_REG(3, 0, 0, 6, 0));
69
err |= read_sys_reg64(fdarray[2], &ahcf->isar.id_aa64isar1,
70
ARM64_SYS_REG(3, 0, 0, 6, 1));
71
+ err |= read_sys_reg64(fdarray[2], &ahcf->isar.id_aa64isar2,
72
+ ARM64_SYS_REG(3, 0, 0, 6, 2));
73
err |= read_sys_reg64(fdarray[2], &ahcf->isar.id_aa64mmfr0,
74
ARM64_SYS_REG(3, 0, 0, 7, 0));
75
err |= read_sys_reg64(fdarray[2], &ahcf->isar.id_aa64mmfr1,
76
--
77
2.34.1
78
79
diff view generated by jsdifflib
[PULL 03/26] target/arm: Add feature detection for FEAT_Pauth2 and extensions
Deleted patch
1
From: Aaron Lindsay <aaron@os.amperecomputing.com>
2
1
3
Rename isar_feature_aa64_pauth_arch to isar_feature_aa64_pauth_qarma5
4
to distinguish the other architectural algorithm qarma3.
5
6
Add ARMPauthFeature and isar_feature_pauth_feature to cover the
7
other pauth conditions.
8
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
11
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
12
Message-id: 20230829232335.965414-4-richard.henderson@linaro.org
13
Message-Id: <20230609172324.982888-3-aaron@os.amperecomputing.com>
14
[rth: Add ARMPauthFeature and eliminate most other predicates]
15
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
18
target/arm/cpu.h | 47 +++++++++++++++++++++++++++++------
19
target/arm/tcg/pauth_helper.c | 2 +-
20
2 files changed, 40 insertions(+), 9 deletions(-)
21
22
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
23
index XXXXXXX..XXXXXXX 100644
24
--- a/target/arm/cpu.h
25
+++ b/target/arm/cpu.h
26
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_fcma(const ARMISARegisters *id)
27
return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, FCMA) != 0;
28
}
29
30
+/*
31
+ * These are the values from APA/API/APA3.
32
+ * In general these must be compared '>=', per the normal Arm ARM
33
+ * treatment of fields in ID registers.
34
+ */
35
+typedef enum {
36
+ PauthFeat_None = 0,
37
+ PauthFeat_1 = 1,
38
+ PauthFeat_EPAC = 2,
39
+ PauthFeat_2 = 3,
40
+ PauthFeat_FPAC = 4,
41
+ PauthFeat_FPACCOMBINED = 5,
42
+} ARMPauthFeature;
43
+
44
+static inline ARMPauthFeature
45
+isar_feature_pauth_feature(const ARMISARegisters *id)
46
+{
47
+ /*
48
+ * Architecturally, only one of {APA,API,APA3} may be active (non-zero)
49
+ * and the other two must be zero. Thus we may avoid conditionals.
50
+ */
51
+ return (FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, APA) |
52
+ FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, API) |
53
+ FIELD_EX64(id->id_aa64isar2, ID_AA64ISAR2, APA3));
54
+}
55
+
56
static inline bool isar_feature_aa64_pauth(const ARMISARegisters *id)
57
{
58
/*
59
* Return true if any form of pauth is enabled, as this
60
* predicate controls migration of the 128-bit keys.
61
*/
62
- return (id->id_aa64isar1 &
63
- (FIELD_DP64(0, ID_AA64ISAR1, APA, 0xf) |
64
- FIELD_DP64(0, ID_AA64ISAR1, API, 0xf) |
65
- FIELD_DP64(0, ID_AA64ISAR1, GPA, 0xf) |
66
- FIELD_DP64(0, ID_AA64ISAR1, GPI, 0xf))) != 0;
67
+ return isar_feature_pauth_feature(id) != PauthFeat_None;
68
}
69
70
-static inline bool isar_feature_aa64_pauth_arch(const ARMISARegisters *id)
71
+static inline bool isar_feature_aa64_pauth_qarma5(const ARMISARegisters *id)
72
{
73
/*
74
- * Return true if pauth is enabled with the architected QARMA algorithm.
75
- * QEMU will always set APA+GPA to the same value.
76
+ * Return true if pauth is enabled with the architected QARMA5 algorithm.
77
+ * QEMU will always enable or disable both APA and GPA.
78
*/
79
return FIELD_EX64(id->id_aa64isar1, ID_AA64ISAR1, APA) != 0;
80
}
81
82
+static inline bool isar_feature_aa64_pauth_qarma3(const ARMISARegisters *id)
83
+{
84
+ /*
85
+ * Return true if pauth is enabled with the architected QARMA3 algorithm.
86
+ * QEMU will always enable or disable both APA3 and GPA3.
87
+ */
88
+ return FIELD_EX64(id->id_aa64isar2, ID_AA64ISAR2, APA3) != 0;
89
+}
90
+
91
static inline bool isar_feature_aa64_tlbirange(const ARMISARegisters *id)
92
{
93
return FIELD_EX64(id->id_aa64isar0, ID_AA64ISAR0, TLB) == 2;
94
diff --git a/target/arm/tcg/pauth_helper.c b/target/arm/tcg/pauth_helper.c
95
index XXXXXXX..XXXXXXX 100644
96
--- a/target/arm/tcg/pauth_helper.c
97
+++ b/target/arm/tcg/pauth_helper.c
98
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac_impdef(uint64_t data, uint64_t modifier,
99
static uint64_t pauth_computepac(CPUARMState *env, uint64_t data,
100
uint64_t modifier, ARMPACKey key)
101
{
102
- if (cpu_isar_feature(aa64_pauth_arch, env_archcpu(env))) {
103
+ if (cpu_isar_feature(aa64_pauth_qarma5, env_archcpu(env))) {
104
return pauth_computepac_architected(data, modifier, key);
105
} else {
106
return pauth_computepac_impdef(data, modifier, key);
107
--
108
2.34.1
diff view generated by jsdifflib
[PULL 04/26] target/arm: Don't change pauth features when changing algorithm
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
We have cpu properties to adjust the pauth algorithm for the
4
purpose of speed of emulation. Retain the set of pauth features
5
supported by the cpu even as the algorithm changes.
6
7
This already affects the neoverse-v1 cpu, which has FEAT_EPAC.
8
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
11
Message-id: 20230829232335.965414-5-richard.henderson@linaro.org
12
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
13
---
14
target/arm/cpu64.c | 70 +++++++++++++++++++++++++++---------------
15
target/arm/tcg/cpu64.c | 2 ++
16
2 files changed, 47 insertions(+), 25 deletions(-)
17
18
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
19
index XXXXXXX..XXXXXXX 100644
20
--- a/target/arm/cpu64.c
21
+++ b/target/arm/cpu64.c
22
@@ -XXX,XX +XXX,XX @@ void aarch64_add_sme_properties(Object *obj)
23
24
void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp)
25
{
26
- int arch_val = 0, impdef_val = 0;
27
- uint64_t t;
28
+ ARMPauthFeature features = cpu_isar_feature(pauth_feature, cpu);
29
+ uint64_t isar1;
30
31
- /* Exit early if PAuth is enabled, and fall through to disable it */
32
- if ((kvm_enabled() || hvf_enabled()) && cpu->prop_pauth) {
33
- if (!cpu_isar_feature(aa64_pauth, cpu)) {
34
- error_setg(errp, "'pauth' feature not supported by %s on this host",
35
- kvm_enabled() ? "KVM" : "hvf");
36
+ /*
37
+ * These properties enable or disable Pauth as a whole, or change
38
+ * the pauth algorithm, but do not change the set of features that
39
+ * are present. We have saved a copy of those features above and
40
+ * will now place it into the field that chooses the algorithm.
41
+ *
42
+ * Begin by disabling all fields.
43
+ */
44
+ isar1 = cpu->isar.id_aa64isar1;
45
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, APA, 0);
46
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPA, 0);
47
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, API, 0);
48
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPI, 0);
49
+
50
+ if (kvm_enabled() || hvf_enabled()) {
51
+ /*
52
+ * Exit early if PAuth is enabled and fall through to disable it.
53
+ * The algorithm selection properties are not present.
54
+ */
55
+ if (cpu->prop_pauth) {
56
+ if (features == 0) {
57
+ error_setg(errp, "'pauth' feature not supported by "
58
+ "%s on this host", current_accel_name());
59
+ }
60
+ return;
61
+ }
62
+ } else {
63
+ /* Pauth properties are only present when the model supports it. */
64
+ if (features == 0) {
65
+ assert(!cpu->prop_pauth);
66
+ return;
67
}
68
69
- return;
70
- }
71
-
72
- /* TODO: Handle HaveEnhancedPAC, HaveEnhancedPAC2, HaveFPAC. */
73
- if (cpu->prop_pauth) {
74
- if (cpu->prop_pauth_impdef) {
75
- impdef_val = 1;
76
- } else {
77
- arch_val = 1;
78
+ if (cpu->prop_pauth) {
79
+ if (cpu->prop_pauth_impdef) {
80
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, API, features);
81
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPI, 1);
82
+ } else {
83
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, APA, features);
84
+ isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPA, 1);
85
+ }
86
+ } else if (cpu->prop_pauth_impdef) {
87
+ error_setg(errp, "cannot enable pauth-impdef without pauth");
88
+ error_append_hint(errp, "Add pauth=on to the CPU property list.\n");
89
}
90
- } else if (cpu->prop_pauth_impdef) {
91
- error_setg(errp, "cannot enable pauth-impdef without pauth");
92
- error_append_hint(errp, "Add pauth=on to the CPU property list.\n");
93
}
94
95
- t = cpu->isar.id_aa64isar1;
96
- t = FIELD_DP64(t, ID_AA64ISAR1, APA, arch_val);
97
- t = FIELD_DP64(t, ID_AA64ISAR1, GPA, arch_val);
98
- t = FIELD_DP64(t, ID_AA64ISAR1, API, impdef_val);
99
- t = FIELD_DP64(t, ID_AA64ISAR1, GPI, impdef_val);
100
- cpu->isar.id_aa64isar1 = t;
101
+ cpu->isar.id_aa64isar1 = isar1;
102
}
103
104
static Property arm_cpu_pauth_property =
105
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
106
index XXXXXXX..XXXXXXX 100644
107
--- a/target/arm/tcg/cpu64.c
108
+++ b/target/arm/tcg/cpu64.c
109
@@ -XXX,XX +XXX,XX @@ void aarch64_max_tcg_initfn(Object *obj)
110
111
t = cpu->isar.id_aa64isar1;
112
t = FIELD_DP64(t, ID_AA64ISAR1, DPB, 2); /* FEAT_DPB2 */
113
+ t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_1);
114
+ t = FIELD_DP64(t, ID_AA64ISAR1, API, 1);
115
t = FIELD_DP64(t, ID_AA64ISAR1, JSCVT, 1); /* FEAT_JSCVT */
116
t = FIELD_DP64(t, ID_AA64ISAR1, FCMA, 1); /* FEAT_FCMA */
117
t = FIELD_DP64(t, ID_AA64ISAR1, LRCPC, 2); /* FEAT_LRCPC2 */
118
--
119
2.34.1
diff view generated by jsdifflib
[PULL 05/26] target/arm: Implement FEAT_PACQARMA3
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
Implement the QARMA3 cryptographic algorithm for PAC calculation.
4
Implement a cpu feature to select the algorithm and document it.
5
6
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
9
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Message-id: 20230829232335.965414-6-richard.henderson@linaro.org
11
Message-Id: <20230609172324.982888-4-aaron@os.amperecomputing.com>
12
[rth: Merge cpu feature addition from another patch.]
13
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
14
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
15
---
16
docs/system/arm/cpu-features.rst | 21 ++++++++-----
17
docs/system/arm/emulation.rst | 3 ++
18
target/arm/cpu.h | 1 +
19
target/arm/arm-qmp-cmds.c | 2 +-
20
target/arm/cpu64.c | 24 ++++++++++++--
21
target/arm/tcg/pauth_helper.c | 54 ++++++++++++++++++++++++++------
22
tests/qtest/arm-cpu-features.c | 12 ++++++-
23
7 files changed, 94 insertions(+), 23 deletions(-)
24
25
diff --git a/docs/system/arm/cpu-features.rst b/docs/system/arm/cpu-features.rst
26
index XXXXXXX..XXXXXXX 100644
27
--- a/docs/system/arm/cpu-features.rst
28
+++ b/docs/system/arm/cpu-features.rst
29
@@ -XXX,XX +XXX,XX @@ TCG VCPU Features
30
TCG VCPU features are CPU features that are specific to TCG.
31
Below is the list of TCG VCPU features and their descriptions.
32
33
-``pauth-impdef``
34
- When ``FEAT_Pauth`` is enabled, either the *impdef* (Implementation
35
- Defined) algorithm is enabled or the *architected* QARMA algorithm
36
- is enabled. By default the impdef algorithm is disabled, and QARMA
37
- is enabled.
38
+``pauth``
39
+ Enable or disable ``FEAT_Pauth`` entirely.
40
41
- The architected QARMA algorithm has good cryptographic properties,
42
- but can be quite slow to emulate. The impdef algorithm used by QEMU
43
- is non-cryptographic but significantly faster.
44
+``pauth-impdef``
45
+ When ``pauth`` is enabled, select the QEMU implementation defined algorithm.
46
+
47
+``pauth-qarma3``
48
+ When ``pauth`` is enabled, select the architected QARMA3 algorithm.
49
+
50
+Without either ``pauth-impdef`` or ``pauth-qarma3`` enabled,
51
+the architected QARMA5 algorithm is used. The architected QARMA5
52
+and QARMA3 algorithms have good cryptographic properties, but can
53
+be quite slow to emulate. The impdef algorithm used by QEMU is
54
+non-cryptographic but significantly faster.
55
56
SVE CPU Properties
57
==================
58
diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst
59
index XXXXXXX..XXXXXXX 100644
60
--- a/docs/system/arm/emulation.rst
61
+++ b/docs/system/arm/emulation.rst
62
@@ -XXX,XX +XXX,XX @@ the following architecture extensions:
63
- FEAT_MTE (Memory Tagging Extension)
64
- FEAT_MTE2 (Memory Tagging Extension)
65
- FEAT_MTE3 (MTE Asymmetric Fault Handling)
66
+- FEAT_PACIMP (Pointer authentication - IMPLEMENTATION DEFINED algorithm)
67
+- FEAT_PACQARMA3 (Pointer authentication - QARMA3 algorithm)
68
+- FEAT_PACQARMA5 (Pointer authentication - QARMA5 algorithm)
69
- FEAT_PAN (Privileged access never)
70
- FEAT_PAN2 (AT S1E1R and AT S1E1W instruction variants affected by PSTATE.PAN)
71
- FEAT_PAN3 (Support for SCTLR_ELx.EPAN)
72
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
73
index XXXXXXX..XXXXXXX 100644
74
--- a/target/arm/cpu.h
75
+++ b/target/arm/cpu.h
76
@@ -XXX,XX +XXX,XX @@ struct ArchCPU {
77
*/
78
bool prop_pauth;
79
bool prop_pauth_impdef;
80
+ bool prop_pauth_qarma3;
81
bool prop_lpa2;
82
83
/* DCZ blocksize, in log_2(words), ie low 4 bits of DCZID_EL0 */
84
diff --git a/target/arm/arm-qmp-cmds.c b/target/arm/arm-qmp-cmds.c
85
index XXXXXXX..XXXXXXX 100644
86
--- a/target/arm/arm-qmp-cmds.c
87
+++ b/target/arm/arm-qmp-cmds.c
88
@@ -XXX,XX +XXX,XX @@ static const char *cpu_model_advertised_features[] = {
89
"sve640", "sve768", "sve896", "sve1024", "sve1152", "sve1280",
90
"sve1408", "sve1536", "sve1664", "sve1792", "sve1920", "sve2048",
91
"kvm-no-adjvtime", "kvm-steal-time",
92
- "pauth", "pauth-impdef",
93
+ "pauth", "pauth-impdef", "pauth-qarma3",
94
NULL
95
};
96
97
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
98
index XXXXXXX..XXXXXXX 100644
99
--- a/target/arm/cpu64.c
100
+++ b/target/arm/cpu64.c
101
@@ -XXX,XX +XXX,XX @@ void aarch64_add_sme_properties(Object *obj)
102
void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp)
103
{
104
ARMPauthFeature features = cpu_isar_feature(pauth_feature, cpu);
105
- uint64_t isar1;
106
+ uint64_t isar1, isar2;
107
108
/*
109
* These properties enable or disable Pauth as a whole, or change
110
@@ -XXX,XX +XXX,XX @@ void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp)
111
isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, API, 0);
112
isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPI, 0);
113
114
+ isar2 = cpu->isar.id_aa64isar2;
115
+ isar2 = FIELD_DP64(isar2, ID_AA64ISAR2, APA3, 0);
116
+ isar2 = FIELD_DP64(isar2, ID_AA64ISAR2, GPA3, 0);
117
+
118
if (kvm_enabled() || hvf_enabled()) {
119
/*
120
* Exit early if PAuth is enabled and fall through to disable it.
121
@@ -XXX,XX +XXX,XX @@ void arm_cpu_pauth_finalize(ARMCPU *cpu, Error **errp)
122
}
123
124
if (cpu->prop_pauth) {
125
+ if (cpu->prop_pauth_impdef && cpu->prop_pauth_qarma3) {
126
+ error_setg(errp,
127
+ "cannot enable both pauth-impdef and pauth-qarma3");
128
+ return;
129
+ }
130
+
131
if (cpu->prop_pauth_impdef) {
132
isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, API, features);
133
isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPI, 1);
134
+ } else if (cpu->prop_pauth_qarma3) {
135
+ isar2 = FIELD_DP64(isar2, ID_AA64ISAR2, APA3, features);
136
+ isar2 = FIELD_DP64(isar2, ID_AA64ISAR2, GPA3, 1);
137
} else {
138
isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, APA, features);
139
isar1 = FIELD_DP64(isar1, ID_AA64ISAR1, GPA, 1);
140
}
141
- } else if (cpu->prop_pauth_impdef) {
142
- error_setg(errp, "cannot enable pauth-impdef without pauth");
143
+ } else if (cpu->prop_pauth_impdef || cpu->prop_pauth_qarma3) {
144
+ error_setg(errp, "cannot enable pauth-impdef or "
145
+ "pauth-qarma3 without pauth");
146
error_append_hint(errp, "Add pauth=on to the CPU property list.\n");
147
}
148
}
149
150
cpu->isar.id_aa64isar1 = isar1;
151
+ cpu->isar.id_aa64isar2 = isar2;
152
}
153
154
static Property arm_cpu_pauth_property =
155
DEFINE_PROP_BOOL("pauth", ARMCPU, prop_pauth, true);
156
static Property arm_cpu_pauth_impdef_property =
157
DEFINE_PROP_BOOL("pauth-impdef", ARMCPU, prop_pauth_impdef, false);
158
+static Property arm_cpu_pauth_qarma3_property =
159
+ DEFINE_PROP_BOOL("pauth-qarma3", ARMCPU, prop_pauth_qarma3, false);
160
161
void aarch64_add_pauth_properties(Object *obj)
162
{
163
@@ -XXX,XX +XXX,XX @@ void aarch64_add_pauth_properties(Object *obj)
164
cpu->prop_pauth = cpu_isar_feature(aa64_pauth, cpu);
165
} else {
166
qdev_property_add_static(DEVICE(obj), &arm_cpu_pauth_impdef_property);
167
+ qdev_property_add_static(DEVICE(obj), &arm_cpu_pauth_qarma3_property);
168
}
169
}
170
171
diff --git a/target/arm/tcg/pauth_helper.c b/target/arm/tcg/pauth_helper.c
172
index XXXXXXX..XXXXXXX 100644
173
--- a/target/arm/tcg/pauth_helper.c
174
+++ b/target/arm/tcg/pauth_helper.c
175
@@ -XXX,XX +XXX,XX @@ static uint64_t pac_sub(uint64_t i)
176
return o;
177
}
178
179
+static uint64_t pac_sub1(uint64_t i)
180
+{
181
+ static const uint8_t sub1[16] = {
182
+ 0xa, 0xd, 0xe, 0x6, 0xf, 0x7, 0x3, 0x5,
183
+ 0x9, 0x8, 0x0, 0xc, 0xb, 0x1, 0x2, 0x4,
184
+ };
185
+ uint64_t o = 0;
186
+ int b;
187
+
188
+ for (b = 0; b < 64; b += 4) {
189
+ o |= (uint64_t)sub1[(i >> b) & 0xf] << b;
190
+ }
191
+ return o;
192
+}
193
+
194
static uint64_t pac_inv_sub(uint64_t i)
195
{
196
static const uint8_t inv_sub[16] = {
197
@@ -XXX,XX +XXX,XX @@ static uint64_t tweak_inv_shuffle(uint64_t i)
198
}
199
200
static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier,
201
- ARMPACKey key)
202
+ ARMPACKey key, bool isqarma3)
203
{
204
static const uint64_t RC[5] = {
205
0x0000000000000000ull,
206
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier,
207
0x452821E638D01377ull,
208
};
209
const uint64_t alpha = 0xC0AC29B7C97C50DDull;
210
+ int iterations = isqarma3 ? 2 : 4;
211
/*
212
* Note that in the ARM pseudocode, key0 contains bits <127:64>
213
* and key1 contains bits <63:0> of the 128-bit key.
214
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier,
215
runningmod = modifier;
216
workingval = data ^ key0;
217
218
- for (i = 0; i <= 4; ++i) {
219
+ for (i = 0; i <= iterations; ++i) {
220
roundkey = key1 ^ runningmod;
221
workingval ^= roundkey;
222
workingval ^= RC[i];
223
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac_architected(uint64_t data, uint64_t modifier,
224
workingval = pac_cell_shuffle(workingval);
225
workingval = pac_mult(workingval);
226
}
227
- workingval = pac_sub(workingval);
228
+ if (isqarma3) {
229
+ workingval = pac_sub1(workingval);
230
+ } else {
231
+ workingval = pac_sub(workingval);
232
+ }
233
runningmod = tweak_shuffle(runningmod);
234
}
235
roundkey = modk0 ^ runningmod;
236
workingval ^= roundkey;
237
workingval = pac_cell_shuffle(workingval);
238
workingval = pac_mult(workingval);
239
- workingval = pac_sub(workingval);
240
+ if (isqarma3) {
241
+ workingval = pac_sub1(workingval);
242
+ } else {
243
+ workingval = pac_sub(workingval);
244
+ }
245
workingval = pac_cell_shuffle(workingval);
246
workingval = pac_mult(workingval);
247
workingval ^= key1;
248
workingval = pac_cell_inv_shuffle(workingval);
249
- workingval = pac_inv_sub(workingval);
250
+ if (isqarma3) {
251
+ workingval = pac_sub1(workingval);
252
+ } else {
253
+ workingval = pac_inv_sub(workingval);
254
+ }
255
workingval = pac_mult(workingval);
256
workingval = pac_cell_inv_shuffle(workingval);
257
workingval ^= key0;
258
workingval ^= runningmod;
259
- for (i = 0; i <= 4; ++i) {
260
- workingval = pac_inv_sub(workingval);
261
- if (i < 4) {
262
+ for (i = 0; i <= iterations; ++i) {
263
+ if (isqarma3) {
264
+ workingval = pac_sub1(workingval);
265
+ } else {
266
+ workingval = pac_inv_sub(workingval);
267
+ }
268
+ if (i < iterations) {
269
workingval = pac_mult(workingval);
270
workingval = pac_cell_inv_shuffle(workingval);
271
}
272
runningmod = tweak_inv_shuffle(runningmod);
273
roundkey = key1 ^ runningmod;
274
- workingval ^= RC[4 - i];
275
+ workingval ^= RC[iterations - i];
276
workingval ^= roundkey;
277
workingval ^= alpha;
278
}
279
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac(CPUARMState *env, uint64_t data,
280
uint64_t modifier, ARMPACKey key)
281
{
282
if (cpu_isar_feature(aa64_pauth_qarma5, env_archcpu(env))) {
283
- return pauth_computepac_architected(data, modifier, key);
284
+ return pauth_computepac_architected(data, modifier, key, false);
285
+ } else if (cpu_isar_feature(aa64_pauth_qarma3, env_archcpu(env))) {
286
+ return pauth_computepac_architected(data, modifier, key, true);
287
} else {
288
return pauth_computepac_impdef(data, modifier, key);
289
}
290
diff --git a/tests/qtest/arm-cpu-features.c b/tests/qtest/arm-cpu-features.c
291
index XXXXXXX..XXXXXXX 100644
292
--- a/tests/qtest/arm-cpu-features.c
293
+++ b/tests/qtest/arm-cpu-features.c
294
@@ -XXX,XX +XXX,XX @@ static void pauth_tests_default(QTestState *qts, const char *cpu_type)
295
{
296
assert_has_feature_enabled(qts, cpu_type, "pauth");
297
assert_has_feature_disabled(qts, cpu_type, "pauth-impdef");
298
+ assert_has_feature_disabled(qts, cpu_type, "pauth-qarma3");
299
assert_set_feature(qts, cpu_type, "pauth", false);
300
assert_set_feature(qts, cpu_type, "pauth", true);
301
assert_set_feature(qts, cpu_type, "pauth-impdef", true);
302
assert_set_feature(qts, cpu_type, "pauth-impdef", false);
303
- assert_error(qts, cpu_type, "cannot enable pauth-impdef without pauth",
304
+ assert_set_feature(qts, cpu_type, "pauth-qarma3", true);
305
+ assert_set_feature(qts, cpu_type, "pauth-qarma3", false);
306
+ assert_error(qts, cpu_type,
307
+ "cannot enable pauth-impdef or pauth-qarma3 without pauth",
308
"{ 'pauth': false, 'pauth-impdef': true }");
309
+ assert_error(qts, cpu_type,
310
+ "cannot enable pauth-impdef or pauth-qarma3 without pauth",
311
+ "{ 'pauth': false, 'pauth-qarma3': true }");
312
+ assert_error(qts, cpu_type,
313
+ "cannot enable both pauth-impdef and pauth-qarma3",
314
+ "{ 'pauth': true, 'pauth-impdef': true, 'pauth-qarma3': true }");
315
}
316
317
static void test_query_cpu_model_expansion(const void *data)
318
--
319
2.34.1
diff view generated by jsdifflib
[PULL 06/26] target/arm: Implement FEAT_EPAC
Deleted patch
1
From: Aaron Lindsay <aaron@os.amperecomputing.com>
2
1
3
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
4
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
5
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Message-id: 20230829232335.965414-7-richard.henderson@linaro.org
8
Message-Id: <20230609172324.982888-5-aaron@os.amperecomputing.com>
9
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
docs/system/arm/emulation.rst | 1 +
13
target/arm/tcg/cpu64.c | 2 +-
14
target/arm/tcg/pauth_helper.c | 16 +++++++++++-----
15
3 files changed, 13 insertions(+), 6 deletions(-)
16
17
diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst
18
index XXXXXXX..XXXXXXX 100644
19
--- a/docs/system/arm/emulation.rst
20
+++ b/docs/system/arm/emulation.rst
21
@@ -XXX,XX +XXX,XX @@ the following architecture extensions:
22
- FEAT_DotProd (Advanced SIMD dot product instructions)
23
- FEAT_DoubleFault (Double Fault Extension)
24
- FEAT_E0PD (Preventing EL0 access to halves of address maps)
25
+- FEAT_EPAC (Enhanced pointer authentication)
26
- FEAT_ETS (Enhanced Translation Synchronization)
27
- FEAT_EVT (Enhanced Virtualization Traps)
28
- FEAT_FCMA (Floating-point complex number instructions)
29
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
30
index XXXXXXX..XXXXXXX 100644
31
--- a/target/arm/tcg/cpu64.c
32
+++ b/target/arm/tcg/cpu64.c
33
@@ -XXX,XX +XXX,XX @@ void aarch64_max_tcg_initfn(Object *obj)
34
35
t = cpu->isar.id_aa64isar1;
36
t = FIELD_DP64(t, ID_AA64ISAR1, DPB, 2); /* FEAT_DPB2 */
37
- t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_1);
38
+ t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_EPAC);
39
t = FIELD_DP64(t, ID_AA64ISAR1, API, 1);
40
t = FIELD_DP64(t, ID_AA64ISAR1, JSCVT, 1); /* FEAT_JSCVT */
41
t = FIELD_DP64(t, ID_AA64ISAR1, FCMA, 1); /* FEAT_FCMA */
42
diff --git a/target/arm/tcg/pauth_helper.c b/target/arm/tcg/pauth_helper.c
43
index XXXXXXX..XXXXXXX 100644
44
--- a/target/arm/tcg/pauth_helper.c
45
+++ b/target/arm/tcg/pauth_helper.c
46
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_computepac(CPUARMState *env, uint64_t data,
47
static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,
48
ARMPACKey *key, bool data)
49
{
50
+ ARMCPU *cpu = env_archcpu(env);
51
ARMMMUIdx mmu_idx = arm_stage1_mmu_idx(env);
52
ARMVAParameters param = aa64_va_parameters(env, ptr, mmu_idx, data, false);
53
+ ARMPauthFeature pauth_feature = cpu_isar_feature(pauth_feature, cpu);
54
uint64_t pac, ext_ptr, ext, test;
55
int bot_bit, top_bit;
56
57
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,
58
*/
59
test = sextract64(ptr, bot_bit, top_bit - bot_bit);
60
if (test != 0 && test != -1) {
61
- /*
62
- * Note that our top_bit is one greater than the pseudocode's
63
- * version, hence "- 2" here.
64
- */
65
- pac ^= MAKE_64BIT_MASK(top_bit - 2, 1);
66
+ if (pauth_feature == PauthFeat_EPAC) {
67
+ pac = 0;
68
+ } else {
69
+ /*
70
+ * Note that our top_bit is one greater than the pseudocode's
71
+ * version, hence "- 2" here.
72
+ */
73
+ pac ^= MAKE_64BIT_MASK(top_bit - 2, 1);
74
+ }
75
}
76
77
/*
78
--
79
2.34.1
diff view generated by jsdifflib
[PULL 07/26] target/arm: Implement FEAT_Pauth2
Deleted patch
1
From: Aaron Lindsay <aaron@os.amperecomputing.com>
2
1
3
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
4
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
5
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Message-id: 20230829232335.965414-8-richard.henderson@linaro.org
8
Message-Id: <20230609172324.982888-6-aaron@os.amperecomputing.com>
9
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
docs/system/arm/emulation.rst | 1 +
13
target/arm/tcg/cpu64.c | 2 +-
14
target/arm/tcg/pauth_helper.c | 21 +++++++++++++++++----
15
3 files changed, 19 insertions(+), 5 deletions(-)
16
17
diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst
18
index XXXXXXX..XXXXXXX 100644
19
--- a/docs/system/arm/emulation.rst
20
+++ b/docs/system/arm/emulation.rst
21
@@ -XXX,XX +XXX,XX @@ the following architecture extensions:
22
- FEAT_PAN2 (AT S1E1R and AT S1E1W instruction variants affected by PSTATE.PAN)
23
- FEAT_PAN3 (Support for SCTLR_ELx.EPAN)
24
- FEAT_PAuth (Pointer authentication)
25
+- FEAT_PAuth2 (Enhacements to pointer authentication)
26
- FEAT_PMULL (PMULL, PMULL2 instructions)
27
- FEAT_PMUv3p1 (PMU Extensions v3.1)
28
- FEAT_PMUv3p4 (PMU Extensions v3.4)
29
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
30
index XXXXXXX..XXXXXXX 100644
31
--- a/target/arm/tcg/cpu64.c
32
+++ b/target/arm/tcg/cpu64.c
33
@@ -XXX,XX +XXX,XX @@ void aarch64_max_tcg_initfn(Object *obj)
34
35
t = cpu->isar.id_aa64isar1;
36
t = FIELD_DP64(t, ID_AA64ISAR1, DPB, 2); /* FEAT_DPB2 */
37
- t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_EPAC);
38
+ t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_2);
39
t = FIELD_DP64(t, ID_AA64ISAR1, API, 1);
40
t = FIELD_DP64(t, ID_AA64ISAR1, JSCVT, 1); /* FEAT_JSCVT */
41
t = FIELD_DP64(t, ID_AA64ISAR1, FCMA, 1); /* FEAT_FCMA */
42
diff --git a/target/arm/tcg/pauth_helper.c b/target/arm/tcg/pauth_helper.c
43
index XXXXXXX..XXXXXXX 100644
44
--- a/target/arm/tcg/pauth_helper.c
45
+++ b/target/arm/tcg/pauth_helper.c
46
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,
47
*/
48
test = sextract64(ptr, bot_bit, top_bit - bot_bit);
49
if (test != 0 && test != -1) {
50
- if (pauth_feature == PauthFeat_EPAC) {
51
+ if (pauth_feature >= PauthFeat_2) {
52
+ /* No action required */
53
+ } else if (pauth_feature == PauthFeat_EPAC) {
54
pac = 0;
55
} else {
56
/*
57
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_addpac(CPUARMState *env, uint64_t ptr, uint64_t modifier,
58
* Preserve the determination between upper and lower at bit 55,
59
* and insert pointer authentication code.
