1
Another lump of target-arm patches. I still have some patches in
1
The following changes since commit ad1b4ec39caa5b3f17cbd8160283a03a3dcfe2ae:
2
my to-review queue, but this is a big enough set that I wanted
3
to send it out.
4
2
5
thanks
3
Merge remote-tracking branch 'remotes/kraxel/tags/input-20180515-pull-request' into staging (2018-05-15 12:50:06 +0100)
6
-- PMM
7
8
The following changes since commit 04bb7fe2bf55bdf66d5b7a5a719b40bbb4048178:
9
10
Merge remote-tracking branch 'remotes/rth/tags/pull-tcg-20180208' into staging (2018-02-08 17:41:15 +0000)
11
4
12
are available in the Git repository at:
5
are available in the Git repository at:
13
6
14
git://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20180209
7
git://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20180515
15
8
16
for you to fetch changes up to bbba7757bacc9f890a3f028d328b4b429dbe78ec:
9
for you to fetch changes up to ae7651804748c6b479d5ae09aeac4edb9c44f76e:
17
10
18
hw/core/generic-loader: Allow PC to be set on command line (2018-02-09 10:55:40 +0000)
11
tcg: Optionally log FPU state in TCG -d cpu logging (2018-05-15 14:58:44 +0100)
19
12
20
----------------------------------------------------------------
13
----------------------------------------------------------------
21
target-arm queue:
14
target-arm queue:
22
* Support M profile derived exceptions on exception entry and exit
15
* Fix coverity nit in int_to_float code
23
* Implement AArch64 v8.2 crypto insns (SHA-512, SHA-3, SM3, SM4)
16
* Don't set Invalid for float-to-int(MAXINT)
24
* Implement working i.MX6 SD controller
17
* Fix fp_status_f16 tininess before rounding
25
* Various devices preparatory to i.MX7 support
18
* Add various missing insns from the v8.2-FP16 extension
26
* Preparatory patches for SVE emulation
19
* Fix sqrt_f16 exception raising
27
* v8M: Fix bug in implementation of 'TT' insn
20
* sdcard: Correct CRC16 offset in sd_function_switch()
28
* Give useful error if user tries to use userspace GICv3 with KVM
21
* tcg: Optionally log FPU state in TCG -d cpu logging
29
22
30
----------------------------------------------------------------
23
----------------------------------------------------------------
31
Andrey Smirnov (10):
24
Alex Bennée (5):
32
sdhci: Add i.MX specific subtype of SDHCI
25
fpu/softfloat: int_to_float ensure r fully initialised
33
hw: i.MX: Convert i.MX6 to use TYPE_IMX_USDHC
26
target/arm: Implement FCMP for fp16
34
i.MX: Add code to emulate i.MX7 CCM, PMU and ANALOG IP blocks
27
target/arm: Implement FCSEL for fp16
35
i.MX: Add code to emulate i.MX2 watchdog IP block
28
target/arm: Implement FMOV (immediate) for fp16
36
i.MX: Add code to emulate i.MX7 SNVS IP-block
29
target/arm: Fix sqrt_f16 exception raising
37
i.MX: Add code to emulate GPCv2 IP block
38
i.MX: Add i.MX7 GPT variant
39
i.MX: Add implementation of i.MX7 GPR IP block
40
usb: Add basic code to emulate Chipidea USB IP
41
hw/arm: Move virt's PSCI DT fixup code to arm/boot.c
42
30
43
Ard Biesheuvel (5):
31
Peter Maydell (3):
44
target/arm: implement SHA-512 instructions
32
fpu/softfloat: Don't set Invalid for float-to-int(MAXINT)
45
target/arm: implement SHA-3 instructions
33
target/arm: Fix fp_status_f16 tininess before rounding
46
target/arm: implement SM3 instructions
34
tcg: Optionally log FPU state in TCG -d cpu logging
47
target/arm: implement SM4 instructions
48
target/arm: enable user-mode SHA-3, SM3, SM4 and SHA-512 instruction support
49
35
50
Christoffer Dall (1):
36
Philippe Mathieu-Daudé (1):
51
target/arm/kvm: gic: Prevent creating userspace GICv3 with KVM
37
sdcard: Correct CRC16 offset in sd_function_switch()
52
38
53
Peter Maydell (9):
39
Richard Henderson (7):
54
target/arm: Add armv7m_nvic_set_pending_derived()
40
target/arm: Implement FMOV (general) for fp16
55
target/arm: Split "get pending exception info" from "acknowledge it"
41
target/arm: Early exit after unallocated_encoding in disas_fp_int_conv
56
target/arm: Add ignore_stackfaults argument to v7m_exception_taken()
42
target/arm: Implement FCVT (scalar, integer) for fp16
57
target/arm: Make v7M exception entry stack push check MPU
43
target/arm: Implement FCVT (scalar, fixed-point) for fp16
58
target/arm: Make v7m_push_callee_stack() honour MPU
44
target/arm: Introduce and use read_fp_hreg
59
target/arm: Make exception vector loads honour the SAU
45
target/arm: Implement FP data-processing (2 source) for fp16
60
target/arm: Handle exceptions during exception stack pop
46
target/arm: Implement FP data-processing (3 source) for fp16
61
target/arm/translate.c: Fix missing 'break' for TT insns
62
hw/core/generic-loader: Allow PC to be set on command line
63
47
64
Richard Henderson (5):
48
include/qemu/log.h | 1 +
65
target/arm: Expand vector registers for SVE
49
target/arm/helper-a64.h | 2 +
66
target/arm: Add predicate registers for SVE
50
target/arm/helper.h | 6 +
67
target/arm: Add SVE to migration state
51
accel/tcg/cpu-exec.c | 9 +-
68
target/arm: Add ZCR_ELx
52
fpu/softfloat.c | 6 +-
69
target/arm: Add SVE state to TB->FLAGS
53
hw/sd/sd.c | 2 +-
54
target/arm/cpu.c | 2 +
55
target/arm/helper-a64.c | 10 ++
56
target/arm/helper.c | 38 +++-
57
target/arm/translate-a64.c | 421 ++++++++++++++++++++++++++++++++++++++-------
58
util/log.c | 2 +
59
11 files changed, 428 insertions(+), 71 deletions(-)
70
60
71
hw/intc/Makefile.objs | 2 +-
72
hw/misc/Makefile.objs | 4 +
73
hw/usb/Makefile.objs | 1 +
74
hw/sd/sdhci-internal.h | 23 ++
75
include/hw/intc/imx_gpcv2.h | 22 ++
76
include/hw/misc/imx2_wdt.h | 33 +++
77
include/hw/misc/imx7_ccm.h | 139 +++++++++++
78
include/hw/misc/imx7_gpr.h | 28 +++
79
include/hw/misc/imx7_snvs.h | 35 +++
80
include/hw/sd/sdhci.h | 13 ++
81
include/hw/timer/imx_gpt.h | 1 +
82
include/hw/usb/chipidea.h | 16 ++
83
target/arm/cpu.h | 120 ++++++++--
84
target/arm/helper.h | 12 +
85
target/arm/kvm_arm.h | 4 +
86
target/arm/translate.h | 2 +
87
hw/arm/boot.c | 65 ++++++
88
hw/arm/fsl-imx6.c | 2 +-
89
hw/arm/virt.c | 61 -----
90
hw/core/generic-loader.c | 2 +-
91
hw/intc/armv7m_nvic.c | 98 +++++++-
92
hw/intc/imx_gpcv2.c | 125 ++++++++++
93
hw/misc/imx2_wdt.c | 89 +++++++
94
hw/misc/imx7_ccm.c | 277 ++++++++++++++++++++++
95
hw/misc/imx7_gpr.c | 124 ++++++++++
96
hw/misc/imx7_snvs.c | 83 +++++++
97
hw/sd/sdhci.c | 230 ++++++++++++++++++-
98
hw/timer/imx_gpt.c | 25 ++
99
hw/usb/chipidea.c | 176 ++++++++++++++
100
linux-user/elfload.c | 19 ++
101
target/arm/cpu64.c | 4 +
102
target/arm/crypto_helper.c | 277 +++++++++++++++++++++-
103
target/arm/helper.c | 548 +++++++++++++++++++++++++++++++++++++-------
104
target/arm/machine.c | 88 ++++++-
105
target/arm/translate-a64.c | 350 +++++++++++++++++++++++++++-
106
target/arm/translate.c | 8 +-
107
hw/intc/trace-events | 5 +-
108
hw/misc/trace-events | 4 +
109
38 files changed, 2928 insertions(+), 187 deletions(-)
110
create mode 100644 include/hw/intc/imx_gpcv2.h
111
create mode 100644 include/hw/misc/imx2_wdt.h
112
create mode 100644 include/hw/misc/imx7_ccm.h
113
create mode 100644 include/hw/misc/imx7_gpr.h
114
create mode 100644 include/hw/misc/imx7_snvs.h
115
create mode 100644 include/hw/usb/chipidea.h
116
create mode 100644 hw/intc/imx_gpcv2.c
117
create mode 100644 hw/misc/imx2_wdt.c
118
create mode 100644 hw/misc/imx7_ccm.c
119
create mode 100644 hw/misc/imx7_gpr.c
120
create mode 100644 hw/misc/imx7_snvs.c
121
create mode 100644 hw/usb/chipidea.c
122
diff view generated by jsdifflib
Deleted patch
1
In order to support derived exceptions (exceptions generated in
2
the course of trying to take an exception), we need to be able
3
to handle prioritizing whether to take the original exception
4
or the derived exception.
5
1
6
We do this by introducing a new function
7
armv7m_nvic_set_pending_derived() which the exception-taking code in
8
helper.c will call when a derived exception occurs. Derived
9
exceptions are dealt with mostly like normal pending exceptions, so
10
we share the implementation with the armv7m_nvic_set_pending()
11
function.
12
13
Note that the way we structure this is significantly different
14
from the v8M Arm ARM pseudocode: that does all the prioritization
15
logic in the DerivedLateArrival() function, whereas we choose to
16
let the existing "identify highest priority exception" logic
17
do the prioritization for us. The effect is the same, though.
18
19
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
20
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
21
Message-id: 1517324542-6607-2-git-send-email-peter.maydell@linaro.org
22
---
23
target/arm/cpu.h | 13 ++++++++++
24
hw/intc/armv7m_nvic.c | 68 +++++++++++++++++++++++++++++++++++++++++++++++++--
25
hw/intc/trace-events | 2 +-
26
3 files changed, 80 insertions(+), 3 deletions(-)
27
28
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
29
index XXXXXXX..XXXXXXX 100644
30
--- a/target/arm/cpu.h
31
+++ b/target/arm/cpu.h
32
@@ -XXX,XX +XXX,XX @@ static inline bool armv7m_nvic_can_take_pending_exception(void *opaque)
33
* of architecturally banked exceptions.
34
*/
35
void armv7m_nvic_set_pending(void *opaque, int irq, bool secure);
36
+/**
37
+ * armv7m_nvic_set_pending_derived: mark this derived exception as pending
38
+ * @opaque: the NVIC
39
+ * @irq: the exception number to mark pending
40
+ * @secure: false for non-banked exceptions or for the nonsecure
41
+ * version of a banked exception, true for the secure version of a banked
42
+ * exception.
43
+ *
44
+ * Similar to armv7m_nvic_set_pending(), but specifically for derived
45
+ * exceptions (exceptions generated in the course of trying to take
46
+ * a different exception).
47
+ */
48
+void armv7m_nvic_set_pending_derived(void *opaque, int irq, bool secure);
49
/**
50
* armv7m_nvic_acknowledge_irq: make highest priority pending exception active
51
* @opaque: the NVIC
52
diff --git a/hw/intc/armv7m_nvic.c b/hw/intc/armv7m_nvic.c
53
index XXXXXXX..XXXXXXX 100644
54
--- a/hw/intc/armv7m_nvic.c
55
+++ b/hw/intc/armv7m_nvic.c
56
@@ -XXX,XX +XXX,XX @@ static void armv7m_nvic_clear_pending(void *opaque, int irq, bool secure)
57
}
58
}
59
60
-void armv7m_nvic_set_pending(void *opaque, int irq, bool secure)
61
+static void do_armv7m_nvic_set_pending(void *opaque, int irq, bool secure,
62
+ bool derived)
63
{
64
+ /* Pend an exception, including possibly escalating it to HardFault.
65
+ *
66
+ * This function handles both "normal" pending of interrupts and
67
+ * exceptions, and also derived exceptions (ones which occur as
68
+ * a result of trying to take some other exception).
69
+ *
70
+ * If derived == true, the caller guarantees that we are part way through
71
+ * trying to take an exception (but have not yet called
72
+ * armv7m_nvic_acknowledge_irq() to make it active), and so:
73
+ * - s->vectpending is the "original exception" we were trying to take
74
+ * - irq is the "derived exception"
75
+ * - nvic_exec_prio(s) gives the priority before exception entry
76
+ * Here we handle the prioritization logic which the pseudocode puts
77
+ * in the DerivedLateArrival() function.
78
+ */
79
+
80
NVICState *s = (NVICState *)opaque;
81
bool banked = exc_is_banked(irq);
82
VecInfo *vec;
83
@@ -XXX,XX +XXX,XX @@ void armv7m_nvic_set_pending(void *opaque, int irq, bool secure)
84
85
vec = (banked && secure) ? &s->sec_vectors[irq] : &s->vectors[irq];
86
87
- trace_nvic_set_pending(irq, secure, vec->enabled, vec->prio);
88
+ trace_nvic_set_pending(irq, secure, derived, vec->enabled, vec->prio);
89
+
90
+ if (derived) {
91
+ /* Derived exceptions are always synchronous. */
92
+ assert(irq >= ARMV7M_EXCP_HARD && irq < ARMV7M_EXCP_PENDSV);
93
+
94
+ if (irq == ARMV7M_EXCP_DEBUG &&
95
+ exc_group_prio(s, vec->prio, secure) >= nvic_exec_prio(s)) {
96
+ /* DebugMonitorFault, but its priority is lower than the
97
+ * preempted exception priority: just ignore it.
98
+ */
99
+ return;
100
+ }
101
+
102
+ if (irq == ARMV7M_EXCP_HARD && vec->prio >= s->vectpending_prio) {
103
+ /* If this is a terminal exception (one which means we cannot
104
+ * take the original exception, like a failure to read its
105
+ * vector table entry), then we must take the derived exception.
106
+ * If the derived exception can't take priority over the
107
+ * original exception, then we go into Lockup.
108
+ *
109
+ * For QEMU, we rely on the fact that a derived exception is
110
+ * terminal if and only if it's reported to us as HardFault,
111
+ * which saves having to have an extra argument is_terminal
112
+ * that we'd only use in one place.
113
+ */
114
+ cpu_abort(&s->cpu->parent_obj,
115
+ "Lockup: can't take terminal derived exception "
116
+ "(original exception priority %d)\n",
117
+ s->vectpending_prio);
118
+ }
119
+ /* We now continue with the same code as for a normal pending
120
+ * exception, which will cause us to pend the derived exception.
121
+ * We'll then take either the original or the derived exception
122
+ * based on which is higher priority by the usual mechanism
123
+ * for selecting the highest priority pending interrupt.
124
+ */
125
+ }
126
127
if (irq >= ARMV7M_EXCP_HARD && irq < ARMV7M_EXCP_PENDSV) {
128
/* If a synchronous exception is pending then it may be
129
@@ -XXX,XX +XXX,XX @@ void armv7m_nvic_set_pending(void *opaque, int irq, bool secure)
130
}
131
}
132
133
+void armv7m_nvic_set_pending(void *opaque, int irq, bool secure)
134
+{
135
+ do_armv7m_nvic_set_pending(opaque, irq, secure, false);
136
+}
137
+
138
+void armv7m_nvic_set_pending_derived(void *opaque, int irq, bool secure)
139
+{
140
+ do_armv7m_nvic_set_pending(opaque, irq, secure, true);
141
+}
142
+
143
/* Make pending IRQ active. */
144
bool armv7m_nvic_acknowledge_irq(void *opaque)
145
{
146
diff --git a/hw/intc/trace-events b/hw/intc/trace-events
147
index XXXXXXX..XXXXXXX 100644
148
--- a/hw/intc/trace-events
149
+++ b/hw/intc/trace-events
150
@@ -XXX,XX +XXX,XX @@ nvic_set_prio(int irq, bool secure, uint8_t prio) "NVIC set irq %d secure-bank %
151
nvic_irq_update(int vectpending, int pendprio, int exception_prio, int level) "NVIC vectpending %d pending prio %d exception_prio %d: setting irq line to %d"
152
nvic_escalate_prio(int irq, int irqprio, int runprio) "NVIC escalating irq %d to HardFault: insufficient priority %d >= %d"
153
nvic_escalate_disabled(int irq) "NVIC escalating irq %d to HardFault: disabled"
154
-nvic_set_pending(int irq, bool secure, int en, int prio) "NVIC set pending irq %d secure-bank %d (enabled: %d priority %d)"
155
+nvic_set_pending(int irq, bool secure, bool derived, int en, int prio) "NVIC set pending irq %d secure-bank %d derived %d (enabled: %d priority %d)"
156
nvic_clear_pending(int irq, bool secure, int en, int prio) "NVIC clear pending irq %d secure-bank %d (enabled: %d priority %d)"
157
nvic_set_pending_level(int irq) "NVIC set pending: irq %d higher prio than vectpending: setting irq line to 1"
158
nvic_acknowledge_irq(int irq, int prio, bool targets_secure) "NVIC acknowledge IRQ: %d now active (prio %d targets_secure %d)"
159
--
160
2.16.1
161
162
diff view generated by jsdifflib
Deleted patch
1
Currently armv7m_nvic_acknowledge_irq() does three things:
2
* make the current highest priority pending interrupt active
3
* return a bool indicating whether that interrupt is targeting
4
Secure or NonSecure state
5
* implicitly tell the caller which is the highest priority
6
pending interrupt by setting env->v7m.exception
7
1
8
We need to split these jobs, because v7m_exception_taken()
9
needs to know whether the pending interrupt targets Secure so
10
it can choose to stack callee-saves registers or not, but it
11
must not make the interrupt active until after it has done
12
that stacking, in case the stacking causes a derived exception.
13
Similarly, it needs to know the number of the pending interrupt
14
so it can read the correct vector table entry before the
15
interrupt is made active, because vector table reads might
16
also cause a derived exception.
17
18
Create a new armv7m_nvic_get_pending_irq_info() function which simply
19
returns information about the highest priority pending interrupt, and
20
use it to rearrange the v7m_exception_taken() code so we don't
21
acknowledge the exception until we've done all the things which could
22
possibly cause a derived exception.
23
24
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
25
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
26
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
27
Message-id: 1517324542-6607-3-git-send-email-peter.maydell@linaro.org
28
---
29
target/arm/cpu.h | 19 ++++++++++++++++---
30
hw/intc/armv7m_nvic.c | 30 +++++++++++++++++++++++-------
31
target/arm/helper.c | 16 ++++++++++++----
32
hw/intc/trace-events | 3 ++-
33
4 files changed, 53 insertions(+), 15 deletions(-)
34
35
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
36
index XXXXXXX..XXXXXXX 100644
37
--- a/target/arm/cpu.h
38
+++ b/target/arm/cpu.h
39
@@ -XXX,XX +XXX,XX @@ void armv7m_nvic_set_pending(void *opaque, int irq, bool secure);
40
* a different exception).
41
*/
42
void armv7m_nvic_set_pending_derived(void *opaque, int irq, bool secure);
43
+/**
44
+ * armv7m_nvic_get_pending_irq_info: return highest priority pending
45
+ * exception, and whether it targets Secure state
46
+ * @opaque: the NVIC
47
+ * @pirq: set to pending exception number
48
+ * @ptargets_secure: set to whether pending exception targets Secure
49
+ *
50
+ * This function writes the number of the highest priority pending
51
+ * exception (the one which would be made active by
52
+ * armv7m_nvic_acknowledge_irq()) to @pirq, and sets @ptargets_secure
53
+ * to true if the current highest priority pending exception should
54
+ * be taken to Secure state, false for NS.
55
+ */
56
+void armv7m_nvic_get_pending_irq_info(void *opaque, int *pirq,
57
+ bool *ptargets_secure);
58
/**
59
* armv7m_nvic_acknowledge_irq: make highest priority pending exception active
60
* @opaque: the NVIC
61
@@ -XXX,XX +XXX,XX @@ void armv7m_nvic_set_pending_derived(void *opaque, int irq, bool secure);
62
* Move the current highest priority pending exception from the pending
63
* state to the active state, and update v7m.exception to indicate that
64
* it is the exception currently being handled.
65
- *
66
- * Returns: true if exception should be taken to Secure state, false for NS
67
*/
68
-bool armv7m_nvic_acknowledge_irq(void *opaque);
69
+void armv7m_nvic_acknowledge_irq(void *opaque);
70
/**
71
* armv7m_nvic_complete_irq: complete specified interrupt or exception
72
* @opaque: the NVIC
73
diff --git a/hw/intc/armv7m_nvic.c b/hw/intc/armv7m_nvic.c
74
index XXXXXXX..XXXXXXX 100644
75
--- a/hw/intc/armv7m_nvic.c
76
+++ b/hw/intc/armv7m_nvic.c
77
@@ -XXX,XX +XXX,XX @@ void armv7m_nvic_set_pending_derived(void *opaque, int irq, bool secure)
78
}
79
80
/* Make pending IRQ active. */
81
-bool armv7m_nvic_acknowledge_irq(void *opaque)
82
+void armv7m_nvic_acknowledge_irq(void *opaque)
83
{
84
NVICState *s = (NVICState *)opaque;
85
CPUARMState *env = &s->cpu->env;
86
const int pending = s->vectpending;
87
const int running = nvic_exec_prio(s);
88
VecInfo *vec;
89
- bool targets_secure;
90
91
assert(pending > ARMV7M_EXCP_RESET && pending < s->num_irq);
92
93
if (s->vectpending_is_s_banked) {
94
vec = &s->sec_vectors[pending];
95
- targets_secure = true;
96
} else {
97
vec = &s->vectors[pending];
98
- targets_secure = !exc_is_banked(s->vectpending) &&
99
- exc_targets_secure(s, s->vectpending);
100
}
101
102
assert(vec->enabled);
103
@@ -XXX,XX +XXX,XX @@ bool armv7m_nvic_acknowledge_irq(void *opaque)
104
105
assert(s->vectpending_prio < running);
106
107
- trace_nvic_acknowledge_irq(pending, s->vectpending_prio, targets_secure);
108
+ trace_nvic_acknowledge_irq(pending, s->vectpending_prio);
109
110
vec->active = 1;
111
vec->pending = 0;
112
@@ -XXX,XX +XXX,XX @@ bool armv7m_nvic_acknowledge_irq(void *opaque)
113
write_v7m_exception(env, s->vectpending);
114
115
nvic_irq_update(s);
116
+}
117
+
118
+void armv7m_nvic_get_pending_irq_info(void *opaque,
119
+ int *pirq, bool *ptargets_secure)
120
+{
121
+ NVICState *s = (NVICState *)opaque;
122
+ const int pending = s->vectpending;
123
+ bool targets_secure;
124
+
125
+ assert(pending > ARMV7M_EXCP_RESET && pending < s->num_irq);
126
+
127
+ if (s->vectpending_is_s_banked) {
128
+ targets_secure = true;
129
+ } else {
130
+ targets_secure = !exc_is_banked(pending) &&
131
+ exc_targets_secure(s, pending);
132
+ }
133
+
134
+ trace_nvic_get_pending_irq_info(pending, targets_secure);
135
136
- return targets_secure;
137
+ *ptargets_secure = targets_secure;
138
+ *pirq = pending;
139
}
140
141
int armv7m_nvic_complete_irq(void *opaque, int irq, bool secure)
142
diff --git a/target/arm/helper.c b/target/arm/helper.c
143
index XXXXXXX..XXXXXXX 100644
144
--- a/target/arm/helper.c
145
+++ b/target/arm/helper.c
146
@@ -XXX,XX +XXX,XX @@ static uint32_t *get_v7m_sp_ptr(CPUARMState *env, bool secure, bool threadmode,
147
}
148
}
149
150
-static uint32_t arm_v7m_load_vector(ARMCPU *cpu, bool targets_secure)
151
+static uint32_t arm_v7m_load_vector(ARMCPU *cpu, int exc, bool targets_secure)
152
{
153
CPUState *cs = CPU(cpu);
154
CPUARMState *env = &cpu->env;
155
MemTxResult result;
156
- hwaddr vec = env->v7m.vecbase[targets_secure] + env->v7m.exception * 4;
157
+ hwaddr vec = env->v7m.vecbase[targets_secure] + exc * 4;
158
uint32_t addr;
159
160
addr = address_space_ldl(cs->as, vec,
161
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain)
162
CPUARMState *env = &cpu->env;
163
uint32_t addr;
164
bool targets_secure;
165
+ int exc;
166
167
- targets_secure = armv7m_nvic_acknowledge_irq(env->nvic);
168
+ armv7m_nvic_get_pending_irq_info(env->nvic, &exc, &targets_secure);
169
170
if (arm_feature(env, ARM_FEATURE_V8)) {
171
if (arm_feature(env, ARM_FEATURE_M_SECURITY) &&
172
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain)
173
}
174
}
175
176
+ addr = arm_v7m_load_vector(cpu, exc, targets_secure);
177
+
178
+ /* Now we've done everything that might cause a derived exception
179
+ * we can go ahead and activate whichever exception we're going to
180
+ * take (which might now be the derived exception).