60
*/
61
+ if (pauth_feature >= PauthFeat_2) {
62
+ pac ^= ptr;
63
+ }
64
if (param.tbi) {
65
ptr &= ~MAKE_64BIT_MASK(bot_bit, 55 - bot_bit + 1);
66
pac &= MAKE_64BIT_MASK(bot_bit, 54 - bot_bit + 1);
67
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param)
68
static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier,
69
ARMPACKey *key, bool data, int keynumber)
70
{
71
+ ARMCPU *cpu = env_archcpu(env);
72
ARMMMUIdx mmu_idx = arm_stage1_mmu_idx(env);
73
ARMVAParameters param = aa64_va_parameters(env, ptr, mmu_idx, data, false);
74
+ ARMPauthFeature pauth_feature = cpu_isar_feature(pauth_feature, cpu);
75
int bot_bit, top_bit;
76
- uint64_t pac, orig_ptr, test;
77
+ uint64_t pac, orig_ptr, cmp_mask;
78
79
orig_ptr = pauth_original_ptr(ptr, param);
80
pac = pauth_computepac(env, orig_ptr, modifier, *key);
81
bot_bit = 64 - param.tsz;
82
top_bit = 64 - 8 * param.tbi;
83
84
- test = (pac ^ ptr) & ~MAKE_64BIT_MASK(55, 1);
85
- if (unlikely(extract64(test, bot_bit, top_bit - bot_bit))) {
86
+ cmp_mask = MAKE_64BIT_MASK(bot_bit, top_bit - bot_bit);
87
+ cmp_mask &= ~MAKE_64BIT_MASK(55, 1);
88
+
89
+ if (pauth_feature >= PauthFeat_2) {
90
+ return ptr ^ (pac & cmp_mask);
91
+ }
92
+
93
+ if ((pac ^ ptr) & cmp_mask) {
94
int error_code = (keynumber << 1) | (keynumber ^ 1);
95
if (param.tbi) {
96
return deposit64(orig_ptr, 53, 2, error_code);
97
--
98
2.34.1
diff view generated by jsdifflib
[PULL 08/26] target/arm: Inform helpers whether a PAC instruction is 'combined'
Deleted patch
1
From: Aaron Lindsay <aaron@os.amperecomputing.com>
2
1
3
An instruction is a 'combined' Pointer Authentication instruction
4
if it does something in addition to PAC -- for instance, branching
5
to or loading an address from the authenticated pointer.
6
7
Knowing whether a PAC operation is 'combined' is needed to
8
implement FEAT_FPACCOMBINE.
9
10
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
11
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
12
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
13
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
14
Message-id: 20230829232335.965414-9-richard.henderson@linaro.org
15
Message-Id: <20230609172324.982888-7-aaron@os.amperecomputing.com>
16
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
17
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18
---
19
target/arm/tcg/helper-a64.h | 4 ++
20
target/arm/tcg/pauth_helper.c | 71 +++++++++++++++++++++++++++-------
21
target/arm/tcg/translate-a64.c | 12 +++---
22
3 files changed, 68 insertions(+), 19 deletions(-)
23
24
diff --git a/target/arm/tcg/helper-a64.h b/target/arm/tcg/helper-a64.h
25
index XXXXXXX..XXXXXXX 100644
26
--- a/target/arm/tcg/helper-a64.h
27
+++ b/target/arm/tcg/helper-a64.h
28
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_3(pacda, TCG_CALL_NO_WG, i64, env, i64, i64)
29
DEF_HELPER_FLAGS_3(pacdb, TCG_CALL_NO_WG, i64, env, i64, i64)
30
DEF_HELPER_FLAGS_3(pacga, TCG_CALL_NO_WG, i64, env, i64, i64)
31
DEF_HELPER_FLAGS_3(autia, TCG_CALL_NO_WG, i64, env, i64, i64)
32
+DEF_HELPER_FLAGS_3(autia_combined, TCG_CALL_NO_WG, i64, env, i64, i64)
33
DEF_HELPER_FLAGS_3(autib, TCG_CALL_NO_WG, i64, env, i64, i64)
34
+DEF_HELPER_FLAGS_3(autib_combined, TCG_CALL_NO_WG, i64, env, i64, i64)
35
DEF_HELPER_FLAGS_3(autda, TCG_CALL_NO_WG, i64, env, i64, i64)
36
+DEF_HELPER_FLAGS_3(autda_combined, TCG_CALL_NO_WG, i64, env, i64, i64)
37
DEF_HELPER_FLAGS_3(autdb, TCG_CALL_NO_WG, i64, env, i64, i64)
38
+DEF_HELPER_FLAGS_3(autdb_combined, TCG_CALL_NO_WG, i64, env, i64, i64)
39
DEF_HELPER_FLAGS_2(xpaci, TCG_CALL_NO_RWG_SE, i64, env, i64)
40
DEF_HELPER_FLAGS_2(xpacd, TCG_CALL_NO_RWG_SE, i64, env, i64)
41
42
diff --git a/target/arm/tcg/pauth_helper.c b/target/arm/tcg/pauth_helper.c
43
index XXXXXXX..XXXXXXX 100644
44
--- a/target/arm/tcg/pauth_helper.c
45
+++ b/target/arm/tcg/pauth_helper.c
46
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param)
47
}
48
49
static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier,
50
- ARMPACKey *key, bool data, int keynumber)
51
+ ARMPACKey *key, bool data, int keynumber,
52
+ uintptr_t ra, bool is_combined)
53
{
54
ARMCPU *cpu = env_archcpu(env);
55
ARMMMUIdx mmu_idx = arm_stage1_mmu_idx(env);
56
@@ -XXX,XX +XXX,XX @@ uint64_t HELPER(pacga)(CPUARMState *env, uint64_t x, uint64_t y)
57
return pac & 0xffffffff00000000ull;
58
}
59
60
-uint64_t HELPER(autia)(CPUARMState *env, uint64_t x, uint64_t y)
61
+static uint64_t pauth_autia(CPUARMState *env, uint64_t x, uint64_t y,
62
+ uintptr_t ra, bool is_combined)
63
{
64
int el = arm_current_el(env);
65
if (!pauth_key_enabled(env, el, SCTLR_EnIA)) {
66
return x;
67
}
68
- pauth_check_trap(env, el, GETPC());
69
- return pauth_auth(env, x, y, &env->keys.apia, false, 0);
70
+ pauth_check_trap(env, el, ra);
71
+ return pauth_auth(env, x, y, &env->keys.apia, false, 0, ra, is_combined);
72
}
73
74
-uint64_t HELPER(autib)(CPUARMState *env, uint64_t x, uint64_t y)
75
+uint64_t HELPER(autia)(CPUARMState *env, uint64_t x, uint64_t y)
76
+{
77
+ return pauth_autia(env, x, y, GETPC(), false);
78
+}
79
+
80
+uint64_t HELPER(autia_combined)(CPUARMState *env, uint64_t x, uint64_t y)
81
+{
82
+ return pauth_autia(env, x, y, GETPC(), true);
83
+}
84
+
85
+static uint64_t pauth_autib(CPUARMState *env, uint64_t x, uint64_t y,
86
+ uintptr_t ra, bool is_combined)
87
{
88
int el = arm_current_el(env);
89
if (!pauth_key_enabled(env, el, SCTLR_EnIB)) {
90
return x;
91
}
92
- pauth_check_trap(env, el, GETPC());
93
- return pauth_auth(env, x, y, &env->keys.apib, false, 1);
94
+ pauth_check_trap(env, el, ra);
95
+ return pauth_auth(env, x, y, &env->keys.apib, false, 1, ra, is_combined);
96
}
97
98
-uint64_t HELPER(autda)(CPUARMState *env, uint64_t x, uint64_t y)
99
+uint64_t HELPER(autib)(CPUARMState *env, uint64_t x, uint64_t y)
100
+{
101
+ return pauth_autib(env, x, y, GETPC(), false);
102
+}
103
+
104
+uint64_t HELPER(autib_combined)(CPUARMState *env, uint64_t x, uint64_t y)
105
+{
106
+ return pauth_autib(env, x, y, GETPC(), true);
107
+}
108
+
109
+static uint64_t pauth_autda(CPUARMState *env, uint64_t x, uint64_t y,
110
+ uintptr_t ra, bool is_combined)
111
{
112
int el = arm_current_el(env);
113
if (!pauth_key_enabled(env, el, SCTLR_EnDA)) {
114
return x;
115
}
116
- pauth_check_trap(env, el, GETPC());
117
- return pauth_auth(env, x, y, &env->keys.apda, true, 0);
118
+ pauth_check_trap(env, el, ra);
119
+ return pauth_auth(env, x, y, &env->keys.apda, true, 0, ra, is_combined);
120
}
121
122
-uint64_t HELPER(autdb)(CPUARMState *env, uint64_t x, uint64_t y)
123
+uint64_t HELPER(autda)(CPUARMState *env, uint64_t x, uint64_t y)
124
+{
125
+ return pauth_autda(env, x, y, GETPC(), false);
126
+}
127
+
128
+uint64_t HELPER(autda_combined)(CPUARMState *env, uint64_t x, uint64_t y)
129
+{
130
+ return pauth_autda(env, x, y, GETPC(), true);
131
+}
132
+
133
+static uint64_t pauth_autdb(CPUARMState *env, uint64_t x, uint64_t y,
134
+ uintptr_t ra, bool is_combined)
135
{
136
int el = arm_current_el(env);
137
if (!pauth_key_enabled(env, el, SCTLR_EnDB)) {
138
return x;
139
}
140
- pauth_check_trap(env, el, GETPC());
141
- return pauth_auth(env, x, y, &env->keys.apdb, true, 1);
142
+ pauth_check_trap(env, el, ra);
143
+ return pauth_auth(env, x, y, &env->keys.apdb, true, 1, ra, is_combined);
144
+}
145
+
146
+uint64_t HELPER(autdb)(CPUARMState *env, uint64_t x, uint64_t y)
147
+{
148
+ return pauth_autdb(env, x, y, GETPC(), false);
149
+}
150
+
151
+uint64_t HELPER(autdb_combined)(CPUARMState *env, uint64_t x, uint64_t y)
152
+{
153
+ return pauth_autdb(env, x, y, GETPC(), true);
154
}
155
156
uint64_t HELPER(xpaci)(CPUARMState *env, uint64_t a)
157
diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c
158
index XXXXXXX..XXXXXXX 100644
159
--- a/target/arm/tcg/translate-a64.c
160
+++ b/target/arm/tcg/translate-a64.c
161
@@ -XXX,XX +XXX,XX @@ static TCGv_i64 auth_branch_target(DisasContext *s, TCGv_i64 dst,
162
163
truedst = tcg_temp_new_i64();
164
if (use_key_a) {
165
- gen_helper_autia(truedst, cpu_env, dst, modifier);
166
+ gen_helper_autia_combined(truedst, cpu_env, dst, modifier);
167
} else {
168
- gen_helper_autib(truedst, cpu_env, dst, modifier);
169
+ gen_helper_autib_combined(truedst, cpu_env, dst, modifier);
170
}
171
return truedst;
172
}
173
@@ -XXX,XX +XXX,XX @@ static bool trans_LDRA(DisasContext *s, arg_LDRA *a)
174
175
if (s->pauth_active) {
176
if (!a->m) {
177
- gen_helper_autda(dirty_addr, cpu_env, dirty_addr,
178
- tcg_constant_i64(0));
179
+ gen_helper_autda_combined(dirty_addr, cpu_env, dirty_addr,
180
+ tcg_constant_i64(0));
181
} else {
182
- gen_helper_autdb(dirty_addr, cpu_env, dirty_addr,
183
- tcg_constant_i64(0));
184
+ gen_helper_autdb_combined(dirty_addr, cpu_env, dirty_addr,
185
+ tcg_constant_i64(0));
186
}
187
}
188
189
--
190
2.34.1
191
192
diff view generated by jsdifflib
[PULL 09/26] target/arm: Implement FEAT_FPAC and FEAT_FPACCOMBINE
Deleted patch
1
From: Aaron Lindsay <aaron@os.amperecomputing.com>
2
1
3
Signed-off-by: Aaron Lindsay <aaron@os.amperecomputing.com>
4
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Message-id: 20230829232335.965414-10-richard.henderson@linaro.org
7
Message-Id: <20230609172324.982888-8-aaron@os.amperecomputing.com>
8
[rth: Simplify fpac comparison, reusing cmp_mask]
9
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
docs/system/arm/emulation.rst | 2 ++
13
target/arm/syndrome.h | 7 +++++++
14
target/arm/tcg/cpu64.c | 2 +-
15
target/arm/tcg/pauth_helper.c | 18 +++++++++++++++++-
16
4 files changed, 27 insertions(+), 2 deletions(-)
17
18
diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst
19
index XXXXXXX..XXXXXXX 100644
20
--- a/docs/system/arm/emulation.rst
21
+++ b/docs/system/arm/emulation.rst
22
@@ -XXX,XX +XXX,XX @@ the following architecture extensions:
23
- FEAT_FGT (Fine-Grained Traps)
24
- FEAT_FHM (Floating-point half-precision multiplication instructions)
25
- FEAT_FP16 (Half-precision floating-point data processing)
26
+- FEAT_FPAC (Faulting on AUT* instructions)
27
+- FEAT_FPACCOMBINE (Faulting on combined pointer authentication instructions)
28
- FEAT_FRINTTS (Floating-point to integer instructions)
29
- FEAT_FlagM (Flag manipulation instructions v2)
30
- FEAT_FlagM2 (Enhancements to flag manipulation instructions)
31
diff --git a/target/arm/syndrome.h b/target/arm/syndrome.h
32
index XXXXXXX..XXXXXXX 100644
33
--- a/target/arm/syndrome.h
34
+++ b/target/arm/syndrome.h
35
@@ -XXX,XX +XXX,XX @@ enum arm_exception_class {
36
EC_SYSTEMREGISTERTRAP = 0x18,
37
EC_SVEACCESSTRAP = 0x19,
38
EC_ERETTRAP = 0x1a,
39
+ EC_PACFAIL = 0x1c,
40
EC_SMETRAP = 0x1d,
41
EC_GPC = 0x1e,
42
EC_INSNABORT = 0x20,
43
@@ -XXX,XX +XXX,XX @@ static inline uint32_t syn_smetrap(SMEExceptionType etype, bool is_16bit)
44
| (is_16bit ? 0 : ARM_EL_IL) | etype;
45
}
46
47
+static inline uint32_t syn_pacfail(bool data, int keynumber)
48
+{
49
+ int error_code = (data << 1) | keynumber;
50
+ return (EC_PACFAIL << ARM_EL_EC_SHIFT) | ARM_EL_IL | error_code;
51
+}
52
+
53
static inline uint32_t syn_pactrap(void)
54
{
55
return EC_PACTRAP << ARM_EL_EC_SHIFT;
56
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
57
index XXXXXXX..XXXXXXX 100644
58
--- a/target/arm/tcg/cpu64.c
59
+++ b/target/arm/tcg/cpu64.c
60
@@ -XXX,XX +XXX,XX @@ void aarch64_max_tcg_initfn(Object *obj)
61
62
t = cpu->isar.id_aa64isar1;
63
t = FIELD_DP64(t, ID_AA64ISAR1, DPB, 2); /* FEAT_DPB2 */
64
- t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_2);
65
+ t = FIELD_DP64(t, ID_AA64ISAR1, APA, PauthFeat_FPACCOMBINED);
66
t = FIELD_DP64(t, ID_AA64ISAR1, API, 1);
67
t = FIELD_DP64(t, ID_AA64ISAR1, JSCVT, 1); /* FEAT_JSCVT */
68
t = FIELD_DP64(t, ID_AA64ISAR1, FCMA, 1); /* FEAT_FCMA */
69
diff --git a/target/arm/tcg/pauth_helper.c b/target/arm/tcg/pauth_helper.c
70
index XXXXXXX..XXXXXXX 100644
71
--- a/target/arm/tcg/pauth_helper.c
72
+++ b/target/arm/tcg/pauth_helper.c
73
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_original_ptr(uint64_t ptr, ARMVAParameters param)
74
}
75
}
76
77
+static G_NORETURN
78
+void pauth_fail_exception(CPUARMState *env, bool data,
79
+ int keynumber, uintptr_t ra)
80
+{
81
+ raise_exception_ra(env, EXCP_UDEF, syn_pacfail(data, keynumber),
82
+ exception_target_el(env), ra);
83
+}
84
+
85
static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier,
86
ARMPACKey *key, bool data, int keynumber,
87
uintptr_t ra, bool is_combined)
88
@@ -XXX,XX +XXX,XX @@ static uint64_t pauth_auth(CPUARMState *env, uint64_t ptr, uint64_t modifier,
89
cmp_mask &= ~MAKE_64BIT_MASK(55, 1);
90
91
if (pauth_feature >= PauthFeat_2) {
92
- return ptr ^ (pac & cmp_mask);
93
+ ARMPauthFeature fault_feature =
94
+ is_combined ? PauthFeat_FPACCOMBINED : PauthFeat_FPAC;
95
+ uint64_t result = ptr ^ (pac & cmp_mask);
96
+
97
+ if (pauth_feature >= fault_feature
98
+ && ((result ^ sextract64(result, 55, 1)) & cmp_mask)) {
99
+ pauth_fail_exception(env, data, keynumber, ra);
100
+ }
101
+ return result;
102
}
103
104
if ((pac ^ ptr) & cmp_mask) {
105
--
106
2.34.1
diff view generated by jsdifflib
[PULL 10/26] hw/intc/arm_gicv3_its: Avoid maybe-uninitialized error in get_vte()
Deleted patch
1
From: Philippe Mathieu-Daudé <philmd@linaro.org>
2
1
3
Fix when using GCC v11.4 (Ubuntu 11.4.0-1ubuntu1~22.04) with CFLAGS=-Og:
4
5
[4/6] Compiling C object libcommon.fa.p/hw_intc_arm_gicv3_its.c.o
6
FAILED: libcommon.fa.p/hw_intc_arm_gicv3_its.c.o
7
inlined from ‘lookup_vte’ at hw/intc/arm_gicv3_its.c:453:9,
8
inlined from ‘vmovp_callback’ at hw/intc/arm_gicv3_its.c:1039:14:
9
hw/intc/arm_gicv3_its.c:347:9: error: ‘vte.rdbase’ may be used uninitialized [-Werror=maybe-uninitialized]
10
347 | trace_gicv3_its_vte_read(vpeid, vte->valid, vte->vptsize,
11
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
12
348 | vte->vptaddr, vte->rdbase);
13
| ~~~~~~~~~~~~~~~~~~~~~~~~~~
14
hw/intc/arm_gicv3_its.c: In function ‘vmovp_callback’:
15
hw/intc/arm_gicv3_its.c:1036:13: note: ‘vte’ declared here
16
1036 | VTEntry vte;
17
| ^~~
18
19
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
20
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
21
Message-id: 20230831131348.69032-1-philmd@linaro.org
22
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
23
---
24
hw/intc/arm_gicv3_its.c | 15 ++++++---------
25
1 file changed, 6 insertions(+), 9 deletions(-)
26
27
diff --git a/hw/intc/arm_gicv3_its.c b/hw/intc/arm_gicv3_its.c
28
index XXXXXXX..XXXXXXX 100644
29
--- a/hw/intc/arm_gicv3_its.c
30
+++ b/hw/intc/arm_gicv3_its.c
31
@@ -XXX,XX +XXX,XX @@ static MemTxResult get_vte(GICv3ITSState *s, uint32_t vpeid, VTEntry *vte)
32
if (entry_addr == -1) {
33
/* No L2 table entry, i.e. no valid VTE, or a memory error */
34
vte->valid = false;
35
- goto out;
36
+ trace_gicv3_its_vte_read_fault(vpeid);
37
+ return MEMTX_OK;
38
}
39
vteval = address_space_ldq_le(as, entry_addr, MEMTXATTRS_UNSPECIFIED, &res);
40
if (res != MEMTX_OK) {
41
- goto out;
42
+ trace_gicv3_its_vte_read_fault(vpeid);
43
+ return res;
44
}
45
vte->valid = FIELD_EX64(vteval, VTE, VALID);
46
vte->vptsize = FIELD_EX64(vteval, VTE, VPTSIZE);
47
vte->vptaddr = FIELD_EX64(vteval, VTE, VPTADDR);
48
vte->rdbase = FIELD_EX64(vteval, VTE, RDBASE);
49
-out:
50
- if (res != MEMTX_OK) {
51
- trace_gicv3_its_vte_read_fault(vpeid);
52
- } else {
53
- trace_gicv3_its_vte_read(vpeid, vte->valid, vte->vptsize,
54
- vte->vptaddr, vte->rdbase);
55
- }
56
+ trace_gicv3_its_vte_read(vpeid, vte->valid, vte->vptsize,
57
+ vte->vptaddr, vte->rdbase);
58
return res;
59
}
60
61
--
62
2.34.1
63
64
diff view generated by jsdifflib
[PULL 11/26] hw/misc: Introduce the Xilinx CFI interface
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Introduce the Xilinx Configuration Frame Interface (CFI) for transmitting
4
CFI data packets between the Xilinx Configuration Frame Unit models
5
(CFU_APB, CFU_FDRO and CFU_SFR), the Xilinx CFRAME controller (CFRAME_REG)
6
and the Xilinx CFRAME broadcast controller (CFRAME_BCAST_REG) models (when
7
emulating bitstream programming and readback).
8
9
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
10
Reviewed-by: Sai Pavan Boddu <sai.pavan.boddu@amd.com>
11
Acked-by: Edgar E. Iglesias <edgar@zeroasic.com>
12
Message-id: 20230831165701.2016397-2-francisco.iglesias@amd.com
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
---
15
MAINTAINERS | 6 ++++
16
include/hw/misc/xlnx-cfi-if.h | 59 +++++++++++++++++++++++++++++++++++
17
hw/misc/xlnx-cfi-if.c | 34 ++++++++++++++++++++
18
hw/misc/meson.build | 1 +
19
4 files changed, 100 insertions(+)
20
create mode 100644 include/hw/misc/xlnx-cfi-if.h
21
create mode 100644 hw/misc/xlnx-cfi-if.c
22
23
diff --git a/MAINTAINERS b/MAINTAINERS
24
index XXXXXXX..XXXXXXX 100644
25
--- a/MAINTAINERS
26
+++ b/MAINTAINERS
27
@@ -XXX,XX +XXX,XX @@ S: Maintained
28
F: hw/ssi/xlnx-versal-ospi.c
29
F: include/hw/ssi/xlnx-versal-ospi.h
30
31
+Xilinx Versal CFI
32
+M: Francisco Iglesias <francisco.iglesias@amd.com>
33
+S: Maintained
34
+F: hw/misc/xlnx-cfi-if.c
35
+F: include/hw/misc/xlnx-cfi-if.h
36
+
37
STM32F100
38
M: Alexandre Iooss <erdnaxe@crans.org>
39
L: qemu-arm@nongnu.org
40
diff --git a/include/hw/misc/xlnx-cfi-if.h b/include/hw/misc/xlnx-cfi-if.h
41
new file mode 100644
42
index XXXXXXX..XXXXXXX
43
--- /dev/null
44
+++ b/include/hw/misc/xlnx-cfi-if.h
45
@@ -XXX,XX +XXX,XX @@
46
+/*
47
+ * Xilinx CFI interface
48
+ *
49
+ * Copyright (C) 2023, Advanced Micro Devices, Inc.
50
+ *
51
+ * Written by Francisco Iglesias <francisco.iglesias@amd.com>
52
+ *
53
+ * SPDX-License-Identifier: GPL-2.0-or-later
54
+ */
55
+#ifndef XLNX_CFI_IF_H
56
+#define XLNX_CFI_IF_H 1
57
+
58
+#include "qemu/help-texts.h"
59
+#include "hw/hw.h"
60
+#include "qom/object.h"
61
+
62
+#define TYPE_XLNX_CFI_IF "xlnx-cfi-if"
63
+typedef struct XlnxCfiIfClass XlnxCfiIfClass;
64
+DECLARE_CLASS_CHECKERS(XlnxCfiIfClass, XLNX_CFI_IF, TYPE_XLNX_CFI_IF)
65
+
66
+#define XLNX_CFI_IF(obj) \
67
+ INTERFACE_CHECK(XlnxCfiIf, (obj), TYPE_XLNX_CFI_IF)
68
+
69
+typedef enum {
70
+ PACKET_TYPE_CFU = 0x52,
71
+ PACKET_TYPE_CFRAME = 0xA1,
72
+} xlnx_cfi_packet_type;
73
+
74
+typedef enum {
75
+ CFRAME_FAR = 1,
76
+ CFRAME_SFR = 2,
77
+ CFRAME_FDRI = 4,
78
+ CFRAME_CMD = 6,
79
+} xlnx_cfi_reg_addr;
80
+
81
+typedef struct XlnxCfiPacket {
82
+ uint8_t reg_addr;
83
+ uint32_t data[4];
84
+} XlnxCfiPacket;
85
+
86
+typedef struct XlnxCfiIf {
87
+ Object Parent;
88
+} XlnxCfiIf;
89
+
90
+typedef struct XlnxCfiIfClass {
91
+ InterfaceClass parent;
92
+
93
+ void (*cfi_transfer_packet)(XlnxCfiIf *cfi_if, XlnxCfiPacket *pkt);
94
+} XlnxCfiIfClass;
95
+
96
+/**
97
+ * Transfer a XlnxCfiPacket.
98
+ *
99
+ * @cfi_if: the object implementing this interface
100
+ * @XlnxCfiPacket: a pointer to the XlnxCfiPacket to transfer
101
+ */
102
+void xlnx_cfi_transfer_packet(XlnxCfiIf *cfi_if, XlnxCfiPacket *pkt);
103
+
104
+#endif /* XLNX_CFI_IF_H */
105
diff --git a/hw/misc/xlnx-cfi-if.c b/hw/misc/xlnx-cfi-if.c
106
new file mode 100644
107
index XXXXXXX..XXXXXXX
108
--- /dev/null
109
+++ b/hw/misc/xlnx-cfi-if.c
110
@@ -XXX,XX +XXX,XX @@
111
+/*
112
+ * Xilinx CFI interface
113
+ *
114
+ * Copyright (C) 2023, Advanced Micro Devices, Inc.
115
+ *
116
+ * Written by Francisco Iglesias <francisco.iglesias@amd.com>
117
+ *
118
+ * SPDX-License-Identifier: GPL-2.0-or-later
119
+ */
120
+#include "qemu/osdep.h"
121
+#include "hw/misc/xlnx-cfi-if.h"
122
+
123
+void xlnx_cfi_transfer_packet(XlnxCfiIf *cfi_if, XlnxCfiPacket *pkt)
124
+{
125
+ XlnxCfiIfClass *xcic = XLNX_CFI_IF_GET_CLASS(cfi_if);
126
+
127
+ if (xcic->cfi_transfer_packet) {
128
+ xcic->cfi_transfer_packet(cfi_if, pkt);
129
+ }
130
+}
131
+
132
+static const TypeInfo xlnx_cfi_if_info = {
133
+ .name = TYPE_XLNX_CFI_IF,
134
+ .parent = TYPE_INTERFACE,
135
+ .class_size = sizeof(XlnxCfiIfClass),
136
+};
137
+
138
+static void xlnx_cfi_if_register_types(void)
139
+{
140
+ type_register_static(&xlnx_cfi_if_info);
141
+}
142
+
143
+type_init(xlnx_cfi_if_register_types)
144
+
145
diff --git a/hw/misc/meson.build b/hw/misc/meson.build
146
index XXXXXXX..XXXXXXX 100644
147
--- a/hw/misc/meson.build
148
+++ b/hw/misc/meson.build
149
@@ -XXX,XX +XXX,XX @@ specific_ss.add(when: 'CONFIG_XLNX_VERSAL', if_true: files('xlnx-versal-crl.c'))
150
system_ss.add(when: 'CONFIG_XLNX_VERSAL', if_true: files(
151
'xlnx-versal-xramc.c',
152
'xlnx-versal-pmc-iou-slcr.c',
153
+ 'xlnx-cfi-if.c',
154
))
155
system_ss.add(when: 'CONFIG_STM32F2XX_SYSCFG', if_true: files('stm32f2xx_syscfg.c'))
156
system_ss.add(when: 'CONFIG_STM32F4XX_SYSCFG', if_true: files('stm32f4xx_syscfg.c'))
157
--
158
2.34.1
diff view generated by jsdifflib
[PULL 12/26] hw/misc: Introduce a model of Xilinx Versal's CFU_APB
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Introduce a model of the software programming interface (CFU_APB) of
4
Xilinx Versal's Configuration Frame Unit.
5
6
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Message-id: 20230831165701.2016397-3-francisco.iglesias@amd.com
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
MAINTAINERS | 2 +
12
include/hw/misc/xlnx-versal-cfu.h | 231 ++++++++++++++++++
13
hw/misc/xlnx-versal-cfu.c | 380 ++++++++++++++++++++++++++++++
14
hw/misc/meson.build | 1 +
15
4 files changed, 614 insertions(+)
16
create mode 100644 include/hw/misc/xlnx-versal-cfu.h
17
create mode 100644 hw/misc/xlnx-versal-cfu.c
18
19
diff --git a/MAINTAINERS b/MAINTAINERS
20
index XXXXXXX..XXXXXXX 100644
21
--- a/MAINTAINERS
22
+++ b/MAINTAINERS
23
@@ -XXX,XX +XXX,XX @@ M: Francisco Iglesias <francisco.iglesias@amd.com>
24
S: Maintained
25
F: hw/misc/xlnx-cfi-if.c
26
F: include/hw/misc/xlnx-cfi-if.h
27
+F: hw/misc/xlnx-versal-cfu.c
28
+F: include/hw/misc/xlnx-versal-cfu.h
29
30
STM32F100
31
M: Alexandre Iooss <erdnaxe@crans.org>
32
diff --git a/include/hw/misc/xlnx-versal-cfu.h b/include/hw/misc/xlnx-versal-cfu.h
33
new file mode 100644
34
index XXXXXXX..XXXXXXX
35
--- /dev/null
36
+++ b/include/hw/misc/xlnx-versal-cfu.h
37
@@ -XXX,XX +XXX,XX @@
38
+/*
39
+ * QEMU model of the CFU Configuration Unit.
40
+ *
41
+ * Copyright (C) 2023, Advanced Micro Devices, Inc.