181
+ */
182
+ armv7m_nvic_acknowledge_irq(env->nvic);
183
+
184
/* Switch to target security state -- must do this before writing SPSEL */
185
switch_v7m_security_state(env, targets_secure);
186
write_v7m_control_spsel(env, 0);
187
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain)
188
/* Clear IT bits */
189
env->condexec_bits = 0;
190
env->regs[14] = lr;
191
- addr = arm_v7m_load_vector(cpu, targets_secure);
192
env->regs[15] = addr & 0xfffffffe;
193
env->thumb = addr & 1;
194
}
195
diff --git a/hw/intc/trace-events b/hw/intc/trace-events
196
index XXXXXXX..XXXXXXX 100644
197
--- a/hw/intc/trace-events
198
+++ b/hw/intc/trace-events
199
@@ -XXX,XX +XXX,XX @@ nvic_escalate_disabled(int irq) "NVIC escalating irq %d to HardFault: disabled"
200
nvic_set_pending(int irq, bool secure, bool derived, int en, int prio) "NVIC set pending irq %d secure-bank %d derived %d (enabled: %d priority %d)"
201
nvic_clear_pending(int irq, bool secure, int en, int prio) "NVIC clear pending irq %d secure-bank %d (enabled: %d priority %d)"
202
nvic_set_pending_level(int irq) "NVIC set pending: irq %d higher prio than vectpending: setting irq line to 1"
203
-nvic_acknowledge_irq(int irq, int prio, bool targets_secure) "NVIC acknowledge IRQ: %d now active (prio %d targets_secure %d)"
204
+nvic_acknowledge_irq(int irq, int prio) "NVIC acknowledge IRQ: %d now active (prio %d)"
205
+nvic_get_pending_irq_info(int irq, bool secure) "NVIC next IRQ %d: targets_secure: %d"
206
nvic_complete_irq(int irq, bool secure) "NVIC complete IRQ %d (secure %d)"
207
nvic_set_irq_level(int irq, int level) "NVIC external irq %d level set to %d"
208
nvic_sysreg_read(uint64_t addr, uint32_t value, unsigned size) "NVIC sysreg read addr 0x%" PRIx64 " data 0x%" PRIx32 " size %u"
209
--
210
2.16.1
211
212
diff view generated by jsdifflib
1
From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
1
From: Alex Bennée <alex.bennee@linaro.org>
2
2
3
Add support for the new ARMv8.2 SHA-3, SM3, SM4 and SHA-512 instructions to
3
Reported by Coverity (CID1390635). We ensure this for uint_to_float
4
AArch64 user mode emulation.
4
later on so we might as well mirror that.
5
5
6
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
6
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
7
Message-id: 20180207111729.15737-6-ard.biesheuvel@linaro.org
7
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
---
10
---
11
linux-user/elfload.c | 19 +++++++++++++++++++
11
fpu/softfloat.c | 2 +-
12
target/arm/cpu64.c | 4 ++++
12
1 file changed, 1 insertion(+), 1 deletion(-)
13
2 files changed, 23 insertions(+)
14
13
15
diff --git a/linux-user/elfload.c b/linux-user/elfload.c
14
diff --git a/fpu/softfloat.c b/fpu/softfloat.c
16
index XXXXXXX..XXXXXXX 100644
15
index XXXXXXX..XXXXXXX 100644
17
--- a/linux-user/elfload.c
16
--- a/fpu/softfloat.c
18
+++ b/linux-user/elfload.c
17
+++ b/fpu/softfloat.c
19
@@ -XXX,XX +XXX,XX @@ enum {
18
@@ -XXX,XX +XXX,XX @@ FLOAT_TO_UINT(64, 64)
20
ARM_HWCAP_A64_SHA1 = 1 << 5,
19
21
ARM_HWCAP_A64_SHA2 = 1 << 6,
20
static FloatParts int_to_float(int64_t a, float_status *status)
22
ARM_HWCAP_A64_CRC32 = 1 << 7,
21
{
23
+ ARM_HWCAP_A64_ATOMICS = 1 << 8,
22
- FloatParts r;
24
+ ARM_HWCAP_A64_FPHP = 1 << 9,
23
+ FloatParts r = {};
25
+ ARM_HWCAP_A64_ASIMDHP = 1 << 10,
24
if (a == 0) {
26
+ ARM_HWCAP_A64_CPUID = 1 << 11,
25
r.cls = float_class_zero;
27
+ ARM_HWCAP_A64_ASIMDRDM = 1 << 12,
26
r.sign = false;
28
+ ARM_HWCAP_A64_JSCVT = 1 << 13,
29
+ ARM_HWCAP_A64_FCMA = 1 << 14,
30
+ ARM_HWCAP_A64_LRCPC = 1 << 15,
31
+ ARM_HWCAP_A64_DCPOP = 1 << 16,
32
+ ARM_HWCAP_A64_SHA3 = 1 << 17,
33
+ ARM_HWCAP_A64_SM3 = 1 << 18,
34
+ ARM_HWCAP_A64_SM4 = 1 << 19,
35
+ ARM_HWCAP_A64_ASIMDDP = 1 << 20,
36
+ ARM_HWCAP_A64_SHA512 = 1 << 21,
37
+ ARM_HWCAP_A64_SVE = 1 << 22,
38
};
39
40
#define ELF_HWCAP get_elf_hwcap()
41
@@ -XXX,XX +XXX,XX @@ static uint32_t get_elf_hwcap(void)
42
GET_FEATURE(ARM_FEATURE_V8_SHA1, ARM_HWCAP_A64_SHA1);
43
GET_FEATURE(ARM_FEATURE_V8_SHA256, ARM_HWCAP_A64_SHA2);
44
GET_FEATURE(ARM_FEATURE_CRC, ARM_HWCAP_A64_CRC32);
45
+ GET_FEATURE(ARM_FEATURE_V8_SHA3, ARM_HWCAP_A64_SHA3);
46
+ GET_FEATURE(ARM_FEATURE_V8_SM3, ARM_HWCAP_A64_SM3);
47
+ GET_FEATURE(ARM_FEATURE_V8_SM4, ARM_HWCAP_A64_SM4);
48
+ GET_FEATURE(ARM_FEATURE_V8_SHA512, ARM_HWCAP_A64_SHA512);
49
#undef GET_FEATURE
50
51
return hwcaps;
52
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
53
index XXXXXXX..XXXXXXX 100644
54
--- a/target/arm/cpu64.c
55
+++ b/target/arm/cpu64.c
56
@@ -XXX,XX +XXX,XX @@ static void aarch64_any_initfn(Object *obj)
57
set_feature(&cpu->env, ARM_FEATURE_V8_AES);
58
set_feature(&cpu->env, ARM_FEATURE_V8_SHA1);
59
set_feature(&cpu->env, ARM_FEATURE_V8_SHA256);
60
+ set_feature(&cpu->env, ARM_FEATURE_V8_SHA512);
61
+ set_feature(&cpu->env, ARM_FEATURE_V8_SHA3);
62
+ set_feature(&cpu->env, ARM_FEATURE_V8_SM3);
63
+ set_feature(&cpu->env, ARM_FEATURE_V8_SM4);
64
set_feature(&cpu->env, ARM_FEATURE_V8_PMULL);
65
set_feature(&cpu->env, ARM_FEATURE_CRC);
66
cpu->ctr = 0x80038003; /* 32 byte I and D cacheline size, VIPT icache */
67
--
27
--
68
2.16.1
28
2.17.0
69
29
70
30
diff view generated by jsdifflib
1
Handle possible MPU faults, SAU faults or bus errors when
1
In float-to-integer conversion, if the floating point input
2
popping register state off the stack during exception return.
2
converts exactly to the largest or smallest integer that
3
fits in to the result type, this is not an overflow.
4
In this situation we were producing the correct result value,
5
but were incorrectly setting the Invalid flag.
6
For example for Arm A64, "FCVTAS w0, d0" on an input of
7
0x41dfffffffc00000 should produce 0x7fffffff and set no flags.
3
8
9
Fix the boundary case to take the right half of the if()
10
statements.
11
12
This fixes a regression from 2.11 introduced by the softfloat
13
refactoring.
14
15
Cc: qemu-stable@nongnu.org
16
Fixes: ab52f973a50
4
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
5
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
18
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
6
Message-id: 1517324542-6607-8-git-send-email-peter.maydell@linaro.org
19
Message-id: 20180510140141.12120-1-peter.maydell@linaro.org
7
---
20
---
8
target/arm/helper.c | 115 ++++++++++++++++++++++++++++++++++++++++++----------
21
fpu/softfloat.c | 4 ++--
9
1 file changed, 94 insertions(+), 21 deletions(-)
22
1 file changed, 2 insertions(+), 2 deletions(-)
10
23
11
diff --git a/target/arm/helper.c b/target/arm/helper.c
24
diff --git a/fpu/softfloat.c b/fpu/softfloat.c
12
index XXXXXXX..XXXXXXX 100644
25
index XXXXXXX..XXXXXXX 100644
13
--- a/target/arm/helper.c
26
--- a/fpu/softfloat.c
14
+++ b/target/arm/helper.c
27
+++ b/fpu/softfloat.c
15
@@ -XXX,XX +XXX,XX @@ pend_fault:
28
@@ -XXX,XX +XXX,XX @@ static int64_t round_to_int_and_pack(FloatParts in, int rmode,
16
return false;
29
r = UINT64_MAX;
17
}
30
}
18
31
if (p.sign) {
19
+static bool v7m_stack_read(ARMCPU *cpu, uint32_t *dest, uint32_t addr,
32
- if (r < -(uint64_t) min) {
20
+ ARMMMUIdx mmu_idx)
33
+ if (r <= -(uint64_t) min) {
21
+{
34
return -r;
22
+ CPUState *cs = CPU(cpu);
35
} else {
23
+ CPUARMState *env = &cpu->env;
36
s->float_exception_flags = orig_flags | float_flag_invalid;
24
+ MemTxAttrs attrs = {};
37
return min;
25
+ MemTxResult txres;
26
+ target_ulong page_size;
27
+ hwaddr physaddr;
28
+ int prot;
29
+ ARMMMUFaultInfo fi;
30
+ bool secure = mmu_idx & ARM_MMU_IDX_M_S;
31
+ int exc;
32
+ bool exc_secure;
33
+ uint32_t value;
34
+
35
+ if (get_phys_addr(env, addr, MMU_DATA_LOAD, mmu_idx, &physaddr,
36
+ &attrs, &prot, &page_size, &fi, NULL)) {
37
+ /* MPU/SAU lookup failed */
38
+ if (fi.type == ARMFault_QEMU_SFault) {
39
+ qemu_log_mask(CPU_LOG_INT,
40
+ "...SecureFault with SFSR.AUVIOL during unstack\n");
41
+ env->v7m.sfsr |= R_V7M_SFSR_AUVIOL_MASK | R_V7M_SFSR_SFARVALID_MASK;
42
+ env->v7m.sfar = addr;
43
+ exc = ARMV7M_EXCP_SECURE;
44
+ exc_secure = false;
45
+ } else {
46
+ qemu_log_mask(CPU_LOG_INT,
47
+ "...MemManageFault with CFSR.MUNSTKERR\n");
48
+ env->v7m.cfsr[secure] |= R_V7M_CFSR_MUNSTKERR_MASK;
49
+ exc = ARMV7M_EXCP_MEM;
50
+ exc_secure = secure;
51
+ }
52
+ goto pend_fault;
53
+ }
54
+
55
+ value = address_space_ldl(arm_addressspace(cs, attrs), physaddr,
56
+ attrs, &txres);
57
+ if (txres != MEMTX_OK) {
58
+ /* BusFault trying to read the data */
59
+ qemu_log_mask(CPU_LOG_INT, "...BusFault with BFSR.UNSTKERR\n");
60
+ env->v7m.cfsr[M_REG_NS] |= R_V7M_CFSR_UNSTKERR_MASK;
61
+ exc = ARMV7M_EXCP_BUS;
62
+ exc_secure = false;
63
+ goto pend_fault;
64
+ }
65
+
66
+ *dest = value;
67
+ return true;
68
+
69
+pend_fault:
70
+ /* By pending the exception at this point we are making
71
+ * the IMPDEF choice "overridden exceptions pended" (see the
72
+ * MergeExcInfo() pseudocode). The other choice would be to not
73
+ * pend them now and then make a choice about which to throw away
74
+ * later if we have two derived exceptions.
75
+ */
76
+ armv7m_nvic_set_pending(env->nvic, exc, exc_secure);
77
+ return false;
78
+}
79
+
80
/* Return true if we're using the process stack pointer (not the MSP) */
81
static bool v7m_using_psp(CPUARMState *env)
82
{
83
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
84
!return_to_handler,
85
return_to_sp_process);
86
uint32_t frameptr = *frame_sp_p;
87
+ bool pop_ok = true;
88
+ ARMMMUIdx mmu_idx;
89
+
90
+ mmu_idx = arm_v7m_mmu_idx_for_secstate_and_priv(env, return_to_secure,
91
+ !return_to_handler);
92
93
if (!QEMU_IS_ALIGNED(frameptr, 8) &&
94
arm_feature(env, ARM_FEATURE_V8)) {
95
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
96
return;
97
}
38
}
98
39
} else {
99
- env->regs[4] = ldl_phys(cs->as, frameptr + 0x8);
40
- if (r < max) {
100
- env->regs[5] = ldl_phys(cs->as, frameptr + 0xc);
41
+ if (r <= max) {
101
- env->regs[6] = ldl_phys(cs->as, frameptr + 0x10);
42
return r;
102
- env->regs[7] = ldl_phys(cs->as, frameptr + 0x14);
43
} else {
103
- env->regs[8] = ldl_phys(cs->as, frameptr + 0x18);
44
s->float_exception_flags = orig_flags | float_flag_invalid;
104
- env->regs[9] = ldl_phys(cs->as, frameptr + 0x1c);
105
- env->regs[10] = ldl_phys(cs->as, frameptr + 0x20);
106
- env->regs[11] = ldl_phys(cs->as, frameptr + 0x24);
107
+ pop_ok =
108
+ v7m_stack_read(cpu, &env->regs[4], frameptr + 0x8, mmu_idx) &&
109
+ v7m_stack_read(cpu, &env->regs[4], frameptr + 0x8, mmu_idx) &&
110
+ v7m_stack_read(cpu, &env->regs[5], frameptr + 0xc, mmu_idx) &&
111
+ v7m_stack_read(cpu, &env->regs[6], frameptr + 0x10, mmu_idx) &&
112
+ v7m_stack_read(cpu, &env->regs[7], frameptr + 0x14, mmu_idx) &&
113
+ v7m_stack_read(cpu, &env->regs[8], frameptr + 0x18, mmu_idx) &&
114
+ v7m_stack_read(cpu, &env->regs[9], frameptr + 0x1c, mmu_idx) &&
115
+ v7m_stack_read(cpu, &env->regs[10], frameptr + 0x20, mmu_idx) &&
116
+ v7m_stack_read(cpu, &env->regs[11], frameptr + 0x24, mmu_idx);
117
118
frameptr += 0x28;
119
}
120
121
- /* Pop registers. TODO: make these accesses use the correct
122
- * attributes and address space (S/NS, priv/unpriv) and handle
123
- * memory transaction failures.
124
- */
125
- env->regs[0] = ldl_phys(cs->as, frameptr);
126
- env->regs[1] = ldl_phys(cs->as, frameptr + 0x4);
127
- env->regs[2] = ldl_phys(cs->as, frameptr + 0x8);
128
- env->regs[3] = ldl_phys(cs->as, frameptr + 0xc);
129
- env->regs[12] = ldl_phys(cs->as, frameptr + 0x10);
130
- env->regs[14] = ldl_phys(cs->as, frameptr + 0x14);
131
- env->regs[15] = ldl_phys(cs->as, frameptr + 0x18);
132
+ /* Pop registers */
133
+ pop_ok = pop_ok &&
134
+ v7m_stack_read(cpu, &env->regs[0], frameptr, mmu_idx) &&
135
+ v7m_stack_read(cpu, &env->regs[1], frameptr + 0x4, mmu_idx) &&
136
+ v7m_stack_read(cpu, &env->regs[2], frameptr + 0x8, mmu_idx) &&
137
+ v7m_stack_read(cpu, &env->regs[3], frameptr + 0xc, mmu_idx) &&
138
+ v7m_stack_read(cpu, &env->regs[12], frameptr + 0x10, mmu_idx) &&
139
+ v7m_stack_read(cpu, &env->regs[14], frameptr + 0x14, mmu_idx) &&
140
+ v7m_stack_read(cpu, &env->regs[15], frameptr + 0x18, mmu_idx) &&
141
+ v7m_stack_read(cpu, &xpsr, frameptr + 0x1c, mmu_idx);
142
+
143
+ if (!pop_ok) {
144
+ /* v7m_stack_read() pended a fault, so take it (as a tail
145
+ * chained exception on the same stack frame)
146
+ */
147
+ v7m_exception_taken(cpu, excret, true, false);
148
+ return;
149
+ }
150
151
/* Returning from an exception with a PC with bit 0 set is defined
152
* behaviour on v8M (bit 0 is ignored), but for v7M it was specified
153
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
154
}
155
}
156
157
- xpsr = ldl_phys(cs->as, frameptr + 0x1c);
158
-
159
if (arm_feature(env, ARM_FEATURE_V8)) {
160
/* For v8M we have to check whether the xPSR exception field
161
* matches the EXCRET value for return to handler/thread
162
--
45
--
163
2.16.1
46
2.17.0
164
47
165
48
diff view generated by jsdifflib
1
The documentation for the generic loader claims that you can
1
In commit d81ce0ef2c4f105 we added an extra float_status field
2
set the PC for a CPU with an option of the form
2
fp_status_fp16 for Arm, but forgot to initialize it correctly
3
-device loader,cpu-num=0,addr=0x10000004
3
by setting it to float_tininess_before_rounding. This currently
4
will only cause problems for the new V8_FP16 feature, since the
5
float-to-float conversion code doesn't use it yet. The effect
6
would be that we failed to set the Underflow IEEE exception flag
7
in all the cases where we should.
4
8
5
However if you try this QEMU complains:
9
Add the missing initialization.
6
cpu_num must be specified when setting a program counter
7
10
8
This is because we were testing against 0 rather than CPU_NONE.
11
Fixes: d81ce0ef2c4f105
12
Cc: qemu-stable@nongnu.org
13
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
14
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
15
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
Message-id: 20180512004311.9299-16-richard.henderson@linaro.org
17
---
18
target/arm/cpu.c | 2 ++
19
1 file changed, 2 insertions(+)
9
20
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
21
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
11
Reviewed-by: Alistair Francis <alistair.francis@xilinx.com>
12
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
13
Message-id: 20180205150426.20542-1-peter.maydell@linaro.org
14
---
15
hw/core/generic-loader.c | 2 +-
16
1 file changed, 1 insertion(+), 1 deletion(-)
17
18
diff --git a/hw/core/generic-loader.c b/hw/core/generic-loader.c
19
index XXXXXXX..XXXXXXX 100644
22
index XXXXXXX..XXXXXXX 100644
20
--- a/hw/core/generic-loader.c
23
--- a/target/arm/cpu.c
21
+++ b/hw/core/generic-loader.c
24
+++ b/target/arm/cpu.c
22
@@ -XXX,XX +XXX,XX @@ static void generic_loader_realize(DeviceState *dev, Error **errp)
25
@@ -XXX,XX +XXX,XX @@ static void arm_cpu_reset(CPUState *s)
23
error_setg(errp, "data can not be specified when setting a "
26
&env->vfp.fp_status);
24
"program counter");
27
set_float_detect_tininess(float_tininess_before_rounding,
25
return;
28
&env->vfp.standard_fp_status);
26
- } else if (!s->cpu_num) {
29
+ set_float_detect_tininess(float_tininess_before_rounding,
27
+ } else if (s->cpu_num == CPU_NONE) {
30
+ &env->vfp.fp_status_f16);
28
error_setg(errp, "cpu_num must be specified when setting a "
31
#ifndef CONFIG_USER_ONLY
29
"program counter");
32
if (kvm_enabled()) {
30
return;
33
kvm_arm_reset_vcpu(cpu);
31
--
34
--
32
2.16.1
35
2.17.0
33
36
34
37
diff view generated by jsdifflib
1
From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
This implements emulation of the new SM3 instructions that have
3
Adding the fp16 moves to/from general registers.
4
been added as an optional extension to the ARMv8 Crypto Extensions
5
in ARM v8.2.
6
4
7
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
5
Cc: qemu-stable@nongnu.org
8
Message-id: 20180207111729.15737-4-ard.biesheuvel@linaro.org
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Tested-by: Alex Bennée <alex.bennee@linaro.org>
8
Message-id: 20180512003217.9105-2-richard.henderson@linaro.org
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
11
---
12
target/arm/cpu.h | 1 +
12
target/arm/translate-a64.c | 21 +++++++++++++++++++++
13
target/arm/helper.h | 4 ++
13
1 file changed, 21 insertions(+)
14
target/arm/crypto_helper.c | 96 ++++++++++++++++++++++++++++++++++++++++++++++
15
target/arm/translate-a64.c | 88 ++++++++++++++++++++++++++++++++++++++++--
16
4 files changed, 186 insertions(+), 3 deletions(-)
17
14
18
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
19
index XXXXXXX..XXXXXXX 100644
20
--- a/target/arm/cpu.h
21
+++ b/target/arm/cpu.h
22
@@ -XXX,XX +XXX,XX @@ enum arm_features {
23
ARM_FEATURE_SVE, /* has Scalable Vector Extension */
24
ARM_FEATURE_V8_SHA512, /* implements SHA512 part of v8 Crypto Extensions */
25
ARM_FEATURE_V8_SHA3, /* implements SHA3 part of v8 Crypto Extensions */
26
+ ARM_FEATURE_V8_SM3, /* implements SM3 part of v8 Crypto Extensions */
27
};
28
29
static inline int arm_feature(CPUARMState *env, int feature)
30
diff --git a/target/arm/helper.h b/target/arm/helper.h
31
index XXXXXXX..XXXXXXX 100644
32
--- a/target/arm/helper.h
33
+++ b/target/arm/helper.h
34
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_3(crypto_sha512h2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
35
DEF_HELPER_FLAGS_2(crypto_sha512su0, TCG_CALL_NO_RWG, void, ptr, ptr)
36
DEF_HELPER_FLAGS_3(crypto_sha512su1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
37
38
+DEF_HELPER_FLAGS_5(crypto_sm3tt, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32, i32)
39
+DEF_HELPER_FLAGS_3(crypto_sm3partw1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
40
+DEF_HELPER_FLAGS_3(crypto_sm3partw2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
41
+
42
DEF_HELPER_FLAGS_3(crc32, TCG_CALL_NO_RWG_SE, i32, i32, i32, i32)
43
DEF_HELPER_FLAGS_3(crc32c, TCG_CALL_NO_RWG_SE, i32, i32, i32, i32)
44
DEF_HELPER_2(dc_zva, void, env, i64)
45
diff --git a/target/arm/crypto_helper.c b/target/arm/crypto_helper.c
46
index XXXXXXX..XXXXXXX 100644
47
--- a/target/arm/crypto_helper.c
48
+++ b/target/arm/crypto_helper.c
49
@@ -XXX,XX +XXX,XX @@ void HELPER(crypto_sha512su1)(void *vd, void *vn, void *vm)
50
rd[0] += s1_512(rn[0]) + rm[0];
51
rd[1] += s1_512(rn[1]) + rm[1];
52
}
53
+
54
+void HELPER(crypto_sm3partw1)(void *vd, void *vn, void *vm)
55
+{
56
+ uint64_t *rd = vd;
57
+ uint64_t *rn = vn;
58
+ uint64_t *rm = vm;
59
+ union CRYPTO_STATE d = { .l = { rd[0], rd[1] } };
60
+ union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
61
+ union CRYPTO_STATE m = { .l = { rm[0], rm[1] } };
62
+ uint32_t t;
63
+
64
+ t = CR_ST_WORD(d, 0) ^ CR_ST_WORD(n, 0) ^ ror32(CR_ST_WORD(m, 1), 17);
65
+ CR_ST_WORD(d, 0) = t ^ ror32(t, 17) ^ ror32(t, 9);
66
+
67
+ t = CR_ST_WORD(d, 1) ^ CR_ST_WORD(n, 1) ^ ror32(CR_ST_WORD(m, 2), 17);
68
+ CR_ST_WORD(d, 1) = t ^ ror32(t, 17) ^ ror32(t, 9);
69
+
70
+ t = CR_ST_WORD(d, 2) ^ CR_ST_WORD(n, 2) ^ ror32(CR_ST_WORD(m, 3), 17);
71
+ CR_ST_WORD(d, 2) = t ^ ror32(t, 17) ^ ror32(t, 9);
72
+
73
+ t = CR_ST_WORD(d, 3) ^ CR_ST_WORD(n, 3) ^ ror32(CR_ST_WORD(d, 0), 17);
74
+ CR_ST_WORD(d, 3) = t ^ ror32(t, 17) ^ ror32(t, 9);
75
+
76
+ rd[0] = d.l[0];
77
+ rd[1] = d.l[1];
78
+}
79
+
80
+void HELPER(crypto_sm3partw2)(void *vd, void *vn, void *vm)
81
+{
82
+ uint64_t *rd = vd;
83
+ uint64_t *rn = vn;
84
+ uint64_t *rm = vm;
85
+ union CRYPTO_STATE d = { .l = { rd[0], rd[1] } };
86
+ union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
87
+ union CRYPTO_STATE m = { .l = { rm[0], rm[1] } };
88
+ uint32_t t = CR_ST_WORD(n, 0) ^ ror32(CR_ST_WORD(m, 0), 25);
89
+
90
+ CR_ST_WORD(d, 0) ^= t;
91
+ CR_ST_WORD(d, 1) ^= CR_ST_WORD(n, 1) ^ ror32(CR_ST_WORD(m, 1), 25);
92
+ CR_ST_WORD(d, 2) ^= CR_ST_WORD(n, 2) ^ ror32(CR_ST_WORD(m, 2), 25);
93
+ CR_ST_WORD(d, 3) ^= CR_ST_WORD(n, 3) ^ ror32(CR_ST_WORD(m, 3), 25) ^
94
+ ror32(t, 17) ^ ror32(t, 2) ^ ror32(t, 26);
95
+
96
+ rd[0] = d.l[0];
97
+ rd[1] = d.l[1];
98
+}
99
+
100
+void HELPER(crypto_sm3tt)(void *vd, void *vn, void *vm, uint32_t imm2,
101
+ uint32_t opcode)
102
+{
103
+ uint64_t *rd = vd;
104
+ uint64_t *rn = vn;
105
+ uint64_t *rm = vm;
106
+ union CRYPTO_STATE d = { .l = { rd[0], rd[1] } };
107
+ union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
108
+ union CRYPTO_STATE m = { .l = { rm[0], rm[1] } };
109
+ uint32_t t;
110
+
111
+ assert(imm2 < 4);
112
+
113
+ if (opcode == 0 || opcode == 2) {
114
+ /* SM3TT1A, SM3TT2A */
115
+ t = par(CR_ST_WORD(d, 3), CR_ST_WORD(d, 2), CR_ST_WORD(d, 1));
116
+ } else if (opcode == 1) {
117
+ /* SM3TT1B */
118
+ t = maj(CR_ST_WORD(d, 3), CR_ST_WORD(d, 2), CR_ST_WORD(d, 1));
119
+ } else if (opcode == 3) {
120
+ /* SM3TT2B */
121
+ t = cho(CR_ST_WORD(d, 3), CR_ST_WORD(d, 2), CR_ST_WORD(d, 1));
122
+ } else {
123
+ g_assert_not_reached();
124
+ }
125
+
126
+ t += CR_ST_WORD(d, 0) + CR_ST_WORD(m, imm2);
127
+
128
+ CR_ST_WORD(d, 0) = CR_ST_WORD(d, 1);
129
+
130
+ if (opcode < 2) {
131
+ /* SM3TT1A, SM3TT1B */
132
+ t += CR_ST_WORD(n, 3) ^ ror32(CR_ST_WORD(d, 3), 20);
133
+
134
+ CR_ST_WORD(d, 1) = ror32(CR_ST_WORD(d, 2), 23);
135
+ } else {
136
+ /* SM3TT2A, SM3TT2B */
137
+ t += CR_ST_WORD(n, 3);
138
+ t ^= rol32(t, 9) ^ rol32(t, 17);
139
+
140
+ CR_ST_WORD(d, 1) = ror32(CR_ST_WORD(d, 2), 13);
141
+ }
142
+
143
+ CR_ST_WORD(d, 2) = CR_ST_WORD(d, 3);
144
+ CR_ST_WORD(d, 3) = t;
145
+
146
+ rd[0] = d.l[0];
147
+ rd[1] = d.l[1];
148
+}
149
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
15
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
150
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
151
--- a/target/arm/translate-a64.c
17
--- a/target/arm/translate-a64.c
152
+++ b/target/arm/translate-a64.c
18
+++ b/target/arm/translate-a64.c
153
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
19
@@ -XXX,XX +XXX,XX @@ static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof)
20
tcg_gen_st_i64(tcg_rn, cpu_env, fp_reg_hi_offset(s, rd));
21
clear_vec_high(s, true, rd);
154
break;
22
break;
23
+ case 3:
24
+ /* 16 bit */
25
+ tmp = tcg_temp_new_i64();
26
+ tcg_gen_ext16u_i64(tmp, tcg_rn);
27
+ write_fp_dreg(s, rd, tmp);
28
+ tcg_temp_free_i64(tmp);
29
+ break;
30
+ default:
31
+ g_assert_not_reached();
155
}
32
}
156
} else {
33
} else {
157
- unallocated_encoding(s);
34
TCGv_i64 tcg_rd = cpu_reg(s, rd);
158
- return;
35
@@ -XXX,XX +XXX,XX @@ static void handle_fmov(DisasContext *s, int rd, int rn, int type, bool itof)
159
+ switch (opcode) {
36
/* 64 bits from top half */
160
+ case 0: /* SM3PARTW1 */
37
tcg_gen_ld_i64(tcg_rd, cpu_env, fp_reg_hi_offset(s, rn));
161
+ feature = ARM_FEATURE_V8_SM3;
38
break;
162
+ genfn = gen_helper_crypto_sm3partw1;
39
+ case 3:
163
+ break;
40
+ /* 16 bit */
164
+ case 1: /* SM3PARTW2 */
41
+ tcg_gen_ld16u_i64(tcg_rd, cpu_env, fp_reg_offset(s, rn, MO_16));
165
+ feature = ARM_FEATURE_V8_SM3;
166
+ genfn = gen_helper_crypto_sm3partw2;
167
+ break;
42
+ break;
168
+ default:
43
+ default:
169
+ unallocated_encoding(s);
44
+ g_assert_not_reached();
170
+ return;
45
}
171
+ }
172
}
173
174
if (!arm_dc_feature(s, feature)) {
175
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_four_reg(DisasContext *s, uint32_t insn)
176
case 1: /* BCAX */
177
feature = ARM_FEATURE_V8_SHA3;
178
break;
179
+ case 2: /* SM3SS1 */
180
+ feature = ARM_FEATURE_V8_SM3;
181
+ break;
182
default:
183
unallocated_encoding(s);
184
return;
185
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_four_reg(DisasContext *s, uint32_t insn)
186
tcg_temp_free_i64(tcg_res[0]);
187
tcg_temp_free_i64(tcg_res[1]);
188
} else {
189
- g_assert_not_reached();
190
+ TCGv_i32 tcg_op1, tcg_op2, tcg_op3, tcg_res, tcg_zero;
191
+
192
+ tcg_op1 = tcg_temp_new_i32();
193
+ tcg_op2 = tcg_temp_new_i32();
194
+ tcg_op3 = tcg_temp_new_i32();
195
+ tcg_res = tcg_temp_new_i32();
196
+ tcg_zero = tcg_const_i32(0);
197
+
198
+ read_vec_element_i32(s, tcg_op1, rn, 3, MO_32);
199
+ read_vec_element_i32(s, tcg_op2, rm, 3, MO_32);
200
+ read_vec_element_i32(s, tcg_op3, ra, 3, MO_32);
201
+
202
+ tcg_gen_rotri_i32(tcg_res, tcg_op1, 20);
203
+ tcg_gen_add_i32(tcg_res, tcg_res, tcg_op2);
204
+ tcg_gen_add_i32(tcg_res, tcg_res, tcg_op3);
205
+ tcg_gen_rotri_i32(tcg_res, tcg_res, 25);
206
+
207
+ write_vec_element_i32(s, tcg_zero, rd, 0, MO_32);
208
+ write_vec_element_i32(s, tcg_zero, rd, 1, MO_32);
209
+ write_vec_element_i32(s, tcg_zero, rd, 2, MO_32);
210
+ write_vec_element_i32(s, tcg_res, rd, 3, MO_32);
211
+
212
+ tcg_temp_free_i32(tcg_op1);
213
+ tcg_temp_free_i32(tcg_op2);
214
+ tcg_temp_free_i32(tcg_op3);
215
+ tcg_temp_free_i32(tcg_res);
216
+ tcg_temp_free_i32(tcg_zero);
217
}
46
}
218
}
47
}
219
48
@@ -XXX,XX +XXX,XX @@ static void disas_fp_int_conv(DisasContext *s, uint32_t insn)
220
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_xar(DisasContext *s, uint32_t insn)
49
case 0xa: /* 64 bit */
221
tcg_temp_free_i64(tcg_res[1]);
50
case 0xd: /* 64 bit to top half of quad */
222
}
51
break;
223
52
+ case 0x6: /* 16-bit float, 32-bit int */
224
+/* Crypto three-reg imm2
53
+ case 0xe: /* 16-bit float, 64-bit int */
225
+ * 31 21 20 16 15 14 13 12 11 10 9 5 4 0
54
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
226
+ * +-----------------------+------+-----+------+--------+------+------+
55
+ break;
227
+ * | 1 1 0 0 1 1 1 0 0 1 0 | Rm | 1 0 | imm2 | opcode | Rn | Rd |
56
+ }
228
+ * +-----------------------+------+-----+------+--------+------+------+
57
+ /* fallthru */
229
+ */
58
default:
230
+static void disas_crypto_three_reg_imm2(DisasContext *s, uint32_t insn)
59
/* all other sf/type/rmode combinations are invalid */
231
+{
60
unallocated_encoding(s);
232
+ int opcode = extract32(insn, 10, 2);
233
+ int imm2 = extract32(insn, 12, 2);
234
+ int rm = extract32(insn, 16, 5);
235
+ int rn = extract32(insn, 5, 5);
236
+ int rd = extract32(insn, 0, 5);
237
+ TCGv_ptr tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr;
238
+ TCGv_i32 tcg_imm2, tcg_opcode;
239
+
240
+ if (!arm_dc_feature(s, ARM_FEATURE_V8_SM3)) {
241
+ unallocated_encoding(s);
242
+ return;
243
+ }
244
+
245
+ if (!fp_access_check(s)) {
246
+ return;
247
+ }
248
+
249
+ tcg_rd_ptr = vec_full_reg_ptr(s, rd);
250
+ tcg_rn_ptr = vec_full_reg_ptr(s, rn);
251
+ tcg_rm_ptr = vec_full_reg_ptr(s, rm);
252
+ tcg_imm2 = tcg_const_i32(imm2);
253
+ tcg_opcode = tcg_const_i32(opcode);
254
+
255
+ gen_helper_crypto_sm3tt(tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr, tcg_imm2,
256
+ tcg_opcode);
257
+
258
+ tcg_temp_free_ptr(tcg_rd_ptr);
259
+ tcg_temp_free_ptr(tcg_rn_ptr);
260
+ tcg_temp_free_ptr(tcg_rm_ptr);
261
+ tcg_temp_free_i32(tcg_imm2);
262
+ tcg_temp_free_i32(tcg_opcode);
263
+}
264
+
265
/* C3.6 Data processing - SIMD, inc Crypto
266
*
267
* As the decode gets a little complex we are using a table based
268
@@ -XXX,XX +XXX,XX @@ static const AArch64DecodeTable data_proc_simd[] = {
269
{ 0xcec08000, 0xfffff000, disas_crypto_two_reg_sha512 },
270
{ 0xce000000, 0xff808000, disas_crypto_four_reg },
271
{ 0xce800000, 0xffe00000, disas_crypto_xar },
272
+ { 0xce408000, 0xffe0c000, disas_crypto_three_reg_imm2 },
273
{ 0x00000000, 0x00000000, NULL }
274
};
275
276
--
61
--
277
2.16.1
62
2.17.0
278
63
279
64
diff view generated by jsdifflib
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
Add both SVE exception state and vector length.
3
No sense in emitting code after the exception.
4
4
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Tested-by: Alex Bennée <alex.bennee@linaro.org>
7
Message-id: 20180512003217.9105-3-richard.henderson@linaro.org
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Message-id: 20180123035349.24538-6-richard.henderson@linaro.org
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
---
10
---
10
target/arm/cpu.h | 8 ++++++++
11
target/arm/translate-a64.c | 2 +-
11
target/arm/translate.h | 2 ++
12
1 file changed, 1 insertion(+), 1 deletion(-)
12
target/arm/helper.c | 25 ++++++++++++++++++++++++-
13
target/arm/translate-a64.c | 2 ++
14
4 files changed, 36 insertions(+), 1 deletion(-)
15
13
16
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
17
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/cpu.h
19
+++ b/target/arm/cpu.h
20
@@ -XXX,XX +XXX,XX @@ static inline bool arm_cpu_data_is_big_endian(CPUARMState *env)
21
#define ARM_TBFLAG_TBI0_MASK (0x1ull << ARM_TBFLAG_TBI0_SHIFT)
22
#define ARM_TBFLAG_TBI1_SHIFT 1 /* TBI1 for EL0/1 */
23
#define ARM_TBFLAG_TBI1_MASK (0x1ull << ARM_TBFLAG_TBI1_SHIFT)
24
+#define ARM_TBFLAG_SVEEXC_EL_SHIFT 2
25
+#define ARM_TBFLAG_SVEEXC_EL_MASK (0x3 << ARM_TBFLAG_SVEEXC_EL_SHIFT)
26
+#define ARM_TBFLAG_ZCR_LEN_SHIFT 4
27
+#define ARM_TBFLAG_ZCR_LEN_MASK (0xf << ARM_TBFLAG_ZCR_LEN_SHIFT)
28
29
/* some convenience accessor macros */
30
#define ARM_TBFLAG_AARCH64_STATE(F) \
31
@@ -XXX,XX +XXX,XX @@ static inline bool arm_cpu_data_is_big_endian(CPUARMState *env)
32
(((F) & ARM_TBFLAG_TBI0_MASK) >> ARM_TBFLAG_TBI0_SHIFT)
33
#define ARM_TBFLAG_TBI1(F) \
34
(((F) & ARM_TBFLAG_TBI1_MASK) >> ARM_TBFLAG_TBI1_SHIFT)
35
+#define ARM_TBFLAG_SVEEXC_EL(F) \
36
+ (((F) & ARM_TBFLAG_SVEEXC_EL_MASK) >> ARM_TBFLAG_SVEEXC_EL_SHIFT)
37
+#define ARM_TBFLAG_ZCR_LEN(F) \
38
+ (((F) & ARM_TBFLAG_ZCR_LEN_MASK) >> ARM_TBFLAG_ZCR_LEN_SHIFT)
39
40
static inline bool bswap_code(bool sctlr_b)
41
{
42
diff --git a/target/arm/translate.h b/target/arm/translate.h
43
index XXXXXXX..XXXXXXX 100644
44
--- a/target/arm/translate.h
45
+++ b/target/arm/translate.h
46
@@ -XXX,XX +XXX,XX @@ typedef struct DisasContext {
47
bool tbi1; /* TBI1 for EL0/1, not used for EL2/3 */
48
bool ns; /* Use non-secure CPREG bank on access */
49
int fp_excp_el; /* FP exception EL or 0 if enabled */
50
+ int sve_excp_el; /* SVE exception EL or 0 if enabled */
51
+ int sve_len; /* SVE vector length in bytes */
52
/* Flag indicating that exceptions from secure mode are routed to EL3. */
53
bool secure_routed_to_el3;
54
bool vfp_enabled; /* FP enabled via FPSCR.EN */
55
diff --git a/target/arm/helper.c b/target/arm/helper.c
56
index XXXXXXX..XXXXXXX 100644
57
--- a/target/arm/helper.c
58
+++ b/target/arm/helper.c
59
@@ -XXX,XX +XXX,XX @@ void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc,
60
target_ulong *cs_base, uint32_t *pflags)
61
{
62
ARMMMUIdx mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false));
63
+ int fp_el = fp_exception_el(env);
64
uint32_t flags;
65
66
if (is_a64(env)) {
67
+ int sve_el = sve_exception_el(env);
68
+ uint32_t zcr_len;
69
+
70
*pc = env->pc;
71
flags = ARM_TBFLAG_AARCH64_STATE_MASK;
72
/* Get control bits for tagged addresses */
73
flags |= (arm_regime_tbi0(env, mmu_idx) << ARM_TBFLAG_TBI0_SHIFT);
74
flags |= (arm_regime_tbi1(env, mmu_idx) << ARM_TBFLAG_TBI1_SHIFT);
75
+ flags |= sve_el << ARM_TBFLAG_SVEEXC_EL_SHIFT;
76
+
77
+ /* If SVE is disabled, but FP is enabled,
78
+ then the effective len is 0. */
79
+ if (sve_el != 0 && fp_el == 0) {
80
+ zcr_len = 0;
81
+ } else {
82
+ int current_el = arm_current_el(env);
83
+
84
+ zcr_len = env->vfp.zcr_el[current_el <= 1 ? 1 : current_el];
85
+ zcr_len &= 0xf;
86
+ if (current_el < 2 && arm_feature(env, ARM_FEATURE_EL2)) {
87
+ zcr_len = MIN(zcr_len, 0xf & (uint32_t)env->vfp.zcr_el[2]);
88
+ }
89
+ if (current_el < 3 && arm_feature(env, ARM_FEATURE_EL3)) {
90
+ zcr_len = MIN(zcr_len, 0xf & (uint32_t)env->vfp.zcr_el[3]);
91
+ }
92
+ }
93
+ flags |= zcr_len << ARM_TBFLAG_ZCR_LEN_SHIFT;
94
} else {
95
*pc = env->regs[15];
96
flags = (env->thumb << ARM_TBFLAG_THUMB_SHIFT)
97
@@ -XXX,XX +XXX,XX @@ void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc,
98
if (arm_cpu_data_is_big_endian(env)) {
99
flags |= ARM_TBFLAG_BE_DATA_MASK;
100
}
101
- flags |= fp_exception_el(env) << ARM_TBFLAG_FPEXC_EL_SHIFT;
102
+ flags |= fp_el << ARM_TBFLAG_FPEXC_EL_SHIFT;
103
104
if (arm_v7m_is_handler_mode(env)) {
105
flags |= ARM_TBFLAG_HANDLER_MASK;
106
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
14
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
107
index XXXXXXX..XXXXXXX 100644
15
index XXXXXXX..XXXXXXX 100644
108
--- a/target/arm/translate-a64.c
16
--- a/target/arm/translate-a64.c
109
+++ b/target/arm/translate-a64.c
17
+++ b/target/arm/translate-a64.c
110
@@ -XXX,XX +XXX,XX @@ static int aarch64_tr_init_disas_context(DisasContextBase *dcbase,
18
@@ -XXX,XX +XXX,XX @@ static void disas_fp_int_conv(DisasContext *s, uint32_t insn)
111
dc->user = (dc->current_el == 0);
19
default:
112
#endif
20
/* all other sf/type/rmode combinations are invalid */
113
dc->fp_excp_el = ARM_TBFLAG_FPEXC_EL(dc->base.tb->flags);
21
unallocated_encoding(s);
114
+ dc->sve_excp_el = ARM_TBFLAG_SVEEXC_EL(dc->base.tb->flags);
22
- break;
115
+ dc->sve_len = (ARM_TBFLAG_ZCR_LEN(dc->base.tb->flags) + 1) * 16;
23
+ return;
116
dc->vec_len = 0;
24
}
117
dc->vec_stride = 0;
25
118
dc->cp_regs = arm_cpu->cp_regs;
26
if (!fp_access_check(s)) {
119
--
27
--
120
2.16.1
28
2.17.0
121
29
122
30
diff view generated by jsdifflib
1
In the v8M architecture, if the process of taking an exception
1
From: Richard Henderson <richard.henderson@linaro.org>
2
results in a further exception this is called a derived exception
2
3
(for example, an MPU exception when writing the exception frame to
3
Cc: qemu-stable@nongnu.org
4
memory). If the derived exception happens while pushing the initial
4
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
5
stack frame, we must ignore any subsequent possible exception
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
pushing the callee-saves registers.
6
Tested-by: Alex Bennée <alex.bennee@linaro.org>
7
7
Message-id: 20180512003217.9105-4-richard.henderson@linaro.org
8
In preparation for making the stack writes check for exceptions,
9
add a return value from v7m_push_stack() and a new parameter to
10
v7m_exception_taken(), so that the former can tell the latter that
11
it needs to ignore failures to write to the stack. We also plumb
12
the argument through to v7m_push_callee_stack(), which is where
13
the code to ignore the failures will be.
14
15
(Note that the v8M ARM pseudocode structures this slightly differently:
16
derived exceptions cause the attempt to process the original
17
exception to be abandoned; then at the top level it calls
18
DerivedLateArrival to prioritize the derived exception and call
19
TakeException from there. We choose to let the NVIC do the prioritization
20
and continue forward with a call to TakeException which will then
21
take either the original or the derived exception. The effect is
22
the same, but this structure works better for QEMU because we don't
23
have a convenient top level place to do the abandon-and-retry logic.)
24
25
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
26
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
27
Message-id: 1517324542-6607-4-git-send-email-peter.maydell@linaro.org
28
---
9
---
29
target/arm/helper.c | 35 +++++++++++++++++++++++------------
10
target/arm/helper.h | 6 +++
30
1 file changed, 23 insertions(+), 12 deletions(-)
11
target/arm/helper.c | 38 ++++++++++++++-
31
12
target/arm/translate-a64.c | 96 +++++++++++++++++++++++++++++++-------
13
3 files changed, 122 insertions(+), 18 deletions(-)
14
15
diff --git a/target/arm/helper.h b/target/arm/helper.h
16
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/helper.h
18
+++ b/target/arm/helper.h
19
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(vfp_touhd_round_to_zero, i64, f64, i32, ptr)
20
DEF_HELPER_3(vfp_tould_round_to_zero, i64, f64, i32, ptr)
21
DEF_HELPER_3(vfp_touhh, i32, f16, i32, ptr)
22
DEF_HELPER_3(vfp_toshh, i32, f16, i32, ptr)
23
+DEF_HELPER_3(vfp_toulh, i32, f16, i32, ptr)
24
+DEF_HELPER_3(vfp_toslh, i32, f16, i32, ptr)
25
+DEF_HELPER_3(vfp_touqh, i64, f16, i32, ptr)
26
+DEF_HELPER_3(vfp_tosqh, i64, f16, i32, ptr)
27
DEF_HELPER_3(vfp_toshs, i32, f32, i32, ptr)
28
DEF_HELPER_3(vfp_tosls, i32, f32, i32, ptr)
29
DEF_HELPER_3(vfp_tosqs, i64, f32, i32, ptr)
30
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_3(vfp_ultod, f64, i64, i32, ptr)
31
DEF_HELPER_3(vfp_uqtod, f64, i64, i32, ptr)
32
DEF_HELPER_3(vfp_sltoh, f16, i32, i32, ptr)
33
DEF_HELPER_3(vfp_ultoh, f16, i32, i32, ptr)
34
+DEF_HELPER_3(vfp_sqtoh, f16, i64, i32, ptr)
35
+DEF_HELPER_3(vfp_uqtoh, f16, i64, i32, ptr)
36
37
DEF_HELPER_FLAGS_2(set_rmode, TCG_CALL_NO_RWG, i32, i32, ptr)
38
DEF_HELPER_FLAGS_2(set_neon_rmode, TCG_CALL_NO_RWG, i32, i32, env)
32
diff --git a/target/arm/helper.c b/target/arm/helper.c
39
diff --git a/target/arm/helper.c b/target/arm/helper.c
33
index XXXXXXX..XXXXXXX 100644
40
index XXXXXXX..XXXXXXX 100644
34
--- a/target/arm/helper.c
41
--- a/target/arm/helper.c
35
+++ b/target/arm/helper.c
42
+++ b/target/arm/helper.c
36
@@ -XXX,XX +XXX,XX @@ static uint32_t arm_v7m_load_vector(ARMCPU *cpu, int exc, bool targets_secure)
43
@@ -XXX,XX +XXX,XX @@ VFP_CONV_FIX_A64(uq, s, 32, 64, uint64)
37
return addr;
44
#undef VFP_CONV_FIX_A64
45
46
/* Conversion to/from f16 can overflow to infinity before/after scaling.
47
- * Therefore we convert to f64 (which does not round), scale,
48
- * and then convert f64 to f16 (which may round).
49
+ * Therefore we convert to f64, scale, and then convert f64 to f16; or
50
+ * vice versa for conversion to integer.
51
+ *
52
+ * For 16- and 32-bit integers, the conversion to f64 never rounds.
53
+ * For 64-bit integers, any integer that would cause rounding will also
54
+ * overflow to f16 infinity, so there is no double rounding problem.
55
*/
56
57
static float16 do_postscale_fp16(float64 f, int shift, float_status *fpst)
58
@@ -XXX,XX +XXX,XX @@ float16 HELPER(vfp_ultoh)(uint32_t x, uint32_t shift, void *fpst)
59
return do_postscale_fp16(uint32_to_float64(x, fpst), shift, fpst);
38
}
60
}
39
61
40
-static void v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain)
62
+float16 HELPER(vfp_sqtoh)(uint64_t x, uint32_t shift, void *fpst)
41
+static void v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain,
63
+{
42
+ bool ignore_faults)
64
+ return do_postscale_fp16(int64_to_float64(x, fpst), shift, fpst);
65
+}
66
+
67
+float16 HELPER(vfp_uqtoh)(uint64_t x, uint32_t shift, void *fpst)
68
+{
69
+ return do_postscale_fp16(uint64_to_float64(x, fpst), shift, fpst);
70
+}
71
+
72
static float64 do_prescale_fp16(float16 f, int shift, float_status *fpst)
43
{
73
{
44
/* For v8M, push the callee-saves register part of the stack frame.
74
if (unlikely(float16_is_any_nan(f))) {
45
* Compare the v8M pseudocode PushCalleeStack().
75
@@ -XXX,XX +XXX,XX @@ uint32_t HELPER(vfp_touhh)(float16 x, uint32_t shift, void *fpst)
46
@@ -XXX,XX +XXX,XX @@ static void v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain)
76
return float64_to_uint16(do_prescale_fp16(x, shift, fpst), fpst);
47
*frame_sp_p = frameptr;
48
}
77
}
49
78
50
-static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain)
79
+uint32_t HELPER(vfp_toslh)(float16 x, uint32_t shift, void *fpst)
51
+static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain,
80
+{
52
+ bool ignore_stackfaults)
81
+ return float64_to_int32(do_prescale_fp16(x, shift, fpst), fpst);
82
+}
83
+
84
+uint32_t HELPER(vfp_toulh)(float16 x, uint32_t shift, void *fpst)
85
+{
86
+ return float64_to_uint32(do_prescale_fp16(x, shift, fpst), fpst);
87
+}
88
+
89
+uint64_t HELPER(vfp_tosqh)(float16 x, uint32_t shift, void *fpst)
90
+{
91
+ return float64_to_int64(do_prescale_fp16(x, shift, fpst), fpst);
92
+}
93
+
94
+uint64_t HELPER(vfp_touqh)(float16 x, uint32_t shift, void *fpst)
95
+{
96
+ return float64_to_uint64(do_prescale_fp16(x, shift, fpst), fpst);
97
+}
98
+
99
/* Set the current fp rounding mode and return the old one.
100
* The argument is a softfloat float_round_ value.