42
+ *
43
+ * Written by Francisco Iglesias <francisco.iglesias@amd.com>
44
+ *
45
+ * SPDX-License-Identifier: GPL-2.0-or-later
46
+ *
47
+ * References:
48
+ * [1] Versal ACAP Technical Reference Manual,
49
+ * https://www.xilinx.com/support/documentation/architecture-manuals/am011-versal-acap-trm.pdf
50
+ *
51
+ * [2] Versal ACAP Register Reference,
52
+ * https://www.xilinx.com/htmldocs/registers/am012/am012-versal-register-reference.html
53
+ */
54
+#ifndef HW_MISC_XLNX_VERSAL_CFU_APB_H
55
+#define HW_MISC_XLNX_VERSAL_CFU_APB_H
56
+
57
+#include "hw/sysbus.h"
58
+#include "hw/register.h"
59
+#include "hw/misc/xlnx-cfi-if.h"
60
+
61
+#define TYPE_XLNX_VERSAL_CFU_APB "xlnx,versal-cfu-apb"
62
+OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFUAPB, XLNX_VERSAL_CFU_APB)
63
+
64
+REG32(CFU_ISR, 0x0)
65
+ FIELD(CFU_ISR, USR_GTS_EVENT, 9, 1)
66
+ FIELD(CFU_ISR, USR_GSR_EVENT, 8, 1)
67
+ FIELD(CFU_ISR, SLVERR, 7, 1)
68
+ FIELD(CFU_ISR, DECOMP_ERROR, 6, 1)
69
+ FIELD(CFU_ISR, BAD_CFI_PACKET, 5, 1)
70
+ FIELD(CFU_ISR, AXI_ALIGN_ERROR, 4, 1)
71
+ FIELD(CFU_ISR, CFI_ROW_ERROR, 3, 1)
72
+ FIELD(CFU_ISR, CRC32_ERROR, 2, 1)
73
+ FIELD(CFU_ISR, CRC8_ERROR, 1, 1)
74
+ FIELD(CFU_ISR, SEU_ENDOFCALIB, 0, 1)
75
+REG32(CFU_IMR, 0x4)
76
+ FIELD(CFU_IMR, USR_GTS_EVENT, 9, 1)
77
+ FIELD(CFU_IMR, USR_GSR_EVENT, 8, 1)
78
+ FIELD(CFU_IMR, SLVERR, 7, 1)
79
+ FIELD(CFU_IMR, DECOMP_ERROR, 6, 1)
80
+ FIELD(CFU_IMR, BAD_CFI_PACKET, 5, 1)
81
+ FIELD(CFU_IMR, AXI_ALIGN_ERROR, 4, 1)
82
+ FIELD(CFU_IMR, CFI_ROW_ERROR, 3, 1)
83
+ FIELD(CFU_IMR, CRC32_ERROR, 2, 1)
84
+ FIELD(CFU_IMR, CRC8_ERROR, 1, 1)
85
+ FIELD(CFU_IMR, SEU_ENDOFCALIB, 0, 1)
86
+REG32(CFU_IER, 0x8)
87
+ FIELD(CFU_IER, USR_GTS_EVENT, 9, 1)
88
+ FIELD(CFU_IER, USR_GSR_EVENT, 8, 1)
89
+ FIELD(CFU_IER, SLVERR, 7, 1)
90
+ FIELD(CFU_IER, DECOMP_ERROR, 6, 1)
91
+ FIELD(CFU_IER, BAD_CFI_PACKET, 5, 1)
92
+ FIELD(CFU_IER, AXI_ALIGN_ERROR, 4, 1)
93
+ FIELD(CFU_IER, CFI_ROW_ERROR, 3, 1)
94
+ FIELD(CFU_IER, CRC32_ERROR, 2, 1)
95
+ FIELD(CFU_IER, CRC8_ERROR, 1, 1)
96
+ FIELD(CFU_IER, SEU_ENDOFCALIB, 0, 1)
97
+REG32(CFU_IDR, 0xc)
98
+ FIELD(CFU_IDR, USR_GTS_EVENT, 9, 1)
99
+ FIELD(CFU_IDR, USR_GSR_EVENT, 8, 1)
100
+ FIELD(CFU_IDR, SLVERR, 7, 1)
101
+ FIELD(CFU_IDR, DECOMP_ERROR, 6, 1)
102
+ FIELD(CFU_IDR, BAD_CFI_PACKET, 5, 1)
103
+ FIELD(CFU_IDR, AXI_ALIGN_ERROR, 4, 1)
104
+ FIELD(CFU_IDR, CFI_ROW_ERROR, 3, 1)
105
+ FIELD(CFU_IDR, CRC32_ERROR, 2, 1)
106
+ FIELD(CFU_IDR, CRC8_ERROR, 1, 1)
107
+ FIELD(CFU_IDR, SEU_ENDOFCALIB, 0, 1)
108
+REG32(CFU_ITR, 0x10)
109
+ FIELD(CFU_ITR, USR_GTS_EVENT, 9, 1)
110
+ FIELD(CFU_ITR, USR_GSR_EVENT, 8, 1)
111
+ FIELD(CFU_ITR, SLVERR, 7, 1)
112
+ FIELD(CFU_ITR, DECOMP_ERROR, 6, 1)
113
+ FIELD(CFU_ITR, BAD_CFI_PACKET, 5, 1)
114
+ FIELD(CFU_ITR, AXI_ALIGN_ERROR, 4, 1)
115
+ FIELD(CFU_ITR, CFI_ROW_ERROR, 3, 1)
116
+ FIELD(CFU_ITR, CRC32_ERROR, 2, 1)
117
+ FIELD(CFU_ITR, CRC8_ERROR, 1, 1)
118
+ FIELD(CFU_ITR, SEU_ENDOFCALIB, 0, 1)
119
+REG32(CFU_PROTECT, 0x14)
120
+ FIELD(CFU_PROTECT, ACTIVE, 0, 1)
121
+REG32(CFU_FGCR, 0x18)
122
+ FIELD(CFU_FGCR, GCLK_CAL, 14, 1)
123
+ FIELD(CFU_FGCR, SC_HBC_TRIGGER, 13, 1)
124
+ FIELD(CFU_FGCR, GLOW, 12, 1)
125
+ FIELD(CFU_FGCR, GPWRDWN, 11, 1)
126
+ FIELD(CFU_FGCR, GCAP, 10, 1)
127
+ FIELD(CFU_FGCR, GSCWE, 9, 1)
128
+ FIELD(CFU_FGCR, GHIGH_B, 8, 1)
129
+ FIELD(CFU_FGCR, GMC_B, 7, 1)
130
+ FIELD(CFU_FGCR, GWE, 6, 1)
131
+ FIELD(CFU_FGCR, GRESTORE, 5, 1)
132
+ FIELD(CFU_FGCR, GTS_CFG_B, 4, 1)
133
+ FIELD(CFU_FGCR, GLUTMASK, 3, 1)
134
+ FIELD(CFU_FGCR, EN_GLOBS_B, 2, 1)
135
+ FIELD(CFU_FGCR, EOS, 1, 1)
136
+ FIELD(CFU_FGCR, INIT_COMPLETE, 0, 1)
137
+REG32(CFU_CTL, 0x1c)
138
+ FIELD(CFU_CTL, GSR_GSC, 15, 1)
139
+ FIELD(CFU_CTL, SLVERR_EN, 14, 1)
140
+ FIELD(CFU_CTL, CRC32_RESET, 13, 1)
141
+ FIELD(CFU_CTL, AXI_ERROR_EN, 12, 1)
142
+ FIELD(CFU_CTL, FLUSH_AXI, 11, 1)
143
+ FIELD(CFU_CTL, SSI_PER_SLR_PR, 10, 1)
144
+ FIELD(CFU_CTL, GCAP_CLK_EN, 9, 1)
145
+ FIELD(CFU_CTL, STATUS_SYNC_DISABLE, 8, 1)
146
+ FIELD(CFU_CTL, IGNORE_CFI_ERROR, 7, 1)
147
+ FIELD(CFU_CTL, CFRAME_DISABLE, 6, 1)
148
+ FIELD(CFU_CTL, QWORD_CNT_RESET, 5, 1)
149
+ FIELD(CFU_CTL, CRC8_DISABLE, 4, 1)
150
+ FIELD(CFU_CTL, CRC32_CHECK, 3, 1)
151
+ FIELD(CFU_CTL, DECOMPRESS, 2, 1)
152
+ FIELD(CFU_CTL, SEU_GO, 1, 1)
153
+ FIELD(CFU_CTL, CFI_LOCAL_RESET, 0, 1)
154
+REG32(CFU_CRAM_RW, 0x20)
155
+ FIELD(CFU_CRAM_RW, RFIFO_AFULL_DEPTH, 18, 9)
156
+ FIELD(CFU_CRAM_RW, RD_WAVE_CNT_LEFT, 12, 6)
157
+ FIELD(CFU_CRAM_RW, RD_WAVE_CNT, 6, 6)
158
+ FIELD(CFU_CRAM_RW, WR_WAVE_CNT, 0, 6)
159
+REG32(CFU_MASK, 0x28)
160
+REG32(CFU_CRC_EXPECT, 0x2c)
161
+REG32(CFU_CFRAME_LEFT_T0, 0x60)
162
+ FIELD(CFU_CFRAME_LEFT_T0, NUM, 0, 20)
163
+REG32(CFU_CFRAME_LEFT_T1, 0x64)
164
+ FIELD(CFU_CFRAME_LEFT_T1, NUM, 0, 20)
165
+REG32(CFU_CFRAME_LEFT_T2, 0x68)
166
+ FIELD(CFU_CFRAME_LEFT_T2, NUM, 0, 20)
167
+REG32(CFU_ROW_RANGE, 0x6c)
168
+ FIELD(CFU_ROW_RANGE, HALF_FSR, 5, 1)
169
+ FIELD(CFU_ROW_RANGE, NUM, 0, 5)
170
+REG32(CFU_STATUS, 0x100)
171
+ FIELD(CFU_STATUS, SEU_WRITE_ERROR, 30, 1)
172
+ FIELD(CFU_STATUS, FRCNT_ERROR, 29, 1)
173
+ FIELD(CFU_STATUS, RSVD_ERROR, 28, 1)
174
+ FIELD(CFU_STATUS, FDRO_ERROR, 27, 1)
175
+ FIELD(CFU_STATUS, FDRI_ERROR, 26, 1)
176
+ FIELD(CFU_STATUS, FDRI_READ_ERROR, 25, 1)
177
+ FIELD(CFU_STATUS, READ_FDRI_ERROR, 24, 1)
178
+ FIELD(CFU_STATUS, READ_SFR_ERROR, 23, 1)
179
+ FIELD(CFU_STATUS, READ_STREAM_ERROR, 22, 1)
180
+ FIELD(CFU_STATUS, UNKNOWN_STREAM_PKT, 21, 1)
181
+ FIELD(CFU_STATUS, USR_GTS, 20, 1)
182
+ FIELD(CFU_STATUS, USR_GSR, 19, 1)
183
+ FIELD(CFU_STATUS, AXI_BAD_WSTRB, 18, 1)
184
+ FIELD(CFU_STATUS, AXI_BAD_AR_SIZE, 17, 1)
185
+ FIELD(CFU_STATUS, AXI_BAD_AW_SIZE, 16, 1)
186
+ FIELD(CFU_STATUS, AXI_BAD_ARADDR, 15, 1)
187
+ FIELD(CFU_STATUS, AXI_BAD_AWADDR, 14, 1)
188
+ FIELD(CFU_STATUS, SCAN_CLEAR_PASS, 13, 1)
189
+ FIELD(CFU_STATUS, HC_SEC_ERROR, 12, 1)
190
+ FIELD(CFU_STATUS, GHIGH_B_ISHIGH, 11, 1)
191
+ FIELD(CFU_STATUS, GHIGH_B_ISLOW, 10, 1)
192
+ FIELD(CFU_STATUS, GMC_B_ISHIGH, 9, 1)
193
+ FIELD(CFU_STATUS, GMC_B_ISLOW, 8, 1)
194
+ FIELD(CFU_STATUS, GPWRDWN_B_ISHIGH, 7, 1)
195
+ FIELD(CFU_STATUS, CFI_SEU_CRC_ERROR, 6, 1)
196
+ FIELD(CFU_STATUS, CFI_SEU_ECC_ERROR, 5, 1)
197
+ FIELD(CFU_STATUS, CFI_SEU_HEARTBEAT, 4, 1)
198
+ FIELD(CFU_STATUS, SCAN_CLEAR_DONE, 3, 1)
199
+ FIELD(CFU_STATUS, HC_COMPLETE, 2, 1)
200
+ FIELD(CFU_STATUS, CFI_CFRAME_BUSY, 1, 1)
201
+ FIELD(CFU_STATUS, CFU_STREAM_BUSY, 0, 1)
202
+REG32(CFU_INTERNAL_STATUS, 0x104)
203
+ FIELD(CFU_INTERNAL_STATUS, SSI_EOS, 22, 1)
204
+ FIELD(CFU_INTERNAL_STATUS, SSI_GWE, 21, 1)
205
+ FIELD(CFU_INTERNAL_STATUS, RFIFO_EMPTY, 20, 1)
206
+ FIELD(CFU_INTERNAL_STATUS, RFIFO_FULL, 19, 1)
207
+ FIELD(CFU_INTERNAL_STATUS, SEL_SFR, 18, 1)
208
+ FIELD(CFU_INTERNAL_STATUS, STREAM_CFRAME, 17, 1)
209
+ FIELD(CFU_INTERNAL_STATUS, FDRI_PHASE, 16, 1)
210
+ FIELD(CFU_INTERNAL_STATUS, CFI_PIPE_EN, 15, 1)
211
+ FIELD(CFU_INTERNAL_STATUS, AWFIFO_DCNT, 10, 5)
212
+ FIELD(CFU_INTERNAL_STATUS, WFIFO_DCNT, 5, 5)
213
+ FIELD(CFU_INTERNAL_STATUS, REPAIR_BUSY, 4, 1)
214
+ FIELD(CFU_INTERNAL_STATUS, TRIMU_BUSY, 3, 1)
215
+ FIELD(CFU_INTERNAL_STATUS, TRIMB_BUSY, 2, 1)
216
+ FIELD(CFU_INTERNAL_STATUS, HCLEANR_BUSY, 1, 1)
217
+ FIELD(CFU_INTERNAL_STATUS, HCLEAN_BUSY, 0, 1)
218
+REG32(CFU_QWORD_CNT, 0x108)
219
+REG32(CFU_CRC_LIVE, 0x10c)
220
+REG32(CFU_PENDING_READ_CNT, 0x110)
221
+ FIELD(CFU_PENDING_READ_CNT, NUM, 0, 25)
222
+REG32(CFU_FDRI_CNT, 0x114)
223
+REG32(CFU_ECO1, 0x118)
224
+REG32(CFU_ECO2, 0x11c)
225
+
226
+#define R_MAX (R_CFU_ECO2 + 1)
227
+
228
+#define NUM_STREAM 2
229
+#define WFIFO_SZ 4
230
+
231
+struct XlnxVersalCFUAPB {
232
+ SysBusDevice parent_obj;
233
+ MemoryRegion iomem;
234
+ MemoryRegion iomem_stream[NUM_STREAM];
235
+ qemu_irq irq_cfu_imr;
236
+
237
+ /* 128-bit wfifo. */
238
+ uint32_t wfifo[WFIFO_SZ];
239
+
240
+ uint32_t regs[R_MAX];
241
+ RegisterInfo regs_info[R_MAX];
242
+
243
+ uint8_t fdri_row_addr;
244
+
245
+ struct {
246
+ XlnxCfiIf *cframe[15];
247
+ } cfg;
248
+};
249
+
250
+/**
251
+ * This is a helper function for updating a CFI data write fifo, an array of 4
252
+ * uint32_t and 128 bits of data that are allowed to be written through 4
253
+ * sequential 32 bit accesses. After the last index has been written into the
254
+ * write fifo (wfifo), the data is copied to and returned in a secondary fifo
255
+ * provided to the function (wfifo_ret), and the write fifo is cleared
256
+ * (zeroized).
257
+ *
258
+ * @addr: the address used when calculating the wfifo array index to update
259
+ * @value: the value to write into the wfifo array
260
+ * @wfifo: the wfifo to update
261
+ * @wfifo_out: will return the wfifo data when all 128 bits have been written
262
+ *
263
+ * @return: true if all 128 bits have been updated.
264
+ */
265
+bool update_wfifo(hwaddr addr, uint64_t value,
266
+ uint32_t *wfifo, uint32_t *wfifo_ret);
267
+
268
+#endif
269
diff --git a/hw/misc/xlnx-versal-cfu.c b/hw/misc/xlnx-versal-cfu.c
270
new file mode 100644
271
index XXXXXXX..XXXXXXX
272
--- /dev/null
273
+++ b/hw/misc/xlnx-versal-cfu.c
274
@@ -XXX,XX +XXX,XX @@
275
+/*
276
+ * QEMU model of the CFU Configuration Unit.
277
+ *
278
+ * Copyright (C) 2023, Advanced Micro Devices, Inc.
279
+ *
280
+ * Written by Edgar E. Iglesias <edgar.iglesias@gmail.com>,
281
+ * Sai Pavan Boddu <sai.pavan.boddu@amd.com>,
282
+ * Francisco Iglesias <francisco.iglesias@amd.com>
283
+ *
284
+ * SPDX-License-Identifier: GPL-2.0-or-later
285
+ */
286
+
287
+#include "qemu/osdep.h"
288
+#include "hw/sysbus.h"
289
+#include "hw/register.h"
290
+#include "hw/irq.h"
291
+#include "qemu/bitops.h"
292
+#include "qemu/log.h"
293
+#include "qemu/units.h"
294
+#include "migration/vmstate.h"
295
+#include "hw/qdev-properties.h"
296
+#include "hw/qdev-properties-system.h"
297
+#include "hw/misc/xlnx-versal-cfu.h"
298
+
299
+#ifndef XLNX_VERSAL_CFU_APB_ERR_DEBUG
300
+#define XLNX_VERSAL_CFU_APB_ERR_DEBUG 0
301
+#endif
302
+
303
+#define KEYHOLE_STREAM_4K (4 * KiB)
304
+#define KEYHOLE_STREAM_256K (256 * KiB)
305
+#define CFRAME_BROADCAST_ROW 0x1F
306
+
307
+bool update_wfifo(hwaddr addr, uint64_t value,
308
+ uint32_t *wfifo, uint32_t *wfifo_ret)
309
+{
310
+ unsigned int idx = extract32(addr, 2, 2);
311
+
312
+ wfifo[idx] = value;
313
+
314
+ if (idx == 3) {
315
+ memcpy(wfifo_ret, wfifo, WFIFO_SZ * sizeof(uint32_t));
316
+ memset(wfifo, 0, WFIFO_SZ * sizeof(uint32_t));
317
+ return true;
318
+ }
319
+
320
+ return false;
321
+}
322
+
323
+static void cfu_imr_update_irq(XlnxVersalCFUAPB *s)
324
+{
325
+ bool pending = s->regs[R_CFU_ISR] & ~s->regs[R_CFU_IMR];
326
+ qemu_set_irq(s->irq_cfu_imr, pending);
327
+}
328
+
329
+static void cfu_isr_postw(RegisterInfo *reg, uint64_t val64)
330
+{
331
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(reg->opaque);
332
+ cfu_imr_update_irq(s);
333
+}
334
+
335
+static uint64_t cfu_ier_prew(RegisterInfo *reg, uint64_t val64)
336
+{
337
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(reg->opaque);
338
+ uint32_t val = val64;
339
+
340
+ s->regs[R_CFU_IMR] &= ~val;
341
+ cfu_imr_update_irq(s);
342
+ return 0;
343
+}
344
+
345
+static uint64_t cfu_idr_prew(RegisterInfo *reg, uint64_t val64)
346
+{
347
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(reg->opaque);
348
+ uint32_t val = val64;
349
+
350
+ s->regs[R_CFU_IMR] |= val;
351
+ cfu_imr_update_irq(s);
352
+ return 0;
353
+}
354
+
355
+static uint64_t cfu_itr_prew(RegisterInfo *reg, uint64_t val64)
356
+{
357
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(reg->opaque);
358
+ uint32_t val = val64;
359
+
360
+ s->regs[R_CFU_ISR] |= val;
361
+ cfu_imr_update_irq(s);
362
+ return 0;
363
+}
364
+
365
+static void cfu_fgcr_postw(RegisterInfo *reg, uint64_t val64)
366
+{
367
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(reg->opaque);
368
+ uint32_t val = (uint32_t)val64;
369
+
370
+ /* Do a scan. It always looks good. */
371
+ if (FIELD_EX32(val, CFU_FGCR, SC_HBC_TRIGGER)) {
372
+ ARRAY_FIELD_DP32(s->regs, CFU_STATUS, SCAN_CLEAR_PASS, 1);
373
+ ARRAY_FIELD_DP32(s->regs, CFU_STATUS, SCAN_CLEAR_DONE, 1);
374
+ }
375
+}
376
+
377
+static const RegisterAccessInfo cfu_apb_regs_info[] = {
378
+ { .name = "CFU_ISR", .addr = A_CFU_ISR,
379
+ .rsvd = 0xfffffc00,
380
+ .w1c = 0x3ff,
381
+ .post_write = cfu_isr_postw,
382
+ },{ .name = "CFU_IMR", .addr = A_CFU_IMR,
383
+ .reset = 0x3ff,
384
+ .rsvd = 0xfffffc00,
385
+ .ro = 0x3ff,
386
+ },{ .name = "CFU_IER", .addr = A_CFU_IER,
387
+ .rsvd = 0xfffffc00,
388
+ .pre_write = cfu_ier_prew,
389
+ },{ .name = "CFU_IDR", .addr = A_CFU_IDR,
390
+ .rsvd = 0xfffffc00,
391
+ .pre_write = cfu_idr_prew,
392
+ },{ .name = "CFU_ITR", .addr = A_CFU_ITR,
393
+ .rsvd = 0xfffffc00,
394
+ .pre_write = cfu_itr_prew,
395
+ },{ .name = "CFU_PROTECT", .addr = A_CFU_PROTECT,
396
+ .reset = 0x1,
397
+ },{ .name = "CFU_FGCR", .addr = A_CFU_FGCR,
398
+ .rsvd = 0xffff8000,
399
+ .post_write = cfu_fgcr_postw,
400
+ },{ .name = "CFU_CTL", .addr = A_CFU_CTL,
401
+ .rsvd = 0xffff0000,
402
+ },{ .name = "CFU_CRAM_RW", .addr = A_CFU_CRAM_RW,
403
+ .reset = 0x401f7d9,
404
+ .rsvd = 0xf8000000,
405
+ },{ .name = "CFU_MASK", .addr = A_CFU_MASK,
406
+ },{ .name = "CFU_CRC_EXPECT", .addr = A_CFU_CRC_EXPECT,
407
+ },{ .name = "CFU_CFRAME_LEFT_T0", .addr = A_CFU_CFRAME_LEFT_T0,
408
+ .rsvd = 0xfff00000,
409
+ },{ .name = "CFU_CFRAME_LEFT_T1", .addr = A_CFU_CFRAME_LEFT_T1,
410
+ .rsvd = 0xfff00000,
411
+ },{ .name = "CFU_CFRAME_LEFT_T2", .addr = A_CFU_CFRAME_LEFT_T2,
412
+ .rsvd = 0xfff00000,
413
+ },{ .name = "CFU_ROW_RANGE", .addr = A_CFU_ROW_RANGE,
414
+ .rsvd = 0xffffffc0,
415
+ .ro = 0x3f,
416
+ },{ .name = "CFU_STATUS", .addr = A_CFU_STATUS,
417
+ .rsvd = 0x80000000,
418
+ .ro = 0x7fffffff,
419
+ },{ .name = "CFU_INTERNAL_STATUS", .addr = A_CFU_INTERNAL_STATUS,
420
+ .rsvd = 0xff800000,
421
+ .ro = 0x7fffff,
422
+ },{ .name = "CFU_QWORD_CNT", .addr = A_CFU_QWORD_CNT,
423
+ .ro = 0xffffffff,
424
+ },{ .name = "CFU_CRC_LIVE", .addr = A_CFU_CRC_LIVE,
425
+ .ro = 0xffffffff,
426
+ },{ .name = "CFU_PENDING_READ_CNT", .addr = A_CFU_PENDING_READ_CNT,
427
+ .rsvd = 0xfe000000,
428
+ .ro = 0x1ffffff,
429
+ },{ .name = "CFU_FDRI_CNT", .addr = A_CFU_FDRI_CNT,
430
+ .ro = 0xffffffff,
431
+ },{ .name = "CFU_ECO1", .addr = A_CFU_ECO1,
432
+ },{ .name = "CFU_ECO2", .addr = A_CFU_ECO2,
433
+ }
434
+};
435
+
436
+static void cfu_apb_reset(DeviceState *dev)
437
+{
438
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(dev);
439
+ unsigned int i;
440
+
441
+ for (i = 0; i < ARRAY_SIZE(s->regs_info); ++i) {
442
+ register_reset(&s->regs_info[i]);
443
+ }
444
+ memset(s->wfifo, 0, WFIFO_SZ * sizeof(uint32_t));
445
+
446
+ s->regs[R_CFU_STATUS] |= R_CFU_STATUS_HC_COMPLETE_MASK;
447
+ cfu_imr_update_irq(s);
448
+}
449
+
450
+static const MemoryRegionOps cfu_apb_ops = {
451
+ .read = register_read_memory,
452
+ .write = register_write_memory,
453
+ .endianness = DEVICE_LITTLE_ENDIAN,
454
+ .valid = {
455
+ .min_access_size = 4,
456
+ .max_access_size = 4,
457
+ },
458
+};
459
+
460
+static void cfu_transfer_cfi_packet(XlnxVersalCFUAPB *s, uint8_t row_addr,
461
+ XlnxCfiPacket *pkt)
462
+{
463
+ if (row_addr == CFRAME_BROADCAST_ROW) {
464
+ for (int i = 0; i < ARRAY_SIZE(s->cfg.cframe); i++) {
465
+ if (s->cfg.cframe[i]) {
466
+ xlnx_cfi_transfer_packet(s->cfg.cframe[i], pkt);
467
+ }
468
+ }
469
+ } else {
470
+ assert(row_addr < ARRAY_SIZE(s->cfg.cframe));
471
+
472
+ if (s->cfg.cframe[row_addr]) {
473
+ xlnx_cfi_transfer_packet(s->cfg.cframe[row_addr], pkt);
474
+ }
475
+ }
476
+}
477
+
478
+static uint64_t cfu_stream_read(void *opaque, hwaddr addr, unsigned size)
479
+{
480
+ qemu_log_mask(LOG_GUEST_ERROR, "%s: Unsupported read from addr=%"
481
+ HWADDR_PRIx "\n", __func__, addr);
482
+ return 0;
483
+}
484
+
485
+static void cfu_stream_write(void *opaque, hwaddr addr, uint64_t value,
486
+ unsigned size)
487
+{
488
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(opaque);
489
+ uint32_t wfifo[WFIFO_SZ];
490
+
491
+ if (update_wfifo(addr, value, s->wfifo, wfifo)) {
492
+ uint8_t packet_type, row_addr, reg_addr;
493
+
494
+ packet_type = extract32(wfifo[0], 24, 8);
495
+ row_addr = extract32(wfifo[0], 16, 5);
496
+ reg_addr = extract32(wfifo[0], 8, 6);
497
+
498
+ /* Compressed bitstreams are not supported yet. */
499
+ if (ARRAY_FIELD_EX32(s->regs, CFU_CTL, DECOMPRESS) == 0) {
500
+ if (s->regs[R_CFU_FDRI_CNT]) {
501
+ XlnxCfiPacket pkt = {
502
+ .reg_addr = CFRAME_FDRI,
503
+ .data[0] = wfifo[0],
504
+ .data[1] = wfifo[1],
505
+ .data[2] = wfifo[2],
506
+ .data[3] = wfifo[3]
507
+ };
508
+
509
+ cfu_transfer_cfi_packet(s, s->fdri_row_addr, &pkt);
510
+
511
+ s->regs[R_CFU_FDRI_CNT]--;
512
+
513
+ } else if (packet_type == PACKET_TYPE_CFU &&
514
+ reg_addr == CFRAME_FDRI) {
515
+
516
+ /* Load R_CFU_FDRI_CNT, must be multiple of 25 */
517
+ s->regs[R_CFU_FDRI_CNT] = wfifo[1];
518
+
519
+ /* Store target row_addr */
520
+ s->fdri_row_addr = row_addr;
521
+
522
+ if (wfifo[1] % 25 != 0) {
523
+ qemu_log_mask(LOG_GUEST_ERROR,
524
+ "CFU FDRI_CNT is not loaded with "
525
+ "a multiple of 25 value\n");
526
+ }
527
+
528
+ } else if (packet_type == PACKET_TYPE_CFRAME) {
529
+ XlnxCfiPacket pkt = {
530
+ .reg_addr = reg_addr,
531
+ .data[0] = wfifo[1],
532
+ .data[1] = wfifo[2],
533
+ .data[2] = wfifo[3],
534
+ };
535
+ cfu_transfer_cfi_packet(s, row_addr, &pkt);
536
+ }
537
+ }
538
+ }
539
+}
540
+
541
+static const MemoryRegionOps cfu_stream_ops = {
542
+ .read = cfu_stream_read,
543
+ .write = cfu_stream_write,
544
+ .endianness = DEVICE_LITTLE_ENDIAN,
545
+ .valid = {
546
+ .min_access_size = 4,
547
+ .max_access_size = 8,
548
+ },
549
+};
550
+
551
+static void cfu_apb_init(Object *obj)
552
+{
553
+ XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(obj);
554
+ SysBusDevice *sbd = SYS_BUS_DEVICE(obj);
555
+ RegisterInfoArray *reg_array;
556
+ unsigned int i;
557
+ char *name;
558
+
559
+ memory_region_init(&s->iomem, obj, TYPE_XLNX_VERSAL_CFU_APB, R_MAX * 4);
560
+ reg_array =
561
+ register_init_block32(DEVICE(obj), cfu_apb_regs_info,
562
+ ARRAY_SIZE(cfu_apb_regs_info),
563
+ s->regs_info, s->regs,
564
+ &cfu_apb_ops,
565
+ XLNX_VERSAL_CFU_APB_ERR_DEBUG,
566
+ R_MAX * 4);
567
+ memory_region_add_subregion(&s->iomem,
568
+ 0x0,
569
+ &reg_array->mem);
570
+ sysbus_init_mmio(sbd, &s->iomem);
571
+ for (i = 0; i < NUM_STREAM; i++) {
572
+ name = g_strdup_printf(TYPE_XLNX_VERSAL_CFU_APB "-stream%d", i);
573
+ memory_region_init_io(&s->iomem_stream[i], obj, &cfu_stream_ops, s,
574
+ name, i == 0 ? KEYHOLE_STREAM_4K :
575
+ KEYHOLE_STREAM_256K);
576
+ sysbus_init_mmio(sbd, &s->iomem_stream[i]);
577
+ g_free(name);
578
+ }
579
+ sysbus_init_irq(sbd, &s->irq_cfu_imr);
580
+}
581
+
582
+static Property cfu_props[] = {
583
+ DEFINE_PROP_LINK("cframe0", XlnxVersalCFUAPB, cfg.cframe[0],
584
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
585
+ DEFINE_PROP_LINK("cframe1", XlnxVersalCFUAPB, cfg.cframe[1],
586
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
587
+ DEFINE_PROP_LINK("cframe2", XlnxVersalCFUAPB, cfg.cframe[2],
588
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
589
+ DEFINE_PROP_LINK("cframe3", XlnxVersalCFUAPB, cfg.cframe[3],
590
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
591
+ DEFINE_PROP_LINK("cframe4", XlnxVersalCFUAPB, cfg.cframe[4],
592
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
593
+ DEFINE_PROP_LINK("cframe5", XlnxVersalCFUAPB, cfg.cframe[5],
594
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
595
+ DEFINE_PROP_LINK("cframe6", XlnxVersalCFUAPB, cfg.cframe[6],
596
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
597
+ DEFINE_PROP_LINK("cframe7", XlnxVersalCFUAPB, cfg.cframe[7],
598
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
599
+ DEFINE_PROP_LINK("cframe8", XlnxVersalCFUAPB, cfg.cframe[8],
600
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
601
+ DEFINE_PROP_LINK("cframe9", XlnxVersalCFUAPB, cfg.cframe[9],
602
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
603
+ DEFINE_PROP_LINK("cframe10", XlnxVersalCFUAPB, cfg.cframe[10],
604
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
605
+ DEFINE_PROP_LINK("cframe11", XlnxVersalCFUAPB, cfg.cframe[11],
606
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
607
+ DEFINE_PROP_LINK("cframe12", XlnxVersalCFUAPB, cfg.cframe[12],
608
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
609
+ DEFINE_PROP_LINK("cframe13", XlnxVersalCFUAPB, cfg.cframe[13],
610
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
611
+ DEFINE_PROP_LINK("cframe14", XlnxVersalCFUAPB, cfg.cframe[14],
612
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
613
+ DEFINE_PROP_END_OF_LIST(),
614
+};
615
+
616
+static const VMStateDescription vmstate_cfu_apb = {
617
+ .name = TYPE_XLNX_VERSAL_CFU_APB,
618
+ .version_id = 1,
619
+ .minimum_version_id = 1,
620
+ .fields = (VMStateField[]) {
621
+ VMSTATE_UINT32_ARRAY(wfifo, XlnxVersalCFUAPB, 4),
622
+ VMSTATE_UINT32_ARRAY(regs, XlnxVersalCFUAPB, R_MAX),
623
+ VMSTATE_UINT8(fdri_row_addr, XlnxVersalCFUAPB),
624
+ VMSTATE_END_OF_LIST(),
625
+ }
626
+};
627
+
628
+static void cfu_apb_class_init(ObjectClass *klass, void *data)
629
+{
630
+ DeviceClass *dc = DEVICE_CLASS(klass);
631
+
632
+ dc->reset = cfu_apb_reset;
633
+ dc->vmsd = &vmstate_cfu_apb;
634
+ device_class_set_props(dc, cfu_props);
635
+}
636
+
637
+static const TypeInfo cfu_apb_info = {
638
+ .name = TYPE_XLNX_VERSAL_CFU_APB,
639
+ .parent = TYPE_SYS_BUS_DEVICE,
640
+ .instance_size = sizeof(XlnxVersalCFUAPB),
641
+ .class_init = cfu_apb_class_init,
642
+ .instance_init = cfu_apb_init,
643
+ .interfaces = (InterfaceInfo[]) {
644
+ { TYPE_XLNX_CFI_IF },
645
+ { }
646
+ }
647
+};
648
+
649
+static void cfu_apb_register_types(void)
650
+{
651
+ type_register_static(&cfu_apb_info);
652
+}
653
+
654
+type_init(cfu_apb_register_types)
655
diff --git a/hw/misc/meson.build b/hw/misc/meson.build
656
index XXXXXXX..XXXXXXX 100644
657
--- a/hw/misc/meson.build
658
+++ b/hw/misc/meson.build
659
@@ -XXX,XX +XXX,XX @@ specific_ss.add(when: 'CONFIG_XLNX_VERSAL', if_true: files('xlnx-versal-crl.c'))
660
system_ss.add(when: 'CONFIG_XLNX_VERSAL', if_true: files(
661
'xlnx-versal-xramc.c',
662
'xlnx-versal-pmc-iou-slcr.c',
663
+ 'xlnx-versal-cfu.c',
664
'xlnx-cfi-if.c',
665
))
666
system_ss.add(when: 'CONFIG_STM32F2XX_SYSCFG', if_true: files('stm32f2xx_syscfg.c'))
667
--
668
2.34.1
diff view generated by jsdifflib
[PULL 13/26] hw/misc/xlnx-versal-cfu: Introduce a model of Xilinx Versal CFU_FDRO
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Introduce a model of Xilinx Versal's Configuration Frame Unit's data out
4
port (CFU_FDRO).