101
*/
102
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
103
index XXXXXXX..XXXXXXX 100644
104
--- a/target/arm/translate-a64.c
105
+++ b/target/arm/translate-a64.c
106
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
107
bool itof, int rmode, int scale, int sf, int type)
53
{
108
{
54
/* Do the "take the exception" parts of exception entry,
109
bool is_signed = !(opcode & 1);
55
* but not the pushing of state to the stack. This is
110
- bool is_double = type;
56
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain)
111
TCGv_ptr tcg_fpstatus;
57
*/
112
- TCGv_i32 tcg_shift;
58
if (lr & R_V7M_EXCRET_DCRS_MASK &&
113
+ TCGv_i32 tcg_shift, tcg_single;
59
!(dotailchain && (lr & R_V7M_EXCRET_ES_MASK))) {
114
+ TCGv_i64 tcg_double;
60
- v7m_push_callee_stack(cpu, lr, dotailchain);
115
61
+ v7m_push_callee_stack(cpu, lr, dotailchain,
116
- tcg_fpstatus = get_fpstatus_ptr(false);
62
+ ignore_stackfaults);
117
+ tcg_fpstatus = get_fpstatus_ptr(type == 3);
118
119
tcg_shift = tcg_const_i32(64 - scale);
120
121
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
122
tcg_int = tcg_extend;
123
}
124
125
- if (is_double) {
126
- TCGv_i64 tcg_double = tcg_temp_new_i64();
127
+ switch (type) {
128
+ case 1: /* float64 */
129
+ tcg_double = tcg_temp_new_i64();
130
if (is_signed) {
131
gen_helper_vfp_sqtod(tcg_double, tcg_int,
132
tcg_shift, tcg_fpstatus);
133
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
134
}
135
write_fp_dreg(s, rd, tcg_double);
136
tcg_temp_free_i64(tcg_double);
137
- } else {
138
- TCGv_i32 tcg_single = tcg_temp_new_i32();
139
+ break;
140
+
141
+ case 0: /* float32 */
142
+ tcg_single = tcg_temp_new_i32();
143
if (is_signed) {
144
gen_helper_vfp_sqtos(tcg_single, tcg_int,
145
tcg_shift, tcg_fpstatus);
146
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
147
}
148
write_fp_sreg(s, rd, tcg_single);
149
tcg_temp_free_i32(tcg_single);
150
+ break;
151
+
152
+ case 3: /* float16 */
153
+ tcg_single = tcg_temp_new_i32();
154
+ if (is_signed) {
155
+ gen_helper_vfp_sqtoh(tcg_single, tcg_int,
156
+ tcg_shift, tcg_fpstatus);
157
+ } else {
158
+ gen_helper_vfp_uqtoh(tcg_single, tcg_int,
159
+ tcg_shift, tcg_fpstatus);
160
+ }
161
+ write_fp_sreg(s, rd, tcg_single);
162
+ tcg_temp_free_i32(tcg_single);
163
+ break;
164
+
165
+ default:
166
+ g_assert_not_reached();
167
}
168
} else {
169
TCGv_i64 tcg_int = cpu_reg(s, rd);
170
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
171
172
gen_helper_set_rmode(tcg_rmode, tcg_rmode, tcg_fpstatus);
173
174
- if (is_double) {
175
- TCGv_i64 tcg_double = read_fp_dreg(s, rn);
176
+ switch (type) {
177
+ case 1: /* float64 */
178
+ tcg_double = read_fp_dreg(s, rn);
179
if (is_signed) {
180
if (!sf) {
181
gen_helper_vfp_tosld(tcg_int, tcg_double,
182
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
183
tcg_shift, tcg_fpstatus);
63
}
184
}
64
lr |= R_V7M_EXCRET_DCRS_MASK;
185
}
65
}
186
+ if (!sf) {
66
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain)
187
+ tcg_gen_ext32u_i64(tcg_int, tcg_int);
67
env->thumb = addr & 1;
188
+ }
68
}
189
tcg_temp_free_i64(tcg_double);
69
190
- } else {
70
-static void v7m_push_stack(ARMCPU *cpu)
191
- TCGv_i32 tcg_single = read_fp_sreg(s, rn);
71
+static bool v7m_push_stack(ARMCPU *cpu)
192
+ break;
72
{
193
+
73
/* Do the "set up stack frame" part of exception entry,
194
+ case 0: /* float32 */
74
* similar to pseudocode PushStack().
195
+ tcg_single = read_fp_sreg(s, rn);
75
+ * Return true if we generate a derived exception (and so
196
if (sf) {
76
+ * should ignore further stack faults trying to process
197
if (is_signed) {
77
+ * that derived exception.)
198
gen_helper_vfp_tosqs(tcg_int, tcg_single,
78
*/
199
@@ -XXX,XX +XXX,XX @@ static void handle_fpfpcvt(DisasContext *s, int rd, int rn, int opcode,
79
CPUARMState *env = &cpu->env;
200
tcg_temp_free_i32(tcg_dest);
80
uint32_t xpsr = xpsr_read(env);
201
}
81
@@ -XXX,XX +XXX,XX @@ static void v7m_push_stack(ARMCPU *cpu)
202
tcg_temp_free_i32(tcg_single);
82
v7m_push(env, env->regs[2]);
203
+ break;
83
v7m_push(env, env->regs[1]);
204
+
84
v7m_push(env, env->regs[0]);
205
+ case 3: /* float16 */
85
+
206
+ tcg_single = read_fp_sreg(s, rn);
86
+ return false;
207
+ if (sf) {
87
}
208
+ if (is_signed) {
88
209
+ gen_helper_vfp_tosqh(tcg_int, tcg_single,
89
static void do_v7m_exception_exit(ARMCPU *cpu)
210
+ tcg_shift, tcg_fpstatus);
90
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
211
+ } else {
91
if (sfault) {
212
+ gen_helper_vfp_touqh(tcg_int, tcg_single,
92
env->v7m.sfsr |= R_V7M_SFSR_INVER_MASK;
213
+ tcg_shift, tcg_fpstatus);
93
armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_SECURE, false);
214
+ }
94
- v7m_exception_taken(cpu, excret, true);
215
+ } else {
95
+ v7m_exception_taken(cpu, excret, true, false);
216
+ TCGv_i32 tcg_dest = tcg_temp_new_i32();
96
qemu_log_mask(CPU_LOG_INT, "...taking SecureFault on existing "
217
+ if (is_signed) {
97
"stackframe: failed EXC_RETURN.ES validity check\n");
218
+ gen_helper_vfp_toslh(tcg_dest, tcg_single,
98
return;
219
+ tcg_shift, tcg_fpstatus);
99
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
220
+ } else {
100
*/
221
+ gen_helper_vfp_toulh(tcg_dest, tcg_single,
101
env->v7m.cfsr[env->v7m.secure] |= R_V7M_CFSR_INVPC_MASK;
222
+ tcg_shift, tcg_fpstatus);
102
armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_USAGE, env->v7m.secure);
223
+ }
103
- v7m_exception_taken(cpu, excret, true);
224
+ tcg_gen_extu_i32_i64(tcg_int, tcg_dest);
104
+ v7m_exception_taken(cpu, excret, true, false);
225
+ tcg_temp_free_i32(tcg_dest);
105
qemu_log_mask(CPU_LOG_INT, "...taking UsageFault on existing "
226
+ }
106
"stackframe: failed exception return integrity check\n");
227
+ tcg_temp_free_i32(tcg_single);
107
return;
228
+ break;
108
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
229
+
109
/* Take a SecureFault on the current stack */
230
+ default:
110
env->v7m.sfsr |= R_V7M_SFSR_INVIS_MASK;
231
+ g_assert_not_reached();
111
armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_SECURE, false);
232
}
112
- v7m_exception_taken(cpu, excret, true);
233
113
+ v7m_exception_taken(cpu, excret, true, false);
234
gen_helper_set_rmode(tcg_rmode, tcg_rmode, tcg_fpstatus);
114
qemu_log_mask(CPU_LOG_INT, "...taking SecureFault on existing "
235
tcg_temp_free_i32(tcg_rmode);
115
"stackframe: failed exception return integrity "
236
-
116
"signature check\n");
237
- if (!sf) {
117
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
238
- tcg_gen_ext32u_i64(tcg_int, tcg_int);
118
armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_USAGE,
239
- }
119
env->v7m.secure);
120
env->v7m.cfsr[env->v7m.secure] |= R_V7M_CFSR_INVPC_MASK;
121
- v7m_exception_taken(cpu, excret, true);
122
+ v7m_exception_taken(cpu, excret, true, false);
123
qemu_log_mask(CPU_LOG_INT, "...taking UsageFault on existing "
124
"stackframe: failed exception return integrity "
125
"check\n");
126
@@ -XXX,XX +XXX,XX @@ static void do_v7m_exception_exit(ARMCPU *cpu)
127
/* Take an INVPC UsageFault by pushing the stack again;
128
* we know we're v7M so this is never a Secure UsageFault.
129
*/
130
+ bool ignore_stackfaults;
131
+
132
assert(!arm_feature(env, ARM_FEATURE_V8));
133
armv7m_nvic_set_pending(env->nvic, ARMV7M_EXCP_USAGE, false);
134
env->v7m.cfsr[env->v7m.secure] |= R_V7M_CFSR_INVPC_MASK;
135
- v7m_push_stack(cpu);
136
- v7m_exception_taken(cpu, excret, false);
137
+ ignore_stackfaults = v7m_push_stack(cpu);
138
+ v7m_exception_taken(cpu, excret, false, ignore_stackfaults);
139
qemu_log_mask(CPU_LOG_INT, "...taking UsageFault on new stackframe: "
140
"failed exception return integrity check\n");
141
return;
142
@@ -XXX,XX +XXX,XX @@ void arm_v7m_cpu_do_interrupt(CPUState *cs)
143
ARMCPU *cpu = ARM_CPU(cs);
144
CPUARMState *env = &cpu->env;
145
uint32_t lr;
146
+ bool ignore_stackfaults;
147
148
arm_log_exception(cs->exception_index);
149
150
@@ -XXX,XX +XXX,XX @@ void arm_v7m_cpu_do_interrupt(CPUState *cs)
151
lr |= R_V7M_EXCRET_MODE_MASK;
152
}
240
}
153
241
154
- v7m_push_stack(cpu);
242
tcg_temp_free_ptr(tcg_fpstatus);
155
- v7m_exception_taken(cpu, lr, false);
243
@@ -XXX,XX +XXX,XX @@ static void disas_fp_int_conv(DisasContext *s, uint32_t insn)
156
+ ignore_stackfaults = v7m_push_stack(cpu);
244
/* actual FP conversions */
157
+ v7m_exception_taken(cpu, lr, false, ignore_stackfaults);
245
bool itof = extract32(opcode, 1, 1);
158
qemu_log_mask(CPU_LOG_INT, "... as %d\n", env->v7m.exception);
246
159
}
247
- if (type > 1 || (rmode != 0 && opcode > 1)) {
160
248
+ if (rmode != 0 && opcode > 1) {
249
+ unallocated_encoding(s);
250
+ return;
251
+ }
252
+ switch (type) {
253
+ case 0: /* float32 */
254
+ case 1: /* float64 */
255
+ break;
256
+ case 3: /* float16 */
257
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
258
+ break;
259
+ }
260
+ /* fallthru */
261
+ default:
262
unallocated_encoding(s);
263
return;
264
}
161
--
265
--
162
2.16.1
266
2.17.0
163
267
164
268
diff view generated by jsdifflib
Deleted patch
1
The memory writes done to push registers on the stack
2
on exception entry in M profile CPUs are supposed to
3
go via MPU permissions checks, which may cause us to
4
take a derived exception instead of the original one of
5
the MPU lookup fails. We were implementing these as
6
always-succeeds direct writes to physical memory.
7
Rewrite v7m_push_stack() to do the necessary checks.
8
1
9
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
11
Message-id: 1517324542-6607-5-git-send-email-peter.maydell@linaro.org
12
---
13
target/arm/helper.c | 103 ++++++++++++++++++++++++++++++++++++++++++++--------
14
1 file changed, 87 insertions(+), 16 deletions(-)
15
16
diff --git a/target/arm/helper.c b/target/arm/helper.c
17
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/helper.c
19
+++ b/target/arm/helper.c
20
@@ -XXX,XX +XXX,XX @@ uint32_t arm_phys_excp_target_el(CPUState *cs, uint32_t excp_idx,
21
return target_el;
22
}
23
24
-static void v7m_push(CPUARMState *env, uint32_t val)
25
+static bool v7m_stack_write(ARMCPU *cpu, uint32_t addr, uint32_t value,
26
+ ARMMMUIdx mmu_idx, bool ignfault)
27
{
28
- CPUState *cs = CPU(arm_env_get_cpu(env));
29
+ CPUState *cs = CPU(cpu);
30
+ CPUARMState *env = &cpu->env;
31
+ MemTxAttrs attrs = {};
32
+ MemTxResult txres;
33
+ target_ulong page_size;
34
+ hwaddr physaddr;
35
+ int prot;
36
+ ARMMMUFaultInfo fi;
37
+ bool secure = mmu_idx & ARM_MMU_IDX_M_S;
38
+ int exc;
39
+ bool exc_secure;
40
41
- env->regs[13] -= 4;
42
- stl_phys(cs->as, env->regs[13], val);
43
+ if (get_phys_addr(env, addr, MMU_DATA_STORE, mmu_idx, &physaddr,
44
+ &attrs, &prot, &page_size, &fi, NULL)) {
45
+ /* MPU/SAU lookup failed */
46
+ if (fi.type == ARMFault_QEMU_SFault) {
47
+ qemu_log_mask(CPU_LOG_INT,
48
+ "...SecureFault with SFSR.AUVIOL during stacking\n");
49
+ env->v7m.sfsr |= R_V7M_SFSR_AUVIOL_MASK | R_V7M_SFSR_SFARVALID_MASK;
50
+ env->v7m.sfar = addr;
51
+ exc = ARMV7M_EXCP_SECURE;
52
+ exc_secure = false;
53
+ } else {
54
+ qemu_log_mask(CPU_LOG_INT, "...MemManageFault with CFSR.MSTKERR\n");
55
+ env->v7m.cfsr[secure] |= R_V7M_CFSR_MSTKERR_MASK;
56
+ exc = ARMV7M_EXCP_MEM;
57
+ exc_secure = secure;
58
+ }
59
+ goto pend_fault;
60
+ }
61
+ address_space_stl_le(arm_addressspace(cs, attrs), physaddr, value,
62
+ attrs, &txres);
63
+ if (txres != MEMTX_OK) {
64
+ /* BusFault trying to write the data */
65
+ qemu_log_mask(CPU_LOG_INT, "...BusFault with BFSR.STKERR\n");
66
+ env->v7m.cfsr[M_REG_NS] |= R_V7M_CFSR_STKERR_MASK;
67
+ exc = ARMV7M_EXCP_BUS;
68
+ exc_secure = false;
69
+ goto pend_fault;
70
+ }
71
+ return true;
72
+
73
+pend_fault:
74
+ /* By pending the exception at this point we are making
75
+ * the IMPDEF choice "overridden exceptions pended" (see the
76
+ * MergeExcInfo() pseudocode). The other choice would be to not
77
+ * pend them now and then make a choice about which to throw away
78
+ * later if we have two derived exceptions.
79
+ * The only case when we must not pend the exception but instead
80
+ * throw it away is if we are doing the push of the callee registers
81
+ * and we've already generated a derived exception. Even in this
82
+ * case we will still update the fault status registers.
83
+ */
84
+ if (!ignfault) {
85
+ armv7m_nvic_set_pending_derived(env->nvic, exc, exc_secure);
86
+ }
87
+ return false;
88
}
89
90
/* Return true if we're using the process stack pointer (not the MSP) */
91
@@ -XXX,XX +XXX,XX @@ static bool v7m_push_stack(ARMCPU *cpu)
92
* should ignore further stack faults trying to process
93
* that derived exception.)
94
*/
95
+ bool stacked_ok;
96
CPUARMState *env = &cpu->env;
97
uint32_t xpsr = xpsr_read(env);
98
+ uint32_t frameptr = env->regs[13];
99
+ ARMMMUIdx mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false));
100
101
/* Align stack pointer if the guest wants that */
102
- if ((env->regs[13] & 4) &&
103
+ if ((frameptr & 4) &&
104
(env->v7m.ccr[env->v7m.secure] & R_V7M_CCR_STKALIGN_MASK)) {
105
- env->regs[13] -= 4;
106
+ frameptr -= 4;
107
xpsr |= XPSR_SPREALIGN;
108
}
109
- /* Switch to the handler mode. */
110
- v7m_push(env, xpsr);
111
- v7m_push(env, env->regs[15]);
112
- v7m_push(env, env->regs[14]);
113
- v7m_push(env, env->regs[12]);
114
- v7m_push(env, env->regs[3]);
115
- v7m_push(env, env->regs[2]);
116
- v7m_push(env, env->regs[1]);
117
- v7m_push(env, env->regs[0]);
118
119
- return false;
120
+ frameptr -= 0x20;
121
+
122
+ /* Write as much of the stack frame as we can. If we fail a stack
123
+ * write this will result in a derived exception being pended
124
+ * (which may be taken in preference to the one we started with
125
+ * if it has higher priority).
126
+ */
127
+ stacked_ok =
128
+ v7m_stack_write(cpu, frameptr, env->regs[0], mmu_idx, false) &&
129
+ v7m_stack_write(cpu, frameptr + 4, env->regs[1], mmu_idx, false) &&
130
+ v7m_stack_write(cpu, frameptr + 8, env->regs[2], mmu_idx, false) &&
131
+ v7m_stack_write(cpu, frameptr + 12, env->regs[3], mmu_idx, false) &&
132
+ v7m_stack_write(cpu, frameptr + 16, env->regs[12], mmu_idx, false) &&
133
+ v7m_stack_write(cpu, frameptr + 20, env->regs[14], mmu_idx, false) &&
134
+ v7m_stack_write(cpu, frameptr + 24, env->regs[15], mmu_idx, false) &&
135
+ v7m_stack_write(cpu, frameptr + 28, xpsr, mmu_idx, false);
136
+
137
+ /* Update SP regardless of whether any of the stack accesses failed.
138
+ * When we implement v8M stack limit checking then this attempt to
139
+ * update SP might also fail and result in a derived exception.
140
+ */
141
+ env->regs[13] = frameptr;
142
+
143
+ return !stacked_ok;
144
}
145
146
static void do_v7m_exception_exit(ARMCPU *cpu)
147
--
148
2.16.1
149
150
diff view generated by jsdifflib
Deleted patch
1
Make v7m_push_callee_stack() honour the MPU by using the
2
new v7m_stack_write() function. We return a flag to indicate
3
whether the pushes failed, which we can then use in
4
v7m_exception_taken() to cause us to handle the derived
5
exception correctly.
6
1
7
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
9
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
10
Message-id: 1517324542-6607-6-git-send-email-peter.maydell@linaro.org
11
---
12
target/arm/helper.c | 64 ++++++++++++++++++++++++++++++++++++++++-------------
13
1 file changed, 49 insertions(+), 15 deletions(-)
14
15
diff --git a/target/arm/helper.c b/target/arm/helper.c
16
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/helper.c
18
+++ b/target/arm/helper.c
19
@@ -XXX,XX +XXX,XX @@ static uint32_t arm_v7m_load_vector(ARMCPU *cpu, int exc, bool targets_secure)
20
return addr;
21
}
22
23
-static void v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain,
24
+static bool v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain,
25
bool ignore_faults)
26
{
27
/* For v8M, push the callee-saves register part of the stack frame.
28
@@ -XXX,XX +XXX,XX @@ static void v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain,
29
* In the tailchaining case this may not be the current stack.
30
*/
31
CPUARMState *env = &cpu->env;
32
- CPUState *cs = CPU(cpu);
33
uint32_t *frame_sp_p;
34
uint32_t frameptr;
35
+ ARMMMUIdx mmu_idx;
36
+ bool stacked_ok;
37
38
if (dotailchain) {
39
- frame_sp_p = get_v7m_sp_ptr(env, true,
40
- lr & R_V7M_EXCRET_MODE_MASK,
41
+ bool mode = lr & R_V7M_EXCRET_MODE_MASK;
42
+ bool priv = !(env->v7m.control[M_REG_S] & R_V7M_CONTROL_NPRIV_MASK) ||
43
+ !mode;
44
+
45
+ mmu_idx = arm_v7m_mmu_idx_for_secstate_and_priv(env, M_REG_S, priv);
46
+ frame_sp_p = get_v7m_sp_ptr(env, M_REG_S, mode,
47
lr & R_V7M_EXCRET_SPSEL_MASK);
48
} else {
49
+ mmu_idx = core_to_arm_mmu_idx(env, cpu_mmu_index(env, false));
50
frame_sp_p = &env->regs[13];
51
}
52
53
frameptr = *frame_sp_p - 0x28;
54
55
- stl_phys(cs->as, frameptr, 0xfefa125b);
56
- stl_phys(cs->as, frameptr + 0x8, env->regs[4]);
57
- stl_phys(cs->as, frameptr + 0xc, env->regs[5]);
58
- stl_phys(cs->as, frameptr + 0x10, env->regs[6]);
59
- stl_phys(cs->as, frameptr + 0x14, env->regs[7]);
60
- stl_phys(cs->as, frameptr + 0x18, env->regs[8]);
61
- stl_phys(cs->as, frameptr + 0x1c, env->regs[9]);
62
- stl_phys(cs->as, frameptr + 0x20, env->regs[10]);
63
- stl_phys(cs->as, frameptr + 0x24, env->regs[11]);
64
+ /* Write as much of the stack frame as we can. A write failure may
65
+ * cause us to pend a derived exception.
66
+ */
67
+ stacked_ok =
68
+ v7m_stack_write(cpu, frameptr, 0xfefa125b, mmu_idx, ignore_faults) &&
69
+ v7m_stack_write(cpu, frameptr + 0x8, env->regs[4], mmu_idx,
70
+ ignore_faults) &&
71
+ v7m_stack_write(cpu, frameptr + 0xc, env->regs[5], mmu_idx,
72
+ ignore_faults) &&
73
+ v7m_stack_write(cpu, frameptr + 0x10, env->regs[6], mmu_idx,
74
+ ignore_faults) &&
75
+ v7m_stack_write(cpu, frameptr + 0x14, env->regs[7], mmu_idx,
76
+ ignore_faults) &&
77
+ v7m_stack_write(cpu, frameptr + 0x18, env->regs[8], mmu_idx,
78
+ ignore_faults) &&
79
+ v7m_stack_write(cpu, frameptr + 0x1c, env->regs[9], mmu_idx,
80
+ ignore_faults) &&
81
+ v7m_stack_write(cpu, frameptr + 0x20, env->regs[10], mmu_idx,
82
+ ignore_faults) &&
83
+ v7m_stack_write(cpu, frameptr + 0x24, env->regs[11], mmu_idx,
84
+ ignore_faults);
85
86
+ /* Update SP regardless of whether any of the stack accesses failed.
87
+ * When we implement v8M stack limit checking then this attempt to
88
+ * update SP might also fail and result in a derived exception.
89
+ */
90
*frame_sp_p = frameptr;
91
+
92
+ return !stacked_ok;
93
}
94
95
static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain,
96
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain,
97
uint32_t addr;
98
bool targets_secure;
99
int exc;
100
+ bool push_failed = false;
101
102
armv7m_nvic_get_pending_irq_info(env->nvic, &exc, &targets_secure);
103
104
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain,
105
*/
106
if (lr & R_V7M_EXCRET_DCRS_MASK &&
107
!(dotailchain && (lr & R_V7M_EXCRET_ES_MASK))) {
108
- v7m_push_callee_stack(cpu, lr, dotailchain,
109
- ignore_stackfaults);
110
+ push_failed = v7m_push_callee_stack(cpu, lr, dotailchain,
111
+ ignore_stackfaults);
112
}
113
lr |= R_V7M_EXCRET_DCRS_MASK;
114
}
115
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain,
116
}
117
}
118
119
+ if (push_failed && !ignore_stackfaults) {
120
+ /* Derived exception on callee-saves register stacking:
121
+ * we might now want to take a different exception which
122
+ * targets a different security state, so try again from the top.
123
+ */
124
+ v7m_exception_taken(cpu, lr, true, true);
125
+ return;
126
+ }
127
+
128
addr = arm_v7m_load_vector(cpu, exc, targets_secure);
129
130
/* Now we've done everything that might cause a derived exception
131
--
132
2.16.1
133
134
diff view generated by jsdifflib
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
Change vfp.regs as a uint64_t to vfp.zregs as an ARMVectorReg.
3
Cc: qemu-stable@nongnu.org
4
The previous patches have made the change in representation
5
relatively painless.
6
7
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
4
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
8
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Tested-by: Alex Bennée <alex.bennee@linaro.org>
10
Message-id: 20180123035349.24538-2-richard.henderson@linaro.org
7
Message-id: 20180512003217.9105-5-richard.henderson@linaro.org
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
---
9
---
13
target/arm/cpu.h | 59 +++++++++++++++++++++++++++++++---------------
10
target/arm/translate-a64.c | 17 +++++++++++++++--
14
target/arm/machine.c | 35 ++++++++++++++++++++++++++-
11
1 file changed, 15 insertions(+), 2 deletions(-)
15
target/arm/translate-a64.c | 8 +++----
16
target/arm/translate.c | 7 +++---
17
4 files changed, 81 insertions(+), 28 deletions(-)
18
12
19
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
20
index XXXXXXX..XXXXXXX 100644
21
--- a/target/arm/cpu.h
22
+++ b/target/arm/cpu.h
23
@@ -XXX,XX +XXX,XX @@ typedef struct {
24
uint32_t base_mask;
25
} TCR;
26
27
+/* Define a maximum sized vector register.
28
+ * For 32-bit, this is a 128-bit NEON/AdvSIMD register.
29
+ * For 64-bit, this is a 2048-bit SVE register.
30
+ *
31
+ * Note that the mapping between S, D, and Q views of the register bank
32
+ * differs between AArch64 and AArch32.
33
+ * In AArch32:
34
+ * Qn = regs[n].d[1]:regs[n].d[0]
35
+ * Dn = regs[n / 2].d[n & 1]
36
+ * Sn = regs[n / 4].d[n % 4 / 2],
37
+ * bits 31..0 for even n, and bits 63..32 for odd n
38
+ * (and regs[16] to regs[31] are inaccessible)
39
+ * In AArch64:
40
+ * Zn = regs[n].d[*]
41
+ * Qn = regs[n].d[1]:regs[n].d[0]
42
+ * Dn = regs[n].d[0]
43
+ * Sn = regs[n].d[0] bits 31..0
44
+ *
45
+ * This corresponds to the architecturally defined mapping between
46
+ * the two execution states, and means we do not need to explicitly
47
+ * map these registers when changing states.
48
+ *
49
+ * Align the data for use with TCG host vector operations.
50
+ */
51
+
52
+#ifdef TARGET_AARCH64
53
+# define ARM_MAX_VQ 16
54
+#else
55
+# define ARM_MAX_VQ 1
56
+#endif
57
+
58
+typedef struct ARMVectorReg {
59
+ uint64_t d[2 * ARM_MAX_VQ] QEMU_ALIGNED(16);
60
+} ARMVectorReg;
61
+
62
+
63
typedef struct CPUARMState {
64
/* Regs for current mode. */
65
uint32_t regs[16];
66
@@ -XXX,XX +XXX,XX @@ typedef struct CPUARMState {
67
68
/* VFP coprocessor state. */
69
struct {
70
- /* VFP/Neon register state. Note that the mapping between S, D and Q
71
- * views of the register bank differs between AArch64 and AArch32:
72
- * In AArch32:
73
- * Qn = regs[2n+1]:regs[2n]
74
- * Dn = regs[n]
75
- * Sn = regs[n/2] bits 31..0 for even n, and bits 63..32 for odd n
76
- * (and regs[32] to regs[63] are inaccessible)
77
- * In AArch64:
78
- * Qn = regs[2n+1]:regs[2n]
79
- * Dn = regs[2n]
80
- * Sn = regs[2n] bits 31..0
81
- * This corresponds to the architecturally defined mapping between
82
- * the two execution states, and means we do not need to explicitly
83
- * map these registers when changing states.
84
- */
85
- uint64_t regs[64] QEMU_ALIGNED(16);
86
+ ARMVectorReg zregs[32];
87
88
uint32_t xregs[16];
89
/* We store these fpcsr fields separately for convenience. */
90
@@ -XXX,XX +XXX,XX @@ static inline void *arm_get_el_change_hook_opaque(ARMCPU *cpu)
91
*/
92
static inline uint64_t *aa32_vfp_dreg(CPUARMState *env, unsigned regno)
93
{
94
- return &env->vfp.regs[regno];
95
+ return &env->vfp.zregs[regno >> 1].d[regno & 1];
96
}
97
98
/**
99
@@ -XXX,XX +XXX,XX @@ static inline uint64_t *aa32_vfp_dreg(CPUARMState *env, unsigned regno)
100
*/
101
static inline uint64_t *aa32_vfp_qreg(CPUARMState *env, unsigned regno)
102
{
103
- return &env->vfp.regs[2 * regno];
104
+ return &env->vfp.zregs[regno].d[0];
105
}
106
107
/**
108
@@ -XXX,XX +XXX,XX @@ static inline uint64_t *aa32_vfp_qreg(CPUARMState *env, unsigned regno)
109
*/
110
static inline uint64_t *aa64_vfp_qreg(CPUARMState *env, unsigned regno)
111
{
112
- return &env->vfp.regs[2 * regno];
113
+ return &env->vfp.zregs[regno].d[0];
114
}
115
116
#endif
117
diff --git a/target/arm/machine.c b/target/arm/machine.c
118
index XXXXXXX..XXXXXXX 100644
119
--- a/target/arm/machine.c
120
+++ b/target/arm/machine.c
121
@@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_vfp = {
122
.minimum_version_id = 3,
123
.needed = vfp_needed,
124
.fields = (VMStateField[]) {
125
- VMSTATE_UINT64_ARRAY(env.vfp.regs, ARMCPU, 64),
126
+ /* For compatibility, store Qn out of Zn here. */
127
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[0].d, ARMCPU, 0, 2),
128
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[1].d, ARMCPU, 0, 2),
129
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[2].d, ARMCPU, 0, 2),
130
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[3].d, ARMCPU, 0, 2),
131
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[4].d, ARMCPU, 0, 2),
132
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[5].d, ARMCPU, 0, 2),
133
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[6].d, ARMCPU, 0, 2),
134
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[7].d, ARMCPU, 0, 2),
135
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[8].d, ARMCPU, 0, 2),
136
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[9].d, ARMCPU, 0, 2),
137
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[10].d, ARMCPU, 0, 2),
138
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[11].d, ARMCPU, 0, 2),
139
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[12].d, ARMCPU, 0, 2),
140
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[13].d, ARMCPU, 0, 2),
141
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[14].d, ARMCPU, 0, 2),
142
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[15].d, ARMCPU, 0, 2),
143
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[16].d, ARMCPU, 0, 2),
144
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[17].d, ARMCPU, 0, 2),
145
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[18].d, ARMCPU, 0, 2),
146
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[19].d, ARMCPU, 0, 2),
147
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[20].d, ARMCPU, 0, 2),
148
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[21].d, ARMCPU, 0, 2),
149
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[22].d, ARMCPU, 0, 2),
150
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[23].d, ARMCPU, 0, 2),
151
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[24].d, ARMCPU, 0, 2),
152
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[25].d, ARMCPU, 0, 2),
153
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[26].d, ARMCPU, 0, 2),
154
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[27].d, ARMCPU, 0, 2),
155
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[28].d, ARMCPU, 0, 2),
156
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[29].d, ARMCPU, 0, 2),
157
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[30].d, ARMCPU, 0, 2),
158
+ VMSTATE_UINT64_SUB_ARRAY(env.vfp.zregs[31].d, ARMCPU, 0, 2),
159
+
160
/* The xregs array is a little awkward because element 1 (FPSCR)
161
* requires a specific accessor, so we have to split it up in
162
* the vmstate:
163
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
13
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
164
index XXXXXXX..XXXXXXX 100644
14
index XXXXXXX..XXXXXXX 100644
165
--- a/target/arm/translate-a64.c
15
--- a/target/arm/translate-a64.c
166
+++ b/target/arm/translate-a64.c
16
+++ b/target/arm/translate-a64.c
167
@@ -XXX,XX +XXX,XX @@ static inline int vec_reg_offset(DisasContext *s, int regno,
17
@@ -XXX,XX +XXX,XX @@ static void disas_fp_fixed_conv(DisasContext *s, uint32_t insn)
168
{
18
bool sf = extract32(insn, 31, 1);
169
int offs = 0;
19
bool itof;
170
#ifdef HOST_WORDS_BIGENDIAN
20
171
- /* This is complicated slightly because vfp.regs[2n] is
21
- if (sbit || (type > 1)
172
- * still the low half and vfp.regs[2n+1] the high half
22
- || (!sf && scale < 32)) {
173
+ /* This is complicated slightly because vfp.zregs[n].d[0] is
23
+ if (sbit || (!sf && scale < 32)) {
174
+ * still the low half and vfp.zregs[n].d[1] the high half
24
+ unallocated_encoding(s);
175
* of the 128 bit vector, even on big endian systems.
25
+ return;
176
* Calculate the offset assuming a fully bigendian 128 bits,
26
+ }
177
* then XOR to account for the order of the two 64 bit halves.
27
+
178
@@ -XXX,XX +XXX,XX @@ static inline int vec_reg_offset(DisasContext *s, int regno,
28
+ switch (type) {
179
#else
29
+ case 0: /* float32 */
180
offs += element * (1 << size);
30
+ case 1: /* float64 */
181
#endif
31
+ break;
182
- offs += offsetof(CPUARMState, vfp.regs[regno * 2]);
32
+ case 3: /* float16 */
183
+ offs += offsetof(CPUARMState, vfp.zregs[regno]);
33
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
184
assert_fp_access_checked(s);
34
+ break;
185
return offs;
35
+ }
186
}
36
+ /* fallthru */
187
@@ -XXX,XX +XXX,XX @@ static inline int vec_reg_offset(DisasContext *s, int regno,
37
+ default:
188
static inline int vec_full_reg_offset(DisasContext *s, int regno)
38
unallocated_encoding(s);
189
{
39
return;
190
assert_fp_access_checked(s);
191
- return offsetof(CPUARMState, vfp.regs[regno * 2]);
192
+ return offsetof(CPUARMState, vfp.zregs[regno]);
193
}
194
195
/* Return a newly allocated pointer to the vector register. */
196
diff --git a/target/arm/translate.c b/target/arm/translate.c
197
index XXXXXXX..XXXXXXX 100644
198
--- a/target/arm/translate.c
199
+++ b/target/arm/translate.c
200
@@ -XXX,XX +XXX,XX @@ static inline void gen_vfp_st(DisasContext *s, int dp, TCGv_i32 addr)
201
}
40
}
202
}
203
204
-static inline long
205
-vfp_reg_offset (int dp, int reg)
206
+static inline long vfp_reg_offset(bool dp, unsigned reg)
207
{
208
if (dp) {
209
- return offsetof(CPUARMState, vfp.regs[reg]);
210
+ return offsetof(CPUARMState, vfp.zregs[reg >> 1].d[reg & 1]);
211
} else {
212
- long ofs = offsetof(CPUARMState, vfp.regs[reg >> 1]);
213
+ long ofs = offsetof(CPUARMState, vfp.zregs[reg >> 2].d[(reg >> 1) & 1]);
214
if (reg & 1) {
215
ofs += offsetof(CPU_DoubleU, l.upper);
216
} else {
217
--
41
--
218
2.16.1
42
2.17.0
219
43
220
44
diff view generated by jsdifflib
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
Save the high parts of the Zregs and all of the Pregs.
3
Cc: qemu-stable@nongnu.org
4
The ZCR_ELx registers are migrated via the CP mechanism.
4
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
5
6
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
6
Tested-by: Alex Bennée <alex.bennee@linaro.org>
8
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Message-id: 20180512003217.9105-6-richard.henderson@linaro.org
9
Message-id: 20180123035349.24538-4-richard.henderson@linaro.org
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
9
---
12
target/arm/machine.c | 53 ++++++++++++++++++++++++++++++++++++++++++++++++++++
10
target/arm/translate-a64.c | 30 ++++++++++++++----------------
13
1 file changed, 53 insertions(+)
11
1 file changed, 14 insertions(+), 16 deletions(-)
14
12
15
diff --git a/target/arm/machine.c b/target/arm/machine.c
13
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
16
index XXXXXXX..XXXXXXX 100644
14
index XXXXXXX..XXXXXXX 100644
17
--- a/target/arm/machine.c
15
--- a/target/arm/translate-a64.c
18
+++ b/target/arm/machine.c
16
+++ b/target/arm/translate-a64.c
19
@@ -XXX,XX +XXX,XX @@ static const VMStateDescription vmstate_iwmmxt = {
17
@@ -XXX,XX +XXX,XX @@ static TCGv_i32 read_fp_sreg(DisasContext *s, int reg)
20
}
18
return v;
21
};
19
}
22
20
23
+#ifdef TARGET_AARCH64
21
+static TCGv_i32 read_fp_hreg(DisasContext *s, int reg)
24
+/* The expression ARM_MAX_VQ - 2 is 0 for pure AArch32 build,
22
+{
25
+ * and ARMPredicateReg is actively empty. This triggers errors
23
+ TCGv_i32 v = tcg_temp_new_i32();
26
+ * in the expansion of the VMSTATE macros.
27
+ */
28
+
24
+
29
+static bool sve_needed(void *opaque)
25
+ tcg_gen_ld16u_i32(v, cpu_env, fp_reg_offset(s, reg, MO_16));
30
+{
26
+ return v;
31
+ ARMCPU *cpu = opaque;
32
+ CPUARMState *env = &cpu->env;
33
+
34
+ return arm_feature(env, ARM_FEATURE_SVE);
35
+}
27
+}
36
+
28
+
37
+/* The first two words of each Zreg is stored in VFP state. */
29
/* Clear the bits above an N-bit vector, for N = (is_q ? 128 : 64).
38
+static const VMStateDescription vmstate_zreg_hi_reg = {
30
* If SVE is not enabled, then there are only 128 bits in the vector.
39
+ .name = "cpu/sve/zreg_hi",
31
*/
40
+ .version_id = 1,
32
@@ -XXX,XX +XXX,XX @@ static void disas_fp_csel(DisasContext *s, uint32_t insn)
41
+ .minimum_version_id = 1,
33
static void handle_fp_1src_half(DisasContext *s, int opcode, int rd, int rn)
42
+ .fields = (VMStateField[]) {
43
+ VMSTATE_UINT64_SUB_ARRAY(d, ARMVectorReg, 2, ARM_MAX_VQ - 2),
44
+ VMSTATE_END_OF_LIST()
45
+ }
46
+};
47
+
48
+static const VMStateDescription vmstate_preg_reg = {
49
+ .name = "cpu/sve/preg",
50
+ .version_id = 1,
51
+ .minimum_version_id = 1,
52
+ .fields = (VMStateField[]) {
53
+ VMSTATE_UINT64_ARRAY(p, ARMPredicateReg, 2 * ARM_MAX_VQ / 8),
54
+ VMSTATE_END_OF_LIST()
55
+ }
56
+};
57
+
58
+static const VMStateDescription vmstate_sve = {
59
+ .name = "cpu/sve",
60
+ .version_id = 1,
61
+ .minimum_version_id = 1,
62
+ .needed = sve_needed,
63
+ .fields = (VMStateField[]) {
64
+ VMSTATE_STRUCT_ARRAY(env.vfp.zregs, ARMCPU, 32, 0,
65
+ vmstate_zreg_hi_reg, ARMVectorReg),
66
+ VMSTATE_STRUCT_ARRAY(env.vfp.pregs, ARMCPU, 17, 0,
67
+ vmstate_preg_reg, ARMPredicateReg),
68
+ VMSTATE_END_OF_LIST()
69
+ }
70
+};
71
+#endif /* AARCH64 */
72
+
73
static bool m_needed(void *opaque)
74
{
34
{
75
ARMCPU *cpu = opaque;
35
TCGv_ptr fpst = NULL;
76
@@ -XXX,XX +XXX,XX @@ const VMStateDescription vmstate_arm_cpu = {
36
- TCGv_i32 tcg_op = tcg_temp_new_i32();
77
&vmstate_pmsav7,
37
+ TCGv_i32 tcg_op = read_fp_hreg(s, rn);
78
&vmstate_pmsav8,
38
TCGv_i32 tcg_res = tcg_temp_new_i32();
79
&vmstate_m_security,
39
80
+#ifdef TARGET_AARCH64
40
- read_vec_element_i32(s, tcg_op, rn, 0, MO_16);
81
+ &vmstate_sve,
41
-
82
+#endif
42
switch (opcode) {
83
NULL
43
case 0x0: /* FMOV */
44
tcg_gen_mov_i32(tcg_res, tcg_op);
45
@@ -XXX,XX +XXX,XX @@ static void disas_simd_scalar_three_reg_diff(DisasContext *s, uint32_t insn)
46
tcg_temp_free_i64(tcg_op2);
47
tcg_temp_free_i64(tcg_res);
48
} else {
49
- TCGv_i32 tcg_op1 = tcg_temp_new_i32();
50
- TCGv_i32 tcg_op2 = tcg_temp_new_i32();
51
+ TCGv_i32 tcg_op1 = read_fp_hreg(s, rn);
52
+ TCGv_i32 tcg_op2 = read_fp_hreg(s, rm);
53
TCGv_i64 tcg_res = tcg_temp_new_i64();
54
55
- read_vec_element_i32(s, tcg_op1, rn, 0, MO_16);
56
- read_vec_element_i32(s, tcg_op2, rm, 0, MO_16);
57
-
58
gen_helper_neon_mull_s16(tcg_res, tcg_op1, tcg_op2);
59
gen_helper_neon_addl_saturate_s32(tcg_res, cpu_env, tcg_res, tcg_res);
60
61
@@ -XXX,XX +XXX,XX @@ static void disas_simd_scalar_three_reg_same_fp16(DisasContext *s,
62
63
fpst = get_fpstatus_ptr(true);
64
65
- tcg_op1 = tcg_temp_new_i32();
66
- tcg_op2 = tcg_temp_new_i32();
67
+ tcg_op1 = read_fp_hreg(s, rn);
68
+ tcg_op2 = read_fp_hreg(s, rm);
69
tcg_res = tcg_temp_new_i32();
70
71
- read_vec_element_i32(s, tcg_op1, rn, 0, MO_16);
72
- read_vec_element_i32(s, tcg_op2, rm, 0, MO_16);
73
-
74
switch (fpopcode) {
75
case 0x03: /* FMULX */
76
gen_helper_advsimd_mulxh(tcg_res, tcg_op1, tcg_op2, fpst);
77
@@ -XXX,XX +XXX,XX @@ static void disas_simd_two_reg_misc_fp16(DisasContext *s, uint32_t insn)
84
}
78
}
85
};
79
80
if (is_scalar) {
81
- TCGv_i32 tcg_op = tcg_temp_new_i32();
82
+ TCGv_i32 tcg_op = read_fp_hreg(s, rn);
83
TCGv_i32 tcg_res = tcg_temp_new_i32();
84
85
- read_vec_element_i32(s, tcg_op, rn, 0, MO_16);
86
-
87
switch (fpop) {
88
case 0x1a: /* FCVTNS */
89
case 0x1b: /* FCVTMS */
86
--
90
--
87
2.16.1
91
2.17.0
88
92
89
93
diff view generated by jsdifflib
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
Move virt's PSCI DT fixup code to arm/boot.c and set this fixup to
3
We missed all of the scalar fp16 binary operations.
4
happen automatically for every board that doesn't mark "psci-conduit"
5
as disabled. This way emulated boards other than "virt" that rely on
6
PSIC for SMP could benefit from that code.
7
4
8
Cc: Peter Maydell <peter.maydell@linaro.org>
5
Cc: qemu-stable@nongnu.org
9
Cc: Jason Wang <jasowang@redhat.com>
6
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
10
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
11
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
8
Tested-by: Alex Bennée <alex.bennee@linaro.org>
12
Cc: Michael S. Tsirkin <mst@redhat.com>
9
Message-id: 20180512003217.9105-7-richard.henderson@linaro.org
13
Cc: qemu-devel@nongnu.org
14
Cc: qemu-arm@nongnu.org
15
Cc: yurovsky@gmail.com
16
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
17
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
18
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
19
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
20
---
11
---
21
hw/arm/boot.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
12
target/arm/translate-a64.c | 65 ++++++++++++++++++++++++++++++++++++++
22
hw/arm/virt.c | 61 -------------------------------------------------------
13
1 file changed, 65 insertions(+)
23
2 files changed, 65 insertions(+), 61 deletions(-)
24
14
25
diff --git a/hw/arm/boot.c b/hw/arm/boot.c
15
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
26
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
27
--- a/hw/arm/boot.c
17
--- a/target/arm/translate-a64.c
28
+++ b/hw/arm/boot.c
18
+++ b/target/arm/translate-a64.c
29
@@ -XXX,XX +XXX,XX @@ static void set_kernel_args_old(const struct arm_boot_info *info)
19
@@ -XXX,XX +XXX,XX @@ static void handle_fp_2src_double(DisasContext *s, int opcode,
30
}
20
tcg_temp_free_i64(tcg_res);
31
}
21
}
32
22
33
+static void fdt_add_psci_node(void *fdt)
23
+/* Floating-point data-processing (2 source) - half precision */
24
+static void handle_fp_2src_half(DisasContext *s, int opcode,
25
+ int rd, int rn, int rm)
34
+{
26
+{
35
+ uint32_t cpu_suspend_fn;
27
+ TCGv_i32 tcg_op1;
36
+ uint32_t cpu_off_fn;
28
+ TCGv_i32 tcg_op2;
37
+ uint32_t cpu_on_fn;
29
+ TCGv_i32 tcg_res;
38
+ uint32_t migrate_fn;
30
+ TCGv_ptr fpst;
39
+ ARMCPU *armcpu = ARM_CPU(qemu_get_cpu(0));
40
+ const char *psci_method;
41
+ int64_t psci_conduit;
42
+
31
+
43
+ psci_conduit = object_property_get_int(OBJECT(armcpu),
32
+ tcg_res = tcg_temp_new_i32();
44
+ "psci-conduit",
33
+ fpst = get_fpstatus_ptr(true);
45
+ &error_abort);
34
+ tcg_op1 = read_fp_hreg(s, rn);
46
+ switch (psci_conduit) {
35
+ tcg_op2 = read_fp_hreg(s, rm);
47
+ case QEMU_PSCI_CONDUIT_DISABLED:
36
+
48
+ return;
37
+ switch (opcode) {
49
+ case QEMU_PSCI_CONDUIT_HVC:
38
+ case 0x0: /* FMUL */
50
+ psci_method = "hvc";
39
+ gen_helper_advsimd_mulh(tcg_res, tcg_op1, tcg_op2, fpst);
51
+ break;
40
+ break;
52
+ case QEMU_PSCI_CONDUIT_SMC:
41
+ case 0x1: /* FDIV */
53
+ psci_method = "smc";
42
+ gen_helper_advsimd_divh(tcg_res, tcg_op1, tcg_op2, fpst);
43
+ break;
44
+ case 0x2: /* FADD */
45
+ gen_helper_advsimd_addh(tcg_res, tcg_op1, tcg_op2, fpst);
46
+ break;
47
+ case 0x3: /* FSUB */
48
+ gen_helper_advsimd_subh(tcg_res, tcg_op1, tcg_op2, fpst);
49
+ break;
50
+ case 0x4: /* FMAX */
51
+ gen_helper_advsimd_maxh(tcg_res, tcg_op1, tcg_op2, fpst);
52
+ break;
53
+ case 0x5: /* FMIN */
54
+ gen_helper_advsimd_minh(tcg_res, tcg_op1, tcg_op2, fpst);
55
+ break;
56
+ case 0x6: /* FMAXNM */
57
+ gen_helper_advsimd_maxnumh(tcg_res, tcg_op1, tcg_op2, fpst);
58
+ break;
59
+ case 0x7: /* FMINNM */
60
+ gen_helper_advsimd_minnumh(tcg_res, tcg_op1, tcg_op2, fpst);
61
+ break;
62
+ case 0x8: /* FNMUL */
63
+ gen_helper_advsimd_mulh(tcg_res, tcg_op1, tcg_op2, fpst);
64
+ tcg_gen_xori_i32(tcg_res, tcg_res, 0x8000);
54
+ break;
65
+ break;
55
+ default:
66
+ default:
56
+ g_assert_not_reached();
67
+ g_assert_not_reached();
57
+ }
68
+ }
58
+
69
+
59
+ qemu_fdt_add_subnode(fdt, "/psci");
70
+ write_fp_sreg(s, rd, tcg_res);
60
+ if (armcpu->psci_version == 2) {
61
+ const char comp[] = "arm,psci-0.2\0arm,psci";
62
+ qemu_fdt_setprop(fdt, "/psci", "compatible", comp, sizeof(comp));
63
+
71
+
64
+ cpu_off_fn = QEMU_PSCI_0_2_FN_CPU_OFF;
72
+ tcg_temp_free_ptr(fpst);
65
+ if (arm_feature(&armcpu->env, ARM_FEATURE_AARCH64)) {
73
+ tcg_temp_free_i32(tcg_op1);
66
+ cpu_suspend_fn = QEMU_PSCI_0_2_FN64_CPU_SUSPEND;
74
+ tcg_temp_free_i32(tcg_op2);
67
+ cpu_on_fn = QEMU_PSCI_0_2_FN64_CPU_ON;
75
+ tcg_temp_free_i32(tcg_res);
68
+ migrate_fn = QEMU_PSCI_0_2_FN64_MIGRATE;
69
+ } else {
70
+ cpu_suspend_fn = QEMU_PSCI_0_2_FN_CPU_SUSPEND;
71
+ cpu_on_fn = QEMU_PSCI_0_2_FN_CPU_ON;
72
+ migrate_fn = QEMU_PSCI_0_2_FN_MIGRATE;
73
+ }
74
+ } else {
75
+ qemu_fdt_setprop_string(fdt, "/psci", "compatible", "arm,psci");
76
+
77
+ cpu_suspend_fn = QEMU_PSCI_0_1_FN_CPU_SUSPEND;
78
+ cpu_off_fn = QEMU_PSCI_0_1_FN_CPU_OFF;
79
+ cpu_on_fn = QEMU_PSCI_0_1_FN_CPU_ON;
80
+ migrate_fn = QEMU_PSCI_0_1_FN_MIGRATE;
81
+ }
82
+
83
+ /* We adopt the PSCI spec's nomenclature, and use 'conduit' to refer
84
+ * to the instruction that should be used to invoke PSCI functions.
85
+ * However, the device tree binding uses 'method' instead, so that is
86
+ * what we should use here.