5
6
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Message-id: 20230831165701.2016397-4-francisco.iglesias@amd.com
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
include/hw/misc/xlnx-versal-cfu.h | 12 ++++
12
hw/misc/xlnx-versal-cfu.c | 96 +++++++++++++++++++++++++++++++
13
2 files changed, 108 insertions(+)
14
15
diff --git a/include/hw/misc/xlnx-versal-cfu.h b/include/hw/misc/xlnx-versal-cfu.h
16
index XXXXXXX..XXXXXXX 100644
17
--- a/include/hw/misc/xlnx-versal-cfu.h
18
+++ b/include/hw/misc/xlnx-versal-cfu.h
19
@@ -XXX,XX +XXX,XX @@
20
#include "hw/sysbus.h"
21
#include "hw/register.h"
22
#include "hw/misc/xlnx-cfi-if.h"
23
+#include "qemu/fifo32.h"
24
25
#define TYPE_XLNX_VERSAL_CFU_APB "xlnx,versal-cfu-apb"
26
OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFUAPB, XLNX_VERSAL_CFU_APB)
27
28
+#define TYPE_XLNX_VERSAL_CFU_FDRO "xlnx,versal-cfu-fdro"
29
+OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFUFDRO, XLNX_VERSAL_CFU_FDRO)
30
+
31
REG32(CFU_ISR, 0x0)
32
FIELD(CFU_ISR, USR_GTS_EVENT, 9, 1)
33
FIELD(CFU_ISR, USR_GSR_EVENT, 8, 1)
34
@@ -XXX,XX +XXX,XX @@ struct XlnxVersalCFUAPB {
35
} cfg;
36
};
37
38
+
39
+struct XlnxVersalCFUFDRO {
40
+ SysBusDevice parent_obj;
41
+ MemoryRegion iomem_fdro;
42
+
43
+ Fifo32 fdro_data;
44
+};
45
+
46
/**
47
* This is a helper function for updating a CFI data write fifo, an array of 4
48
* uint32_t and 128 bits of data that are allowed to be written through 4
49
diff --git a/hw/misc/xlnx-versal-cfu.c b/hw/misc/xlnx-versal-cfu.c
50
index XXXXXXX..XXXXXXX 100644
51
--- a/hw/misc/xlnx-versal-cfu.c
52
+++ b/hw/misc/xlnx-versal-cfu.c
53
@@ -XXX,XX +XXX,XX @@ static void cfu_stream_write(void *opaque, hwaddr addr, uint64_t value,
54
}
55
}
56
57
+static uint64_t cfu_fdro_read(void *opaque, hwaddr addr, unsigned size)
58
+{
59
+ XlnxVersalCFUFDRO *s = XLNX_VERSAL_CFU_FDRO(opaque);
60
+ uint64_t ret = 0;
61
+
62
+ if (!fifo32_is_empty(&s->fdro_data)) {
63
+ ret = fifo32_pop(&s->fdro_data);
64
+ }
65
+
66
+ return ret;
67
+}
68
+
69
+static void cfu_fdro_write(void *opaque, hwaddr addr, uint64_t value,
70
+ unsigned size)
71
+{
72
+ qemu_log_mask(LOG_GUEST_ERROR, "%s: Unsupported write from addr=%"
73
+ HWADDR_PRIx "\n", __func__, addr);
74
+}
75
+
76
static const MemoryRegionOps cfu_stream_ops = {
77
.read = cfu_stream_read,
78
.write = cfu_stream_write,
79
@@ -XXX,XX +XXX,XX @@ static const MemoryRegionOps cfu_stream_ops = {
80
},
81
};
82
83
+static const MemoryRegionOps cfu_fdro_ops = {
84
+ .read = cfu_fdro_read,
85
+ .write = cfu_fdro_write,
86
+ .endianness = DEVICE_LITTLE_ENDIAN,
87
+ .valid = {
88
+ .min_access_size = 4,
89
+ .max_access_size = 4,
90
+ },
91
+};
92
+
93
static void cfu_apb_init(Object *obj)
94
{
95
XlnxVersalCFUAPB *s = XLNX_VERSAL_CFU_APB(obj);
96
@@ -XXX,XX +XXX,XX @@ static void cfu_apb_init(Object *obj)
97
sysbus_init_irq(sbd, &s->irq_cfu_imr);
98
}
99
100
+static void cfu_fdro_init(Object *obj)
101
+{
102
+ XlnxVersalCFUFDRO *s = XLNX_VERSAL_CFU_FDRO(obj);
103
+ SysBusDevice *sbd = SYS_BUS_DEVICE(obj);
104
+
105
+ memory_region_init_io(&s->iomem_fdro, obj, &cfu_fdro_ops, s,
106
+ TYPE_XLNX_VERSAL_CFU_FDRO, KEYHOLE_STREAM_4K);
107
+ sysbus_init_mmio(sbd, &s->iomem_fdro);
108
+ fifo32_create(&s->fdro_data, 8 * KiB / sizeof(uint32_t));
109
+}
110
+
111
+static void cfu_fdro_reset_enter(Object *obj, ResetType type)
112
+{
113
+ XlnxVersalCFUFDRO *s = XLNX_VERSAL_CFU_FDRO(obj);
114
+
115
+ fifo32_reset(&s->fdro_data);
116
+}
117
+
118
+static void cfu_fdro_cfi_transfer_packet(XlnxCfiIf *cfi_if, XlnxCfiPacket *pkt)
119
+{
120
+ XlnxVersalCFUFDRO *s = XLNX_VERSAL_CFU_FDRO(cfi_if);
121
+
122
+ if (fifo32_num_free(&s->fdro_data) >= ARRAY_SIZE(pkt->data)) {
123
+ for (int i = 0; i < ARRAY_SIZE(pkt->data); i++) {
124
+ fifo32_push(&s->fdro_data, pkt->data[i]);
125
+ }
126
+ } else {
127
+ /* It is a programming error to fill the fifo. */
128
+ qemu_log_mask(LOG_GUEST_ERROR,
129
+ "CFU_FDRO: CFI data dropped due to full read fifo\n");
130
+ }
131
+}
132
+
133
static Property cfu_props[] = {
134
DEFINE_PROP_LINK("cframe0", XlnxVersalCFUAPB, cfg.cframe[0],
135
TYPE_XLNX_CFI_IF, XlnxCfiIf *),
136
@@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_cfu_apb = {
137
}
138
};
139
140
+static const VMStateDescription vmstate_cfu_fdro = {
141
+ .name = TYPE_XLNX_VERSAL_CFU_FDRO,
142
+ .version_id = 1,
143
+ .minimum_version_id = 1,
144
+ .fields = (VMStateField[]) {
145
+ VMSTATE_FIFO32(fdro_data, XlnxVersalCFUFDRO),
146
+ VMSTATE_END_OF_LIST(),
147
+ }
148
+};
149
+
150
static void cfu_apb_class_init(ObjectClass *klass, void *data)
151
{
152
DeviceClass *dc = DEVICE_CLASS(klass);
153
@@ -XXX,XX +XXX,XX @@ static void cfu_apb_class_init(ObjectClass *klass, void *data)
154
device_class_set_props(dc, cfu_props);
155
}
156
157
+static void cfu_fdro_class_init(ObjectClass *klass, void *data)
158
+{
159
+ DeviceClass *dc = DEVICE_CLASS(klass);
160
+ ResettableClass *rc = RESETTABLE_CLASS(klass);
161
+ XlnxCfiIfClass *xcic = XLNX_CFI_IF_CLASS(klass);
162
+
163
+ dc->vmsd = &vmstate_cfu_fdro;
164
+ xcic->cfi_transfer_packet = cfu_fdro_cfi_transfer_packet;
165
+ rc->phases.enter = cfu_fdro_reset_enter;
166
+}
167
+
168
static const TypeInfo cfu_apb_info = {
169
.name = TYPE_XLNX_VERSAL_CFU_APB,
170
.parent = TYPE_SYS_BUS_DEVICE,
171
@@ -XXX,XX +XXX,XX @@ static const TypeInfo cfu_apb_info = {
172
}
173
};
174
175
+static const TypeInfo cfu_fdro_info = {
176
+ .name = TYPE_XLNX_VERSAL_CFU_FDRO,
177
+ .parent = TYPE_SYS_BUS_DEVICE,
178
+ .instance_size = sizeof(XlnxVersalCFUFDRO),
179
+ .class_init = cfu_fdro_class_init,
180
+ .instance_init = cfu_fdro_init,
181
+ .interfaces = (InterfaceInfo[]) {
182
+ { TYPE_XLNX_CFI_IF },
183
+ { }
184
+ }
185
+};
186
+
187
static void cfu_apb_register_types(void)
188
{
189
type_register_static(&cfu_apb_info);
190
+ type_register_static(&cfu_fdro_info);
191
}
192
193
type_init(cfu_apb_register_types)
194
--
195
2.34.1
diff view generated by jsdifflib
[PULL 14/26] hw/misc/xlnx-versal-cfu: Introduce a model of Xilinx Versal's CFU_SFR
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Introduce a model of Xilinx Versal's Configuration Frame Unit's Single
4
Frame Read port (CFU_SFR).
5
6
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Message-id: 20230831165701.2016397-5-francisco.iglesias@amd.com
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
include/hw/misc/xlnx-versal-cfu.h | 15 ++++++
12
hw/misc/xlnx-versal-cfu.c | 87 +++++++++++++++++++++++++++++++
13
2 files changed, 102 insertions(+)
14
15
diff --git a/include/hw/misc/xlnx-versal-cfu.h b/include/hw/misc/xlnx-versal-cfu.h
16
index XXXXXXX..XXXXXXX 100644
17
--- a/include/hw/misc/xlnx-versal-cfu.h
18
+++ b/include/hw/misc/xlnx-versal-cfu.h
19
@@ -XXX,XX +XXX,XX @@ OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFUAPB, XLNX_VERSAL_CFU_APB)
20
#define TYPE_XLNX_VERSAL_CFU_FDRO "xlnx,versal-cfu-fdro"
21
OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFUFDRO, XLNX_VERSAL_CFU_FDRO)
22
23
+#define TYPE_XLNX_VERSAL_CFU_SFR "xlnx,versal-cfu-sfr"
24
+OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFUSFR, XLNX_VERSAL_CFU_SFR)
25
+
26
REG32(CFU_ISR, 0x0)
27
FIELD(CFU_ISR, USR_GTS_EVENT, 9, 1)
28
FIELD(CFU_ISR, USR_GSR_EVENT, 8, 1)
29
@@ -XXX,XX +XXX,XX @@ struct XlnxVersalCFUFDRO {
30
Fifo32 fdro_data;
31
};
32
33
+struct XlnxVersalCFUSFR {
34
+ SysBusDevice parent_obj;
35
+ MemoryRegion iomem_sfr;
36
+
37
+ /* 128-bit wfifo. */
38
+ uint32_t wfifo[WFIFO_SZ];
39
+
40
+ struct {
41
+ XlnxVersalCFUAPB *cfu;
42
+ } cfg;
43
+};
44
+
45
/**
46
* This is a helper function for updating a CFI data write fifo, an array of 4
47
* uint32_t and 128 bits of data that are allowed to be written through 4
48
diff --git a/hw/misc/xlnx-versal-cfu.c b/hw/misc/xlnx-versal-cfu.c
49
index XXXXXXX..XXXXXXX 100644
50
--- a/hw/misc/xlnx-versal-cfu.c
51
+++ b/hw/misc/xlnx-versal-cfu.c
52
@@ -XXX,XX +XXX,XX @@ static void cfu_stream_write(void *opaque, hwaddr addr, uint64_t value,
53
}
54
}
55
56
+static uint64_t cfu_sfr_read(void *opaque, hwaddr addr, unsigned size)
57
+{
58
+ qemu_log_mask(LOG_GUEST_ERROR, "%s: Unsupported read from addr=%"
59
+ HWADDR_PRIx "\n", __func__, addr);
60
+ return 0;
61
+}
62
+
63
+static void cfu_sfr_write(void *opaque, hwaddr addr, uint64_t value,
64
+ unsigned size)
65
+{
66
+ XlnxVersalCFUSFR *s = XLNX_VERSAL_CFU_SFR(opaque);
67
+ uint32_t wfifo[WFIFO_SZ];
68
+
69
+ if (update_wfifo(addr, value, s->wfifo, wfifo)) {
70
+ uint8_t row_addr = extract32(wfifo[0], 23, 5);
71
+ uint32_t frame_addr = extract32(wfifo[0], 0, 23);
72
+ XlnxCfiPacket pkt = { .reg_addr = CFRAME_SFR,
73
+ .data[0] = frame_addr };
74
+
75
+ if (s->cfg.cfu) {
76
+ cfu_transfer_cfi_packet(s->cfg.cfu, row_addr, &pkt);
77
+ }
78
+ }
79
+}
80
+
81
static uint64_t cfu_fdro_read(void *opaque, hwaddr addr, unsigned size)
82
{
83
XlnxVersalCFUFDRO *s = XLNX_VERSAL_CFU_FDRO(opaque);
84
@@ -XXX,XX +XXX,XX @@ static const MemoryRegionOps cfu_stream_ops = {
85
},
86
};
87
88
+static const MemoryRegionOps cfu_sfr_ops = {
89
+ .read = cfu_sfr_read,
90
+ .write = cfu_sfr_write,
91
+ .endianness = DEVICE_LITTLE_ENDIAN,
92
+ .valid = {
93
+ .min_access_size = 4,
94
+ .max_access_size = 4,
95
+ },
96
+};
97
+
98
static const MemoryRegionOps cfu_fdro_ops = {
99
.read = cfu_fdro_read,
100
.write = cfu_fdro_write,
101
@@ -XXX,XX +XXX,XX @@ static void cfu_apb_init(Object *obj)
102
sysbus_init_irq(sbd, &s->irq_cfu_imr);
103
}
104
105
+static void cfu_sfr_init(Object *obj)
106
+{
107
+ XlnxVersalCFUSFR *s = XLNX_VERSAL_CFU_SFR(obj);
108
+ SysBusDevice *sbd = SYS_BUS_DEVICE(obj);
109
+
110
+ memory_region_init_io(&s->iomem_sfr, obj, &cfu_sfr_ops, s,
111
+ TYPE_XLNX_VERSAL_CFU_SFR, KEYHOLE_STREAM_4K);
112
+ sysbus_init_mmio(sbd, &s->iomem_sfr);
113
+}
114
+
115
+static void cfu_sfr_reset_enter(Object *obj, ResetType type)
116
+{
117
+ XlnxVersalCFUSFR *s = XLNX_VERSAL_CFU_SFR(obj);
118
+
119
+ memset(s->wfifo, 0, WFIFO_SZ * sizeof(uint32_t));
120
+}
121
+
122
static void cfu_fdro_init(Object *obj)
123
{
124
XlnxVersalCFUFDRO *s = XLNX_VERSAL_CFU_FDRO(obj);
125
@@ -XXX,XX +XXX,XX @@ static Property cfu_props[] = {
126
DEFINE_PROP_END_OF_LIST(),
127
};
128
129
+static Property cfu_sfr_props[] = {
130
+ DEFINE_PROP_LINK("cfu", XlnxVersalCFUSFR, cfg.cfu,
131
+ TYPE_XLNX_VERSAL_CFU_APB, XlnxVersalCFUAPB *),
132
+ DEFINE_PROP_END_OF_LIST(),
133
+};
134
+
135
static const VMStateDescription vmstate_cfu_apb = {
136
.name = TYPE_XLNX_VERSAL_CFU_APB,
137
.version_id = 1,
138
@@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_cfu_fdro = {
139
}
140
};
141
142
+static const VMStateDescription vmstate_cfu_sfr = {
143
+ .name = TYPE_XLNX_VERSAL_CFU_SFR,
144
+ .version_id = 1,
145
+ .minimum_version_id = 1,
146
+ .fields = (VMStateField[]) {
147
+ VMSTATE_UINT32_ARRAY(wfifo, XlnxVersalCFUSFR, 4),
148
+ VMSTATE_END_OF_LIST(),
149
+ }
150
+};
151
+
152
static void cfu_apb_class_init(ObjectClass *klass, void *data)
153
{
154
DeviceClass *dc = DEVICE_CLASS(klass);
155
@@ -XXX,XX +XXX,XX @@ static void cfu_fdro_class_init(ObjectClass *klass, void *data)
156
rc->phases.enter = cfu_fdro_reset_enter;
157
}
158
159
+static void cfu_sfr_class_init(ObjectClass *klass, void *data)
160
+{
161
+ DeviceClass *dc = DEVICE_CLASS(klass);
162
+ ResettableClass *rc = RESETTABLE_CLASS(klass);
163
+
164
+ device_class_set_props(dc, cfu_sfr_props);
165
+ dc->vmsd = &vmstate_cfu_sfr;
166
+ rc->phases.enter = cfu_sfr_reset_enter;
167
+}
168
+
169
static const TypeInfo cfu_apb_info = {
170
.name = TYPE_XLNX_VERSAL_CFU_APB,
171
.parent = TYPE_SYS_BUS_DEVICE,
172
@@ -XXX,XX +XXX,XX @@ static const TypeInfo cfu_fdro_info = {
173
}
174
};
175
176
+static const TypeInfo cfu_sfr_info = {
177
+ .name = TYPE_XLNX_VERSAL_CFU_SFR,
178
+ .parent = TYPE_SYS_BUS_DEVICE,
179
+ .instance_size = sizeof(XlnxVersalCFUSFR),
180
+ .class_init = cfu_sfr_class_init,
181
+ .instance_init = cfu_sfr_init,
182
+};
183
+
184
static void cfu_apb_register_types(void)
185
{
186
type_register_static(&cfu_apb_info);
187
type_register_static(&cfu_fdro_info);
188
+ type_register_static(&cfu_sfr_info);
189
}
190
191
type_init(cfu_apb_register_types)
192
--
193
2.34.1
diff view generated by jsdifflib
[PULL 15/26] hw/misc: Introduce a model of Xilinx Versal's CFRAME_REG
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Introduce a model of Xilinx Versal's Configuration Frame controller
4
(CFRAME_REG).
5
6
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
7
Message-id: 20230831165701.2016397-6-francisco.iglesias@amd.com
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
MAINTAINERS | 2 +
12
include/hw/misc/xlnx-versal-cframe-reg.h | 286 ++++++++++
13
hw/misc/xlnx-versal-cframe-reg.c | 697 +++++++++++++++++++++++
14
hw/misc/meson.build | 1 +
15
4 files changed, 986 insertions(+)
16
create mode 100644 include/hw/misc/xlnx-versal-cframe-reg.h
17
create mode 100644 hw/misc/xlnx-versal-cframe-reg.c
18
19
diff --git a/MAINTAINERS b/MAINTAINERS
20
index XXXXXXX..XXXXXXX 100644
21
--- a/MAINTAINERS
22
+++ b/MAINTAINERS
23
@@ -XXX,XX +XXX,XX @@ F: hw/misc/xlnx-cfi-if.c
24
F: include/hw/misc/xlnx-cfi-if.h
25
F: hw/misc/xlnx-versal-cfu.c
26
F: include/hw/misc/xlnx-versal-cfu.h
27
+F: hw/misc/xlnx-versal-cframe-reg.c
28
+F: include/hw/misc/xlnx-versal-cframe-reg.h
29
30
STM32F100
31
M: Alexandre Iooss <erdnaxe@crans.org>
32
diff --git a/include/hw/misc/xlnx-versal-cframe-reg.h b/include/hw/misc/xlnx-versal-cframe-reg.h
33
new file mode 100644
34
index XXXXXXX..XXXXXXX
35
--- /dev/null
36
+++ b/include/hw/misc/xlnx-versal-cframe-reg.h
37
@@ -XXX,XX +XXX,XX @@
38
+/*
39
+ * QEMU model of the Configuration Frame Control module
40
+ *
41
+ * Copyright (C) 2023, Advanced Micro Devices, Inc.
42
+ *
43
+ * Written by Francisco Iglesias <francisco.iglesias@amd.com>
44
+ *
45
+ * SPDX-License-Identifier: GPL-2.0-or-later
46
+ *
47
+ * References:
48
+ * [1] Versal ACAP Technical Reference Manual,
49
+ * https://www.xilinx.com/support/documentation/architecture-manuals/am011-versal-acap-trm.pdf
50
+ *
51
+ * [2] Versal ACAP Register Reference,
52
+ * https://www.xilinx.com/htmldocs/registers/am012/am012-versal-register-reference.html
53
+ */
54
+#ifndef HW_MISC_XLNX_VERSAL_CFRAME_REG_H
55
+#define HW_MISC_XLNX_VERSAL_CFRAME_REG_H
56
+
57
+#include "hw/sysbus.h"
58
+#include "hw/register.h"
59
+#include "hw/misc/xlnx-cfi-if.h"
60
+#include "hw/misc/xlnx-versal-cfu.h"
61
+#include "qemu/fifo32.h"
62
+
63
+#define TYPE_XLNX_VERSAL_CFRAME_REG "xlnx,cframe-reg"
64
+OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFrameReg, XLNX_VERSAL_CFRAME_REG)
65
+
66
+/*
67
+ * The registers in this module are 128 bits wide but it is ok to write
68
+ * and read them through 4 sequential 32 bit accesses (address[3:2] = 0,
69
+ * 1, 2, 3).
70
+ */
71
+REG32(CRC0, 0x0)
72
+ FIELD(CRC, CRC, 0, 32)
73
+REG32(CRC1, 0x4)
74
+REG32(CRC2, 0x8)
75
+REG32(CRC3, 0xc)
76
+REG32(FAR0, 0x10)
77
+ FIELD(FAR0, SEGMENT, 23, 2)
78
+ FIELD(FAR0, BLOCKTYPE, 20, 3)
79
+ FIELD(FAR0, FRAME_ADDR, 0, 20)
80
+REG32(FAR1, 0x14)
81
+REG32(FAR2, 0x18)
82
+REG32(FAR3, 0x1c)
83
+REG32(FAR_SFR0, 0x20)
84
+ FIELD(FAR_SFR0, BLOCKTYPE, 20, 3)
85
+ FIELD(FAR_SFR0, FRAME_ADDR, 0, 20)
86
+REG32(FAR_SFR1, 0x24)
87
+REG32(FAR_SFR2, 0x28)
88
+REG32(FAR_SFR3, 0x2c)
89
+REG32(FDRI0, 0x40)
90
+REG32(FDRI1, 0x44)
91
+REG32(FDRI2, 0x48)
92
+REG32(FDRI3, 0x4c)
93
+REG32(FRCNT0, 0x50)
94
+ FIELD(FRCNT0, FRCNT, 0, 32)
95
+REG32(FRCNT1, 0x54)
96
+REG32(FRCNT2, 0x58)
97
+REG32(FRCNT3, 0x5c)
98
+REG32(CMD0, 0x60)
99
+ FIELD(CMD0, CMD, 0, 5)
100
+REG32(CMD1, 0x64)
101
+REG32(CMD2, 0x68)
102
+REG32(CMD3, 0x6c)
103
+REG32(CR_MASK0, 0x70)
104
+REG32(CR_MASK1, 0x74)
105
+REG32(CR_MASK2, 0x78)
106
+REG32(CR_MASK3, 0x7c)
107
+REG32(CTL0, 0x80)
108
+ FIELD(CTL, PER_FRAME_CRC, 0, 1)
109
+REG32(CTL1, 0x84)
110
+REG32(CTL2, 0x88)
111
+REG32(CTL3, 0x8c)
112
+REG32(CFRM_ISR0, 0x150)
113
+ FIELD(CFRM_ISR0, READ_BROADCAST_ERROR, 21, 1)
114
+ FIELD(CFRM_ISR0, CMD_MISSING_ERROR, 20, 1)
115
+ FIELD(CFRM_ISR0, RW_ROWOFF_ERROR, 19, 1)
116
+ FIELD(CFRM_ISR0, READ_REG_ADDR_ERROR, 18, 1)
117
+ FIELD(CFRM_ISR0, READ_BLK_TYPE_ERROR, 17, 1)
118
+ FIELD(CFRM_ISR0, READ_FRAME_ADDR_ERROR, 16, 1)
119
+ FIELD(CFRM_ISR0, WRITE_REG_ADDR_ERROR, 15, 1)
120
+ FIELD(CFRM_ISR0, WRITE_BLK_TYPE_ERROR, 13, 1)
121
+ FIELD(CFRM_ISR0, WRITE_FRAME_ADDR_ERROR, 12, 1)
122
+ FIELD(CFRM_ISR0, MFW_OVERRUN_ERROR, 11, 1)
123
+ FIELD(CFRM_ISR0, FAR_FIFO_UNDERFLOW, 10, 1)
124
+ FIELD(CFRM_ISR0, FAR_FIFO_OVERFLOW, 9, 1)
125
+ FIELD(CFRM_ISR0, PER_FRAME_SEQ_ERROR, 8, 1)
126
+ FIELD(CFRM_ISR0, CRC_ERROR, 7, 1)
127
+ FIELD(CFRM_ISR0, WRITE_OVERRUN_ERROR, 6, 1)
128
+ FIELD(CFRM_ISR0, READ_OVERRUN_ERROR, 5, 1)
129
+ FIELD(CFRM_ISR0, CMD_INTERRUPT_ERROR, 4, 1)
130
+ FIELD(CFRM_ISR0, WRITE_INTERRUPT_ERROR, 3, 1)
131
+ FIELD(CFRM_ISR0, READ_INTERRUPT_ERROR, 2, 1)
132
+ FIELD(CFRM_ISR0, SEU_CRC_ERROR, 1, 1)
133
+ FIELD(CFRM_ISR0, SEU_ECC_ERROR, 0, 1)
134
+REG32(CFRM_ISR1, 0x154)
135
+REG32(CFRM_ISR2, 0x158)
136
+REG32(CFRM_ISR3, 0x15c)
137
+REG32(CFRM_IMR0, 0x160)
138
+ FIELD(CFRM_IMR0, READ_BROADCAST_ERROR, 21, 1)
139
+ FIELD(CFRM_IMR0, CMD_MISSING_ERROR, 20, 1)
140
+ FIELD(CFRM_IMR0, RW_ROWOFF_ERROR, 19, 1)
141
+ FIELD(CFRM_IMR0, READ_REG_ADDR_ERROR, 18, 1)
142
+ FIELD(CFRM_IMR0, READ_BLK_TYPE_ERROR, 17, 1)
143
+ FIELD(CFRM_IMR0, READ_FRAME_ADDR_ERROR, 16, 1)
144
+ FIELD(CFRM_IMR0, WRITE_REG_ADDR_ERROR, 15, 1)
145
+ FIELD(CFRM_IMR0, WRITE_BLK_TYPE_ERROR, 13, 1)
146
+ FIELD(CFRM_IMR0, WRITE_FRAME_ADDR_ERROR, 12, 1)
147
+ FIELD(CFRM_IMR0, MFW_OVERRUN_ERROR, 11, 1)
148
+ FIELD(CFRM_IMR0, FAR_FIFO_UNDERFLOW, 10, 1)
149
+ FIELD(CFRM_IMR0, FAR_FIFO_OVERFLOW, 9, 1)
150
+ FIELD(CFRM_IMR0, PER_FRAME_SEQ_ERROR, 8, 1)
151
+ FIELD(CFRM_IMR0, CRC_ERROR, 7, 1)
152
+ FIELD(CFRM_IMR0, WRITE_OVERRUN_ERROR, 6, 1)
153
+ FIELD(CFRM_IMR0, READ_OVERRUN_ERROR, 5, 1)
154
+ FIELD(CFRM_IMR0, CMD_INTERRUPT_ERROR, 4, 1)
155
+ FIELD(CFRM_IMR0, WRITE_INTERRUPT_ERROR, 3, 1)
156
+ FIELD(CFRM_IMR0, READ_INTERRUPT_ERROR, 2, 1)
157
+ FIELD(CFRM_IMR0, SEU_CRC_ERROR, 1, 1)
158
+ FIELD(CFRM_IMR0, SEU_ECC_ERROR, 0, 1)
159
+REG32(CFRM_IMR1, 0x164)
160
+REG32(CFRM_IMR2, 0x168)
161
+REG32(CFRM_IMR3, 0x16c)
162
+REG32(CFRM_IER0, 0x170)
163
+ FIELD(CFRM_IER0, READ_BROADCAST_ERROR, 21, 1)
164
+ FIELD(CFRM_IER0, CMD_MISSING_ERROR, 20, 1)
165
+ FIELD(CFRM_IER0, RW_ROWOFF_ERROR, 19, 1)
166
+ FIELD(CFRM_IER0, READ_REG_ADDR_ERROR, 18, 1)
167
+ FIELD(CFRM_IER0, READ_BLK_TYPE_ERROR, 17, 1)
168
+ FIELD(CFRM_IER0, READ_FRAME_ADDR_ERROR, 16, 1)
169
+ FIELD(CFRM_IER0, WRITE_REG_ADDR_ERROR, 15, 1)
170
+ FIELD(CFRM_IER0, WRITE_BLK_TYPE_ERROR, 13, 1)
171
+ FIELD(CFRM_IER0, WRITE_FRAME_ADDR_ERROR, 12, 1)
172
+ FIELD(CFRM_IER0, MFW_OVERRUN_ERROR, 11, 1)
173
+ FIELD(CFRM_IER0, FAR_FIFO_UNDERFLOW, 10, 1)
174
+ FIELD(CFRM_IER0, FAR_FIFO_OVERFLOW, 9, 1)
175
+ FIELD(CFRM_IER0, PER_FRAME_SEQ_ERROR, 8, 1)
176
+ FIELD(CFRM_IER0, CRC_ERROR, 7, 1)
177
+ FIELD(CFRM_IER0, WRITE_OVERRUN_ERROR, 6, 1)
178
+ FIELD(CFRM_IER0, READ_OVERRUN_ERROR, 5, 1)
179
+ FIELD(CFRM_IER0, CMD_INTERRUPT_ERROR, 4, 1)
180
+ FIELD(CFRM_IER0, WRITE_INTERRUPT_ERROR, 3, 1)
181
+ FIELD(CFRM_IER0, READ_INTERRUPT_ERROR, 2, 1)
182
+ FIELD(CFRM_IER0, SEU_CRC_ERROR, 1, 1)
183
+ FIELD(CFRM_IER0, SEU_ECC_ERROR, 0, 1)
184
+REG32(CFRM_IER1, 0x174)
185
+REG32(CFRM_IER2, 0x178)
186
+REG32(CFRM_IER3, 0x17c)
187
+REG32(CFRM_IDR0, 0x180)
188
+ FIELD(CFRM_IDR0, READ_BROADCAST_ERROR, 21, 1)
189
+ FIELD(CFRM_IDR0, CMD_MISSING_ERROR, 20, 1)
190
+ FIELD(CFRM_IDR0, RW_ROWOFF_ERROR, 19, 1)
191
+ FIELD(CFRM_IDR0, READ_REG_ADDR_ERROR, 18, 1)
192
+ FIELD(CFRM_IDR0, READ_BLK_TYPE_ERROR, 17, 1)
193
+ FIELD(CFRM_IDR0, READ_FRAME_ADDR_ERROR, 16, 1)
194
+ FIELD(CFRM_IDR0, WRITE_REG_ADDR_ERROR, 15, 1)
195
+ FIELD(CFRM_IDR0, WRITE_BLK_TYPE_ERROR, 13, 1)
196
+ FIELD(CFRM_IDR0, WRITE_FRAME_ADDR_ERROR, 12, 1)
197
+ FIELD(CFRM_IDR0, MFW_OVERRUN_ERROR, 11, 1)
198
+ FIELD(CFRM_IDR0, FAR_FIFO_UNDERFLOW, 10, 1)
199
+ FIELD(CFRM_IDR0, FAR_FIFO_OVERFLOW, 9, 1)
200
+ FIELD(CFRM_IDR0, PER_FRAME_SEQ_ERROR, 8, 1)
201
+ FIELD(CFRM_IDR0, CRC_ERROR, 7, 1)
202
+ FIELD(CFRM_IDR0, WRITE_OVERRUN_ERROR, 6, 1)
203
+ FIELD(CFRM_IDR0, READ_OVERRUN_ERROR, 5, 1)
204
+ FIELD(CFRM_IDR0, CMD_INTERRUPT_ERROR, 4, 1)
205
+ FIELD(CFRM_IDR0, WRITE_INTERRUPT_ERROR, 3, 1)
206
+ FIELD(CFRM_IDR0, READ_INTERRUPT_ERROR, 2, 1)
207
+ FIELD(CFRM_IDR0, SEU_CRC_ERROR, 1, 1)
208
+ FIELD(CFRM_IDR0, SEU_ECC_ERROR, 0, 1)
209
+REG32(CFRM_IDR1, 0x184)
210
+REG32(CFRM_IDR2, 0x188)
211
+REG32(CFRM_IDR3, 0x18c)
212
+REG32(CFRM_ITR0, 0x190)
213
+ FIELD(CFRM_ITR0, READ_BROADCAST_ERROR, 21, 1)
214
+ FIELD(CFRM_ITR0, CMD_MISSING_ERROR, 20, 1)
215
+ FIELD(CFRM_ITR0, RW_ROWOFF_ERROR, 19, 1)
216
+ FIELD(CFRM_ITR0, READ_REG_ADDR_ERROR, 18, 1)
217
+ FIELD(CFRM_ITR0, READ_BLK_TYPE_ERROR, 17, 1)
218
+ FIELD(CFRM_ITR0, READ_FRAME_ADDR_ERROR, 16, 1)
219
+ FIELD(CFRM_ITR0, WRITE_REG_ADDR_ERROR, 15, 1)
220
+ FIELD(CFRM_ITR0, WRITE_BLK_TYPE_ERROR, 13, 1)
221
+ FIELD(CFRM_ITR0, WRITE_FRAME_ADDR_ERROR, 12, 1)
222
+ FIELD(CFRM_ITR0, MFW_OVERRUN_ERROR, 11, 1)
223
+ FIELD(CFRM_ITR0, FAR_FIFO_UNDERFLOW, 10, 1)
224
+ FIELD(CFRM_ITR0, FAR_FIFO_OVERFLOW, 9, 1)
225
+ FIELD(CFRM_ITR0, PER_FRAME_SEQ_ERROR, 8, 1)
226
+ FIELD(CFRM_ITR0, CRC_ERROR, 7, 1)
227
+ FIELD(CFRM_ITR0, WRITE_OVERRUN_ERROR, 6, 1)
228
+ FIELD(CFRM_ITR0, READ_OVERRUN_ERROR, 5, 1)
229
+ FIELD(CFRM_ITR0, CMD_INTERRUPT_ERROR, 4, 1)
230
+ FIELD(CFRM_ITR0, WRITE_INTERRUPT_ERROR, 3, 1)
231
+ FIELD(CFRM_ITR0, READ_INTERRUPT_ERROR, 2, 1)
232
+ FIELD(CFRM_ITR0, SEU_CRC_ERROR, 1, 1)
233
+ FIELD(CFRM_ITR0, SEU_ECC_ERROR, 0, 1)
234
+REG32(CFRM_ITR1, 0x194)
235
+REG32(CFRM_ITR2, 0x198)
236
+REG32(CFRM_ITR3, 0x19c)
237
+REG32(SEU_SYNDRM00, 0x1a0)
238
+REG32(SEU_SYNDRM01, 0x1a4)
239
+REG32(SEU_SYNDRM02, 0x1a8)
240
+REG32(SEU_SYNDRM03, 0x1ac)
241
+REG32(SEU_SYNDRM10, 0x1b0)
242
+REG32(SEU_SYNDRM11, 0x1b4)
243
+REG32(SEU_SYNDRM12, 0x1b8)
244
+REG32(SEU_SYNDRM13, 0x1bc)
245
+REG32(SEU_SYNDRM20, 0x1c0)
246
+REG32(SEU_SYNDRM21, 0x1c4)
247
+REG32(SEU_SYNDRM22, 0x1c8)
248
+REG32(SEU_SYNDRM23, 0x1cc)
249
+REG32(SEU_SYNDRM30, 0x1d0)
250
+REG32(SEU_SYNDRM31, 0x1d4)
251
+REG32(SEU_SYNDRM32, 0x1d8)
252
+REG32(SEU_SYNDRM33, 0x1dc)
253
+REG32(SEU_VIRTUAL_SYNDRM0, 0x1e0)
254
+REG32(SEU_VIRTUAL_SYNDRM1, 0x1e4)
255
+REG32(SEU_VIRTUAL_SYNDRM2, 0x1e8)
256
+REG32(SEU_VIRTUAL_SYNDRM3, 0x1ec)
257
+REG32(SEU_CRC0, 0x1f0)
258
+REG32(SEU_CRC1, 0x1f4)
259
+REG32(SEU_CRC2, 0x1f8)
260
+REG32(SEU_CRC3, 0x1fc)
261
+REG32(CFRAME_FAR_BOT0, 0x200)
262
+REG32(CFRAME_FAR_BOT1, 0x204)
263
+REG32(CFRAME_FAR_BOT2, 0x208)
264
+REG32(CFRAME_FAR_BOT3, 0x20c)
265
+REG32(CFRAME_FAR_TOP0, 0x210)
266
+REG32(CFRAME_FAR_TOP1, 0x214)
267
+REG32(CFRAME_FAR_TOP2, 0x218)
268
+REG32(CFRAME_FAR_TOP3, 0x21c)
269
+REG32(LAST_FRAME_BOT0, 0x220)
270
+ FIELD(LAST_FRAME_BOT0, BLOCKTYPE1_LAST_FRAME_LSB, 20, 12)
271
+ FIELD(LAST_FRAME_BOT0, BLOCKTYPE0_LAST_FRAME, 0, 20)
272
+REG32(LAST_FRAME_BOT1, 0x224)
273
+ FIELD(LAST_FRAME_BOT1, BLOCKTYPE3_LAST_FRAME_LSB, 28, 4)
274
+ FIELD(LAST_FRAME_BOT1, BLOCKTYPE2_LAST_FRAME, 8, 20)
275
+ FIELD(LAST_FRAME_BOT1, BLOCKTYPE1_LAST_FRAME_MSB, 0, 8)
276
+REG32(LAST_FRAME_BOT2, 0x228)
277
+ FIELD(LAST_FRAME_BOT2, BLOCKTYPE3_LAST_FRAME_MSB, 0, 16)
278
+REG32(LAST_FRAME_BOT3, 0x22c)
279
+REG32(LAST_FRAME_TOP0, 0x230)
280
+ FIELD(LAST_FRAME_TOP0, BLOCKTYPE5_LAST_FRAME_LSB, 20, 12)
281
+ FIELD(LAST_FRAME_TOP0, BLOCKTYPE4_LAST_FRAME, 0, 20)
282
+REG32(LAST_FRAME_TOP1, 0x234)
283
+ FIELD(LAST_FRAME_TOP1, BLOCKTYPE6_LAST_FRAME, 8, 20)
284
+ FIELD(LAST_FRAME_TOP1, BLOCKTYPE5_LAST_FRAME_MSB, 0, 8)
285
+REG32(LAST_FRAME_TOP2, 0x238)
286
+REG32(LAST_FRAME_TOP3, 0x23c)
287
+
288
+#define CFRAME_REG_R_MAX (R_LAST_FRAME_TOP3 + 1)
289
+
290
+#define FRAME_NUM_QWORDS 25
291
+#define FRAME_NUM_WORDS (FRAME_NUM_QWORDS * 4) /* 25 * 128 bits */
292
+
293
+typedef struct XlnxCFrame {
294
+ uint32_t data[FRAME_NUM_WORDS];
295
+} XlnxCFrame;
296
+
297
+struct XlnxVersalCFrameReg {
298
+ SysBusDevice parent_obj;
299
+ MemoryRegion iomem;
300
+ MemoryRegion iomem_fdri;
301
+ qemu_irq irq_cfrm_imr;
302
+
303
+ /* 128-bit wfifo. */
304
+ uint32_t wfifo[WFIFO_SZ];
305
+
306
+ uint32_t regs[CFRAME_REG_R_MAX];
307
+ RegisterInfo regs_info[CFRAME_REG_R_MAX];
308
+
309
+ bool rowon;
310
+ bool wcfg;
311
+ bool rcfg;
312
+
313
+ GTree *cframes;
314
+ Fifo32 new_f_data;
315
+
316
+ struct {
317
+ XlnxCfiIf *cfu_fdro;
318
+ uint32_t blktype_num_frames[7];
319
+ } cfg;
320
+ bool row_configured;
321
+};
322
+
323
+#endif
324
diff --git a/hw/misc/xlnx-versal-cframe-reg.c b/hw/misc/xlnx-versal-cframe-reg.c
325
new file mode 100644
326
index XXXXXXX..XXXXXXX
327
--- /dev/null
328
+++ b/hw/misc/xlnx-versal-cframe-reg.c
329
@@ -XXX,XX +XXX,XX @@
330
+/*
331
+ * QEMU model of the Configuration Frame Control module
332
+ *
333
+ * Copyright (C) 2023, Advanced Micro Devices, Inc.