87
+ */
88
+ qemu_fdt_setprop_string(fdt, "/psci", "method", psci_method);
89
+
90
+ qemu_fdt_setprop_cell(fdt, "/psci", "cpu_suspend", cpu_suspend_fn);
91
+ qemu_fdt_setprop_cell(fdt, "/psci", "cpu_off", cpu_off_fn);
92
+ qemu_fdt_setprop_cell(fdt, "/psci", "cpu_on", cpu_on_fn);
93
+ qemu_fdt_setprop_cell(fdt, "/psci", "migrate", migrate_fn);
94
+}
76
+}
95
+
77
+
96
/**
78
/* Floating point data-processing (2 source)
97
* load_dtb() - load a device tree binary image into memory
79
* 31 30 29 28 24 23 22 21 20 16 15 12 11 10 9 5 4 0
98
* @addr: the address to load the image at
80
* +---+---+---+-----------+------+---+------+--------+-----+------+------+
99
@@ -XXX,XX +XXX,XX @@ static int load_dtb(hwaddr addr, const struct arm_boot_info *binfo,
81
@@ -XXX,XX +XXX,XX @@ static void disas_fp_2src(DisasContext *s, uint32_t insn)
100
}
82
}
83
handle_fp_2src_double(s, opcode, rd, rn, rm);
84
break;
85
+ case 3:
86
+ if (!arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
87
+ unallocated_encoding(s);
88
+ return;
89
+ }
90
+ if (!fp_access_check(s)) {
91
+ return;
92
+ }
93
+ handle_fp_2src_half(s, opcode, rd, rn, rm);
94
+ break;
95
default:
96
unallocated_encoding(s);
101
}
97
}
102
103
+ fdt_add_psci_node(fdt);
104
+
105
if (binfo->modify_dtb) {
106
binfo->modify_dtb(binfo, fdt);
107
}
108
diff --git a/hw/arm/virt.c b/hw/arm/virt.c
109
index XXXXXXX..XXXXXXX 100644
110
--- a/hw/arm/virt.c
111
+++ b/hw/arm/virt.c
112
@@ -XXX,XX +XXX,XX @@ static void create_fdt(VirtMachineState *vms)
113
}
114
}
115
116
-static void fdt_add_psci_node(const VirtMachineState *vms)
117
-{
118
- uint32_t cpu_suspend_fn;
119
- uint32_t cpu_off_fn;
120
- uint32_t cpu_on_fn;
121
- uint32_t migrate_fn;
122
- void *fdt = vms->fdt;
123
- ARMCPU *armcpu = ARM_CPU(qemu_get_cpu(0));
124
- const char *psci_method;
125
-
126
- switch (vms->psci_conduit) {
127
- case QEMU_PSCI_CONDUIT_DISABLED:
128
- return;
129
- case QEMU_PSCI_CONDUIT_HVC:
130
- psci_method = "hvc";
131
- break;
132
- case QEMU_PSCI_CONDUIT_SMC:
133
- psci_method = "smc";
134
- break;
135
- default:
136
- g_assert_not_reached();
137
- }
138
-
139
- qemu_fdt_add_subnode(fdt, "/psci");
140
- if (armcpu->psci_version == 2) {
141
- const char comp[] = "arm,psci-0.2\0arm,psci";
142
- qemu_fdt_setprop(fdt, "/psci", "compatible", comp, sizeof(comp));
143
-
144
- cpu_off_fn = QEMU_PSCI_0_2_FN_CPU_OFF;
145
- if (arm_feature(&armcpu->env, ARM_FEATURE_AARCH64)) {
146
- cpu_suspend_fn = QEMU_PSCI_0_2_FN64_CPU_SUSPEND;
147
- cpu_on_fn = QEMU_PSCI_0_2_FN64_CPU_ON;
148
- migrate_fn = QEMU_PSCI_0_2_FN64_MIGRATE;
149
- } else {
150
- cpu_suspend_fn = QEMU_PSCI_0_2_FN_CPU_SUSPEND;
151
- cpu_on_fn = QEMU_PSCI_0_2_FN_CPU_ON;
152
- migrate_fn = QEMU_PSCI_0_2_FN_MIGRATE;
153
- }
154
- } else {
155
- qemu_fdt_setprop_string(fdt, "/psci", "compatible", "arm,psci");
156
-
157
- cpu_suspend_fn = QEMU_PSCI_0_1_FN_CPU_SUSPEND;
158
- cpu_off_fn = QEMU_PSCI_0_1_FN_CPU_OFF;
159
- cpu_on_fn = QEMU_PSCI_0_1_FN_CPU_ON;
160
- migrate_fn = QEMU_PSCI_0_1_FN_MIGRATE;
161
- }
162
-
163
- /* We adopt the PSCI spec's nomenclature, and use 'conduit' to refer
164
- * to the instruction that should be used to invoke PSCI functions.
165
- * However, the device tree binding uses 'method' instead, so that is
166
- * what we should use here.
167
- */
168
- qemu_fdt_setprop_string(fdt, "/psci", "method", psci_method);
169
-
170
- qemu_fdt_setprop_cell(fdt, "/psci", "cpu_suspend", cpu_suspend_fn);
171
- qemu_fdt_setprop_cell(fdt, "/psci", "cpu_off", cpu_off_fn);
172
- qemu_fdt_setprop_cell(fdt, "/psci", "cpu_on", cpu_on_fn);
173
- qemu_fdt_setprop_cell(fdt, "/psci", "migrate", migrate_fn);
174
-}
175
-
176
static void fdt_add_timer_nodes(const VirtMachineState *vms)
177
{
178
/* On real hardware these interrupts are level-triggered.
179
@@ -XXX,XX +XXX,XX @@ static void machvirt_init(MachineState *machine)
180
}
181
fdt_add_timer_nodes(vms);
182
fdt_add_cpu_nodes(vms);
183
- fdt_add_psci_node(vms);
184
185
memory_region_allocate_system_memory(ram, NULL, "mach-virt.ram",
186
machine->ram_size);
187
--
98
--
188
2.16.1
99
2.17.0
189
100
190
101
diff view generated by jsdifflib
1
From: Richard Henderson <richard.henderson@linaro.org>
1
From: Richard Henderson <richard.henderson@linaro.org>
2
2
3
Define ZCR_EL[1-3].
3
We missed all of the scalar fp16 fma operations.
4
4
5
Cc: qemu-stable@nongnu.org
6
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
5
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
7
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Tested-by: Alex Bennée <alex.bennee@linaro.org>
7
Message-id: 20180123035349.24538-5-richard.henderson@linaro.org
9
Message-id: 20180512003217.9105-8-richard.henderson@linaro.org
8
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
9
---
11
---
10
target/arm/cpu.h | 5 ++
12
target/arm/translate-a64.c | 48 ++++++++++++++++++++++++++++++++++++++
11
target/arm/helper.c | 131 ++++++++++++++++++++++++++++++++++++++++++++++++++++
13
1 file changed, 48 insertions(+)
12
2 files changed, 136 insertions(+)
13
14
14
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
15
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
15
index XXXXXXX..XXXXXXX 100644
16
index XXXXXXX..XXXXXXX 100644
16
--- a/target/arm/cpu.h
17
--- a/target/arm/translate-a64.c
17
+++ b/target/arm/cpu.h
18
+++ b/target/arm/translate-a64.c
18
@@ -XXX,XX +XXX,XX @@ typedef struct CPUARMState {
19
@@ -XXX,XX +XXX,XX @@ static void handle_fp_3src_double(DisasContext *s, bool o0, bool o1,
19
*/
20
tcg_temp_free_i64(tcg_res);
20
float_status fp_status;
21
}
21
float_status standard_fp_status;
22
23
+/* Floating-point data-processing (3 source) - half precision */
24
+static void handle_fp_3src_half(DisasContext *s, bool o0, bool o1,
25
+ int rd, int rn, int rm, int ra)
26
+{
27
+ TCGv_i32 tcg_op1, tcg_op2, tcg_op3;
28
+ TCGv_i32 tcg_res = tcg_temp_new_i32();
29
+ TCGv_ptr fpst = get_fpstatus_ptr(true);
22
+
30
+
23
+ /* ZCR_EL[1-3] */
31
+ tcg_op1 = read_fp_hreg(s, rn);
24
+ uint64_t zcr_el[4];
32
+ tcg_op2 = read_fp_hreg(s, rm);
25
} vfp;
33
+ tcg_op3 = read_fp_hreg(s, ra);
26
uint64_t exclusive_addr;
27
uint64_t exclusive_val;
28
@@ -XXX,XX +XXX,XX @@ void pmccntr_sync(CPUARMState *env);
29
#define CPTR_TCPAC (1U << 31)
30
#define CPTR_TTA (1U << 20)
31
#define CPTR_TFP (1U << 10)
32
+#define CPTR_TZ (1U << 8) /* CPTR_EL2 */
33
+#define CPTR_EZ (1U << 8) /* CPTR_EL3 */
34
35
#define MDCR_EPMAD (1U << 21)
36
#define MDCR_EDAD (1U << 20)
37
diff --git a/target/arm/helper.c b/target/arm/helper.c
38
index XXXXXXX..XXXXXXX 100644
39
--- a/target/arm/helper.c
40
+++ b/target/arm/helper.c
41
@@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo debug_lpae_cp_reginfo[] = {
42
REGINFO_SENTINEL
43
};
44
45
+/* Return the exception level to which SVE-disabled exceptions should
46
+ * be taken, or 0 if SVE is enabled.
47
+ */
48
+static int sve_exception_el(CPUARMState *env)
49
+{
50
+#ifndef CONFIG_USER_ONLY
51
+ unsigned current_el = arm_current_el(env);
52
+
34
+
53
+ /* The CPACR.ZEN controls traps to EL1:
35
+ /* These are fused multiply-add, and must be done as one
54
+ * 0, 2 : trap EL0 and EL1 accesses
36
+ * floating point operation with no rounding between the
55
+ * 1 : trap only EL0 accesses
37
+ * multiplication and addition steps.
56
+ * 3 : trap no accesses
38
+ * NB that doing the negations here as separate steps is
39
+ * correct : an input NaN should come out with its sign bit
40
+ * flipped if it is a negated-input.
57
+ */
41
+ */
58
+ switch (extract32(env->cp15.cpacr_el1, 16, 2)) {
42
+ if (o1 == true) {
59
+ default:
43
+ tcg_gen_xori_i32(tcg_op3, tcg_op3, 0x8000);
60
+ if (current_el <= 1) {
61
+ /* Trap to PL1, which might be EL1 or EL3 */
62
+ if (arm_is_secure(env) && !arm_el_is_aa64(env, 3)) {
63
+ return 3;
64
+ }
65
+ return 1;
66
+ }
67
+ break;
68
+ case 1:
69
+ if (current_el == 0) {
70
+ return 1;
71
+ }
72
+ break;
73
+ case 3:
74
+ break;
75
+ }
44
+ }
76
+
45
+
77
+ /* Similarly for CPACR.FPEN, after having checked ZEN. */
46
+ if (o0 != o1) {
78
+ switch (extract32(env->cp15.cpacr_el1, 20, 2)) {
47
+ tcg_gen_xori_i32(tcg_op1, tcg_op1, 0x8000);
79
+ default:
80
+ if (current_el <= 1) {
81
+ if (arm_is_secure(env) && !arm_el_is_aa64(env, 3)) {
82
+ return 3;
83
+ }
84
+ return 1;
85
+ }
86
+ break;
87
+ case 1:
88
+ if (current_el == 0) {
89
+ return 1;
90
+ }
91
+ break;
92
+ case 3:
93
+ break;
94
+ }
48
+ }
95
+
49
+
96
+ /* CPTR_EL2. Check both TZ and TFP. */
50
+ gen_helper_advsimd_muladdh(tcg_res, tcg_op1, tcg_op2, tcg_op3, fpst);
97
+ if (current_el <= 2
98
+ && (env->cp15.cptr_el[2] & (CPTR_TFP | CPTR_TZ))
99
+ && !arm_is_secure_below_el3(env)) {
100
+ return 2;
101
+ }
102
+
51
+
103
+ /* CPTR_EL3. Check both EZ and TFP. */
52
+ write_fp_sreg(s, rd, tcg_res);
104
+ if (!(env->cp15.cptr_el[3] & CPTR_EZ)
53
+
105
+ || (env->cp15.cptr_el[3] & CPTR_TFP)) {
54
+ tcg_temp_free_ptr(fpst);
106
+ return 3;
55
+ tcg_temp_free_i32(tcg_op1);
107
+ }
56
+ tcg_temp_free_i32(tcg_op2);
108
+#endif
57
+ tcg_temp_free_i32(tcg_op3);
109
+ return 0;
58
+ tcg_temp_free_i32(tcg_res);
110
+}
59
+}
111
+
60
+
112
+static CPAccessResult zcr_access(CPUARMState *env, const ARMCPRegInfo *ri,
61
/* Floating point data-processing (3 source)
113
+ bool isread)
62
* 31 30 29 28 24 23 22 21 20 16 15 14 10 9 5 4 0
114
+{
63
* +---+---+---+-----------+------+----+------+----+------+------+------+
115
+ switch (sve_exception_el(env)) {
64
@@ -XXX,XX +XXX,XX @@ static void disas_fp_3src(DisasContext *s, uint32_t insn)
65
}
66
handle_fp_3src_double(s, o0, o1, rd, rn, rm, ra);
67
break;
116
+ case 3:
68
+ case 3:
117
+ return CP_ACCESS_TRAP_EL3;
69
+ if (!arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
118
+ case 2:
70
+ unallocated_encoding(s);
119
+ return CP_ACCESS_TRAP_EL2;
71
+ return;
120
+ case 1:
72
+ }
121
+ return CP_ACCESS_TRAP;
73
+ if (!fp_access_check(s)) {
122
+ }
74
+ return;
123
+ return CP_ACCESS_OK;
75
+ }
124
+}
76
+ handle_fp_3src_half(s, o0, o1, rd, rn, rm, ra);
125
+
77
+ break;
126
+static void zcr_write(CPUARMState *env, const ARMCPRegInfo *ri,
78
default:
127
+ uint64_t value)
79
unallocated_encoding(s);
128
+{
129
+ /* Bits other than [3:0] are RAZ/WI. */
130
+ raw_write(env, ri, value & 0xf);
131
+}
132
+
133
+static const ARMCPRegInfo zcr_el1_reginfo = {
134
+ .name = "ZCR_EL1", .state = ARM_CP_STATE_AA64,
135
+ .opc0 = 3, .opc1 = 0, .crn = 1, .crm = 2, .opc2 = 0,
136
+ .access = PL1_RW, .accessfn = zcr_access, .type = ARM_CP_64BIT,
137
+ .fieldoffset = offsetof(CPUARMState, vfp.zcr_el[1]),
138
+ .writefn = zcr_write, .raw_writefn = raw_write
139
+};
140
+
141
+static const ARMCPRegInfo zcr_el2_reginfo = {
142
+ .name = "ZCR_EL2", .state = ARM_CP_STATE_AA64,
143
+ .opc0 = 3, .opc1 = 4, .crn = 1, .crm = 2, .opc2 = 0,
144
+ .access = PL2_RW, .accessfn = zcr_access, .type = ARM_CP_64BIT,
145
+ .fieldoffset = offsetof(CPUARMState, vfp.zcr_el[2]),
146
+ .writefn = zcr_write, .raw_writefn = raw_write
147
+};
148
+
149
+static const ARMCPRegInfo zcr_no_el2_reginfo = {
150
+ .name = "ZCR_EL2", .state = ARM_CP_STATE_AA64,
151
+ .opc0 = 3, .opc1 = 4, .crn = 1, .crm = 2, .opc2 = 0,
152
+ .access = PL2_RW, .type = ARM_CP_64BIT,
153
+ .readfn = arm_cp_read_zero, .writefn = arm_cp_write_ignore
154
+};
155
+
156
+static const ARMCPRegInfo zcr_el3_reginfo = {
157
+ .name = "ZCR_EL3", .state = ARM_CP_STATE_AA64,
158
+ .opc0 = 3, .opc1 = 6, .crn = 1, .crm = 2, .opc2 = 0,
159
+ .access = PL3_RW, .accessfn = zcr_access, .type = ARM_CP_64BIT,
160
+ .fieldoffset = offsetof(CPUARMState, vfp.zcr_el[3]),
161
+ .writefn = zcr_write, .raw_writefn = raw_write
162
+};
163
+
164
void hw_watchpoint_update(ARMCPU *cpu, int n)
165
{
166
CPUARMState *env = &cpu->env;
167
@@ -XXX,XX +XXX,XX @@ void register_cp_regs_for_features(ARMCPU *cpu)
168
}
169
define_one_arm_cp_reg(cpu, &sctlr);
170
}
80
}
171
+
172
+ if (arm_feature(env, ARM_FEATURE_SVE)) {
173
+ define_one_arm_cp_reg(cpu, &zcr_el1_reginfo);
174
+ if (arm_feature(env, ARM_FEATURE_EL2)) {
175
+ define_one_arm_cp_reg(cpu, &zcr_el2_reginfo);
176
+ } else {
177
+ define_one_arm_cp_reg(cpu, &zcr_no_el2_reginfo);
178
+ }
179
+ if (arm_feature(env, ARM_FEATURE_EL3)) {
180
+ define_one_arm_cp_reg(cpu, &zcr_el3_reginfo);
181
+ }
182
+ }
183
}
184
185
void arm_cpu_register_gdb_regs_for_features(ARMCPU *cpu)
186
--
81
--
187
2.16.1
82
2.17.0
188
83
189
84
diff view generated by jsdifflib
1
From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
1
From: Alex Bennée <alex.bennee@linaro.org>
2
2
3
This implements emulation of the new SHA-512 instructions that have
3
These where missed out from the rest of the half-precision work.
4
been added as an optional extensions to the ARMv8 Crypto Extensions
4
5
in ARM v8.2.
5
Cc: qemu-stable@nongnu.org
6
7
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
8
Message-id: 20180207111729.15737-2-ard.biesheuvel@linaro.org
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
8
Tested-by: Alex Bennée <alex.bennee@linaro.org>
9
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Message-id: 20180512003217.9105-9-richard.henderson@linaro.org
11
[rth: Diagnose lack of FP16 before fp_access_check]
12
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
14
---
12
target/arm/cpu.h | 1 +
15
target/arm/helper-a64.h | 2 +
13
target/arm/helper.h | 5 +++
16
target/arm/helper-a64.c | 10 +++++
14
target/arm/crypto_helper.c | 90 ++++++++++++++++++++++++++++++++++++-
17
target/arm/translate-a64.c | 88 ++++++++++++++++++++++++++++++--------
15
target/arm/translate-a64.c | 110 +++++++++++++++++++++++++++++++++++++++++++++
18
3 files changed, 83 insertions(+), 17 deletions(-)
16
4 files changed, 205 insertions(+), 1 deletion(-)
19
17
20
diff --git a/target/arm/helper-a64.h b/target/arm/helper-a64.h
18
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
19
index XXXXXXX..XXXXXXX 100644
21
index XXXXXXX..XXXXXXX 100644
20
--- a/target/arm/cpu.h
22
--- a/target/arm/helper-a64.h
21
+++ b/target/arm/cpu.h
23
+++ b/target/arm/helper-a64.h
22
@@ -XXX,XX +XXX,XX @@ enum arm_features {
24
@@ -XXX,XX +XXX,XX @@
23
ARM_FEATURE_M_SECURITY, /* M profile Security Extension */
25
DEF_HELPER_FLAGS_2(udiv64, TCG_CALL_NO_RWG_SE, i64, i64, i64)
24
ARM_FEATURE_JAZELLE, /* has (trivial) Jazelle implementation */
26
DEF_HELPER_FLAGS_2(sdiv64, TCG_CALL_NO_RWG_SE, s64, s64, s64)
25
ARM_FEATURE_SVE, /* has Scalable Vector Extension */
27
DEF_HELPER_FLAGS_1(rbit64, TCG_CALL_NO_RWG_SE, i64, i64)
26
+ ARM_FEATURE_V8_SHA512, /* implements SHA512 part of v8 Crypto Extensions */
28
+DEF_HELPER_3(vfp_cmph_a64, i64, f16, f16, ptr)
27
};
29
+DEF_HELPER_3(vfp_cmpeh_a64, i64, f16, f16, ptr)
28
30
DEF_HELPER_3(vfp_cmps_a64, i64, f32, f32, ptr)
29
static inline int arm_feature(CPUARMState *env, int feature)
31
DEF_HELPER_3(vfp_cmpes_a64, i64, f32, f32, ptr)
30
diff --git a/target/arm/helper.h b/target/arm/helper.h
32
DEF_HELPER_3(vfp_cmpd_a64, i64, f64, f64, ptr)
33
diff --git a/target/arm/helper-a64.c b/target/arm/helper-a64.c
31
index XXXXXXX..XXXXXXX 100644
34
index XXXXXXX..XXXXXXX 100644
32
--- a/target/arm/helper.h
35
--- a/target/arm/helper-a64.c
33
+++ b/target/arm/helper.h
36
+++ b/target/arm/helper-a64.c
34
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_3(crypto_sha256h2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
37
@@ -XXX,XX +XXX,XX @@ static inline uint32_t float_rel_to_flags(int res)
35
DEF_HELPER_FLAGS_2(crypto_sha256su0, TCG_CALL_NO_RWG, void, ptr, ptr)
38
return flags;
36
DEF_HELPER_FLAGS_3(crypto_sha256su1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
37
38
+DEF_HELPER_FLAGS_3(crypto_sha512h, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
39
+DEF_HELPER_FLAGS_3(crypto_sha512h2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
40
+DEF_HELPER_FLAGS_2(crypto_sha512su0, TCG_CALL_NO_RWG, void, ptr, ptr)
41
+DEF_HELPER_FLAGS_3(crypto_sha512su1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
42
+
43
DEF_HELPER_FLAGS_3(crc32, TCG_CALL_NO_RWG_SE, i32, i32, i32, i32)
44
DEF_HELPER_FLAGS_3(crc32c, TCG_CALL_NO_RWG_SE, i32, i32, i32, i32)
45
DEF_HELPER_2(dc_zva, void, env, i64)
46
diff --git a/target/arm/crypto_helper.c b/target/arm/crypto_helper.c
47
index XXXXXXX..XXXXXXX 100644
48
--- a/target/arm/crypto_helper.c
49
+++ b/target/arm/crypto_helper.c
50
@@ -XXX,XX +XXX,XX @@
51
/*
52
* crypto_helper.c - emulate v8 Crypto Extensions instructions
53
*
54
- * Copyright (C) 2013 - 2014 Linaro Ltd <ard.biesheuvel@linaro.org>
55
+ * Copyright (C) 2013 - 2018 Linaro Ltd <ard.biesheuvel@linaro.org>
56
*
57
* This library is free software; you can redistribute it and/or
58
* modify it under the terms of the GNU Lesser General Public
59
@@ -XXX,XX +XXX,XX @@ void HELPER(crypto_sha256su1)(void *vd, void *vn, void *vm)
60
rd[0] = d.l[0];
61
rd[1] = d.l[1];
62
}
39
}
63
+
40
64
+/*
41
+uint64_t HELPER(vfp_cmph_a64)(float16 x, float16 y, void *fp_status)
65
+ * The SHA-512 logical functions (same as above but using 64-bit operands)
66
+ */
67
+
68
+static uint64_t cho512(uint64_t x, uint64_t y, uint64_t z)
69
+{
42
+{
70
+ return (x & (y ^ z)) ^ z;
43
+ return float_rel_to_flags(float16_compare_quiet(x, y, fp_status));
71
+}
44
+}
72
+
45
+
73
+static uint64_t maj512(uint64_t x, uint64_t y, uint64_t z)
46
+uint64_t HELPER(vfp_cmpeh_a64)(float16 x, float16 y, void *fp_status)
74
+{
47
+{
75
+ return (x & y) | ((x | y) & z);
48
+ return float_rel_to_flags(float16_compare(x, y, fp_status));
76
+}
49
+}
77
+
50
+
78
+static uint64_t S0_512(uint64_t x)
51
uint64_t HELPER(vfp_cmps_a64)(float32 x, float32 y, void *fp_status)
79
+{
52
{
80
+ return ror64(x, 28) ^ ror64(x, 34) ^ ror64(x, 39);
53
return float_rel_to_flags(float32_compare_quiet(x, y, fp_status));
81
+}
82
+
83
+static uint64_t S1_512(uint64_t x)
84
+{
85
+ return ror64(x, 14) ^ ror64(x, 18) ^ ror64(x, 41);
86
+}
87
+
88
+static uint64_t s0_512(uint64_t x)
89
+{
90
+ return ror64(x, 1) ^ ror64(x, 8) ^ (x >> 7);
91
+}
92
+
93
+static uint64_t s1_512(uint64_t x)
94
+{
95
+ return ror64(x, 19) ^ ror64(x, 61) ^ (x >> 6);
96
+}
97
+
98
+void HELPER(crypto_sha512h)(void *vd, void *vn, void *vm)
99
+{
100
+ uint64_t *rd = vd;
101
+ uint64_t *rn = vn;
102
+ uint64_t *rm = vm;
103
+ uint64_t d0 = rd[0];
104
+ uint64_t d1 = rd[1];
105
+
106
+ d1 += S1_512(rm[1]) + cho512(rm[1], rn[0], rn[1]);
107
+ d0 += S1_512(d1 + rm[0]) + cho512(d1 + rm[0], rm[1], rn[0]);
108
+
109
+ rd[0] = d0;
110
+ rd[1] = d1;
111
+}
112
+
113
+void HELPER(crypto_sha512h2)(void *vd, void *vn, void *vm)
114
+{
115
+ uint64_t *rd = vd;
116
+ uint64_t *rn = vn;
117
+ uint64_t *rm = vm;
118
+ uint64_t d0 = rd[0];
119
+ uint64_t d1 = rd[1];
120
+
121
+ d1 += S0_512(rm[0]) + maj512(rn[0], rm[1], rm[0]);
122
+ d0 += S0_512(d1) + maj512(d1, rm[0], rm[1]);
123
+
124
+ rd[0] = d0;
125
+ rd[1] = d1;
126
+}
127
+
128
+void HELPER(crypto_sha512su0)(void *vd, void *vn)
129
+{
130
+ uint64_t *rd = vd;
131
+ uint64_t *rn = vn;
132
+ uint64_t d0 = rd[0];
133
+ uint64_t d1 = rd[1];
134
+
135
+ d0 += s0_512(rd[1]);
136
+ d1 += s0_512(rn[0]);
137
+
138
+ rd[0] = d0;
139
+ rd[1] = d1;
140
+}
141
+
142
+void HELPER(crypto_sha512su1)(void *vd, void *vn, void *vm)
143
+{
144
+ uint64_t *rd = vd;
145
+ uint64_t *rn = vn;
146
+ uint64_t *rm = vm;
147
+
148
+ rd[0] += s1_512(rn[0]) + rm[0];
149
+ rd[1] += s1_512(rn[1]) + rm[1];
150
+}
151
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
54
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
152
index XXXXXXX..XXXXXXX 100644
55
index XXXXXXX..XXXXXXX 100644
153
--- a/target/arm/translate-a64.c
56
--- a/target/arm/translate-a64.c
154
+++ b/target/arm/translate-a64.c
57
+++ b/target/arm/translate-a64.c
155
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_two_reg_sha(DisasContext *s, uint32_t insn)
58
@@ -XXX,XX +XXX,XX @@ static void disas_data_proc_reg(DisasContext *s, uint32_t insn)
156
tcg_temp_free_ptr(tcg_rn_ptr);
59
}
157
}
60
}
158
61
159
+/* Crypto three-reg SHA512
62
-static void handle_fp_compare(DisasContext *s, bool is_double,
160
+ * 31 21 20 16 15 14 13 12 11 10 9 5 4 0
63
+static void handle_fp_compare(DisasContext *s, int size,
161
+ * +-----------------------+------+---+---+-----+--------+------+------+
64
unsigned int rn, unsigned int rm,
162
+ * | 1 1 0 0 1 1 1 0 0 1 1 | Rm | 1 | O | 0 0 | opcode | Rn | Rd |
65
bool cmp_with_zero, bool signal_all_nans)
163
+ * +-----------------------+------+---+---+-----+--------+------+------+
66
{
164
+ */
67
TCGv_i64 tcg_flags = tcg_temp_new_i64();
165
+static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
68
- TCGv_ptr fpst = get_fpstatus_ptr(false);
166
+{
69
+ TCGv_ptr fpst = get_fpstatus_ptr(size == MO_16);
167
+ int opcode = extract32(insn, 10, 2);
70
168
+ int o = extract32(insn, 14, 1);
71
- if (is_double) {
169
+ int rm = extract32(insn, 16, 5);
72
+ if (size == MO_64) {
170
+ int rn = extract32(insn, 5, 5);
73
TCGv_i64 tcg_vn, tcg_vm;
171
+ int rd = extract32(insn, 0, 5);
74
172
+ int feature;
75
tcg_vn = read_fp_dreg(s, rn);
173
+ CryptoThreeOpFn *genfn;
76
@@ -XXX,XX +XXX,XX @@ static void handle_fp_compare(DisasContext *s, bool is_double,
174
+
77
tcg_temp_free_i64(tcg_vn);
175
+ if (o == 0) {
78
tcg_temp_free_i64(tcg_vm);
176
+ switch (opcode) {
79
} else {
177
+ case 0: /* SHA512H */
80
- TCGv_i32 tcg_vn, tcg_vm;
178
+ feature = ARM_FEATURE_V8_SHA512;
81
+ TCGv_i32 tcg_vn = tcg_temp_new_i32();
179
+ genfn = gen_helper_crypto_sha512h;
82
+ TCGv_i32 tcg_vm = tcg_temp_new_i32();
180
+ break;
83
181
+ case 1: /* SHA512H2 */
84
- tcg_vn = read_fp_sreg(s, rn);
182
+ feature = ARM_FEATURE_V8_SHA512;
85
+ read_vec_element_i32(s, tcg_vn, rn, 0, size);
183
+ genfn = gen_helper_crypto_sha512h2;
86
if (cmp_with_zero) {
184
+ break;
87
- tcg_vm = tcg_const_i32(0);
185
+ case 2: /* SHA512SU1 */
88
+ tcg_gen_movi_i32(tcg_vm, 0);
186
+ feature = ARM_FEATURE_V8_SHA512;
89
} else {
187
+ genfn = gen_helper_crypto_sha512su1;
90
- tcg_vm = read_fp_sreg(s, rm);
91
+ read_vec_element_i32(s, tcg_vm, rm, 0, size);
92
}
93
- if (signal_all_nans) {
94
- gen_helper_vfp_cmpes_a64(tcg_flags, tcg_vn, tcg_vm, fpst);
95
- } else {
96
- gen_helper_vfp_cmps_a64(tcg_flags, tcg_vn, tcg_vm, fpst);
97
+
98
+ switch (size) {
99
+ case MO_32:
100
+ if (signal_all_nans) {
101
+ gen_helper_vfp_cmpes_a64(tcg_flags, tcg_vn, tcg_vm, fpst);
102
+ } else {
103
+ gen_helper_vfp_cmps_a64(tcg_flags, tcg_vn, tcg_vm, fpst);
104
+ }
105
+ break;
106
+ case MO_16:
107
+ if (signal_all_nans) {
108
+ gen_helper_vfp_cmpeh_a64(tcg_flags, tcg_vn, tcg_vm, fpst);
109
+ } else {
110
+ gen_helper_vfp_cmph_a64(tcg_flags, tcg_vn, tcg_vm, fpst);
111
+ }
188
+ break;
112
+ break;
189
+ default:
113
+ default:
190
+ unallocated_encoding(s);
114
+ g_assert_not_reached();
191
+ return;
115
}
192
+ }
116
+
193
+ } else {
117
tcg_temp_free_i32(tcg_vn);
118
tcg_temp_free_i32(tcg_vm);
119
}
120
@@ -XXX,XX +XXX,XX @@ static void handle_fp_compare(DisasContext *s, bool is_double,
121
static void disas_fp_compare(DisasContext *s, uint32_t insn)
122
{
123
unsigned int mos, type, rm, op, rn, opc, op2r;
124
+ int size;
125
126
mos = extract32(insn, 29, 3);
127
- type = extract32(insn, 22, 2); /* 0 = single, 1 = double */
128
+ type = extract32(insn, 22, 2);
129
rm = extract32(insn, 16, 5);
130
op = extract32(insn, 14, 2);
131
rn = extract32(insn, 5, 5);
132
opc = extract32(insn, 3, 2);
133
op2r = extract32(insn, 0, 3);
134
135
- if (mos || op || op2r || type > 1) {
136
+ if (mos || op || op2r) {
194
+ unallocated_encoding(s);
137
+ unallocated_encoding(s);
195
+ return;
138
+ return;
196
+ }
139
+ }
197
+
140
+
198
+ if (!arm_dc_feature(s, feature)) {
141
+ switch (type) {
142
+ case 0:
143
+ size = MO_32;
144
+ break;
145
+ case 1:
146
+ size = MO_64;
147
+ break;
148
+ case 3:
149
+ size = MO_16;
150
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
151
+ break;
152
+ }
153
+ /* fallthru */
154
+ default:
155
unallocated_encoding(s);
156
return;
157
}
158
@@ -XXX,XX +XXX,XX @@ static void disas_fp_compare(DisasContext *s, uint32_t insn)
159
return;
160
}
161
162
- handle_fp_compare(s, type, rn, rm, opc & 1, opc & 2);
163
+ handle_fp_compare(s, size, rn, rm, opc & 1, opc & 2);
164
}
165
166
/* Floating point conditional compare
167
@@ -XXX,XX +XXX,XX @@ static void disas_fp_ccomp(DisasContext *s, uint32_t insn)
168
unsigned int mos, type, rm, cond, rn, op, nzcv;
169
TCGv_i64 tcg_flags;
170
TCGLabel *label_continue = NULL;
171
+ int size;
172
173
mos = extract32(insn, 29, 3);
174
- type = extract32(insn, 22, 2); /* 0 = single, 1 = double */
175
+ type = extract32(insn, 22, 2);
176
rm = extract32(insn, 16, 5);
177
cond = extract32(insn, 12, 4);
178
rn = extract32(insn, 5, 5);
179
op = extract32(insn, 4, 1);
180
nzcv = extract32(insn, 0, 4);
181
182
- if (mos || type > 1) {
183
+ if (mos) {
199
+ unallocated_encoding(s);
184
+ unallocated_encoding(s);
200
+ return;
185
+ return;
201
+ }
186
+ }
202
+
187
+
203
+ if (!fp_access_check(s)) {
188
+ switch (type) {
204
+ return;
189
+ case 0:
205
+ }
190
+ size = MO_32;
206
+
191
+ break;
207
+ if (genfn) {
192
+ case 1:
208
+ TCGv_ptr tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr;
193
+ size = MO_64;
209
+
194
+ break;
210
+ tcg_rd_ptr = vec_full_reg_ptr(s, rd);
195
+ case 3:
211
+ tcg_rn_ptr = vec_full_reg_ptr(s, rn);
196
+ size = MO_16;
212
+ tcg_rm_ptr = vec_full_reg_ptr(s, rm);
197
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
213
+
198
+ break;
214
+ genfn(tcg_rd_ptr, tcg_rn_ptr, tcg_rm_ptr);
199
+ }
215
+
200
+ /* fallthru */
216
+ tcg_temp_free_ptr(tcg_rd_ptr);
217
+ tcg_temp_free_ptr(tcg_rn_ptr);
218
+ tcg_temp_free_ptr(tcg_rm_ptr);
219
+ } else {
220
+ g_assert_not_reached();
221
+ }
222
+}
223
+
224
+/* Crypto two-reg SHA512
225
+ * 31 12 11 10 9 5 4 0
226
+ * +-----------------------------------------+--------+------+------+
227
+ * | 1 1 0 0 1 1 1 0 1 1 0 0 0 0 0 0 1 0 0 0 | opcode | Rn | Rd |
228
+ * +-----------------------------------------+--------+------+------+
229
+ */
230
+static void disas_crypto_two_reg_sha512(DisasContext *s, uint32_t insn)
231
+{
232
+ int opcode = extract32(insn, 10, 2);
233
+ int rn = extract32(insn, 5, 5);
234
+ int rd = extract32(insn, 0, 5);
235
+ TCGv_ptr tcg_rd_ptr, tcg_rn_ptr;
236
+ int feature;
237
+ CryptoTwoOpFn *genfn;
238
+
239
+ switch (opcode) {
240
+ case 0: /* SHA512SU0 */
241
+ feature = ARM_FEATURE_V8_SHA512;
242
+ genfn = gen_helper_crypto_sha512su0;
243
+ break;
244
+ default:
201
+ default:
245
+ unallocated_encoding(s);
202
unallocated_encoding(s);
246
+ return;
203
return;
247
+ }
204
}
248
+
205
@@ -XXX,XX +XXX,XX @@ static void disas_fp_ccomp(DisasContext *s, uint32_t insn)
249
+ if (!arm_dc_feature(s, feature)) {
206
gen_set_label(label_match);
250
+ unallocated_encoding(s);
207
}
251
+ return;
208
252
+ }
209
- handle_fp_compare(s, type, rn, rm, false, op);
253
+
210
+ handle_fp_compare(s, size, rn, rm, false, op);
254
+ if (!fp_access_check(s)) {
211
255
+ return;
212
if (cond < 0x0e) {
256
+ }
213
gen_set_label(label_continue);
257
+
258
+ tcg_rd_ptr = vec_full_reg_ptr(s, rd);
259
+ tcg_rn_ptr = vec_full_reg_ptr(s, rn);
260
+
261
+ genfn(tcg_rd_ptr, tcg_rn_ptr);
262
+
263
+ tcg_temp_free_ptr(tcg_rd_ptr);
264
+ tcg_temp_free_ptr(tcg_rn_ptr);
265
+}
266
+
267
/* C3.6 Data processing - SIMD, inc Crypto
268
*
269
* As the decode gets a little complex we are using a table based
270
@@ -XXX,XX +XXX,XX @@ static const AArch64DecodeTable data_proc_simd[] = {
271
{ 0x4e280800, 0xff3e0c00, disas_crypto_aes },
272
{ 0x5e000000, 0xff208c00, disas_crypto_three_reg_sha },
273
{ 0x5e280800, 0xff3e0c00, disas_crypto_two_reg_sha },
274
+ { 0xce608000, 0xffe0b000, disas_crypto_three_reg_sha512 },
275
+ { 0xcec08000, 0xfffff000, disas_crypto_two_reg_sha512 },
276
{ 0x00000000, 0x00000000, NULL }
277
};
278
279
--
214
--
280
2.16.1
215
2.17.0
281
216
282
217
diff view generated by jsdifflib
1
From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
1
From: Alex Bennée <alex.bennee@linaro.org>
2
2
3
This implements emulation of the new SHA-3 instructions that have
3
These were missed out from the rest of the half-precision work.
4
been added as an optional extensions to the ARMv8 Crypto Extensions
5
in ARM v8.2.
6
4
7
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
5
Cc: qemu-stable@nongnu.org
8
Message-id: 20180207111729.15737-3-ard.biesheuvel@linaro.org
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
8
Tested-by: Alex Bennée <alex.bennee@linaro.org>
9
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Message-id: 20180512003217.9105-10-richard.henderson@linaro.org
11
[rth: Fix erroneous check vs type]
12
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
13
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
14
---
12
target/arm/cpu.h | 1 +
15
target/arm/translate-a64.c | 31 +++++++++++++++++++++++++------
13
target/arm/translate-a64.c | 148 +++++++++++++++++++++++++++++++++++++++++++--
16
1 file changed, 25 insertions(+), 6 deletions(-)
14
2 files changed, 145 insertions(+), 4 deletions(-)
15
17
16
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
17
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/cpu.h
19
+++ b/target/arm/cpu.h
20
@@ -XXX,XX +XXX,XX @@ enum arm_features {
21
ARM_FEATURE_JAZELLE, /* has (trivial) Jazelle implementation */
22
ARM_FEATURE_SVE, /* has Scalable Vector Extension */
23
ARM_FEATURE_V8_SHA512, /* implements SHA512 part of v8 Crypto Extensions */
24
+ ARM_FEATURE_V8_SHA3, /* implements SHA3 part of v8 Crypto Extensions */
25
};
26
27
static inline int arm_feature(CPUARMState *env, int feature)
28
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
18
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
29
index XXXXXXX..XXXXXXX 100644
19
index XXXXXXX..XXXXXXX 100644
30
--- a/target/arm/translate-a64.c
20
--- a/target/arm/translate-a64.c
31
+++ b/target/arm/translate-a64.c
21
+++ b/target/arm/translate-a64.c
32
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
22
@@ -XXX,XX +XXX,XX @@ static void disas_fp_csel(DisasContext *s, uint32_t insn)
33
feature = ARM_FEATURE_V8_SHA512;
23
unsigned int mos, type, rm, cond, rn, rd;
34
genfn = gen_helper_crypto_sha512su1;
24
TCGv_i64 t_true, t_false, t_zero;
35
break;
25
DisasCompare64 c;
36
- default:
26
+ TCGMemOp sz;
37
- unallocated_encoding(s);
27
38
- return;
28
mos = extract32(insn, 29, 3);
39
+ case 3: /* RAX1 */
29
- type = extract32(insn, 22, 2); /* 0 = single, 1 = double */
40
+ feature = ARM_FEATURE_V8_SHA3;
30
+ type = extract32(insn, 22, 2);
41
+ genfn = NULL;
31
rm = extract32(insn, 16, 5);
42
+ break;
32
cond = extract32(insn, 12, 4);
43
}
33
rn = extract32(insn, 5, 5);
44
} else {
34
rd = extract32(insn, 0, 5);
45
unallocated_encoding(s);
35
46
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
36
- if (mos || type > 1) {
47
tcg_temp_free_ptr(tcg_rn_ptr);
37
+ if (mos) {
48
tcg_temp_free_ptr(tcg_rm_ptr);
49
} else {
50
- g_assert_not_reached();
51
+ TCGv_i64 tcg_op1, tcg_op2, tcg_res[2];
52
+ int pass;
53
+
54
+ tcg_op1 = tcg_temp_new_i64();
55
+ tcg_op2 = tcg_temp_new_i64();
56
+ tcg_res[0] = tcg_temp_new_i64();
57
+ tcg_res[1] = tcg_temp_new_i64();
58
+
59
+ for (pass = 0; pass < 2; pass++) {
60
+ read_vec_element(s, tcg_op1, rn, pass, MO_64);
61
+ read_vec_element(s, tcg_op2, rm, pass, MO_64);
62
+
63
+ tcg_gen_rotli_i64(tcg_res[pass], tcg_op2, 1);
64
+ tcg_gen_xor_i64(tcg_res[pass], tcg_res[pass], tcg_op1);
65
+ }
66
+ write_vec_element(s, tcg_res[0], rd, 0, MO_64);
67
+ write_vec_element(s, tcg_res[1], rd, 1, MO_64);
68
+
69
+ tcg_temp_free_i64(tcg_op1);
70
+ tcg_temp_free_i64(tcg_op2);
71
+ tcg_temp_free_i64(tcg_res[0]);
72
+ tcg_temp_free_i64(tcg_res[1]);
73
}
74
}
75
76
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_two_reg_sha512(DisasContext *s, uint32_t insn)
77
tcg_temp_free_ptr(tcg_rn_ptr);
78
}
79
80
+/* Crypto four-register
81
+ * 31 23 22 21 20 16 15 14 10 9 5 4 0
82
+ * +-------------------+-----+------+---+------+------+------+
83
+ * | 1 1 0 0 1 1 1 0 0 | Op0 | Rm | 0 | Ra | Rn | Rd |
84
+ * +-------------------+-----+------+---+------+------+------+
85
+ */
86
+static void disas_crypto_four_reg(DisasContext *s, uint32_t insn)
87
+{
88
+ int op0 = extract32(insn, 21, 2);
89
+ int rm = extract32(insn, 16, 5);
90
+ int ra = extract32(insn, 10, 5);
91
+ int rn = extract32(insn, 5, 5);
92
+ int rd = extract32(insn, 0, 5);
93
+ int feature;
94
+
95
+ switch (op0) {
96
+ case 0: /* EOR3 */
97
+ case 1: /* BCAX */
98
+ feature = ARM_FEATURE_V8_SHA3;
99
+ break;
100
+ default:
101
+ unallocated_encoding(s);
38
+ unallocated_encoding(s);
102
+ return;
39
+ return;
103
+ }
40
+ }
104
+
41
+
105
+ if (!arm_dc_feature(s, feature)) {
42
+ switch (type) {
106
+ unallocated_encoding(s);
43
+ case 0:
107
+ return;
44
+ sz = MO_32;
108
+ }
45
+ break;
109
+
46
+ case 1:
110
+ if (!fp_access_check(s)) {
47
+ sz = MO_64;
111
+ return;
48
+ break;
112
+ }
49
+ case 3:
113
+
50
+ sz = MO_16;
114
+ if (op0 < 2) {
51
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
115
+ TCGv_i64 tcg_op1, tcg_op2, tcg_op3, tcg_res[2];
52
+ break;
116
+ int pass;
117
+
118
+ tcg_op1 = tcg_temp_new_i64();
119
+ tcg_op2 = tcg_temp_new_i64();
120
+ tcg_op3 = tcg_temp_new_i64();
121
+ tcg_res[0] = tcg_temp_new_i64();
122
+ tcg_res[1] = tcg_temp_new_i64();
123
+
124
+ for (pass = 0; pass < 2; pass++) {
125
+ read_vec_element(s, tcg_op1, rn, pass, MO_64);
126
+ read_vec_element(s, tcg_op2, rm, pass, MO_64);
127
+ read_vec_element(s, tcg_op3, ra, pass, MO_64);
128
+
129
+ if (op0 == 0) {
130
+ /* EOR3 */
131
+ tcg_gen_xor_i64(tcg_res[pass], tcg_op2, tcg_op3);
132
+ } else {
133
+ /* BCAX */
134
+ tcg_gen_andc_i64(tcg_res[pass], tcg_op2, tcg_op3);
135
+ }
136
+ tcg_gen_xor_i64(tcg_res[pass], tcg_res[pass], tcg_op1);
137
+ }
53
+ }
138
+ write_vec_element(s, tcg_res[0], rd, 0, MO_64);
54
+ /* fallthru */
139
+ write_vec_element(s, tcg_res[1], rd, 1, MO_64);
55
+ default:
140
+
56
unallocated_encoding(s);
141
+ tcg_temp_free_i64(tcg_op1);
57
return;
142
+ tcg_temp_free_i64(tcg_op2);
58
}
143
+ tcg_temp_free_i64(tcg_op3);
59
@@ -XXX,XX +XXX,XX @@ static void disas_fp_csel(DisasContext *s, uint32_t insn)
144
+ tcg_temp_free_i64(tcg_res[0]);
60
return;
145
+ tcg_temp_free_i64(tcg_res[1]);
61
}
146
+ } else {
62
147
+ g_assert_not_reached();
63
- /* Zero extend sreg inputs to 64 bits now. */
148
+ }
64
+ /* Zero extend sreg & hreg inputs to 64 bits now. */
149
+}
65
t_true = tcg_temp_new_i64();
150
+
66
t_false = tcg_temp_new_i64();
151
+/* Crypto XAR
67
- read_vec_element(s, t_true, rn, 0, type ? MO_64 : MO_32);
152
+ * 31 21 20 16 15 10 9 5 4 0
68
- read_vec_element(s, t_false, rm, 0, type ? MO_64 : MO_32);
153
+ * +-----------------------+------+--------+------+------+
69
+ read_vec_element(s, t_true, rn, 0, sz);
154
+ * | 1 1 0 0 1 1 1 0 1 0 0 | Rm | imm6 | Rn | Rd |
70
+ read_vec_element(s, t_false, rm, 0, sz);
155
+ * +-----------------------+------+--------+------+------+
71
156
+ */
72
a64_test_cc(&c, cond);
157
+static void disas_crypto_xar(DisasContext *s, uint32_t insn)
73
t_zero = tcg_const_i64(0);
158
+{
74
@@ -XXX,XX +XXX,XX @@ static void disas_fp_csel(DisasContext *s, uint32_t insn)
159
+ int rm = extract32(insn, 16, 5);
75
tcg_temp_free_i64(t_false);
160
+ int imm6 = extract32(insn, 10, 6);
76
a64_free_cc(&c);
161
+ int rn = extract32(insn, 5, 5);
77
162
+ int rd = extract32(insn, 0, 5);
78
- /* Note that sregs write back zeros to the high bits,
163
+ TCGv_i64 tcg_op1, tcg_op2, tcg_res[2];
79
+ /* Note that sregs & hregs write back zeros to the high bits,
164
+ int pass;
80
and we've already done the zero-extension. */
165
+
81
write_fp_dreg(s, rd, t_true);
166
+ if (!arm_dc_feature(s, ARM_FEATURE_V8_SHA3)) {
82
tcg_temp_free_i64(t_true);
167
+ unallocated_encoding(s);
168
+ return;
169
+ }
170
+
171
+ if (!fp_access_check(s)) {
172
+ return;
173
+ }
174
+
175
+ tcg_op1 = tcg_temp_new_i64();
176
+ tcg_op2 = tcg_temp_new_i64();
177
+ tcg_res[0] = tcg_temp_new_i64();
178
+ tcg_res[1] = tcg_temp_new_i64();
179
+
180
+ for (pass = 0; pass < 2; pass++) {
181
+ read_vec_element(s, tcg_op1, rn, pass, MO_64);
182
+ read_vec_element(s, tcg_op2, rm, pass, MO_64);
183
+
184
+ tcg_gen_xor_i64(tcg_res[pass], tcg_op1, tcg_op2);
185
+ tcg_gen_rotri_i64(tcg_res[pass], tcg_res[pass], imm6);
186
+ }
187
+ write_vec_element(s, tcg_res[0], rd, 0, MO_64);
188
+ write_vec_element(s, tcg_res[1], rd, 1, MO_64);
189
+
190
+ tcg_temp_free_i64(tcg_op1);
191
+ tcg_temp_free_i64(tcg_op2);
192
+ tcg_temp_free_i64(tcg_res[0]);
193
+ tcg_temp_free_i64(tcg_res[1]);
194
+}
195
+
196
/* C3.6 Data processing - SIMD, inc Crypto
197
*
198
* As the decode gets a little complex we are using a table based
199
@@ -XXX,XX +XXX,XX @@ static const AArch64DecodeTable data_proc_simd[] = {
200
{ 0x5e280800, 0xff3e0c00, disas_crypto_two_reg_sha },
201
{ 0xce608000, 0xffe0b000, disas_crypto_three_reg_sha512 },
202
{ 0xcec08000, 0xfffff000, disas_crypto_two_reg_sha512 },
203
+ { 0xce000000, 0xff808000, disas_crypto_four_reg },
204
+ { 0xce800000, 0xffe00000, disas_crypto_xar },
205
{ 0x00000000, 0x00000000, NULL }
206
};
207
208
--
83
--
209
2.16.1
84
2.17.0
210
85
211
86
diff view generated by jsdifflib
1
From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
1
From: Alex Bennée <alex.bennee@linaro.org>
2
2
3
This implements emulation of the new SM4 instructions that have
3
All the hard work is already done by vfp_expand_imm, we just need to
4
been added as an optional extension to the ARMv8 Crypto Extensions
4
make sure we pick up the correct size.
5
in ARM v8.2.
6
5
7
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
6
Cc: qemu-stable@nongnu.org
8
Message-id: 20180207111729.15737-5-ard.biesheuvel@linaro.org
9
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
7
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
8
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
9
Tested-by: Alex Bennée <alex.bennee@linaro.org>
10
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
11
Message-id: 20180512003217.9105-11-richard.henderson@linaro.org
12
[rth: Merge unallocated_encoding check with TCGMemOp conversion.]