334
+ *
335
+ * Written by Francisco Iglesias <francisco.iglesias@amd.com>
336
+ *
337
+ * SPDX-License-Identifier: GPL-2.0-or-later
338
+ */
339
+
340
+#include "qemu/osdep.h"
341
+#include "hw/sysbus.h"
342
+#include "hw/register.h"
343
+#include "hw/registerfields.h"
344
+#include "qemu/bitops.h"
345
+#include "qemu/log.h"
346
+#include "qemu/units.h"
347
+#include "qapi/error.h"
348
+#include "hw/qdev-properties.h"
349
+#include "migration/vmstate.h"
350
+#include "hw/irq.h"
351
+#include "hw/misc/xlnx-versal-cframe-reg.h"
352
+
353
+#ifndef XLNX_VERSAL_CFRAME_REG_ERR_DEBUG
354
+#define XLNX_VERSAL_CFRAME_REG_ERR_DEBUG 0
355
+#endif
356
+
357
+#define KEYHOLE_STREAM_4K (4 * KiB)
358
+#define N_WORDS_128BIT 4
359
+
360
+#define MAX_BLOCKTYPE 6
361
+#define MAX_BLOCKTYPE_FRAMES 0xFFFFF
362
+
363
+enum {
364
+ CFRAME_CMD_WCFG = 1,
365
+ CFRAME_CMD_ROWON = 2,
366
+ CFRAME_CMD_ROWOFF = 3,
367
+ CFRAME_CMD_RCFG = 4,
368
+ CFRAME_CMD_DLPARK = 5,
369
+};
370
+
371
+static gint int_cmp(gconstpointer a, gconstpointer b, gpointer user_data)
372
+{
373
+ guint ua = GPOINTER_TO_UINT(a);
374
+ guint ub = GPOINTER_TO_UINT(b);
375
+ return (ua > ub) - (ua < ub);
376
+}
377
+
378
+static void cfrm_imr_update_irq(XlnxVersalCFrameReg *s)
379
+{
380
+ bool pending = s->regs[R_CFRM_ISR0] & ~s->regs[R_CFRM_IMR0];
381
+ qemu_set_irq(s->irq_cfrm_imr, pending);
382
+}
383
+
384
+static void cfrm_isr_postw(RegisterInfo *reg, uint64_t val64)
385
+{
386
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
387
+ cfrm_imr_update_irq(s);
388
+}
389
+
390
+static uint64_t cfrm_ier_prew(RegisterInfo *reg, uint64_t val64)
391
+{
392
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
393
+
394
+ s->regs[R_CFRM_IMR0] &= ~s->regs[R_CFRM_IER0];
395
+ s->regs[R_CFRM_IER0] = 0;
396
+ cfrm_imr_update_irq(s);
397
+ return 0;
398
+}
399
+
400
+static uint64_t cfrm_idr_prew(RegisterInfo *reg, uint64_t val64)
401
+{
402
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
403
+
404
+ s->regs[R_CFRM_IMR0] |= s->regs[R_CFRM_IDR0];
405
+ s->regs[R_CFRM_IDR0] = 0;
406
+ cfrm_imr_update_irq(s);
407
+ return 0;
408
+}
409
+
410
+static uint64_t cfrm_itr_prew(RegisterInfo *reg, uint64_t val64)
411
+{
412
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
413
+
414
+ s->regs[R_CFRM_ISR0] |= s->regs[R_CFRM_ITR0];
415
+ s->regs[R_CFRM_ITR0] = 0;
416
+ cfrm_imr_update_irq(s);
417
+ return 0;
418
+}
419
+
420
+static void cframe_incr_far(XlnxVersalCFrameReg *s)
421
+{
422
+ uint32_t faddr = ARRAY_FIELD_EX32(s->regs, FAR0, FRAME_ADDR);
423
+ uint32_t blktype = ARRAY_FIELD_EX32(s->regs, FAR0, BLOCKTYPE);
424
+
425
+ assert(blktype <= MAX_BLOCKTYPE);
426
+
427
+ faddr++;
428
+ if (faddr > s->cfg.blktype_num_frames[blktype]) {
429
+ /* Restart from 0 and increment block type */
430
+ faddr = 0;
431
+ blktype++;
432
+
433
+ assert(blktype <= MAX_BLOCKTYPE);
434
+
435
+ ARRAY_FIELD_DP32(s->regs, FAR0, BLOCKTYPE, blktype);
436
+ }
437
+
438
+ ARRAY_FIELD_DP32(s->regs, FAR0, FRAME_ADDR, faddr);
439
+}
440
+
441
+static void cfrm_fdri_post_write(RegisterInfo *reg, uint64_t val)
442
+{
443
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
444
+
445
+ if (s->row_configured && s->rowon && s->wcfg) {
446
+
447
+ if (fifo32_num_free(&s->new_f_data) >= N_WORDS_128BIT) {
448
+ fifo32_push(&s->new_f_data, s->regs[R_FDRI0]);
449
+ fifo32_push(&s->new_f_data, s->regs[R_FDRI1]);
450
+ fifo32_push(&s->new_f_data, s->regs[R_FDRI2]);
451
+ fifo32_push(&s->new_f_data, s->regs[R_FDRI3]);
452
+ }
453
+
454
+ if (fifo32_is_full(&s->new_f_data)) {
455
+ uint32_t addr = extract32(s->regs[R_FAR0], 0, 23);
456
+ XlnxCFrame *f = g_new(XlnxCFrame, 1);
457
+
458
+ for (int i = 0; i < FRAME_NUM_WORDS; i++) {
459
+ f->data[i] = fifo32_pop(&s->new_f_data);
460
+ }
461
+
462
+ g_tree_replace(s->cframes, GUINT_TO_POINTER(addr), f);
463
+
464
+ cframe_incr_far(s);
465
+
466
+ fifo32_reset(&s->new_f_data);
467
+ }
468
+ }
469
+}
470
+
471
+static void cfrm_readout_frames(XlnxVersalCFrameReg *s, uint32_t start_addr,
472
+ uint32_t end_addr)
473
+{
474
+ /*
475
+ * NB: when our minimum glib version is at least 2.68 we can improve the
476
+ * performance of the cframe traversal by using g_tree_lookup_node and
477
+ * g_tree_node_next (instead of calling g_tree_lookup for finding each
478
+ * cframe).
479
+ */
480
+ for (uint32_t addr = start_addr; addr < end_addr; addr++) {
481
+ XlnxCFrame *f = g_tree_lookup(s->cframes, GUINT_TO_POINTER(addr));
482
+
483
+ /* Transmit the data if a frame was found */
484
+ if (f) {
485
+ for (int i = 0; i < FRAME_NUM_WORDS; i += 4) {
486
+ XlnxCfiPacket pkt = {};
487
+
488
+ pkt.data[0] = f->data[i];
489
+ pkt.data[1] = f->data[i + 1];
490
+ pkt.data[2] = f->data[i + 2];
491
+ pkt.data[3] = f->data[i + 3];
492
+
493
+ if (s->cfg.cfu_fdro) {
494
+ xlnx_cfi_transfer_packet(s->cfg.cfu_fdro, &pkt);
495
+ }
496
+ }
497
+ }
498
+ }
499
+}
500
+
501
+static void cfrm_frcnt_post_write(RegisterInfo *reg, uint64_t val)
502
+{
503
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
504
+
505
+ if (s->row_configured && s->rowon && s->rcfg) {
506
+ uint32_t start_addr = extract32(s->regs[R_FAR0], 0, 23);
507
+ uint32_t end_addr = start_addr + s->regs[R_FRCNT0] / FRAME_NUM_QWORDS;
508
+
509
+ cfrm_readout_frames(s, start_addr, end_addr);
510
+ }
511
+}
512
+
513
+static void cfrm_cmd_post_write(RegisterInfo *reg, uint64_t val)
514
+{
515
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
516
+
517
+ if (s->row_configured) {
518
+ uint8_t cmd = ARRAY_FIELD_EX32(s->regs, CMD0, CMD);
519
+
520
+ switch (cmd) {
521
+ case CFRAME_CMD_WCFG:
522
+ s->wcfg = true;
523
+ break;
524
+ case CFRAME_CMD_ROWON:
525
+ s->rowon = true;
526
+ break;
527
+ case CFRAME_CMD_ROWOFF:
528
+ s->rowon = false;
529
+ break;
530
+ case CFRAME_CMD_RCFG:
531
+ s->rcfg = true;
532
+ break;
533
+ case CFRAME_CMD_DLPARK:
534
+ s->wcfg = false;
535
+ s->rcfg = false;
536
+ break;
537
+ default:
538
+ break;
539
+ };
540
+ }
541
+}
542
+
543
+static uint64_t cfrm_last_frame_bot_post_read(RegisterInfo *reg,
544
+ uint64_t val64)
545
+{
546
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
547
+ uint64_t val = 0;
548
+
549
+ switch (reg->access->addr) {
550
+ case A_LAST_FRAME_BOT0:
551
+ val = FIELD_DP32(val, LAST_FRAME_BOT0, BLOCKTYPE1_LAST_FRAME_LSB,
552
+ s->cfg.blktype_num_frames[1]);
553
+ val = FIELD_DP32(val, LAST_FRAME_BOT0, BLOCKTYPE0_LAST_FRAME,
554
+ s->cfg.blktype_num_frames[0]);
555
+ break;
556
+ case A_LAST_FRAME_BOT1:
557
+ val = FIELD_DP32(val, LAST_FRAME_BOT1, BLOCKTYPE3_LAST_FRAME_LSB,
558
+ s->cfg.blktype_num_frames[3]);
559
+ val = FIELD_DP32(val, LAST_FRAME_BOT1, BLOCKTYPE2_LAST_FRAME,
560
+ s->cfg.blktype_num_frames[2]);
561
+ val = FIELD_DP32(val, LAST_FRAME_BOT1, BLOCKTYPE1_LAST_FRAME_MSB,
562
+ (s->cfg.blktype_num_frames[1] >> 12));
563
+ break;
564
+ case A_LAST_FRAME_BOT2:
565
+ val = FIELD_DP32(val, LAST_FRAME_BOT2, BLOCKTYPE3_LAST_FRAME_MSB,
566
+ (s->cfg.blktype_num_frames[3] >> 4));
567
+ break;
568
+ case A_LAST_FRAME_BOT3:
569
+ default:
570
+ break;
571
+ }
572
+
573
+ return val;
574
+}
575
+
576
+static uint64_t cfrm_last_frame_top_post_read(RegisterInfo *reg,
577
+ uint64_t val64)
578
+{
579
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
580
+ uint64_t val = 0;
581
+
582
+ switch (reg->access->addr) {
583
+ case A_LAST_FRAME_TOP0:
584
+ val = FIELD_DP32(val, LAST_FRAME_TOP0, BLOCKTYPE5_LAST_FRAME_LSB,
585
+ s->cfg.blktype_num_frames[5]);
586
+ val = FIELD_DP32(val, LAST_FRAME_TOP0, BLOCKTYPE4_LAST_FRAME,
587
+ s->cfg.blktype_num_frames[4]);
588
+ break;
589
+ case A_LAST_FRAME_TOP1:
590
+ val = FIELD_DP32(val, LAST_FRAME_TOP1, BLOCKTYPE6_LAST_FRAME,
591
+ s->cfg.blktype_num_frames[6]);
592
+ val = FIELD_DP32(val, LAST_FRAME_TOP1, BLOCKTYPE5_LAST_FRAME_MSB,
593
+ (s->cfg.blktype_num_frames[5] >> 12));
594
+ break;
595
+ case A_LAST_FRAME_TOP2:
596
+ case A_LAST_FRAME_BOT3:
597
+ default:
598
+ break;
599
+ }
600
+
601
+ return val;
602
+}
603
+
604
+static void cfrm_far_sfr_post_write(RegisterInfo *reg, uint64_t val)
605
+{
606
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(reg->opaque);
607
+
608
+ if (s->row_configured && s->rowon && s->rcfg) {
609
+ uint32_t start_addr = extract32(s->regs[R_FAR_SFR0], 0, 23);
610
+
611
+ /* Readback 1 frame */
612
+ cfrm_readout_frames(s, start_addr, start_addr + 1);
613
+ }
614
+}
615
+
616
+static const RegisterAccessInfo cframe_reg_regs_info[] = {
617
+ { .name = "CRC0", .addr = A_CRC0,
618
+ .rsvd = 0x00000000,
619
+ },{ .name = "CRC1", .addr = A_CRC0,
620
+ .rsvd = 0xffffffff,
621
+ },{ .name = "CRC2", .addr = A_CRC0,
622
+ .rsvd = 0xffffffff,
623
+ },{ .name = "CRC3", .addr = A_CRC0,
624
+ .rsvd = 0xffffffff,
625
+ },{ .name = "FAR0", .addr = A_FAR0,
626
+ .rsvd = 0xfe000000,
627
+ },{ .name = "FAR1", .addr = A_FAR1,
628
+ .rsvd = 0xffffffff,
629
+ },{ .name = "FAR2", .addr = A_FAR2,
630
+ .rsvd = 0xffffffff,
631
+ },{ .name = "FAR3", .addr = A_FAR3,
632
+ .rsvd = 0xffffffff,
633
+ },{ .name = "FAR_SFR0", .addr = A_FAR_SFR0,
634
+ .rsvd = 0xff800000,
635
+ },{ .name = "FAR_SFR1", .addr = A_FAR_SFR1,
636
+ .rsvd = 0xffffffff,
637
+ },{ .name = "FAR_SFR2", .addr = A_FAR_SFR2,
638
+ .rsvd = 0xffffffff,
639
+ },{ .name = "FAR_SFR3", .addr = A_FAR_SFR3,
640
+ .rsvd = 0xffffffff,
641
+ .post_write = cfrm_far_sfr_post_write,
642
+ },{ .name = "FDRI0", .addr = A_FDRI0,
643
+ },{ .name = "FDRI1", .addr = A_FDRI1,
644
+ },{ .name = "FDRI2", .addr = A_FDRI2,
645
+ },{ .name = "FDRI3", .addr = A_FDRI3,
646
+ .post_write = cfrm_fdri_post_write,
647
+ },{ .name = "FRCNT0", .addr = A_FRCNT0,
648
+ .rsvd = 0x00000000,
649
+ },{ .name = "FRCNT1", .addr = A_FRCNT1,
650
+ .rsvd = 0xffffffff,
651
+ },{ .name = "FRCNT2", .addr = A_FRCNT2,
652
+ .rsvd = 0xffffffff,
653
+ },{ .name = "FRCNT3", .addr = A_FRCNT3,
654
+ .rsvd = 0xffffffff,
655
+ .post_write = cfrm_frcnt_post_write
656
+ },{ .name = "CMD0", .addr = A_CMD0,
657
+ .rsvd = 0xffffffe0,
658
+ },{ .name = "CMD1", .addr = A_CMD1,
659
+ .rsvd = 0xffffffff,
660
+ },{ .name = "CMD2", .addr = A_CMD2,
661
+ .rsvd = 0xffffffff,
662
+ },{ .name = "CMD3", .addr = A_CMD3,
663
+ .rsvd = 0xffffffff,
664
+ .post_write = cfrm_cmd_post_write
665
+ },{ .name = "CR_MASK0", .addr = A_CR_MASK0,
666
+ .rsvd = 0x00000000,
667
+ },{ .name = "CR_MASK1", .addr = A_CR_MASK1,
668
+ .rsvd = 0x00000000,
669
+ },{ .name = "CR_MASK2", .addr = A_CR_MASK2,
670
+ .rsvd = 0x00000000,
671
+ },{ .name = "CR_MASK3", .addr = A_CR_MASK3,
672
+ .rsvd = 0xffffffff,
673
+ },{ .name = "CTL0", .addr = A_CTL0,
674
+ .rsvd = 0xfffffff8,
675
+ },{ .name = "CTL1", .addr = A_CTL1,
676
+ .rsvd = 0xffffffff,
677
+ },{ .name = "CTL2", .addr = A_CTL2,
678
+ .rsvd = 0xffffffff,
679
+ },{ .name = "CTL3", .addr = A_CTL3,
680
+ .rsvd = 0xffffffff,
681
+ },{ .name = "CFRM_ISR0", .addr = A_CFRM_ISR0,
682
+ .rsvd = 0xffc04000,
683
+ .w1c = 0x3bfff,
684
+ },{ .name = "CFRM_ISR1", .addr = A_CFRM_ISR1,
685
+ .rsvd = 0xffffffff,
686
+ },{ .name = "CFRM_ISR2", .addr = A_CFRM_ISR2,
687
+ .rsvd = 0xffffffff,
688
+ },{ .name = "CFRM_ISR3", .addr = A_CFRM_ISR3,
689
+ .rsvd = 0xffffffff,
690
+ .post_write = cfrm_isr_postw,
691
+ },{ .name = "CFRM_IMR0", .addr = A_CFRM_IMR0,
692
+ .rsvd = 0xffc04000,
693
+ .ro = 0xfffff,
694
+ .reset = 0x3bfff,
695
+ },{ .name = "CFRM_IMR1", .addr = A_CFRM_IMR1,
696
+ .rsvd = 0xffffffff,
697
+ },{ .name = "CFRM_IMR2", .addr = A_CFRM_IMR2,
698
+ .rsvd = 0xffffffff,
699
+ },{ .name = "CFRM_IMR3", .addr = A_CFRM_IMR3,
700
+ .rsvd = 0xffffffff,
701
+ },{ .name = "CFRM_IER0", .addr = A_CFRM_IER0,
702
+ .rsvd = 0xffc04000,
703
+ },{ .name = "CFRM_IER1", .addr = A_CFRM_IER1,
704
+ .rsvd = 0xffffffff,
705
+ },{ .name = "CFRM_IER2", .addr = A_CFRM_IER2,
706
+ .rsvd = 0xffffffff,
707
+ },{ .name = "CFRM_IER3", .addr = A_CFRM_IER3,
708
+ .rsvd = 0xffffffff,
709
+ .pre_write = cfrm_ier_prew,
710
+ },{ .name = "CFRM_IDR0", .addr = A_CFRM_IDR0,
711
+ .rsvd = 0xffc04000,
712
+ },{ .name = "CFRM_IDR1", .addr = A_CFRM_IDR1,
713
+ .rsvd = 0xffffffff,
714
+ },{ .name = "CFRM_IDR2", .addr = A_CFRM_IDR2,
715
+ .rsvd = 0xffffffff,
716
+ },{ .name = "CFRM_IDR3", .addr = A_CFRM_IDR3,
717
+ .rsvd = 0xffffffff,
718
+ .pre_write = cfrm_idr_prew,
719
+ },{ .name = "CFRM_ITR0", .addr = A_CFRM_ITR0,
720
+ .rsvd = 0xffc04000,
721
+ },{ .name = "CFRM_ITR1", .addr = A_CFRM_ITR1,
722
+ .rsvd = 0xffffffff,
723
+ },{ .name = "CFRM_ITR2", .addr = A_CFRM_ITR2,
724
+ .rsvd = 0xffffffff,
725
+ },{ .name = "CFRM_ITR3", .addr = A_CFRM_ITR3,
726
+ .rsvd = 0xffffffff,
727
+ .pre_write = cfrm_itr_prew,
728
+ },{ .name = "SEU_SYNDRM00", .addr = A_SEU_SYNDRM00,
729
+ },{ .name = "SEU_SYNDRM01", .addr = A_SEU_SYNDRM01,
730
+ },{ .name = "SEU_SYNDRM02", .addr = A_SEU_SYNDRM02,
731
+ },{ .name = "SEU_SYNDRM03", .addr = A_SEU_SYNDRM03,
732
+ },{ .name = "SEU_SYNDRM10", .addr = A_SEU_SYNDRM10,
733
+ },{ .name = "SEU_SYNDRM11", .addr = A_SEU_SYNDRM11,
734
+ },{ .name = "SEU_SYNDRM12", .addr = A_SEU_SYNDRM12,
735
+ },{ .name = "SEU_SYNDRM13", .addr = A_SEU_SYNDRM13,
736
+ },{ .name = "SEU_SYNDRM20", .addr = A_SEU_SYNDRM20,
737
+ },{ .name = "SEU_SYNDRM21", .addr = A_SEU_SYNDRM21,
738
+ },{ .name = "SEU_SYNDRM22", .addr = A_SEU_SYNDRM22,
739
+ },{ .name = "SEU_SYNDRM23", .addr = A_SEU_SYNDRM23,
740
+ },{ .name = "SEU_SYNDRM30", .addr = A_SEU_SYNDRM30,
741
+ },{ .name = "SEU_SYNDRM31", .addr = A_SEU_SYNDRM31,
742
+ },{ .name = "SEU_SYNDRM32", .addr = A_SEU_SYNDRM32,
743
+ },{ .name = "SEU_SYNDRM33", .addr = A_SEU_SYNDRM33,
744
+ },{ .name = "SEU_VIRTUAL_SYNDRM0", .addr = A_SEU_VIRTUAL_SYNDRM0,
745
+ },{ .name = "SEU_VIRTUAL_SYNDRM1", .addr = A_SEU_VIRTUAL_SYNDRM1,
746
+ },{ .name = "SEU_VIRTUAL_SYNDRM2", .addr = A_SEU_VIRTUAL_SYNDRM2,
747
+ },{ .name = "SEU_VIRTUAL_SYNDRM3", .addr = A_SEU_VIRTUAL_SYNDRM3,
748
+ },{ .name = "SEU_CRC0", .addr = A_SEU_CRC0,
749
+ },{ .name = "SEU_CRC1", .addr = A_SEU_CRC1,
750
+ },{ .name = "SEU_CRC2", .addr = A_SEU_CRC2,
751
+ },{ .name = "SEU_CRC3", .addr = A_SEU_CRC3,
752
+ },{ .name = "CFRAME_FAR_BOT0", .addr = A_CFRAME_FAR_BOT0,
753
+ },{ .name = "CFRAME_FAR_BOT1", .addr = A_CFRAME_FAR_BOT1,
754
+ },{ .name = "CFRAME_FAR_BOT2", .addr = A_CFRAME_FAR_BOT2,
755
+ },{ .name = "CFRAME_FAR_BOT3", .addr = A_CFRAME_FAR_BOT3,
756
+ },{ .name = "CFRAME_FAR_TOP0", .addr = A_CFRAME_FAR_TOP0,
757
+ },{ .name = "CFRAME_FAR_TOP1", .addr = A_CFRAME_FAR_TOP1,
758
+ },{ .name = "CFRAME_FAR_TOP2", .addr = A_CFRAME_FAR_TOP2,
759
+ },{ .name = "CFRAME_FAR_TOP3", .addr = A_CFRAME_FAR_TOP3,
760
+ },{ .name = "LAST_FRAME_BOT0", .addr = A_LAST_FRAME_BOT0,
761
+ .ro = 0xffffffff,
762
+ .post_read = cfrm_last_frame_bot_post_read,
763
+ },{ .name = "LAST_FRAME_BOT1", .addr = A_LAST_FRAME_BOT1,
764
+ .ro = 0xffffffff,
765
+ .post_read = cfrm_last_frame_bot_post_read,
766
+ },{ .name = "LAST_FRAME_BOT2", .addr = A_LAST_FRAME_BOT2,
767
+ .ro = 0xffffffff,
768
+ .post_read = cfrm_last_frame_bot_post_read,
769
+ },{ .name = "LAST_FRAME_BOT3", .addr = A_LAST_FRAME_BOT3,
770
+ .ro = 0xffffffff,
771
+ .post_read = cfrm_last_frame_bot_post_read,
772
+ },{ .name = "LAST_FRAME_TOP0", .addr = A_LAST_FRAME_TOP0,
773
+ .ro = 0xffffffff,
774
+ .post_read = cfrm_last_frame_top_post_read,
775
+ },{ .name = "LAST_FRAME_TOP1", .addr = A_LAST_FRAME_TOP1,
776
+ .ro = 0xffffffff,
777
+ .post_read = cfrm_last_frame_top_post_read,
778
+ },{ .name = "LAST_FRAME_TOP2", .addr = A_LAST_FRAME_TOP2,
779
+ .ro = 0xffffffff,
780
+ .post_read = cfrm_last_frame_top_post_read,
781
+ },{ .name = "LAST_FRAME_TOP3", .addr = A_LAST_FRAME_TOP3,
782
+ .ro = 0xffffffff,
783
+ .post_read = cfrm_last_frame_top_post_read,
784
+ }
785
+};
786
+
787
+static void cframe_reg_cfi_transfer_packet(XlnxCfiIf *cfi_if,
788
+ XlnxCfiPacket *pkt)
789
+{
790
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(cfi_if);
791
+ uint64_t we = MAKE_64BIT_MASK(0, 4 * 8);
792
+
793
+ if (!s->row_configured) {
794
+ return;
795
+ }
796
+
797
+ switch (pkt->reg_addr) {
798
+ case CFRAME_FAR:
799
+ s->regs[R_FAR0] = pkt->data[0];
800
+ break;
801
+ case CFRAME_SFR:
802
+ s->regs[R_FAR_SFR0] = pkt->data[0];
803
+ register_write(&s->regs_info[R_FAR_SFR3], 0,
804
+ we, object_get_typename(OBJECT(s)),
805
+ XLNX_VERSAL_CFRAME_REG_ERR_DEBUG);
806
+ break;
807
+ case CFRAME_FDRI:
808
+ s->regs[R_FDRI0] = pkt->data[0];
809
+ s->regs[R_FDRI1] = pkt->data[1];
810
+ s->regs[R_FDRI2] = pkt->data[2];
811
+ register_write(&s->regs_info[R_FDRI3], pkt->data[3],
812
+ we, object_get_typename(OBJECT(s)),
813
+ XLNX_VERSAL_CFRAME_REG_ERR_DEBUG);
814
+ break;
815
+ case CFRAME_CMD:
816
+ ARRAY_FIELD_DP32(s->regs, CMD0, CMD, pkt->data[0]);
817
+
818
+ register_write(&s->regs_info[R_CMD3], 0,
819
+ we, object_get_typename(OBJECT(s)),
820
+ XLNX_VERSAL_CFRAME_REG_ERR_DEBUG);
821
+ break;
822
+ default:
823
+ break;
824
+ }
825
+}
826
+
827
+static uint64_t cframe_reg_fdri_read(void *opaque, hwaddr addr, unsigned size)
828
+{
829
+ qemu_log_mask(LOG_GUEST_ERROR, "%s: Unsupported read from addr=%"
830
+ HWADDR_PRIx "\n", __func__, addr);
831
+ return 0;
832
+}
833
+
834
+static void cframe_reg_fdri_write(void *opaque, hwaddr addr, uint64_t value,
835
+ unsigned size)
836
+{
837
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(opaque);
838
+ uint32_t wfifo[WFIFO_SZ];
839
+
840
+ if (update_wfifo(addr, value, s->wfifo, wfifo)) {
841
+ uint64_t we = MAKE_64BIT_MASK(0, 4 * 8);
842
+
843
+ s->regs[R_FDRI0] = wfifo[0];
844
+ s->regs[R_FDRI1] = wfifo[1];
845
+ s->regs[R_FDRI2] = wfifo[2];
846
+ register_write(&s->regs_info[R_FDRI3], wfifo[3],
847
+ we, object_get_typename(OBJECT(s)),
848
+ XLNX_VERSAL_CFRAME_REG_ERR_DEBUG);
849
+ }
850
+}
851
+
852
+static void cframe_reg_reset_enter(Object *obj, ResetType type)
853
+{
854
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(obj);
855
+ unsigned int i;
856
+
857
+ for (i = 0; i < ARRAY_SIZE(s->regs_info); ++i) {
858
+ register_reset(&s->regs_info[i]);
859
+ }
860
+ memset(s->wfifo, 0, WFIFO_SZ * sizeof(uint32_t));
861
+ fifo32_reset(&s->new_f_data);
862
+
863
+ if (g_tree_nnodes(s->cframes)) {
864
+ /*
865
+ * Take a reference so when g_tree_destroy() unrefs it we keep the
866
+ * GTree and only destroy its contents. NB: when our minimum
867
+ * glib version is at least 2.70 we could use g_tree_remove_all().
868
+ */
869
+ g_tree_ref(s->cframes);
870
+ g_tree_destroy(s->cframes);
871
+ }
872
+}
873
+
874
+static void cframe_reg_reset_hold(Object *obj)
875
+{
876
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(obj);
877
+
878
+ cfrm_imr_update_irq(s);
879
+}
880
+
881
+static const MemoryRegionOps cframe_reg_ops = {
882
+ .read = register_read_memory,
883
+ .write = register_write_memory,
884
+ .endianness = DEVICE_LITTLE_ENDIAN,
885
+ .valid = {
886
+ .min_access_size = 4,
887
+ .max_access_size = 4,
888
+ },
889
+};
890
+
891
+static const MemoryRegionOps cframe_reg_fdri_ops = {
892
+ .read = cframe_reg_fdri_read,
893
+ .write = cframe_reg_fdri_write,
894
+ .endianness = DEVICE_LITTLE_ENDIAN,
895
+ .valid = {
896
+ .min_access_size = 4,
897
+ .max_access_size = 4,
898
+ },
899
+};
900
+
901
+static void cframe_reg_realize(DeviceState *dev, Error **errp)
902
+{
903
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(dev);
904
+
905
+ for (int i = 0; i < ARRAY_SIZE(s->cfg.blktype_num_frames); i++) {
906
+ if (s->cfg.blktype_num_frames[i] > MAX_BLOCKTYPE_FRAMES) {
907
+ error_setg(errp,
908
+ "blktype-frames%d > 0xFFFFF (max frame per block)",
909
+ i);
910
+ return;
911
+ }
912
+ if (s->cfg.blktype_num_frames[i]) {
913
+ s->row_configured = true;
914
+ }
915
+ }
916
+}
917
+
918
+static void cframe_reg_init(Object *obj)
919
+{
920
+ XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(obj);
921
+ SysBusDevice *sbd = SYS_BUS_DEVICE(obj);
922
+ RegisterInfoArray *reg_array;
923
+
924
+ memory_region_init(&s->iomem, obj, TYPE_XLNX_VERSAL_CFRAME_REG,
925
+ CFRAME_REG_R_MAX * 4);
926
+ reg_array =
927
+ register_init_block32(DEVICE(obj), cframe_reg_regs_info,
928
+ ARRAY_SIZE(cframe_reg_regs_info),
929
+ s->regs_info, s->regs,
930
+ &cframe_reg_ops,
931
+ XLNX_VERSAL_CFRAME_REG_ERR_DEBUG,
932
+ CFRAME_REG_R_MAX * 4);
933
+ memory_region_add_subregion(&s->iomem,
934
+ 0x0,
935
+ &reg_array->mem);
936
+ sysbus_init_mmio(sbd, &s->iomem);
937
+ memory_region_init_io(&s->iomem_fdri, obj, &cframe_reg_fdri_ops, s,
938
+ TYPE_XLNX_VERSAL_CFRAME_REG "-fdri",
939
+ KEYHOLE_STREAM_4K);
940
+ sysbus_init_mmio(sbd, &s->iomem_fdri);
941
+ sysbus_init_irq(sbd, &s->irq_cfrm_imr);
942
+
943
+ s->cframes = g_tree_new_full((GCompareDataFunc)int_cmp, NULL,
944
+ NULL, (GDestroyNotify)g_free);
945
+ fifo32_create(&s->new_f_data, FRAME_NUM_WORDS);
946
+}
947
+
948
+static const VMStateDescription vmstate_cframe = {
949
+ .name = "cframe",
950
+ .version_id = 1,
951
+ .minimum_version_id = 1,
952
+ .fields = (VMStateField[]) {
953
+ VMSTATE_UINT32_ARRAY(data, XlnxCFrame, FRAME_NUM_WORDS),
954
+ VMSTATE_END_OF_LIST()
955
+ }
956
+};
957
+
958
+static const VMStateDescription vmstate_cframe_reg = {
959
+ .name = TYPE_XLNX_VERSAL_CFRAME_REG,
960
+ .version_id = 1,
961
+ .minimum_version_id = 1,
962
+ .fields = (VMStateField[]) {
963
+ VMSTATE_UINT32_ARRAY(wfifo, XlnxVersalCFrameReg, 4),
964
+ VMSTATE_UINT32_ARRAY(regs, XlnxVersalCFrameReg, CFRAME_REG_R_MAX),
965
+ VMSTATE_BOOL(rowon, XlnxVersalCFrameReg),
966
+ VMSTATE_BOOL(wcfg, XlnxVersalCFrameReg),
967
+ VMSTATE_BOOL(rcfg, XlnxVersalCFrameReg),
968
+ VMSTATE_GTREE_DIRECT_KEY_V(cframes, XlnxVersalCFrameReg, 1,
969
+ &vmstate_cframe, XlnxCFrame),
970
+ VMSTATE_FIFO32(new_f_data, XlnxVersalCFrameReg),
971
+ VMSTATE_END_OF_LIST(),
972
+ }
973
+};
974
+
975
+static Property cframe_regs_props[] = {
976
+ DEFINE_PROP_LINK("cfu-fdro", XlnxVersalCFrameReg, cfg.cfu_fdro,
977
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
978
+ DEFINE_PROP_UINT32("blktype0-frames", XlnxVersalCFrameReg,
979
+ cfg.blktype_num_frames[0], 0),
980
+ DEFINE_PROP_UINT32("blktype1-frames", XlnxVersalCFrameReg,
981
+ cfg.blktype_num_frames[1], 0),
982
+ DEFINE_PROP_UINT32("blktype2-frames", XlnxVersalCFrameReg,
983
+ cfg.blktype_num_frames[2], 0),
984
+ DEFINE_PROP_UINT32("blktype3-frames", XlnxVersalCFrameReg,
985
+ cfg.blktype_num_frames[3], 0),
986
+ DEFINE_PROP_UINT32("blktype4-frames", XlnxVersalCFrameReg,
987
+ cfg.blktype_num_frames[4], 0),
988
+ DEFINE_PROP_UINT32("blktype5-frames", XlnxVersalCFrameReg,
989
+ cfg.blktype_num_frames[5], 0),
990
+ DEFINE_PROP_UINT32("blktype6-frames", XlnxVersalCFrameReg,
991
+ cfg.blktype_num_frames[6], 0),
992
+ DEFINE_PROP_END_OF_LIST(),
993
+};
994
+
995
+static void cframe_reg_class_init(ObjectClass *klass, void *data)
996
+{
997
+ ResettableClass *rc = RESETTABLE_CLASS(klass);
998
+ DeviceClass *dc = DEVICE_CLASS(klass);
999
+ XlnxCfiIfClass *xcic = XLNX_CFI_IF_CLASS(klass);
1000
+
1001
+ dc->vmsd = &vmstate_cframe_reg;
1002
+ dc->realize = cframe_reg_realize;
1003
+ rc->phases.enter = cframe_reg_reset_enter;
1004
+ rc->phases.hold = cframe_reg_reset_hold;
1005
+ device_class_set_props(dc, cframe_regs_props);
1006
+ xcic->cfi_transfer_packet = cframe_reg_cfi_transfer_packet;
1007
+}
1008
+
1009
+static const TypeInfo cframe_reg_info = {
1010
+ .name = TYPE_XLNX_VERSAL_CFRAME_REG,
1011
+ .parent = TYPE_SYS_BUS_DEVICE,
1012
+ .instance_size = sizeof(XlnxVersalCFrameReg),
1013
+ .class_init = cframe_reg_class_init,
1014
+ .instance_init = cframe_reg_init,
1015
+ .interfaces = (InterfaceInfo[]) {
1016
+ { TYPE_XLNX_CFI_IF },
1017
+ { }
1018
+ }
1019
+};
1020
+
1021
+static void cframe_reg_register_types(void)
1022
+{
1023
+ type_register_static(&cframe_reg_info);
1024
+}
1025
+
1026
+type_init(cframe_reg_register_types)
1027
diff --git a/hw/misc/meson.build b/hw/misc/meson.build
1028
index XXXXXXX..XXXXXXX 100644
1029
--- a/hw/misc/meson.build
1030
+++ b/hw/misc/meson.build
1031
@@ -XXX,XX +XXX,XX @@ system_ss.add(when: 'CONFIG_XLNX_VERSAL', if_true: files(
1032
'xlnx-versal-pmc-iou-slcr.c',
1033
'xlnx-versal-cfu.c',
1034
'xlnx-cfi-if.c',
1035
+ 'xlnx-versal-cframe-reg.c',
1036
))
1037
system_ss.add(when: 'CONFIG_STM32F2XX_SYSCFG', if_true: files('stm32f2xx_syscfg.c'))
1038
system_ss.add(when: 'CONFIG_STM32F4XX_SYSCFG', if_true: files('stm32f4xx_syscfg.c'))
1039
--
1040
2.34.1
diff view generated by jsdifflib
[PULL 16/26] hw/misc: Introduce a model of Xilinx Versal's CFRAME_BCAST_REG
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Introduce a model of Xilinx Versal's Configuration Frame broadcast
4
controller (CFRAME_BCAST_REG).
5
6
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Message-id: 20230831165701.2016397-7-francisco.iglesias@amd.com
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
include/hw/misc/xlnx-versal-cframe-reg.h | 17 +++
12
hw/misc/xlnx-versal-cframe-reg.c | 161 +++++++++++++++++++++++
13
2 files changed, 178 insertions(+)
14
15
diff --git a/include/hw/misc/xlnx-versal-cframe-reg.h b/include/hw/misc/xlnx-versal-cframe-reg.h
16
index XXXXXXX..XXXXXXX 100644
17
--- a/include/hw/misc/xlnx-versal-cframe-reg.h
18
+++ b/include/hw/misc/xlnx-versal-cframe-reg.h
19
@@ -XXX,XX +XXX,XX @@
20
#define TYPE_XLNX_VERSAL_CFRAME_REG "xlnx,cframe-reg"
21
OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFrameReg, XLNX_VERSAL_CFRAME_REG)
22
23
+#define TYPE_XLNX_VERSAL_CFRAME_BCAST_REG "xlnx.cframe-bcast-reg"
24
+OBJECT_DECLARE_SIMPLE_TYPE(XlnxVersalCFrameBcastReg,
25
+ XLNX_VERSAL_CFRAME_BCAST_REG)
26
+
27
/*
28
* The registers in this module are 128 bits wide but it is ok to write
29
* and read them through 4 sequential 32 bit accesses (address[3:2] = 0,
30
@@ -XXX,XX +XXX,XX @@ struct XlnxVersalCFrameReg {
31
bool row_configured;
32
};
33
34
+struct XlnxVersalCFrameBcastReg {
35
+ SysBusDevice parent_obj;
36
+ MemoryRegion iomem_reg;
37
+ MemoryRegion iomem_fdri;
38
+
39
+ /* 128-bit wfifo. */
40
+ uint32_t wfifo[WFIFO_SZ];
41
+
42
+ struct {
43
+ XlnxCfiIf *cframe[15];
44
+ } cfg;
45
+};
46
+
47
#endif
48
diff --git a/hw/misc/xlnx-versal-cframe-reg.c b/hw/misc/xlnx-versal-cframe-reg.c
49
index XXXXXXX..XXXXXXX 100644
50
--- a/hw/misc/xlnx-versal-cframe-reg.c
51
+++ b/hw/misc/xlnx-versal-cframe-reg.c
52
@@ -XXX,XX +XXX,XX @@ static const MemoryRegionOps cframe_reg_fdri_ops = {
53
},
54
};
55
56
+static uint64_t cframes_bcast_reg_read(void *opaque, hwaddr addr, unsigned size)
57
+{
58
+ qemu_log_mask(LOG_GUEST_ERROR, "%s: Unsupported read from addr=%"
59
+ HWADDR_PRIx "\n", __func__, addr);
60
+ return 0;
61
+}
62
+
63
+static void cframes_bcast_write(XlnxVersalCFrameBcastReg *s, uint8_t reg_addr,
64
+ uint32_t *wfifo)
65
+{
66
+ XlnxCfiPacket pkt = {
67
+ .reg_addr = reg_addr,
68
+ .data[0] = wfifo[0],
69
+ .data[1] = wfifo[1],
70
+ .data[2] = wfifo[2],
71
+ .data[3] = wfifo[3]
72
+ };
73
+
74
+ for (int i = 0; i < ARRAY_SIZE(s->cfg.cframe); i++) {
75
+ if (s->cfg.cframe[i]) {
76
+ xlnx_cfi_transfer_packet(s->cfg.cframe[i], &pkt);
77
+ }
78
+ }
79
+}
80
+
81
+static void cframes_bcast_reg_write(void *opaque, hwaddr addr, uint64_t value,
82
+ unsigned size)
83
+{
84
+ XlnxVersalCFrameBcastReg *s = XLNX_VERSAL_CFRAME_BCAST_REG(opaque);
85
+ uint32_t wfifo[WFIFO_SZ];
86
+
87
+ if (update_wfifo(addr, value, s->wfifo, wfifo)) {
88
+ uint8_t reg_addr = extract32(addr, 4, 6);
89
+
90
+ cframes_bcast_write(s, reg_addr, wfifo);
91
+ }
92
+}
93
+
94
+static uint64_t cframes_bcast_fdri_read(void *opaque, hwaddr addr,
95
+ unsigned size)
96
+{
97
+ qemu_log_mask(LOG_GUEST_ERROR, "%s: Unsupported read from addr=%"
98
+ HWADDR_PRIx "\n", __func__, addr);
99
+ return 0;
100
+}
101
+
102
+static void cframes_bcast_fdri_write(void *opaque, hwaddr addr, uint64_t value,
103
+ unsigned size)
104
+{
105
+ XlnxVersalCFrameBcastReg *s = XLNX_VERSAL_CFRAME_BCAST_REG(opaque);
106
+ uint32_t wfifo[WFIFO_SZ];
107
+
108
+ if (update_wfifo(addr, value, s->wfifo, wfifo)) {
109
+ cframes_bcast_write(s, CFRAME_FDRI, wfifo);
110
+ }
111
+}
112
+
113
+static const MemoryRegionOps cframes_bcast_reg_reg_ops = {
114
+ .read = cframes_bcast_reg_read,
115
+ .write = cframes_bcast_reg_write,
116
+ .endianness = DEVICE_LITTLE_ENDIAN,
117
+ .valid = {
118
+ .min_access_size = 4,
119
+ .max_access_size = 4,
120
+ },
121
+};
122
+
123
+static const MemoryRegionOps cframes_bcast_reg_fdri_ops = {
124
+ .read = cframes_bcast_fdri_read,
125
+ .write = cframes_bcast_fdri_write,
126
+ .endianness = DEVICE_LITTLE_ENDIAN,
127
+ .valid = {
128
+ .min_access_size = 4,
129
+ .max_access_size = 4,
130
+ },
131
+};
132
+
133
static void cframe_reg_realize(DeviceState *dev, Error **errp)
134
{
135
XlnxVersalCFrameReg *s = XLNX_VERSAL_CFRAME_REG(dev);
136
@@ -XXX,XX +XXX,XX @@ static Property cframe_regs_props[] = {
137
DEFINE_PROP_END_OF_LIST(),
138
};
139
140
+static void cframe_bcast_reg_init(Object *obj)
141
+{
142
+ XlnxVersalCFrameBcastReg *s = XLNX_VERSAL_CFRAME_BCAST_REG(obj);
143
+ SysBusDevice *sbd = SYS_BUS_DEVICE(obj);
144
+
145
+ memory_region_init_io(&s->iomem_reg, obj, &cframes_bcast_reg_reg_ops, s,
146
+ TYPE_XLNX_VERSAL_CFRAME_BCAST_REG, KEYHOLE_STREAM_4K);
147
+ memory_region_init_io(&s->iomem_fdri, obj, &cframes_bcast_reg_fdri_ops, s,
148
+ TYPE_XLNX_VERSAL_CFRAME_BCAST_REG "-fdri",
149
+ KEYHOLE_STREAM_4K);
150
+ sysbus_init_mmio(sbd, &s->iomem_reg);
151
+ sysbus_init_mmio(sbd, &s->iomem_fdri);
152
+}
153
+
154
+static void cframe_bcast_reg_reset_enter(Object *obj, ResetType type)
155
+{
156
+ XlnxVersalCFrameBcastReg *s = XLNX_VERSAL_CFRAME_BCAST_REG(obj);
157
+
158
+ memset(s->wfifo, 0, WFIFO_SZ * sizeof(uint32_t));
159
+}
160
+
161
+static const VMStateDescription vmstate_cframe_bcast_reg = {
162
+ .name = TYPE_XLNX_VERSAL_CFRAME_BCAST_REG,
163
+ .version_id = 1,
164
+ .minimum_version_id = 1,
165
+ .fields = (VMStateField[]) {
166
+ VMSTATE_UINT32_ARRAY(wfifo, XlnxVersalCFrameBcastReg, 4),
167
+ VMSTATE_END_OF_LIST(),
168
+ }
169
+};
170
+
171
+static Property cframe_bcast_regs_props[] = {
172
+ DEFINE_PROP_LINK("cframe0", XlnxVersalCFrameBcastReg, cfg.cframe[0],
173
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
174
+ DEFINE_PROP_LINK("cframe1", XlnxVersalCFrameBcastReg, cfg.cframe[1],
175
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
176
+ DEFINE_PROP_LINK("cframe2", XlnxVersalCFrameBcastReg, cfg.cframe[2],
177
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
178
+ DEFINE_PROP_LINK("cframe3", XlnxVersalCFrameBcastReg, cfg.cframe[3],
179
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
180
+ DEFINE_PROP_LINK("cframe4", XlnxVersalCFrameBcastReg, cfg.cframe[4],
181
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
182
+ DEFINE_PROP_LINK("cframe5", XlnxVersalCFrameBcastReg, cfg.cframe[5],
183
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
184
+ DEFINE_PROP_LINK("cframe6", XlnxVersalCFrameBcastReg, cfg.cframe[6],
185
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
186
+ DEFINE_PROP_LINK("cframe7", XlnxVersalCFrameBcastReg, cfg.cframe[7],
187
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
188
+ DEFINE_PROP_LINK("cframe8", XlnxVersalCFrameBcastReg, cfg.cframe[8],
189
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
190
+ DEFINE_PROP_LINK("cframe9", XlnxVersalCFrameBcastReg, cfg.cframe[9],
191
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
192
+ DEFINE_PROP_LINK("cframe10", XlnxVersalCFrameBcastReg, cfg.cframe[10],
193
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
194
+ DEFINE_PROP_LINK("cframe11", XlnxVersalCFrameBcastReg, cfg.cframe[11],
195
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
196
+ DEFINE_PROP_LINK("cframe12", XlnxVersalCFrameBcastReg, cfg.cframe[12],
197
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
198
+ DEFINE_PROP_LINK("cframe13", XlnxVersalCFrameBcastReg, cfg.cframe[13],
199
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
200
+ DEFINE_PROP_LINK("cframe14", XlnxVersalCFrameBcastReg, cfg.cframe[14],
201
+ TYPE_XLNX_CFI_IF, XlnxCfiIf *),
202
+ DEFINE_PROP_END_OF_LIST(),
203
+};
204
+
205
static void cframe_reg_class_init(ObjectClass *klass, void *data)
206
{
207
ResettableClass *rc = RESETTABLE_CLASS(klass);
208
@@ -XXX,XX +XXX,XX @@ static void cframe_reg_class_init(ObjectClass *klass, void *data)
209
xcic->cfi_transfer_packet = cframe_reg_cfi_transfer_packet;
210
}
211
212
+static void cframe_bcast_reg_class_init(ObjectClass *klass, void *data)
213
+{
214
+ DeviceClass *dc = DEVICE_CLASS(klass);
215
+ ResettableClass *rc = RESETTABLE_CLASS(klass);
216
+
217
+ dc->vmsd = &vmstate_cframe_bcast_reg;
218
+ device_class_set_props(dc, cframe_bcast_regs_props);
219
+ rc->phases.enter = cframe_bcast_reg_reset_enter;
220
+}
221
+
222
static const TypeInfo cframe_reg_info = {
223
.name = TYPE_XLNX_VERSAL_CFRAME_REG,
224
.parent = TYPE_SYS_BUS_DEVICE,
225
@@ -XXX,XX +XXX,XX @@ static const TypeInfo cframe_reg_info = {
226
}
227
};
228
229
+static const TypeInfo cframe_bcast_reg_info = {
230
+ .name = TYPE_XLNX_VERSAL_CFRAME_BCAST_REG,
231
+ .parent = TYPE_SYS_BUS_DEVICE,
232
+ .instance_size = sizeof(XlnxVersalCFrameBcastReg),
233
+ .class_init = cframe_bcast_reg_class_init,
234
+ .instance_init = cframe_bcast_reg_init,
235
+};
236
+
237
static void cframe_reg_register_types(void)
238
{
239
type_register_static(&cframe_reg_info);
240
+ type_register_static(&cframe_bcast_reg_info);
241
}
242
243
type_init(cframe_reg_register_types)
244
--
245
2.34.1
diff view generated by jsdifflib
[PULL 17/26] hw/arm/xlnx-versal: Connect the CFU_APB, CFU_FDRO and CFU_SFR
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Connect the Configuration Frame Unit (CFU_APB, CFU_FDRO and CFU_SFR) to
4
the Versal machine.
5
6
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
7
Acked-by: Edgar E. Iglesias <edgar@zeroasic.com>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Message-id: 20230831165701.2016397-8-francisco.iglesias@amd.com
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
include/hw/arm/xlnx-versal.h | 16 ++++++++++++++
13
hw/arm/xlnx-versal.c | 42 ++++++++++++++++++++++++++++++++++++
14
2 files changed, 58 insertions(+)
15
16
diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h
17
index XXXXXXX..XXXXXXX 100644
18
--- a/include/hw/arm/xlnx-versal.h
19
+++ b/include/hw/arm/xlnx-versal.h
20
@@ -XXX,XX +XXX,XX @@
21
#include "hw/misc/xlnx-versal-crl.h"
22
#include "hw/misc/xlnx-versal-pmc-iou-slcr.h"
23
#include "hw/net/xlnx-versal-canfd.h"
24
+#include "hw/misc/xlnx-versal-cfu.h"
25
26
#define TYPE_XLNX_VERSAL "xlnx-versal"
27
OBJECT_DECLARE_SIMPLE_TYPE(Versal, XLNX_VERSAL)
28
@@ -XXX,XX +XXX,XX @@ struct Versal {
29
XlnxEFuse efuse;
30
XlnxVersalEFuseCtrl efuse_ctrl;
31
XlnxVersalEFuseCache efuse_cache;
32
+ XlnxVersalCFUAPB cfu_apb;
33
+ XlnxVersalCFUFDRO cfu_fdro;
34
+ XlnxVersalCFUSFR cfu_sfr;
35
36
OrIRQState apb_irq_orgate;
37
} pmc;
38
@@ -XXX,XX +XXX,XX @@ struct Versal {
39
#define VERSAL_GEM1_WAKE_IRQ_0 59
40
#define VERSAL_ADMA_IRQ_0 60
41
#define VERSAL_XRAM_IRQ_0 79
42
+#define VERSAL_CFU_IRQ_0 120
43
#define VERSAL_PMC_APB_IRQ 121
44
#define VERSAL_OSPI_IRQ 124
45
#define VERSAL_SD0_IRQ_0 126
46
@@ -XXX,XX +XXX,XX @@ struct Versal {
47
#define MM_PMC_EFUSE_CACHE 0xf1250000
48
#define MM_PMC_EFUSE_CACHE_SIZE 0x00C00
49
50
+#define MM_PMC_CFU_APB 0xf12b0000
51
+#define MM_PMC_CFU_APB_SIZE 0x10000
52
+#define MM_PMC_CFU_STREAM 0xf12c0000
53
+#define MM_PMC_CFU_STREAM_SIZE 0x1000
54
+#define MM_PMC_CFU_SFR 0xf12c1000
55
+#define MM_PMC_CFU_SFR_SIZE 0x1000
56
+#define MM_PMC_CFU_FDRO 0xf12c2000
57
+#define MM_PMC_CFU_FDRO_SIZE 0x1000
58
+#define MM_PMC_CFU_STREAM_2 0xf1f80000
59
+#define MM_PMC_CFU_STREAM_2_SIZE 0x40000
60
+
61
#define MM_PMC_CRP 0xf1260000U
62
#define MM_PMC_CRP_SIZE 0x10000
63
#define MM_PMC_RTC 0xf12a0000
64
diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c
65
index XXXXXXX..XXXXXXX 100644
66
--- a/hw/arm/xlnx-versal.c
67
+++ b/hw/arm/xlnx-versal.c
68
@@ -XXX,XX +XXX,XX @@ static void versal_create_ospi(Versal *s, qemu_irq *pic)
69
qdev_connect_gpio_out(orgate, 0, pic[VERSAL_OSPI_IRQ]);
70
}
71
72
+static void versal_create_cfu(Versal *s, qemu_irq *pic)
73
+{
74
+ SysBusDevice *sbd;
75
+
76
+ /* CFU FDRO */
77
+ object_initialize_child(OBJECT(s), "cfu-fdro", &s->pmc.cfu_fdro,
78
+ TYPE_XLNX_VERSAL_CFU_FDRO);
79
+ sbd = SYS_BUS_DEVICE(&s->pmc.cfu_fdro);
80
+
81
+ sysbus_realize(sbd, &error_fatal);
82
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_FDRO,
83
+ sysbus_mmio_get_region(sbd, 0));
84
+
85
+ /* CFU APB */
86
+ object_initialize_child(OBJECT(s), "cfu-apb", &s->pmc.cfu_apb,
87
+ TYPE_XLNX_VERSAL_CFU_APB);
88
+ sbd = SYS_BUS_DEVICE(&s->pmc.cfu_apb);
89
+
90
+ sysbus_realize(sbd, &error_fatal);
91
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_APB,
92
+ sysbus_mmio_get_region(sbd, 0));
93
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_STREAM,
94
+ sysbus_mmio_get_region(sbd, 1));
95
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_STREAM_2,
96
+ sysbus_mmio_get_region(sbd, 2));
97
+ sysbus_connect_irq(sbd, 0, pic[VERSAL_CFU_IRQ_0]);
98
+
99
+ /* CFU SFR */
100
+ object_initialize_child(OBJECT(s), "cfu-sfr", &s->pmc.cfu_sfr,
101
+ TYPE_XLNX_VERSAL_CFU_SFR);
102
+
103
+ sbd = SYS_BUS_DEVICE(&s->pmc.cfu_sfr);
104
+
105
+ object_property_set_link(OBJECT(&s->pmc.cfu_sfr),
106
+ "cfu", OBJECT(&s->pmc.cfu_apb), &error_abort);
107
+
108
+ sysbus_realize(sbd, &error_fatal);
109
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_SFR,
110
+ sysbus_mmio_get_region(sbd, 0));
111
+}
112
+
113
static void versal_create_crl(Versal *s, qemu_irq *pic)
114
{
115
SysBusDevice *sbd;
116
@@ -XXX,XX +XXX,XX @@ static void versal_realize(DeviceState *dev, Error **errp)
117
versal_create_pmc_iou_slcr(s, pic);
118
versal_create_ospi(s, pic);
119
versal_create_crl(s, pic);
120
+ versal_create_cfu(s, pic);
121
versal_map_ddr(s);
122
versal_unimp(s);
123
124
--
125
2.34.1
diff view generated by jsdifflib
[PULL 18/26] hw/arm/versal: Connect the CFRAME_REG and CFRAME_BCAST_REG
Deleted patch
1
From: Francisco Iglesias <francisco.iglesias@amd.com>
2
1
3
Connect the Configuration Frame controller (CFRAME_REG) and the
4
Configuration Frame broadcast controller (CFRAME_BCAST_REG) to the
5
Versal machine.
6
7
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Message-id: 20230831165701.2016397-9-francisco.iglesias@amd.com
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
include/hw/arm/xlnx-versal.h | 69 +++++++++++++++++++++
13
hw/arm/xlnx-versal.c | 113 ++++++++++++++++++++++++++++++++++-
14
2 files changed, 181 insertions(+), 1 deletion(-)
15
16
diff --git a/include/hw/arm/xlnx-versal.h b/include/hw/arm/xlnx-versal.h
17
index XXXXXXX..XXXXXXX 100644
18
--- a/include/hw/arm/xlnx-versal.h
19
+++ b/include/hw/arm/xlnx-versal.h
20
@@ -XXX,XX +XXX,XX @@
21
#include "hw/misc/xlnx-versal-pmc-iou-slcr.h"
22
#include "hw/net/xlnx-versal-canfd.h"
23
#include "hw/misc/xlnx-versal-cfu.h"
24
+#include "hw/misc/xlnx-versal-cframe-reg.h"
25
26
#define TYPE_XLNX_VERSAL "xlnx-versal"
27
OBJECT_DECLARE_SIMPLE_TYPE(Versal, XLNX_VERSAL)
28
@@ -XXX,XX +XXX,XX @@ OBJECT_DECLARE_SIMPLE_TYPE(Versal, XLNX_VERSAL)
29
#define XLNX_VERSAL_NR_IRQS 192
30
#define XLNX_VERSAL_NR_CANFD 2
31
#define XLNX_VERSAL_CANFD_REF_CLK (24 * 1000 * 1000)
32
+#define XLNX_VERSAL_NR_CFRAME 15
33
34
struct Versal {
35
/*< private >*/
36
@@ -XXX,XX +XXX,XX @@ struct Versal {
37
XlnxVersalCFUAPB cfu_apb;
38
XlnxVersalCFUFDRO cfu_fdro;
39
XlnxVersalCFUSFR cfu_sfr;
40
+ XlnxVersalCFrameReg cframe[XLNX_VERSAL_NR_CFRAME];
41
+ XlnxVersalCFrameBcastReg cframe_bcast;
42
43
OrIRQState apb_irq_orgate;
44
} pmc;
45
@@ -XXX,XX +XXX,XX @@ struct Versal {
46
#define MM_PMC_CFU_STREAM_2 0xf1f80000
47
#define MM_PMC_CFU_STREAM_2_SIZE 0x40000
48
49
+#define MM_PMC_CFRAME0_REG 0xf12d0000
50
+#define MM_PMC_CFRAME0_REG_SIZE 0x1000
51
+#define MM_PMC_CFRAME0_FDRI 0xf12d1000
52
+#define MM_PMC_CFRAME0_FDRI_SIZE 0x1000
53
+#define MM_PMC_CFRAME1_REG 0xf12d2000
54
+#define MM_PMC_CFRAME1_REG_SIZE 0x1000
55
+#define MM_PMC_CFRAME1_FDRI 0xf12d3000
56
+#define MM_PMC_CFRAME1_FDRI_SIZE 0x1000
57
+#define MM_PMC_CFRAME2_REG 0xf12d4000
58
+#define MM_PMC_CFRAME2_REG_SIZE 0x1000
59
+#define MM_PMC_CFRAME2_FDRI 0xf12d5000
60
+#define MM_PMC_CFRAME2_FDRI_SIZE 0x1000
61
+#define MM_PMC_CFRAME3_REG 0xf12d6000
62
+#define MM_PMC_CFRAME3_REG_SIZE 0x1000
63
+#define MM_PMC_CFRAME3_FDRI 0xf12d7000
64
+#define MM_PMC_CFRAME3_FDRI_SIZE 0x1000
65
+#define MM_PMC_CFRAME4_REG 0xf12d8000
66
+#define MM_PMC_CFRAME4_REG_SIZE 0x1000
67
+#define MM_PMC_CFRAME4_FDRI 0xf12d9000
68
+#define MM_PMC_CFRAME4_FDRI_SIZE 0x1000
69
+#define MM_PMC_CFRAME5_REG 0xf12da000
70
+#define MM_PMC_CFRAME5_REG_SIZE 0x1000
71
+#define MM_PMC_CFRAME5_FDRI 0xf12db000
72
+#define MM_PMC_CFRAME5_FDRI_SIZE 0x1000
73
+#define MM_PMC_CFRAME6_REG 0xf12dc000
74
+#define MM_PMC_CFRAME6_REG_SIZE 0x1000
75
+#define MM_PMC_CFRAME6_FDRI 0xf12dd000
76
+#define MM_PMC_CFRAME6_FDRI_SIZE 0x1000
77
+#define MM_PMC_CFRAME7_REG 0xf12de000
78
+#define MM_PMC_CFRAME7_REG_SIZE 0x1000
79
+#define MM_PMC_CFRAME7_FDRI 0xf12df000
80
+#define MM_PMC_CFRAME7_FDRI_SIZE 0x1000
81
+#define MM_PMC_CFRAME8_REG 0xf12e0000
82
+#define MM_PMC_CFRAME8_REG_SIZE 0x1000
83
+#define MM_PMC_CFRAME8_FDRI 0xf12e1000
84
+#define MM_PMC_CFRAME8_FDRI_SIZE 0x1000
85
+#define MM_PMC_CFRAME9_REG 0xf12e2000
86
+#define MM_PMC_CFRAME9_REG_SIZE 0x1000
87
+#define MM_PMC_CFRAME9_FDRI 0xf12e3000
88
+#define MM_PMC_CFRAME9_FDRI_SIZE 0x1000
89
+#define MM_PMC_CFRAME10_REG 0xf12e4000
90
+#define MM_PMC_CFRAME10_REG_SIZE 0x1000
91
+#define MM_PMC_CFRAME10_FDRI 0xf12e5000
92
+#define MM_PMC_CFRAME10_FDRI_SIZE 0x1000
93
+#define MM_PMC_CFRAME11_REG 0xf12e6000
94
+#define MM_PMC_CFRAME11_REG_SIZE 0x1000
95
+#define MM_PMC_CFRAME11_FDRI 0xf12e7000
96
+#define MM_PMC_CFRAME11_FDRI_SIZE 0x1000
97
+#define MM_PMC_CFRAME12_REG 0xf12e8000
98
+#define MM_PMC_CFRAME12_REG_SIZE 0x1000
99
+#define MM_PMC_CFRAME12_FDRI 0xf12e9000
100
+#define MM_PMC_CFRAME12_FDRI_SIZE 0x1000
101
+#define MM_PMC_CFRAME13_REG 0xf12ea000
102
+#define MM_PMC_CFRAME13_REG_SIZE 0x1000
103
+#define MM_PMC_CFRAME13_FDRI 0xf12eb000
104
+#define MM_PMC_CFRAME13_FDRI_SIZE 0x1000
105
+#define MM_PMC_CFRAME14_REG 0xf12ec000
106
+#define MM_PMC_CFRAME14_REG_SIZE 0x1000
107
+#define MM_PMC_CFRAME14_FDRI 0xf12ed000
108
+#define MM_PMC_CFRAME14_FDRI_SIZE 0x1000
109
+#define MM_PMC_CFRAME_BCAST_REG 0xf12ee000
110
+#define MM_PMC_CFRAME_BCAST_REG_SIZE 0x1000
111
+#define MM_PMC_CFRAME_BCAST_FDRI 0xf12ef000
112
+#define MM_PMC_CFRAME_BCAST_FDRI_SIZE 0x1000
113
+
114
#define MM_PMC_CRP 0xf1260000U
115
#define MM_PMC_CRP_SIZE 0x10000
116
#define MM_PMC_RTC 0xf12a0000
117
diff --git a/hw/arm/xlnx-versal.c b/hw/arm/xlnx-versal.c
118
index XXXXXXX..XXXXXXX 100644
119
--- a/hw/arm/xlnx-versal.c
120
+++ b/hw/arm/xlnx-versal.c
121
@@ -XXX,XX +XXX,XX @@
122
#define XLNX_VERSAL_RCPU_TYPE ARM_CPU_TYPE_NAME("cortex-r5f")
123
#define GEM_REVISION 0x40070106
124
125
-#define VERSAL_NUM_PMC_APB_IRQS 3
126
+#define VERSAL_NUM_PMC_APB_IRQS 18
127
#define NUM_OSPI_IRQ_LINES 3
128
129
static void versal_create_apu_cpus(Versal *s)
130
@@ -XXX,XX +XXX,XX @@ static void versal_create_pmc_apb_irq_orgate(Versal *s, qemu_irq *pic)
131
* - RTC
132
* - BBRAM
133
* - PMC SLCR
134
+ * - CFRAME regs (input 3 - 17 to the orgate)
135
*/
136
object_initialize_child(OBJECT(s), "pmc-apb-irq-orgate",
137
&s->pmc.apb_irq_orgate, TYPE_OR_IRQ);
138
@@ -XXX,XX +XXX,XX @@ static void versal_create_ospi(Versal *s, qemu_irq *pic)
139
static void versal_create_cfu(Versal *s, qemu_irq *pic)
140
{
141
SysBusDevice *sbd;
142
+ DeviceState *dev;
143
+ int i;
144
+ const struct {
145
+ uint64_t reg_base;
146
+ uint64_t fdri_base;
147
+ } cframe_addr[] = {
148
+ { MM_PMC_CFRAME0_REG, MM_PMC_CFRAME0_FDRI },
149
+ { MM_PMC_CFRAME1_REG, MM_PMC_CFRAME1_FDRI },
150
+ { MM_PMC_CFRAME2_REG, MM_PMC_CFRAME2_FDRI },
151
+ { MM_PMC_CFRAME3_REG, MM_PMC_CFRAME3_FDRI },
152
+ { MM_PMC_CFRAME4_REG, MM_PMC_CFRAME4_FDRI },
153
+ { MM_PMC_CFRAME5_REG, MM_PMC_CFRAME5_FDRI },
154
+ { MM_PMC_CFRAME6_REG, MM_PMC_CFRAME6_FDRI },
155
+ { MM_PMC_CFRAME7_REG, MM_PMC_CFRAME7_FDRI },
156
+ { MM_PMC_CFRAME8_REG, MM_PMC_CFRAME8_FDRI },
157
+ { MM_PMC_CFRAME9_REG, MM_PMC_CFRAME9_FDRI },
158
+ { MM_PMC_CFRAME10_REG, MM_PMC_CFRAME10_FDRI },
159
+ { MM_PMC_CFRAME11_REG, MM_PMC_CFRAME11_FDRI },
160
+ { MM_PMC_CFRAME12_REG, MM_PMC_CFRAME12_FDRI },
161
+ { MM_PMC_CFRAME13_REG, MM_PMC_CFRAME13_FDRI },
162
+ { MM_PMC_CFRAME14_REG, MM_PMC_CFRAME14_FDRI },
163
+ };
164
+ const struct {
165
+ uint32_t blktype0_frames;
166
+ uint32_t blktype1_frames;
167
+ uint32_t blktype2_frames;
168
+ uint32_t blktype3_frames;
169
+ uint32_t blktype4_frames;
170
+ uint32_t blktype5_frames;
171
+ uint32_t blktype6_frames;
172
+ } cframe_cfg[] = {
173
+ [0] = { 34111, 3528, 12800, 11, 5, 1, 1 },
174
+ [1] = { 38498, 3841, 15361, 13, 7, 3, 1 },
175
+ [2] = { 38498, 3841, 15361, 13, 7, 3, 1 },
176
+ [3] = { 38498, 3841, 15361, 13, 7, 3, 1 },
177
+ };
178
179
/* CFU FDRO */
180
object_initialize_child(OBJECT(s), "cfu-fdro", &s->pmc.cfu_fdro,
181
@@ -XXX,XX +XXX,XX @@ static void versal_create_cfu(Versal *s, qemu_irq *pic)
182
memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_FDRO,
183
sysbus_mmio_get_region(sbd, 0));
184
185
+ /* CFRAME REG */
186
+ for (i = 0; i < ARRAY_SIZE(s->pmc.cframe); i++) {
187
+ g_autofree char *name = g_strdup_printf("cframe%d", i);
188
+
189
+ object_initialize_child(OBJECT(s), name, &s->pmc.cframe[i],
190
+ TYPE_XLNX_VERSAL_CFRAME_REG);
191
+
192
+ sbd = SYS_BUS_DEVICE(&s->pmc.cframe[i]);
193
+ dev = DEVICE(&s->pmc.cframe[i]);
194
+
195
+ if (i < ARRAY_SIZE(cframe_cfg)) {
196
+ object_property_set_int(OBJECT(dev), "blktype0-frames",
197
+ cframe_cfg[i].blktype0_frames,
198
+ &error_abort);
199
+ object_property_set_int(OBJECT(dev), "blktype1-frames",
200
+ cframe_cfg[i].blktype1_frames,
201
+ &error_abort);
202
+ object_property_set_int(OBJECT(dev), "blktype2-frames",
203
+ cframe_cfg[i].blktype2_frames,
204
+ &error_abort);
205
+ object_property_set_int(OBJECT(dev), "blktype3-frames",
206
+ cframe_cfg[i].blktype3_frames,
207
+ &error_abort);
208
+ object_property_set_int(OBJECT(dev), "blktype4-frames",
209
+ cframe_cfg[i].blktype4_frames,
210
+ &error_abort);
211
+ object_property_set_int(OBJECT(dev), "blktype5-frames",
212
+ cframe_cfg[i].blktype5_frames,
213
+ &error_abort);
214
+ object_property_set_int(OBJECT(dev), "blktype6-frames",
215
+ cframe_cfg[i].blktype6_frames,
216
+ &error_abort);
217
+ }
218
+ object_property_set_link(OBJECT(dev), "cfu-fdro",
219
+ OBJECT(&s->pmc.cfu_fdro), &error_fatal);
220
+
221
+ sysbus_realize(SYS_BUS_DEVICE(dev), &error_fatal);
222
+
223
+ memory_region_add_subregion(&s->mr_ps, cframe_addr[i].reg_base,
224
+ sysbus_mmio_get_region(sbd, 0));
225
+ memory_region_add_subregion(&s->mr_ps, cframe_addr[i].fdri_base,
226
+ sysbus_mmio_get_region(sbd, 1));
227
+ sysbus_connect_irq(sbd, 0,
228
+ qdev_get_gpio_in(DEVICE(&s->pmc.apb_irq_orgate),
229
+ 3 + i));
230
+ }
231
+
232
+ /* CFRAME BCAST */
233
+ object_initialize_child(OBJECT(s), "cframe_bcast", &s->pmc.cframe_bcast,
234
+ TYPE_XLNX_VERSAL_CFRAME_BCAST_REG);
235
+
236
+ sbd = SYS_BUS_DEVICE(&s->pmc.cframe_bcast);
237
+ dev = DEVICE(&s->pmc.cframe_bcast);
238
+
239
+ for (i = 0; i < ARRAY_SIZE(s->pmc.cframe); i++) {
240
+ g_autofree char *propname = g_strdup_printf("cframe%d", i);
241
+ object_property_set_link(OBJECT(dev), propname,
242
+ OBJECT(&s->pmc.cframe[i]), &error_fatal);
243
+ }
244
+
245
+ sysbus_realize(sbd, &error_fatal);
246
+
247
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFRAME_BCAST_REG,
248
+ sysbus_mmio_get_region(sbd, 0));
249
+ memory_region_add_subregion(&s->mr_ps, MM_PMC_CFRAME_BCAST_FDRI,
250
+ sysbus_mmio_get_region(sbd, 1));
251
+
252
/* CFU APB */
253
object_initialize_child(OBJECT(s), "cfu-apb", &s->pmc.cfu_apb,
254
TYPE_XLNX_VERSAL_CFU_APB);
255
sbd = SYS_BUS_DEVICE(&s->pmc.cfu_apb);
256
+ dev = DEVICE(&s->pmc.cfu_apb);
257
+
258
+ for (i = 0; i < ARRAY_SIZE(s->pmc.cframe); i++) {
259
+ g_autofree char *propname = g_strdup_printf("cframe%d", i);
260
+ object_property_set_link(OBJECT(dev), propname,
261
+ OBJECT(&s->pmc.cframe[i]), &error_fatal);
262
+ }
263
264
sysbus_realize(sbd, &error_fatal);
265
memory_region_add_subregion(&s->mr_ps, MM_PMC_CFU_APB,
266
--
267
2.34.1
diff view generated by jsdifflib
[PULL 19/26] target/arm: Do not use gen_mte_checkN in trans_STGP
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
STGP writes to tag memory, it does not check it.
4
This happened to work because we wrote tag memory first
5
so that the check always succeeded.
6
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
8
Message-id: 20230901203103.136408-1-richard.henderson@linaro.org
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
target/arm/tcg/translate-a64.c | 41 +++++++++++++---------------------
13
1 file changed, 15 insertions(+), 26 deletions(-)
14
15
diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c
16
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/tcg/translate-a64.c
18
+++ b/target/arm/tcg/translate-a64.c
19
@@ -XXX,XX +XXX,XX @@ static bool trans_STGP(DisasContext *s, arg_ldstpair *a)
20
tcg_gen_addi_i64(dirty_addr, dirty_addr, offset);
21
}
22
23
- if (!s->ata) {
24
- /*
25
- * TODO: We could rely on the stores below, at least for
26
- * system mode, if we arrange to add MO_ALIGN_16.
27
- */
28
- gen_helper_stg_stub(cpu_env, dirty_addr);
29
- } else if (tb_cflags(s->base.tb) & CF_PARALLEL) {
30
- gen_helper_stg_parallel(cpu_env, dirty_addr, dirty_addr);
31
- } else {
32
- gen_helper_stg(cpu_env, dirty_addr, dirty_addr);
33
- }
34
-
35
- mop = finalize_memop(s, MO_64);
36
- clean_addr = gen_mte_checkN(s, dirty_addr, true, false, 2 << MO_64, mop);
37
-
38
+ clean_addr = clean_data_tbi(s, dirty_addr);
39
tcg_rt = cpu_reg(s, a->rt);
40
tcg_rt2 = cpu_reg(s, a->rt2);
41
42
/*
43
- * STGP is defined as two 8-byte memory operations and one tag operation.
44
- * We implement it as one single 16-byte memory operation for convenience.
45
- * Rebuild mop as for STP.
46
- * TODO: The atomicity with LSE2 is stronger than required.
47
- * Need a form of MO_ATOM_WITHIN16_PAIR that never requires
48
- * 16-byte atomicity.
49
+ * STGP is defined as two 8-byte memory operations, aligned to TAG_GRANULE,
50
+ * and one tag operation. We implement it as one single aligned 16-byte
51
+ * memory operation for convenience. Note that the alignment ensures
52
+ * MO_ATOM_IFALIGN_PAIR produces 8-byte atomicity for the memory store.
53
*/
54
- mop = MO_128;
55
- if (s->align_mem) {
56
- mop |= MO_ALIGN_8;
57
- }
58
- mop = finalize_memop_pair(s, mop);
59
+ mop = finalize_memop_atom(s, MO_128 | MO_ALIGN, MO_ATOM_IFALIGN_PAIR);
60
61
tmp = tcg_temp_new_i128();
62
if (s->be_data == MO_LE) {
63
@@ -XXX,XX +XXX,XX @@ static bool trans_STGP(DisasContext *s, arg_ldstpair *a)
64
}
65
tcg_gen_qemu_st_i128(tmp, clean_addr, get_mem_index(s), mop);
66
67
+ /* Perform the tag store, if tag access enabled. */
68
+ if (s->ata) {
69
+ if (tb_cflags(s->base.tb) & CF_PARALLEL) {
70
+ gen_helper_stg_parallel(cpu_env, dirty_addr, dirty_addr);
71
+ } else {
72
+ gen_helper_stg(cpu_env, dirty_addr, dirty_addr);
73
+ }
74
+ }
75
+
76
op_addr_ldstpair_post(s, a, dirty_addr, offset);
77
return true;
78
}
79
--
80
2.34.1
diff view generated by jsdifflib
[PULL 20/26] arm64: Restore trapless ptimer access
Deleted patch
1
From: Colton Lewis <coltonlewis@google.com>
2
1
3
Due to recent KVM changes, QEMU is setting a ptimer offset resulting
4
in unintended trap and emulate access and a consequent performance
5
hit. Filter out the PTIMER_CNT register to restore trapless ptimer
6
access.
7
8
Quoting Andrew Jones:
9
10
Simply reading the CNT register and writing back the same value is
11
enough to set an offset, since the timer will have certainly moved
12
past whatever value was read by the time it's written. QEMU
13
frequently saves and restores all registers in the get-reg-list array,
14
unless they've been explicitly filtered out (with Linux commit
15
680232a94c12, KVM_REG_ARM_PTIMER_CNT is now in the array). So, to
16
restore trapless ptimer accesses, we need a QEMU patch to filter out
17
the register.
18
19
See
20
https://lore.kernel.org/kvmarm/gsntttsonus5.fsf@coltonlewis-kvm.c.googlers.com/T/#m0770023762a821db2a3f0dd0a7dc6aa54e0d0da9
21
for additional context.
22
23
Cc: qemu-stable@nongnu.org
24
Signed-off-by: Andrew Jones <andrew.jones@linux.dev>
25
Signed-off-by: Colton Lewis <coltonlewis@google.com>
26
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
27
Tested-by: Colton Lewis <coltonlewis@google.com>
28
Message-id: 20230831190052.129045-1-coltonlewis@google.com
29
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
30
---
31
target/arm/kvm64.c | 1 +
32
1 file changed, 1 insertion(+)
33
34
diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
35
index XXXXXXX..XXXXXXX 100644
36
--- a/target/arm/kvm64.c
37
+++ b/target/arm/kvm64.c
38
@@ -XXX,XX +XXX,XX @@ typedef struct CPRegStateLevel {
39
*/
40
static const CPRegStateLevel non_runtime_cpregs[] = {
41
{ KVM_REG_ARM_TIMER_CNT, KVM_PUT_FULL_STATE },
42
+ { KVM_REG_ARM_PTIMER_CNT, KVM_PUT_FULL_STATE },
43
};
44
45
int kvm_arm_cpreg_level(uint64_t regidx)
46
--
47
2.34.1
diff view generated by jsdifflib
[PULL 21/26] target/arm: Implement RMR_ELx
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
Provide a stub implementation, as a write is a "request".
4
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Message-id: 20230831232441.66020-2-richard.henderson@linaro.org
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
---
10
target/arm/helper.c | 64 +++++++++++++++++++++++++++++----------------
11
1 file changed, 41 insertions(+), 23 deletions(-)
12
13
diff --git a/target/arm/helper.c b/target/arm/helper.c
14
index XXXXXXX..XXXXXXX 100644
15
--- a/target/arm/helper.c
16
+++ b/target/arm/helper.c
17
@@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu)
18
};
19
modify_arm_cp_regs(v8_idregs, v8_user_idregs);
20
#endif
21
- /* RVBAR_EL1 is only implemented if EL1 is the highest EL */
22
+ /*
23
+ * RVBAR_EL1 and RMR_EL1 only implemented if EL1 is the highest EL.
24
+ * TODO: For RMR, a write with bit 1 set should do something with
25
+ * cpu_reset(). In the meantime, "the bit is strictly a request",
26
+ * so we are in spec just ignoring writes.
27
+ */
28
if (!arm_feature(env, ARM_FEATURE_EL3) &&
29
!arm_feature(env, ARM_FEATURE_EL2)) {
30
- ARMCPRegInfo rvbar = {
31
- .name = "RVBAR_EL1", .state = ARM_CP_STATE_BOTH,
32
- .opc0 = 3, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1,
33
- .access = PL1_R,
34
- .fieldoffset = offsetof(CPUARMState, cp15.rvbar),
35
+ ARMCPRegInfo el1_reset_regs[] = {
36
+ { .name = "RVBAR_EL1", .state = ARM_CP_STATE_BOTH,
37
+ .opc0 = 3, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1,
38
+ .access = PL1_R,
39
+ .fieldoffset = offsetof(CPUARMState, cp15.rvbar) },
40
+ { .name = "RMR_EL1", .state = ARM_CP_STATE_BOTH,
41
+ .opc0 = 3, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 2,
42
+ .access = PL1_RW, .type = ARM_CP_CONST,
43
+ .resetvalue = arm_feature(env, ARM_FEATURE_AARCH64) }
44
};
45
- define_one_arm_cp_reg(cpu, &rvbar);
46
+ define_arm_cp_regs(cpu, el1_reset_regs);
47
}
48
define_arm_cp_regs(cpu, v8_idregs);
49
define_arm_cp_regs(cpu, v8_cp_reginfo);
50
@@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu)
51
if (cpu_isar_feature(aa64_sel2, cpu)) {
52
define_arm_cp_regs(cpu, el2_sec_cp_reginfo);
53
}
54
- /* RVBAR_EL2 is only implemented if EL2 is the highest EL */
55
+ /*
56
+ * RVBAR_EL2 and RMR_EL2 only implemented if EL2 is the highest EL.
57
+ * See commentary near RMR_EL1.
58
+ */
59
if (!arm_feature(env, ARM_FEATURE_EL3)) {
60
- ARMCPRegInfo rvbar[] = {
61
- {
62
- .name = "RVBAR_EL2", .state = ARM_CP_STATE_AA64,
63
- .opc0 = 3, .opc1 = 4, .crn = 12, .crm = 0, .opc2 = 1,
64
- .access = PL2_R,
65
- .fieldoffset = offsetof(CPUARMState, cp15.rvbar),
66
- },
67
- { .name = "RVBAR", .type = ARM_CP_ALIAS,
68
- .cp = 15, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1,
69
- .access = PL2_R,
70
- .fieldoffset = offsetof(CPUARMState, cp15.rvbar),
71
- },
72
+ static const ARMCPRegInfo el2_reset_regs[] = {
73
+ { .name = "RVBAR_EL2", .state = ARM_CP_STATE_AA64,
74
+ .opc0 = 3, .opc1 = 4, .crn = 12, .crm = 0, .opc2 = 1,
75
+ .access = PL2_R,
76
+ .fieldoffset = offsetof(CPUARMState, cp15.rvbar) },
77
+ { .name = "RVBAR", .type = ARM_CP_ALIAS,
78
+ .cp = 15, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 1,
79
+ .access = PL2_R,
80
+ .fieldoffset = offsetof(CPUARMState, cp15.rvbar) },
81
+ { .name = "RMR_EL2", .state = ARM_CP_STATE_AA64,
82
+ .opc0 = 3, .opc1 = 4, .crn = 12, .crm = 0, .opc2 = 2,
83
+ .access = PL2_RW, .type = ARM_CP_CONST, .resetvalue = 1 },
84
};
85
- define_arm_cp_regs(cpu, rvbar);
86
+ define_arm_cp_regs(cpu, el2_reset_regs);
87
}
88
}
89
90
@@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu)
91
{ .name = "RVBAR_EL3", .state = ARM_CP_STATE_AA64,
92
.opc0 = 3, .opc1 = 6, .crn = 12, .crm = 0, .opc2 = 1,
93
.access = PL3_R,
94
- .fieldoffset = offsetof(CPUARMState, cp15.rvbar),
95
- },
96
+ .fieldoffset = offsetof(CPUARMState, cp15.rvbar), },
97
+ { .name = "RMR_EL3", .state = ARM_CP_STATE_AA64,
98
+ .opc0 = 3, .opc1 = 6, .crn = 12, .crm = 0, .opc2 = 2,
99
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 1 },
100
+ { .name = "RMR", .state = ARM_CP_STATE_AA32,
101
+ .cp = 15, .opc1 = 0, .crn = 12, .crm = 0, .opc2 = 2,
102
+ .access = PL3_RW, .type = ARM_CP_CONST,
103
+ .resetvalue = arm_feature(env, ARM_FEATURE_AARCH64) },
104
{ .name = "SCTLR_EL3", .state = ARM_CP_STATE_AA64,
105
.opc0 = 3, .opc1 = 6, .crn = 1, .crm = 0, .opc2 = 0,
106
.access = PL3_RW,
107
--
108
2.34.1
diff view generated by jsdifflib
[PULL 22/26] target/arm: Implement cortex-a710
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
The cortex-a710 is a first generation ARMv9.0-A processor.
4
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Message-id: 20230831232441.66020-3-richard.henderson@linaro.org
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
---
10
docs/system/arm/virt.rst | 1 +
11
hw/arm/virt.c | 1 +
12
target/arm/tcg/cpu64.c | 212 +++++++++++++++++++++++++++++++++++++++
13
3 files changed, 214 insertions(+)
14
15
diff --git a/docs/system/arm/virt.rst b/docs/system/arm/virt.rst
16
index XXXXXXX..XXXXXXX 100644
17
--- a/docs/system/arm/virt.rst
18
+++ b/docs/system/arm/virt.rst
19
@@ -XXX,XX +XXX,XX @@ Supported guest CPU types:
20
- ``cortex-a57`` (64-bit)
21
- ``cortex-a72`` (64-bit)
22
- ``cortex-a76`` (64-bit)
23
+- ``cortex-a710`` (64-bit)
24
- ``a64fx`` (64-bit)
25
- ``host`` (with KVM only)
26
- ``neoverse-n1`` (64-bit)
27
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
28
index XXXXXXX..XXXXXXX 100644
29
--- a/hw/arm/virt.c
30
+++ b/hw/arm/virt.c
31
@@ -XXX,XX +XXX,XX @@ static const char *valid_cpus[] = {
32
ARM_CPU_TYPE_NAME("cortex-a55"),
33
ARM_CPU_TYPE_NAME("cortex-a72"),
34
ARM_CPU_TYPE_NAME("cortex-a76"),
35
+ ARM_CPU_TYPE_NAME("cortex-a710"),
36
ARM_CPU_TYPE_NAME("a64fx"),
37
ARM_CPU_TYPE_NAME("neoverse-n1"),
38
ARM_CPU_TYPE_NAME("neoverse-v1"),
39
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
40
index XXXXXXX..XXXXXXX 100644
41
--- a/target/arm/tcg/cpu64.c
42
+++ b/target/arm/tcg/cpu64.c
43
@@ -XXX,XX +XXX,XX @@ static void aarch64_neoverse_v1_initfn(Object *obj)
44
aarch64_add_sve_properties(obj);
45
}
46
47
+static const ARMCPRegInfo cortex_a710_cp_reginfo[] = {
48
+ { .name = "CPUACTLR_EL1", .state = ARM_CP_STATE_AA64,
49
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 1, .opc2 = 0,
50
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
51
+ .accessfn = access_actlr_w },
52
+ { .name = "CPUACTLR2_EL1", .state = ARM_CP_STATE_AA64,
53
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 1, .opc2 = 1,
54
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
55
+ .accessfn = access_actlr_w },
56
+ { .name = "CPUACTLR3_EL1", .state = ARM_CP_STATE_AA64,
57
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 1, .opc2 = 2,
58
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
59
+ .accessfn = access_actlr_w },
60
+ { .name = "CPUACTLR4_EL1", .state = ARM_CP_STATE_AA64,
61
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 1, .opc2 = 3,
62
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
63
+ .accessfn = access_actlr_w },
64
+ { .name = "CPUECTLR_EL1", .state = ARM_CP_STATE_AA64,
65
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 1, .opc2 = 4,
66
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
67
+ .accessfn = access_actlr_w },
68
+ { .name = "CPUECTLR2_EL1", .state = ARM_CP_STATE_AA64,
69
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 1, .opc2 = 5,
70
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
71
+ .accessfn = access_actlr_w },
72
+ { .name = "CPUPPMCR_EL3", .state = ARM_CP_STATE_AA64,
73
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 2, .opc2 = 4,
74
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
75
+ { .name = "CPUPWRCTLR_EL1", .state = ARM_CP_STATE_AA64,
76
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 2, .opc2 = 7,
77
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
78
+ .accessfn = access_actlr_w },
79
+ { .name = "ATCR_EL1", .state = ARM_CP_STATE_AA64,
80
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 7, .opc2 = 0,
81
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
82
+ { .name = "CPUACTLR5_EL1", .state = ARM_CP_STATE_AA64,
83
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 8, .opc2 = 0,
84
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
85
+ .accessfn = access_actlr_w },
86
+ { .name = "CPUACTLR6_EL1", .state = ARM_CP_STATE_AA64,
87
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 8, .opc2 = 1,
88
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
89
+ .accessfn = access_actlr_w },
90
+ { .name = "CPUACTLR7_EL1", .state = ARM_CP_STATE_AA64,
91
+ .opc0 = 3, .opc1 = 0, .crn = 15, .crm = 8, .opc2 = 2,
92
+ .access = PL1_RW, .type = ARM_CP_CONST, .resetvalue = 0,
93
+ .accessfn = access_actlr_w },
94
+ { .name = "ATCR_EL2", .state = ARM_CP_STATE_AA64,
95
+ .opc0 = 3, .opc1 = 4, .crn = 15, .crm = 7, .opc2 = 0,
96
+ .access = PL2_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
97
+ { .name = "AVTCR_EL2", .state = ARM_CP_STATE_AA64,
98
+ .opc0 = 3, .opc1 = 4, .crn = 15, .crm = 7, .opc2 = 1,
99
+ .access = PL2_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
100
+ { .name = "CPUPPMCR_EL3", .state = ARM_CP_STATE_AA64,
101
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 2, .opc2 = 0,
102
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
103
+ { .name = "CPUPPMCR2_EL3", .state = ARM_CP_STATE_AA64,
104
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 2, .opc2 = 1,
105
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
106
+ { .name = "CPUPPMCR4_EL3", .state = ARM_CP_STATE_AA64,
107
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 2, .opc2 = 4,
108
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
109
+ { .name = "CPUPPMCR5_EL3", .state = ARM_CP_STATE_AA64,
110
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 2, .opc2 = 5,
111
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
112
+ { .name = "CPUPPMCR6_EL3", .state = ARM_CP_STATE_AA64,
113
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 2, .opc2 = 6,
114
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
115
+ { .name = "CPUACTLR_EL3", .state = ARM_CP_STATE_AA64,
116
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 4, .opc2 = 0,
117
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
118
+ { .name = "ATCR_EL3", .state = ARM_CP_STATE_AA64,
119
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 7, .opc2 = 0,
120
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
121
+ { .name = "CPUPSELR_EL3", .state = ARM_CP_STATE_AA64,
122
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 0,
123
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
124
+ { .name = "CPUPCR_EL3", .state = ARM_CP_STATE_AA64,
125
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 1,
126
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
127
+ { .name = "CPUPOR_EL3", .state = ARM_CP_STATE_AA64,
128
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 2,
129
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
130
+ { .name = "CPUPMR_EL3", .state = ARM_CP_STATE_AA64,
131
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 3,
132
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
133
+ { .name = "CPUPOR2_EL3", .state = ARM_CP_STATE_AA64,
134
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 4,
135
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
136
+ { .name = "CPUPMR2_EL3", .state = ARM_CP_STATE_AA64,
137
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 5,
138
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
139
+ { .name = "CPUPFR_EL3", .state = ARM_CP_STATE_AA64,
140
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 8, .opc2 = 6,
141
+ .access = PL3_RW, .type = ARM_CP_CONST, .resetvalue = 0 },
142
+
143
+ /*
144
+ * Stub RAMINDEX, as we don't actually implement caches, BTB,
145
+ * or anything else with cpu internal memory.
146
+ * "Read" zeros into the IDATA* and DDATA* output registers.
147
+ */
148
+ { .name = "RAMINDEX_EL3", .state = ARM_CP_STATE_AA64,
149
+ .opc0 = 1, .opc1 = 6, .crn = 15, .crm = 0, .opc2 = 0,
150
+ .access = PL3_W, .type = ARM_CP_CONST, .resetvalue = 0 },
151
+ { .name = "IDATA0_EL3", .state = ARM_CP_STATE_AA64,
152
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 0, .opc2 = 0,
153
+ .access = PL3_R, .type = ARM_CP_CONST, .resetvalue = 0 },
154
+ { .name = "IDATA1_EL3", .state = ARM_CP_STATE_AA64,
155
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 0, .opc2 = 1,
156
+ .access = PL3_R, .type = ARM_CP_CONST, .resetvalue = 0 },
157
+ { .name = "IDATA2_EL3", .state = ARM_CP_STATE_AA64,
158
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 0, .opc2 = 2,
159
+ .access = PL3_R, .type = ARM_CP_CONST, .resetvalue = 0 },
160
+ { .name = "DDATA0_EL3", .state = ARM_CP_STATE_AA64,
161
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 1, .opc2 = 0,
162
+ .access = PL3_R, .type = ARM_CP_CONST, .resetvalue = 0 },
163
+ { .name = "DDATA1_EL3", .state = ARM_CP_STATE_AA64,
164
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 1, .opc2 = 1,
165
+ .access = PL3_R, .type = ARM_CP_CONST, .resetvalue = 0 },
166
+ { .name = "DDATA2_EL3", .state = ARM_CP_STATE_AA64,
167
+ .opc0 = 3, .opc1 = 6, .crn = 15, .crm = 1, .opc2 = 2,
168
+ .access = PL3_R, .type = ARM_CP_CONST, .resetvalue = 0 },
169
+};
170
+
171
+static void aarch64_a710_initfn(Object *obj)
172
+{
173
+ ARMCPU *cpu = ARM_CPU(obj);
174
+
175
+ cpu->dtb_compatible = "arm,cortex-a710";
176
+ set_feature(&cpu->env, ARM_FEATURE_V8);
177
+ set_feature(&cpu->env, ARM_FEATURE_NEON);
178
+ set_feature(&cpu->env, ARM_FEATURE_GENERIC_TIMER);
179
+ set_feature(&cpu->env, ARM_FEATURE_AARCH64);
180
+ set_feature(&cpu->env, ARM_FEATURE_CBAR_RO);
181
+ set_feature(&cpu->env, ARM_FEATURE_EL2);
182
+ set_feature(&cpu->env, ARM_FEATURE_EL3);
183
+ set_feature(&cpu->env, ARM_FEATURE_PMU);
184
+
185
+ /* Ordered by Section B.4: AArch64 registers */
186
+ cpu->midr = 0x412FD471; /* r2p1 */
187
+ cpu->revidr = 0;
188
+ cpu->isar.id_pfr0 = 0x21110131;
189
+ cpu->isar.id_pfr1 = 0x00010000; /* GIC filled in later */
190
+ cpu->isar.id_dfr0 = 0x16011099;
191
+ cpu->id_afr0 = 0;
192
+ cpu->isar.id_mmfr0 = 0x10201105;
193
+ cpu->isar.id_mmfr1 = 0x40000000;
194
+ cpu->isar.id_mmfr2 = 0x01260000;
195
+ cpu->isar.id_mmfr3 = 0x02122211;
196
+ cpu->isar.id_isar0 = 0x02101110;
197
+ cpu->isar.id_isar1 = 0x13112111;
198
+ cpu->isar.id_isar2 = 0x21232042;
199
+ cpu->isar.id_isar3 = 0x01112131;
200
+ cpu->isar.id_isar4 = 0x00010142;
201
+ cpu->isar.id_isar5 = 0x11011121; /* with Crypto */
202
+ cpu->isar.id_mmfr4 = 0x21021110;
203
+ cpu->isar.id_isar6 = 0x01111111;
204
+ cpu->isar.mvfr0 = 0x10110222;
205
+ cpu->isar.mvfr1 = 0x13211111;
206
+ cpu->isar.mvfr2 = 0x00000043;
207
+ cpu->isar.id_pfr2 = 0x00000011;
208
+ cpu->isar.id_aa64pfr0 = 0x1201111120111112ull; /* GIC filled in later */
209
+ cpu->isar.id_aa64pfr1 = 0x0000000000000221ull;
210
+ cpu->isar.id_aa64zfr0 = 0x0000110100110021ull; /* with Crypto */
211
+ cpu->isar.id_aa64dfr0 = 0x000011f010305611ull;
212
+ cpu->isar.id_aa64dfr1 = 0;
213
+ cpu->id_aa64afr0 = 0;
214
+ cpu->id_aa64afr1 = 0;
215
+ cpu->isar.id_aa64isar0 = 0x0221111110212120ull; /* with Crypto */
216
+ cpu->isar.id_aa64isar1 = 0x0010111101211032ull;
217
+ cpu->isar.id_aa64mmfr0 = 0x0000022200101122ull;
218
+ cpu->isar.id_aa64mmfr1 = 0x0000000010212122ull;
219
+ cpu->isar.id_aa64mmfr2 = 0x1221011110101011ull;
220
+ cpu->clidr = 0x0000001482000023ull;
221
+ cpu->gm_blocksize = 4;
222
+ cpu->ctr = 0x000000049444c004ull;
223
+ cpu->dcz_blocksize = 4;
224
+ /* TODO FEAT_MPAM: mpamidr_el1 = 0x0000_0001_0006_003f */
225
+
226
+ /* Section B.5.2: PMCR_EL0 */
227
+ cpu->isar.reset_pmcr_el0 = 0xa000; /* with 20 counters */
228
+
229
+ /* Section B.6.7: ICH_VTR_EL2 */
230
+ cpu->gic_num_lrs = 4;
231
+ cpu->gic_vpribits = 5;
232
+ cpu->gic_vprebits = 5;
233
+ cpu->gic_pribits = 5;
234
+
235
+ /* Section 14: Scalable Vector Extensions support */
236
+ cpu->sve_vq.supported = 1 << 0; /* 128bit */
237
+
238
+ /*
239
+ * The cortex-a710 TRM does not list CCSIDR values. The layout of
240
+ * the caches are in text in Table 7-1, Table 8-1, and Table 9-1.
241
+ *
242
+ * L1: 4-way set associative 64-byte line size, total either 32K or 64K.
243
+ * L2: 8-way set associative 64 byte line size, total either 256K or 512K.
244
+ */
245
+ cpu->ccsidr[0] = make_ccsidr64(4, 64, 64 * KiB); /* L1 dcache */
246
+ cpu->ccsidr[1] = cpu->ccsidr[0]; /* L1 icache */
247
+ cpu->ccsidr[2] = make_ccsidr64(8, 64, 512 * KiB); /* L2 cache */
248
+
249
+ /* FIXME: Not documented -- copied from neoverse-v1 */
250
+ cpu->reset_sctlr = 0x30c50838;
251
+
252
+ define_arm_cp_regs(cpu, cortex_a710_cp_reginfo);
253
+
254
+ aarch64_add_pauth_properties(obj);
255
+ aarch64_add_sve_properties(obj);
256
+}
257
+
258
/*
259
* -cpu max: a CPU with as many features enabled as our emulation supports.
260
* The version of '-cpu max' for qemu-system-arm is defined in cpu32.c;
261
@@ -XXX,XX +XXX,XX @@ static const ARMCPUInfo aarch64_cpus[] = {
262
{ .name = "cortex-a55", .initfn = aarch64_a55_initfn },
263
{ .name = "cortex-a72", .initfn = aarch64_a72_initfn },
264
{ .name = "cortex-a76", .initfn = aarch64_a76_initfn },
265
+ { .name = "cortex-a710", .initfn = aarch64_a710_initfn },
266
{ .name = "a64fx", .initfn = aarch64_a64fx_initfn },
267
{ .name = "neoverse-n1", .initfn = aarch64_neoverse_n1_initfn },
268
{ .name = "neoverse-v1", .initfn = aarch64_neoverse_v1_initfn },
269
--
270
2.34.1
diff view generated by jsdifflib
[PULL 23/26] target/arm: Implement HCR_EL2.TIDCP
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
Perform the check for EL2 enabled in the security space and the
4
TIDCP bit in an out-of-line helper.
5
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Message-id: 20230831232441.66020-4-richard.henderson@linaro.org
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
11
target/arm/helper.h | 1 +
12
target/arm/tcg/op_helper.c | 13 +++++++++++++
13
target/arm/tcg/translate-a64.c | 16 ++++++++++++++--
14
target/arm/tcg/translate.c | 27 +++++++++++++++++++++++++++
15
4 files changed, 55 insertions(+), 2 deletions(-)
16
17
diff --git a/target/arm/helper.h b/target/arm/helper.h
18
index XXXXXXX..XXXXXXX 100644
19
--- a/target/arm/helper.h
20
+++ b/target/arm/helper.h
21
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(check_bxj_trap, TCG_CALL_NO_WG, void, env, i32)
22
23
DEF_HELPER_4(access_check_cp_reg, cptr, env, i32, i32, i32)
24
DEF_HELPER_FLAGS_2(lookup_cp_reg, TCG_CALL_NO_RWG_SE, cptr, env, i32)
25
+DEF_HELPER_FLAGS_2(tidcp_el1, TCG_CALL_NO_WG, void, env, i32)
26
DEF_HELPER_3(set_cp_reg, void, env, cptr, i32)
27
DEF_HELPER_2(get_cp_reg, i32, env, cptr)
28
DEF_HELPER_3(set_cp_reg64, void, env, cptr, i64)
29
diff --git a/target/arm/tcg/op_helper.c b/target/arm/tcg/op_helper.c
30
index XXXXXXX..XXXXXXX 100644
31
--- a/target/arm/tcg/op_helper.c
32
+++ b/target/arm/tcg/op_helper.c
33
@@ -XXX,XX +XXX,XX @@ const void *HELPER(lookup_cp_reg)(CPUARMState *env, uint32_t key)
34
return ri;
35
}
36
37
+/*
38
+ * Test for HCR_EL2.TIDCP at EL1.
39
+ * Since implementation defined registers are rare, and within QEMU
40
+ * most of them are no-op, do not waste HFLAGS space for this and
41
+ * always use a helper.
42
+ */
43
+void HELPER(tidcp_el1)(CPUARMState *env, uint32_t syndrome)
44
+{
45
+ if (arm_hcr_el2_eff(env) & HCR_TIDCP) {
46
+ raise_exception_ra(env, EXCP_UDEF, syndrome, 2, GETPC());
47
+ }
48
+}
49
+
50
void HELPER(set_cp_reg)(CPUARMState *env, const void *rip, uint32_t value)
51
{
52
const ARMCPRegInfo *ri = rip;
53
diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c
54
index XXXXXXX..XXXXXXX 100644
55
--- a/target/arm/tcg/translate-a64.c
56
+++ b/target/arm/tcg/translate-a64.c
57
@@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, bool isread,
58
bool need_exit_tb = false;
59
TCGv_ptr tcg_ri = NULL;
60
TCGv_i64 tcg_rt;
61
+ uint32_t syndrome;
62
+
63
+ if (crn == 11 || crn == 15) {
64
+ /*
65
+ * Check for TIDCP trap, which must take precedence over
66
+ * the UNDEF for "no such register" etc.
67
+ */
68
+ syndrome = syn_aa64_sysregtrap(op0, op1, op2, crn, crm, rt, isread);
69
+ switch (s->current_el) {
70
+ case 1:
71
+ gen_helper_tidcp_el1(cpu_env, tcg_constant_i32(syndrome));
72
+ break;
73
+ }
74
+ }
75
76
if (!ri) {
77
/* Unknown register; this might be a guest error or a QEMU
78
@@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, bool isread,
79
/* Emit code to perform further access permissions checks at
80
* runtime; this may result in an exception.
81
*/
82
- uint32_t syndrome;
83
-
84
syndrome = syn_aa64_sysregtrap(op0, op1, op2, crn, crm, rt, isread);
85
gen_a64_update_pc(s, 0);
86
tcg_ri = tcg_temp_new_ptr();
87
diff --git a/target/arm/tcg/translate.c b/target/arm/tcg/translate.c
88
index XXXXXXX..XXXXXXX 100644
89
--- a/target/arm/tcg/translate.c
90
+++ b/target/arm/tcg/translate.c
91
@@ -XXX,XX +XXX,XX @@ void gen_gvec_uaba(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
92
tcg_gen_gvec_3(rd_ofs, rn_ofs, rm_ofs, opr_sz, max_sz, &ops[vece]);
93
}
94
95
+static bool aa32_cpreg_encoding_in_impdef_space(uint8_t crn, uint8_t crm)
96
+{
97
+ static const uint16_t mask[3] = {
98
+ 0b0000000111100111, /* crn == 9, crm == {c0-c2, c5-c8} */
99
+ 0b0000000100010011, /* crn == 10, crm == {c0, c1, c4, c8} */
100
+ 0b1000000111111111, /* crn == 11, crm == {c0-c8, c15} */
101
+ };
102
+
103
+ if (crn >= 9 && crn <= 11) {
104
+ return (mask[crn - 9] >> crm) & 1;
105
+ }
106
+ return false;
107
+}
108
+
109
static void do_coproc_insn(DisasContext *s, int cpnum, int is64,
110
int opc1, int crn, int crm, int opc2,
111
bool isread, int rt, int rt2)
112
@@ -XXX,XX +XXX,XX @@ static void do_coproc_insn(DisasContext *s, int cpnum, int is64,
113
}
114
}
115
116
+ if (cpnum == 15 && aa32_cpreg_encoding_in_impdef_space(crn, crm)) {
117
+ /*
118
+ * Check for TIDCP trap, which must take precedence over the UNDEF
119
+ * for "no such register" etc. It shares precedence with HSTR,
120
+ * but raises the same exception, so order doesn't matter.
121
+ */
122
+ switch (s->current_el) {
123
+ case 1:
124
+ gen_helper_tidcp_el1(cpu_env, tcg_constant_i32(syndrome));
125
+ break;
126
+ }
127
+ }
128
+
129
if (!ri) {
130
/*
131
* Unknown register; this might be a guest error or a QEMU
132
--
133
2.34.1
diff view generated by jsdifflib
[PULL 24/26] target/arm: Implement FEAT_TIDCP1
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
4
Message-id: 20230831232441.66020-5-richard.henderson@linaro.org
5
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7
---
8
docs/system/arm/emulation.rst | 1 +
9
target/arm/cpu.h | 5 +++++
10
target/arm/helper.h | 1 +
11
target/arm/tcg/cpu64.c | 1 +
12
target/arm/tcg/op_helper.c | 20 ++++++++++++++++++++
13
target/arm/tcg/translate-a64.c | 5 +++++
14
target/arm/tcg/translate.c | 6 ++++++
15
7 files changed, 39 insertions(+)
16
17
diff --git a/docs/system/arm/emulation.rst b/docs/system/arm/emulation.rst
18
index XXXXXXX..XXXXXXX 100644
19
--- a/docs/system/arm/emulation.rst
20
+++ b/docs/system/arm/emulation.rst
21
@@ -XXX,XX +XXX,XX @@ the following architecture extensions:
22
- FEAT_SME_I16I64 (16-bit to 64-bit integer widening outer product instructions)
23
- FEAT_SPECRES (Speculation restriction instructions)
24
- FEAT_SSBS (Speculative Store Bypass Safe)
25
+- FEAT_TIDCP1 (EL0 use of IMPLEMENTATION DEFINED functionality)
26
- FEAT_TLBIOS (TLB invalidate instructions in Outer Shareable domain)
27
- FEAT_TLBIRANGE (TLB invalidate range instructions)
28
- FEAT_TTCNP (Translation table Common not private translations)
29
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
30
index XXXXXXX..XXXXXXX 100644
31
--- a/target/arm/cpu.h
32
+++ b/target/arm/cpu.h
33
@@ -XXX,XX +XXX,XX @@ static inline bool isar_feature_aa64_hcx(const ARMISARegisters *id)
34
return FIELD_EX64(id->id_aa64mmfr1, ID_AA64MMFR1, HCX) != 0;
35
}
36
37
+static inline bool isar_feature_aa64_tidcp1(const ARMISARegisters *id)
38
+{
39
+ return FIELD_EX64(id->id_aa64mmfr2, ID_AA64MMFR1, TIDCP1) != 0;
40
+}
41
+
42
static inline bool isar_feature_aa64_uao(const ARMISARegisters *id)
43
{
44
return FIELD_EX64(id->id_aa64mmfr2, ID_AA64MMFR2, UAO) != 0;
45
diff --git a/target/arm/helper.h b/target/arm/helper.h
46
index XXXXXXX..XXXXXXX 100644
47
--- a/target/arm/helper.h
48
+++ b/target/arm/helper.h
49
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_2(check_bxj_trap, TCG_CALL_NO_WG, void, env, i32)
50
51
DEF_HELPER_4(access_check_cp_reg, cptr, env, i32, i32, i32)
52
DEF_HELPER_FLAGS_2(lookup_cp_reg, TCG_CALL_NO_RWG_SE, cptr, env, i32)
53
+DEF_HELPER_FLAGS_2(tidcp_el0, TCG_CALL_NO_WG, void, env, i32)
54
DEF_HELPER_FLAGS_2(tidcp_el1, TCG_CALL_NO_WG, void, env, i32)
55
DEF_HELPER_3(set_cp_reg, void, env, cptr, i32)
56
DEF_HELPER_2(get_cp_reg, i32, env, cptr)
57
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
58
index XXXXXXX..XXXXXXX 100644
59
--- a/target/arm/tcg/cpu64.c
60
+++ b/target/arm/tcg/cpu64.c
61
@@ -XXX,XX +XXX,XX @@ void aarch64_max_tcg_initfn(Object *obj)
62
t = FIELD_DP64(t, ID_AA64MMFR1, XNX, 1); /* FEAT_XNX */
63
t = FIELD_DP64(t, ID_AA64MMFR1, ETS, 1); /* FEAT_ETS */
64
t = FIELD_DP64(t, ID_AA64MMFR1, HCX, 1); /* FEAT_HCX */
65
+ t = FIELD_DP64(t, ID_AA64MMFR1, TIDCP1, 1); /* FEAT_TIDCP1 */
66
cpu->isar.id_aa64mmfr1 = t;
67
68
t = cpu->isar.id_aa64mmfr2;
69
diff --git a/target/arm/tcg/op_helper.c b/target/arm/tcg/op_helper.c
70
index XXXXXXX..XXXXXXX 100644
71
--- a/target/arm/tcg/op_helper.c
72
+++ b/target/arm/tcg/op_helper.c
73
@@ -XXX,XX +XXX,XX @@ void HELPER(tidcp_el1)(CPUARMState *env, uint32_t syndrome)
74
}
75
}
76
77
+/*
78
+ * Similarly, for FEAT_TIDCP1 at EL0.
79
+ * We have already checked for the presence of the feature.
80
+ */
81
+void HELPER(tidcp_el0)(CPUARMState *env, uint32_t syndrome)
82
+{
83
+ /* See arm_sctlr(), but we also need the sctlr el. */
84
+ ARMMMUIdx mmu_idx = arm_mmu_idx_el(env, 0);
85
+ int target_el = mmu_idx == ARMMMUIdx_E20_0 ? 2 : 1;
86
+
87
+ /*
88
+ * The bit is not valid unless the target el is aa64, but since the
89
+ * bit test is simpler perform that first and check validity after.
90
+ */
91
+ if ((env->cp15.sctlr_el[target_el] & SCTLR_TIDCP)
92
+ && arm_el_is_aa64(env, target_el)) {
93
+ raise_exception_ra(env, EXCP_UDEF, syndrome, target_el, GETPC());
94
+ }
95
+}
96
+
97
void HELPER(set_cp_reg)(CPUARMState *env, const void *rip, uint32_t value)
98
{
99
const ARMCPRegInfo *ri = rip;
100
diff --git a/target/arm/tcg/translate-a64.c b/target/arm/tcg/translate-a64.c
101
index XXXXXXX..XXXXXXX 100644
102
--- a/target/arm/tcg/translate-a64.c
103
+++ b/target/arm/tcg/translate-a64.c
104
@@ -XXX,XX +XXX,XX @@ static void handle_sys(DisasContext *s, bool isread,
105
*/
106
syndrome = syn_aa64_sysregtrap(op0, op1, op2, crn, crm, rt, isread);
107
switch (s->current_el) {
108
+ case 0:
109
+ if (dc_isar_feature(aa64_tidcp1, s)) {
110
+ gen_helper_tidcp_el0(cpu_env, tcg_constant_i32(syndrome));
111
+ }
112
+ break;
113
case 1:
114
gen_helper_tidcp_el1(cpu_env, tcg_constant_i32(syndrome));
115
break;
116
diff --git a/target/arm/tcg/translate.c b/target/arm/tcg/translate.c
117
index XXXXXXX..XXXXXXX 100644
118
--- a/target/arm/tcg/translate.c
119
+++ b/target/arm/tcg/translate.c
120
@@ -XXX,XX +XXX,XX @@ static void do_coproc_insn(DisasContext *s, int cpnum, int is64,
121
* but raises the same exception, so order doesn't matter.
122
*/
123
switch (s->current_el) {
124
+ case 0:
125
+ if (arm_dc_feature(s, ARM_FEATURE_AARCH64)
126
+ && dc_isar_feature(aa64_tidcp1, s)) {
127
+ gen_helper_tidcp_el0(cpu_env, tcg_constant_i32(syndrome));
128
+ }
129
+ break;
130
case 1:
131
gen_helper_tidcp_el1(cpu_env, tcg_constant_i32(syndrome));
132
break;
133
--
134
2.34.1
diff view generated by jsdifflib
[PULL 25/26] target/arm: Enable SCTLR_EL1.TIDCP for user-only
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
The linux kernel detects and enables this bit. Once trapped,
4
EC_SYSTEMREGISTERTRAP is treated like EC_UNCATEGORIZED, so
5
no changes required within linux-user/aarch64/cpu_loop.c.
6
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
8
Message-id: 20230831232441.66020-6-richard.henderson@linaro.org
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
target/arm/cpu.c | 4 ++++
13
1 file changed, 4 insertions(+)
14
15
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
16
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/cpu.c
18
+++ b/target/arm/cpu.c
19
@@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset_hold(Object *obj)
20
SCTLR_EnDA | SCTLR_EnDB);
21
/* Trap on btype=3 for PACIxSP. */
22
env->cp15.sctlr_el[1] |= SCTLR_BT0;
23
+ /* Trap on implementation defined registers. */
24
+ if (cpu_isar_feature(aa64_tidcp1, cpu)) {
25
+ env->cp15.sctlr_el[1] |= SCTLR_TIDCP;
26
+ }
27
/* and to the FP/Neon instructions */
28
env->cp15.cpacr_el1 = FIELD_DP64(env->cp15.cpacr_el1,
29
CPACR_EL1, FPEN, 3);
30
--
31
2.34.1
diff view generated by jsdifflib
[PULL 26/26] arm/kvm: Enable support for KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE
Deleted patch
1
From: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
2
1
3
Now that we have Eager Page Split support added for ARM in the kernel,
4
enable it in Qemu. This adds,
5
-eager-split-size to -accel sub-options to set the eager page split chunk size.
6
-enable KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE.
7
8
The chunk size specifies how many pages to break at a time, using a
9
single allocation. Bigger the chunk size, more pages need to be
10
allocated ahead of time.
11
12
Reviewed-by: Gavin Shan <gshan@redhat.com>
13
Signed-off-by: Shameer Kolothum <shameerali.kolothum.thodi@huawei.com>
14
Message-id: 20230905091246.1931-1-shameerali.kolothum.thodi@huawei.com
15
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
---
17
include/sysemu/kvm_int.h | 1 +
18
accel/kvm/kvm-all.c | 1 +
19
target/arm/kvm.c | 61 ++++++++++++++++++++++++++++++++++++++++
20
qemu-options.hx | 15 ++++++++++
21
4 files changed, 78 insertions(+)
22
23
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
24
index XXXXXXX..XXXXXXX 100644
25
--- a/include/sysemu/kvm_int.h
26
+++ b/include/sysemu/kvm_int.h
27
@@ -XXX,XX +XXX,XX @@ struct KVMState
28
uint64_t kvm_dirty_ring_bytes; /* Size of the per-vcpu dirty ring */
29
uint32_t kvm_dirty_ring_size; /* Number of dirty GFNs per ring */
30
bool kvm_dirty_ring_with_bitmap;
31
+ uint64_t kvm_eager_split_size; /* Eager Page Splitting chunk size */
32
struct KVMDirtyRingReaper reaper;
33
NotifyVmexitOption notify_vmexit;
34
uint32_t notify_window;
35
diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
36
index XXXXXXX..XXXXXXX 100644
37
--- a/accel/kvm/kvm-all.c
38
+++ b/accel/kvm/kvm-all.c
39
@@ -XXX,XX +XXX,XX @@ static void kvm_accel_instance_init(Object *obj)
40
/* KVM dirty ring is by default off */
41
s->kvm_dirty_ring_size = 0;
42
s->kvm_dirty_ring_with_bitmap = false;
43
+ s->kvm_eager_split_size = 0;
44
s->notify_vmexit = NOTIFY_VMEXIT_OPTION_RUN;
45
s->notify_window = 0;
46
s->xen_version = 0;
47
diff --git a/target/arm/kvm.c b/target/arm/kvm.c
48
index XXXXXXX..XXXXXXX 100644
49
--- a/target/arm/kvm.c
50
+++ b/target/arm/kvm.c
51
@@ -XXX,XX +XXX,XX @@
52
#include "exec/address-spaces.h"
53
#include "hw/boards.h"
54
#include "hw/irq.h"
55
+#include "qapi/visitor.h"
56
#include "qemu/log.h"
57
58
const KVMCapabilityInfo kvm_arch_required_capabilities[] = {
59
@@ -XXX,XX +XXX,XX @@ int kvm_arch_init(MachineState *ms, KVMState *s)
60
}
61
}
62
63
+ if (s->kvm_eager_split_size) {
64
+ uint32_t sizes;
65
+
66
+ sizes = kvm_vm_check_extension(s, KVM_CAP_ARM_SUPPORTED_BLOCK_SIZES);
67
+ if (!sizes) {
68
+ s->kvm_eager_split_size = 0;
69
+ warn_report("Eager Page Split support not available");
70
+ } else if (!(s->kvm_eager_split_size & sizes)) {
71
+ error_report("Eager Page Split requested chunk size not valid");
72
+ ret = -EINVAL;
73
+ } else {
74
+ ret = kvm_vm_enable_cap(s, KVM_CAP_ARM_EAGER_SPLIT_CHUNK_SIZE, 0,
75
+ s->kvm_eager_split_size);
76
+ if (ret < 0) {
77
+ error_report("Enabling of Eager Page Split failed: %s",
78
+ strerror(-ret));
79
+ }
80
+ }
81
+ }
82
+
83
kvm_arm_init_debug(s);
84
85
return ret;
86
@@ -XXX,XX +XXX,XX @@ bool kvm_arch_cpu_check_are_resettable(void)
87
return true;
88
}
89
90
+static void kvm_arch_get_eager_split_size(Object *obj, Visitor *v,
91
+ const char *name, void *opaque,
92
+ Error **errp)
93
+{
94
+ KVMState *s = KVM_STATE(obj);
95
+ uint64_t value = s->kvm_eager_split_size;
96
+
97
+ visit_type_size(v, name, &value, errp);
98
+}
99
+
100
+static void kvm_arch_set_eager_split_size(Object *obj, Visitor *v,
101
+ const char *name, void *opaque,
102
+ Error **errp)
103
+{
104
+ KVMState *s = KVM_STATE(obj);
105
+ uint64_t value;
106
+
107
+ if (s->fd != -1) {
108
+ error_setg(errp, "Unable to set early-split-size after KVM has been initialized");
109
+ return;
110
+ }
111
+
112
+ if (!visit_type_size(v, name, &value, errp)) {
113
+ return;
114
+ }
115
+
116
+ if (value && !is_power_of_2(value)) {
117
+ error_setg(errp, "early-split-size must be a power of two");
118
+ return;
119
+ }
120
+
121
+ s->kvm_eager_split_size = value;
122
+}
123
+
124
void kvm_arch_accel_class_init(ObjectClass *oc)
125
{
126
+ object_class_property_add(oc, "eager-split-size", "size",
127
+ kvm_arch_get_eager_split_size,
128
+ kvm_arch_set_eager_split_size, NULL, NULL);
129
+
130
+ object_class_property_set_description(oc, "eager-split-size",
131
+ "Eager Page Split chunk size for hugepages. (default: 0, disabled)");
132
}
133
diff --git a/qemu-options.hx b/qemu-options.hx
134
index XXXXXXX..XXXXXXX 100644
135
--- a/qemu-options.hx
136
+++ b/qemu-options.hx
137
@@ -XXX,XX +XXX,XX @@ DEF("accel", HAS_ARG, QEMU_OPTION_accel,
138
" split-wx=on|off (enable TCG split w^x mapping)\n"
139
" tb-size=n (TCG translation block cache size)\n"
140
" dirty-ring-size=n (KVM dirty ring GFN count, default 0)\n"
141
+ " eager-split-size=n (KVM Eager Page Split chunk size, default 0, disabled. ARM only)\n"
142
" notify-vmexit=run|internal-error|disable,notify-window=n (enable notify VM exit and set notify window, x86 only)\n"
143
" thread=single|multi (enable multi-threaded TCG)\n", QEMU_ARCH_ALL)
144
SRST
145
@@ -XXX,XX +XXX,XX @@ SRST
146
is disabled (dirty-ring-size=0). When enabled, KVM will instead
147
record dirty pages in a bitmap.
148
149
+ ``eager-split-size=n``
150
+ KVM implements dirty page logging at the PAGE_SIZE granularity and
151
+ enabling dirty-logging on a huge-page requires breaking it into
152
+ PAGE_SIZE pages in the first place. KVM on ARM does this splitting
153
+ lazily by default. There are performance benefits in doing huge-page
154
+ split eagerly, especially in situations where TLBI costs associated
155
+ with break-before-make sequences are considerable and also if guest
156
+ workloads are read intensive. The size here specifies how many pages
157
+ to break at a time and needs to be a valid block size which is
158
+ 1GB/2MB/4KB, 32MB/16KB and 512MB/64KB for 4KB/16KB/64KB PAGE_SIZE
159
+ respectively. Be wary of specifying a higher size as it will have an
160
+ impact on the memory. By default, this feature is disabled
161
+ (eager-split-size=0).
162
+
163
``notify-vmexit=run|internal-error|disable,notify-window=n``
164
Enables or disables notify VM exit support on x86 host and specify
165
the corresponding notify window to trigger the VM exit if enabled.
166
--
167
2.34.1
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.