13
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
14
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
15
---
12
target/arm/cpu.h | 1 +
16
target/arm/translate-a64.c | 20 +++++++++++++++++---
13
target/arm/helper.h | 3 ++
17
1 file changed, 17 insertions(+), 3 deletions(-)
14
target/arm/crypto_helper.c | 91 ++++++++++++++++++++++++++++++++++++++++++++++
15
target/arm/translate-a64.c | 8 ++++
16
4 files changed, 103 insertions(+)
17
18
18
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
19
index XXXXXXX..XXXXXXX 100644
20
--- a/target/arm/cpu.h
21
+++ b/target/arm/cpu.h
22
@@ -XXX,XX +XXX,XX @@ enum arm_features {
23
ARM_FEATURE_V8_SHA512, /* implements SHA512 part of v8 Crypto Extensions */
24
ARM_FEATURE_V8_SHA3, /* implements SHA3 part of v8 Crypto Extensions */
25
ARM_FEATURE_V8_SM3, /* implements SM3 part of v8 Crypto Extensions */
26
+ ARM_FEATURE_V8_SM4, /* implements SM4 part of v8 Crypto Extensions */
27
};
28
29
static inline int arm_feature(CPUARMState *env, int feature)
30
diff --git a/target/arm/helper.h b/target/arm/helper.h
31
index XXXXXXX..XXXXXXX 100644
32
--- a/target/arm/helper.h
33
+++ b/target/arm/helper.h
34
@@ -XXX,XX +XXX,XX @@ DEF_HELPER_FLAGS_5(crypto_sm3tt, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32, i32)
35
DEF_HELPER_FLAGS_3(crypto_sm3partw1, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
36
DEF_HELPER_FLAGS_3(crypto_sm3partw2, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
37
38
+DEF_HELPER_FLAGS_2(crypto_sm4e, TCG_CALL_NO_RWG, void, ptr, ptr)
39
+DEF_HELPER_FLAGS_3(crypto_sm4ekey, TCG_CALL_NO_RWG, void, ptr, ptr, ptr)
40
+
41
DEF_HELPER_FLAGS_3(crc32, TCG_CALL_NO_RWG_SE, i32, i32, i32, i32)
42
DEF_HELPER_FLAGS_3(crc32c, TCG_CALL_NO_RWG_SE, i32, i32, i32, i32)
43
DEF_HELPER_2(dc_zva, void, env, i64)
44
diff --git a/target/arm/crypto_helper.c b/target/arm/crypto_helper.c
45
index XXXXXXX..XXXXXXX 100644
46
--- a/target/arm/crypto_helper.c
47
+++ b/target/arm/crypto_helper.c
48
@@ -XXX,XX +XXX,XX @@ void HELPER(crypto_sm3tt)(void *vd, void *vn, void *vm, uint32_t imm2,
49
rd[0] = d.l[0];
50
rd[1] = d.l[1];
51
}
52
+
53
+static uint8_t const sm4_sbox[] = {
54
+ 0xd6, 0x90, 0xe9, 0xfe, 0xcc, 0xe1, 0x3d, 0xb7,
55
+ 0x16, 0xb6, 0x14, 0xc2, 0x28, 0xfb, 0x2c, 0x05,
56
+ 0x2b, 0x67, 0x9a, 0x76, 0x2a, 0xbe, 0x04, 0xc3,
57
+ 0xaa, 0x44, 0x13, 0x26, 0x49, 0x86, 0x06, 0x99,
58
+ 0x9c, 0x42, 0x50, 0xf4, 0x91, 0xef, 0x98, 0x7a,
59
+ 0x33, 0x54, 0x0b, 0x43, 0xed, 0xcf, 0xac, 0x62,
60
+ 0xe4, 0xb3, 0x1c, 0xa9, 0xc9, 0x08, 0xe8, 0x95,
61
+ 0x80, 0xdf, 0x94, 0xfa, 0x75, 0x8f, 0x3f, 0xa6,
62
+ 0x47, 0x07, 0xa7, 0xfc, 0xf3, 0x73, 0x17, 0xba,
63
+ 0x83, 0x59, 0x3c, 0x19, 0xe6, 0x85, 0x4f, 0xa8,
64
+ 0x68, 0x6b, 0x81, 0xb2, 0x71, 0x64, 0xda, 0x8b,
65
+ 0xf8, 0xeb, 0x0f, 0x4b, 0x70, 0x56, 0x9d, 0x35,
66
+ 0x1e, 0x24, 0x0e, 0x5e, 0x63, 0x58, 0xd1, 0xa2,
67
+ 0x25, 0x22, 0x7c, 0x3b, 0x01, 0x21, 0x78, 0x87,
68
+ 0xd4, 0x00, 0x46, 0x57, 0x9f, 0xd3, 0x27, 0x52,
69
+ 0x4c, 0x36, 0x02, 0xe7, 0xa0, 0xc4, 0xc8, 0x9e,
70
+ 0xea, 0xbf, 0x8a, 0xd2, 0x40, 0xc7, 0x38, 0xb5,
71
+ 0xa3, 0xf7, 0xf2, 0xce, 0xf9, 0x61, 0x15, 0xa1,
72
+ 0xe0, 0xae, 0x5d, 0xa4, 0x9b, 0x34, 0x1a, 0x55,
73
+ 0xad, 0x93, 0x32, 0x30, 0xf5, 0x8c, 0xb1, 0xe3,
74
+ 0x1d, 0xf6, 0xe2, 0x2e, 0x82, 0x66, 0xca, 0x60,
75
+ 0xc0, 0x29, 0x23, 0xab, 0x0d, 0x53, 0x4e, 0x6f,
76
+ 0xd5, 0xdb, 0x37, 0x45, 0xde, 0xfd, 0x8e, 0x2f,
77
+ 0x03, 0xff, 0x6a, 0x72, 0x6d, 0x6c, 0x5b, 0x51,
78
+ 0x8d, 0x1b, 0xaf, 0x92, 0xbb, 0xdd, 0xbc, 0x7f,
79
+ 0x11, 0xd9, 0x5c, 0x41, 0x1f, 0x10, 0x5a, 0xd8,
80
+ 0x0a, 0xc1, 0x31, 0x88, 0xa5, 0xcd, 0x7b, 0xbd,
81
+ 0x2d, 0x74, 0xd0, 0x12, 0xb8, 0xe5, 0xb4, 0xb0,
82
+ 0x89, 0x69, 0x97, 0x4a, 0x0c, 0x96, 0x77, 0x7e,
83
+ 0x65, 0xb9, 0xf1, 0x09, 0xc5, 0x6e, 0xc6, 0x84,
84
+ 0x18, 0xf0, 0x7d, 0xec, 0x3a, 0xdc, 0x4d, 0x20,
85
+ 0x79, 0xee, 0x5f, 0x3e, 0xd7, 0xcb, 0x39, 0x48,
86
+};
87
+
88
+void HELPER(crypto_sm4e)(void *vd, void *vn)
89
+{
90
+ uint64_t *rd = vd;
91
+ uint64_t *rn = vn;
92
+ union CRYPTO_STATE d = { .l = { rd[0], rd[1] } };
93
+ union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
94
+ uint32_t t, i;
95
+
96
+ for (i = 0; i < 4; i++) {
97
+ t = CR_ST_WORD(d, (i + 1) % 4) ^
98
+ CR_ST_WORD(d, (i + 2) % 4) ^
99
+ CR_ST_WORD(d, (i + 3) % 4) ^
100
+ CR_ST_WORD(n, i);
101
+
102
+ t = sm4_sbox[t & 0xff] |
103
+ sm4_sbox[(t >> 8) & 0xff] << 8 |
104
+ sm4_sbox[(t >> 16) & 0xff] << 16 |
105
+ sm4_sbox[(t >> 24) & 0xff] << 24;
106
+
107
+ CR_ST_WORD(d, i) ^= t ^ rol32(t, 2) ^ rol32(t, 10) ^ rol32(t, 18) ^
108
+ rol32(t, 24);
109
+ }
110
+
111
+ rd[0] = d.l[0];
112
+ rd[1] = d.l[1];
113
+}
114
+
115
+void HELPER(crypto_sm4ekey)(void *vd, void *vn, void* vm)
116
+{
117
+ uint64_t *rd = vd;
118
+ uint64_t *rn = vn;
119
+ uint64_t *rm = vm;
120
+ union CRYPTO_STATE d;
121
+ union CRYPTO_STATE n = { .l = { rn[0], rn[1] } };
122
+ union CRYPTO_STATE m = { .l = { rm[0], rm[1] } };
123
+ uint32_t t, i;
124
+
125
+ d = n;
126
+ for (i = 0; i < 4; i++) {
127
+ t = CR_ST_WORD(d, (i + 1) % 4) ^
128
+ CR_ST_WORD(d, (i + 2) % 4) ^
129
+ CR_ST_WORD(d, (i + 3) % 4) ^
130
+ CR_ST_WORD(m, i);
131
+
132
+ t = sm4_sbox[t & 0xff] |
133
+ sm4_sbox[(t >> 8) & 0xff] << 8 |
134
+ sm4_sbox[(t >> 16) & 0xff] << 16 |
135
+ sm4_sbox[(t >> 24) & 0xff] << 24;
136
+
137
+ CR_ST_WORD(d, i) ^= t ^ rol32(t, 13) ^ rol32(t, 23);
138
+ }
139
+
140
+ rd[0] = d.l[0];
141
+ rd[1] = d.l[1];
142
+}
143
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
19
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
144
index XXXXXXX..XXXXXXX 100644
20
index XXXXXXX..XXXXXXX 100644
145
--- a/target/arm/translate-a64.c
21
--- a/target/arm/translate-a64.c
146
+++ b/target/arm/translate-a64.c
22
+++ b/target/arm/translate-a64.c
147
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_three_reg_sha512(DisasContext *s, uint32_t insn)
23
@@ -XXX,XX +XXX,XX @@ static void disas_fp_imm(DisasContext *s, uint32_t insn)
148
feature = ARM_FEATURE_V8_SM3;
24
{
149
genfn = gen_helper_crypto_sm3partw2;
25
int rd = extract32(insn, 0, 5);
150
break;
26
int imm8 = extract32(insn, 13, 8);
151
+ case 2: /* SM4EKEY */
27
- int is_double = extract32(insn, 22, 2);
152
+ feature = ARM_FEATURE_V8_SM4;
28
+ int type = extract32(insn, 22, 2);
153
+ genfn = gen_helper_crypto_sm4ekey;
29
uint64_t imm;
30
TCGv_i64 tcg_res;
31
+ TCGMemOp sz;
32
33
- if (is_double > 1) {
34
+ switch (type) {
35
+ case 0:
36
+ sz = MO_32;
37
+ break;
38
+ case 1:
39
+ sz = MO_64;
40
+ break;
41
+ case 3:
42
+ sz = MO_16;
43
+ if (arm_dc_feature(s, ARM_FEATURE_V8_FP16)) {
154
+ break;
44
+ break;
155
default:
45
+ }
156
unallocated_encoding(s);
46
+ /* fallthru */
157
return;
47
+ default:
158
@@ -XXX,XX +XXX,XX @@ static void disas_crypto_two_reg_sha512(DisasContext *s, uint32_t insn)
159
feature = ARM_FEATURE_V8_SHA512;
160
genfn = gen_helper_crypto_sha512su0;
161
break;
162
+ case 1: /* SM4E */
163
+ feature = ARM_FEATURE_V8_SM4;
164
+ genfn = gen_helper_crypto_sm4e;
165
+ break;
166
default:
167
unallocated_encoding(s);
48
unallocated_encoding(s);
168
return;
49
return;
50
}
51
@@ -XXX,XX +XXX,XX @@ static void disas_fp_imm(DisasContext *s, uint32_t insn)
52
return;
53
}
54
55
- imm = vfp_expand_imm(MO_32 + is_double, imm8);
56
+ imm = vfp_expand_imm(sz, imm8);
57
58
tcg_res = tcg_const_i64(imm);
59
write_fp_dreg(s, rd, tcg_res);
169
--
60
--
170
2.16.1
61
2.17.0
171
62
172
63
diff view generated by jsdifflib
1
From: Christoffer Dall <christoffer.dall@linaro.org>
1
From: Alex Bennée <alex.bennee@linaro.org>
2
2
3
KVM doesn't support emulating a GICv3 in userspace, only GICv2. We
3
We are meant to explicitly pass fpst, not cpu_env.
4
currently attempt this anyway, and as a result a KVM guest doesn't
5
receive interrupts and the user is left wondering why. Report an error
6
to the user if this particular combination is requested.
7
4
8
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
5
Cc: qemu-stable@nongnu.org
9
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
6
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
10
Message-id: 20180201205307.30343-1-christoffer.dall@linaro.org
7
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
8
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
9
Tested-by: Alex Bennée <alex.bennee@linaro.org>
10
Message-id: 20180512003217.9105-12-richard.henderson@linaro.org
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
---
12
---
13
target/arm/kvm_arm.h | 4 ++++
13
target/arm/translate-a64.c | 3 ++-
14
1 file changed, 4 insertions(+)
14
1 file changed, 2 insertions(+), 1 deletion(-)
15
15
16
diff --git a/target/arm/kvm_arm.h b/target/arm/kvm_arm.h
16
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
17
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
18
--- a/target/arm/kvm_arm.h
18
--- a/target/arm/translate-a64.c
19
+++ b/target/arm/kvm_arm.h
19
+++ b/target/arm/translate-a64.c
20
@@ -XXX,XX +XXX,XX @@ static inline const char *gicv3_class_name(void)
20
@@ -XXX,XX +XXX,XX @@ static void handle_fp_1src_half(DisasContext *s, int opcode, int rd, int rn)
21
exit(1);
21
tcg_gen_xori_i32(tcg_res, tcg_op, 0x8000);
22
#endif
22
break;
23
} else {
23
case 0x3: /* FSQRT */
24
+ if (kvm_enabled()) {
24
- gen_helper_sqrt_f16(tcg_res, tcg_op, cpu_env);
25
+ error_report("Userspace GICv3 is not supported with KVM");
25
+ fpst = get_fpstatus_ptr(true);
26
+ exit(1);
26
+ gen_helper_sqrt_f16(tcg_res, tcg_op, fpst);
27
+ }
27
break;
28
return "arm-gicv3";
28
case 0x8: /* FRINTN */
29
}
29
case 0x9: /* FRINTP */
30
}
31
--
30
--
32
2.16.1
31
2.17.0
33
32
34
33
diff view generated by jsdifflib
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
1
From: Philippe Mathieu-Daudé <f4bug@amsat.org>
2
2
3
Add minimal code needed to allow upstream Linux guest to boot.
3
Per the Physical Layer Simplified Spec. "4.3.10.4 Switch Function Status":
4
4
5
Cc: Peter Maydell <peter.maydell@linaro.org>
5
The block length is predefined to 512 bits
6
Cc: Jason Wang <jasowang@redhat.com>
6
7
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
7
and "4.10.2 SD Status":
8
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
8
9
Cc: Michael S. Tsirkin <mst@redhat.com>
9
The SD Status contains status bits that are related to the SD Memory Card
10
Cc: qemu-devel@nongnu.org
10
proprietary features and may be used for future application-specific usage.
11
Cc: qemu-arm@nongnu.org
11
The size of the SD Status is one data block of 512 bit. The content of this
12
Cc: yurovsky@gmail.com
12
register is transmitted to the Host over the DAT bus along with a 16-bit CRC.
13
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
13
14
Thus the 16-bit CRC goes at offset 64.
15
16
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
17
Message-id: 20180509060104.4458-3-f4bug@amsat.org
14
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
18
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
19
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
20
---
18
include/hw/timer/imx_gpt.h | 1 +
21
hw/sd/sd.c | 2 +-
19
hw/timer/imx_gpt.c | 25 +++++++++++++++++++++++++
22
1 file changed, 1 insertion(+), 1 deletion(-)
20
2 files changed, 26 insertions(+)
21
23
22
diff --git a/include/hw/timer/imx_gpt.h b/include/hw/timer/imx_gpt.h
24
diff --git a/hw/sd/sd.c b/hw/sd/sd.c
23
index XXXXXXX..XXXXXXX 100644
25
index XXXXXXX..XXXXXXX 100644
24
--- a/include/hw/timer/imx_gpt.h
26
--- a/hw/sd/sd.c
25
+++ b/include/hw/timer/imx_gpt.h
27
+++ b/hw/sd/sd.c
26
@@ -XXX,XX +XXX,XX @@
28
@@ -XXX,XX +XXX,XX @@ static void sd_function_switch(SDState *sd, uint32_t arg)
27
#define TYPE_IMX25_GPT "imx25.gpt"
29
sd->data[14 + (i >> 1)] = new_func << ((i * 4) & 4);
28
#define TYPE_IMX31_GPT "imx31.gpt"
30
}
29
#define TYPE_IMX6_GPT "imx6.gpt"
31
memset(&sd->data[17], 0, 47);
30
+#define TYPE_IMX7_GPT "imx7.gpt"
32
- stw_be_p(sd->data + 65, sd_crc16(sd->data, 64));
31
33
+ stw_be_p(sd->data + 64, sd_crc16(sd->data, 64));
32
#define TYPE_IMX_GPT TYPE_IMX25_GPT
33
34
diff --git a/hw/timer/imx_gpt.c b/hw/timer/imx_gpt.c
35
index XXXXXXX..XXXXXXX 100644
36
--- a/hw/timer/imx_gpt.c
37
+++ b/hw/timer/imx_gpt.c
38
@@ -XXX,XX +XXX,XX @@ static const IMXClk imx6_gpt_clocks[] = {
39
CLK_HIGH, /* 111 reference clock */
40
};
41
42
+static const IMXClk imx7_gpt_clocks[] = {
43
+ CLK_NONE, /* 000 No clock source */
44
+ CLK_IPG, /* 001 ipg_clk, 532MHz*/
45
+ CLK_IPG_HIGH, /* 010 ipg_clk_highfreq */
46
+ CLK_EXT, /* 011 External clock */
47
+ CLK_32k, /* 100 ipg_clk_32k */
48
+ CLK_HIGH, /* 101 reference clock */
49
+ CLK_NONE, /* 110 not defined */
50
+ CLK_NONE, /* 111 not defined */
51
+};
52
+
53
static void imx_gpt_set_freq(IMXGPTState *s)
54
{
55
uint32_t clksrc = extract32(s->cr, GPT_CR_CLKSRC_SHIFT, 3);
56
@@ -XXX,XX +XXX,XX @@ static void imx6_gpt_init(Object *obj)
57
s->clocks = imx6_gpt_clocks;
58
}
34
}
59
35
60
+static void imx7_gpt_init(Object *obj)
36
static inline bool sd_wp_addr(SDState *sd, uint64_t addr)
61
+{
62
+ IMXGPTState *s = IMX_GPT(obj);
63
+
64
+ s->clocks = imx7_gpt_clocks;
65
+}
66
+
67
static const TypeInfo imx25_gpt_info = {
68
.name = TYPE_IMX25_GPT,
69
.parent = TYPE_SYS_BUS_DEVICE,
70
@@ -XXX,XX +XXX,XX @@ static const TypeInfo imx6_gpt_info = {
71
.instance_init = imx6_gpt_init,
72
};
73
74
+static const TypeInfo imx7_gpt_info = {
75
+ .name = TYPE_IMX7_GPT,
76
+ .parent = TYPE_IMX25_GPT,
77
+ .instance_init = imx7_gpt_init,
78
+};
79
+
80
static void imx_gpt_register_types(void)
81
{
82
type_register_static(&imx25_gpt_info);
83
type_register_static(&imx31_gpt_info);
84
type_register_static(&imx6_gpt_info);
85
+ type_register_static(&imx7_gpt_info);
86
}
87
88
type_init(imx_gpt_register_types)
89
--
37
--
90
2.16.1
38
2.17.0
91
39
92
40
diff view generated by jsdifflib
1
Make the load of the exception vector from the vector table honour
1
Usually the logging of the CPU state produced by -d cpu is sufficient
2
the SAU and any bus error on the load (possibly provoking a derived
2
to diagnose problems, but sometimes you want to see the state of
3
exception), rather than simply aborting if the load fails.
3
the floating point registers as well. We don't want to enable that
4
by default as it adds a lot of extra data to the log; instead,
5
allow it to be optionally enabled via -d fpu.
4
6
5
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
6
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
8
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
7
Message-id: 1517324542-6607-7-git-send-email-peter.maydell@linaro.org
9
Message-id: 20180510130024.31678-1-peter.maydell@linaro.org
8
---
10
---
9
target/arm/helper.c | 71 +++++++++++++++++++++++++++++++++++++++++------------
11
include/qemu/log.h | 1 +
10
1 file changed, 55 insertions(+), 16 deletions(-)
12
accel/tcg/cpu-exec.c | 9 ++++++---
13
util/log.c | 2 ++
14
3 files changed, 9 insertions(+), 3 deletions(-)
11
15
12
diff --git a/target/arm/helper.c b/target/arm/helper.c
16
diff --git a/include/qemu/log.h b/include/qemu/log.h
13
index XXXXXXX..XXXXXXX 100644
17
index XXXXXXX..XXXXXXX 100644
14
--- a/target/arm/helper.c
18
--- a/include/qemu/log.h
15
+++ b/target/arm/helper.c
19
+++ b/include/qemu/log.h
16
@@ -XXX,XX +XXX,XX @@ static uint32_t *get_v7m_sp_ptr(CPUARMState *env, bool secure, bool threadmode,
20
@@ -XXX,XX +XXX,XX @@ static inline bool qemu_log_separate(void)
21
#define CPU_LOG_PAGE (1 << 14)
22
/* LOG_TRACE (1 << 15) is defined in log-for-trace.h */
23
#define CPU_LOG_TB_OP_IND (1 << 16)
24
+#define CPU_LOG_TB_FPU (1 << 17)
25
26
/* Lock output for a series of related logs. Since this is not needed
27
* for a single qemu_log / qemu_log_mask / qemu_log_mask_and_addr, we
28
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
29
index XXXXXXX..XXXXXXX 100644
30
--- a/accel/tcg/cpu-exec.c
31
+++ b/accel/tcg/cpu-exec.c
32
@@ -XXX,XX +XXX,XX @@ static inline tcg_target_ulong cpu_tb_exec(CPUState *cpu, TranslationBlock *itb)
33
if (qemu_loglevel_mask(CPU_LOG_TB_CPU)
34
&& qemu_log_in_addr_range(itb->pc)) {
35
qemu_log_lock();
36
+ int flags = 0;
37
+ if (qemu_loglevel_mask(CPU_LOG_TB_FPU)) {
38
+ flags |= CPU_DUMP_FPU;
39
+ }
40
#if defined(TARGET_I386)
41
- log_cpu_state(cpu, CPU_DUMP_CCOP);
42
-#else
43
- log_cpu_state(cpu, 0);
44
+ flags |= CPU_DUMP_CCOP;
45
#endif
46
+ log_cpu_state(cpu, flags);
47
qemu_log_unlock();
17
}
48
}
18
}
49
#endif /* DEBUG_DISAS */
19
50
diff --git a/util/log.c b/util/log.c
20
-static uint32_t arm_v7m_load_vector(ARMCPU *cpu, int exc, bool targets_secure)
51
index XXXXXXX..XXXXXXX 100644
21
+static bool arm_v7m_load_vector(ARMCPU *cpu, int exc, bool targets_secure,
52
--- a/util/log.c
22
+ uint32_t *pvec)
53
+++ b/util/log.c
23
{
54
@@ -XXX,XX +XXX,XX @@ const QEMULogItem qemu_log_items[] = {
24
CPUState *cs = CPU(cpu);
55
"show trace before each executed TB (lots of logs)" },
25
CPUARMState *env = &cpu->env;
56
{ CPU_LOG_TB_CPU, "cpu",
26
MemTxResult result;
57
"show CPU registers before entering a TB (lots of logs)" },
27
- hwaddr vec = env->v7m.vecbase[targets_secure] + exc * 4;
58
+ { CPU_LOG_TB_FPU, "fpu",
28
- uint32_t addr;
59
+ "include FPU registers in the 'cpu' logging" },
29
+ uint32_t addr = env->v7m.vecbase[targets_secure] + exc * 4;
60
{ CPU_LOG_MMU, "mmu",
30
+ uint32_t vector_entry;
61
"log MMU-related activities" },
31
+ MemTxAttrs attrs = {};
62
{ CPU_LOG_PCALL, "pcall",
32
+ ARMMMUIdx mmu_idx;
33
+ bool exc_secure;
34
+
35
+ mmu_idx = arm_v7m_mmu_idx_for_secstate_and_priv(env, targets_secure, true);
36
37
- addr = address_space_ldl(cs->as, vec,
38
- MEMTXATTRS_UNSPECIFIED, &result);
39
+ /* We don't do a get_phys_addr() here because the rules for vector
40
+ * loads are special: they always use the default memory map, and
41
+ * the default memory map permits reads from all addresses.
42
+ * Since there's no easy way to pass through to pmsav8_mpu_lookup()
43
+ * that we want this special case which would always say "yes",
44
+ * we just do the SAU lookup here followed by a direct physical load.
45
+ */
46
+ attrs.secure = targets_secure;
47
+ attrs.user = false;
48
+
49
+ if (arm_feature(env, ARM_FEATURE_M_SECURITY)) {
50
+ V8M_SAttributes sattrs = {};
51
+
52
+ v8m_security_lookup(env, addr, MMU_DATA_LOAD, mmu_idx, &sattrs);
53
+ if (sattrs.ns) {
54
+ attrs.secure = false;
55
+ } else if (!targets_secure) {
56
+ /* NS access to S memory */
57
+ goto load_fail;
58
+ }
59
+ }
60
+
61
+ vector_entry = address_space_ldl(arm_addressspace(cs, attrs), addr,
62
+ attrs, &result);
63
if (result != MEMTX_OK) {
64
- /* Architecturally this should cause a HardFault setting HSFR.VECTTBL,
65
- * which would then be immediately followed by our failing to load
66
- * the entry vector for that HardFault, which is a Lockup case.
67
- * Since we don't model Lockup, we just report this guest error
68
- * via cpu_abort().
69
- */
70
- cpu_abort(cs, "Failed to read from %s exception vector table "
71
- "entry %08x\n", targets_secure ? "secure" : "nonsecure",
72
- (unsigned)vec);
73
+ goto load_fail;
74
}
75
- return addr;
76
+ *pvec = vector_entry;
77
+ return true;
78
+
79
+load_fail:
80
+ /* All vector table fetch fails are reported as HardFault, with
81
+ * HFSR.VECTTBL and .FORCED set. (FORCED is set because
82
+ * technically the underlying exception is a MemManage or BusFault
83
+ * that is escalated to HardFault.) This is a terminal exception,
84
+ * so we will either take the HardFault immediately or else enter
85
+ * lockup (the latter case is handled in armv7m_nvic_set_pending_derived()).
86
+ */
87
+ exc_secure = targets_secure ||
88
+ !(cpu->env.v7m.aircr & R_V7M_AIRCR_BFHFNMINS_MASK);
89
+ env->v7m.hfsr |= R_V7M_HFSR_VECTTBL_MASK | R_V7M_HFSR_FORCED_MASK;
90
+ armv7m_nvic_set_pending_derived(env->nvic, ARMV7M_EXCP_HARD, exc_secure);
91
+ return false;
92
}
93
94
static bool v7m_push_callee_stack(ARMCPU *cpu, uint32_t lr, bool dotailchain,
95
@@ -XXX,XX +XXX,XX @@ static void v7m_exception_taken(ARMCPU *cpu, uint32_t lr, bool dotailchain,
96
return;
97
}
98
99
- addr = arm_v7m_load_vector(cpu, exc, targets_secure);
100
+ if (!arm_v7m_load_vector(cpu, exc, targets_secure, &addr)) {
101
+ /* Vector load failed: derived exception */
102
+ v7m_exception_taken(cpu, lr, true, true);
103
+ return;
104
+ }
105
106
/* Now we've done everything that might cause a derived exception
107
* we can go ahead and activate whichever exception we're going to
108
--
63
--
109
2.16.1
64
2.17.0
110
65
111
66
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
IP block found on several generations of i.MX family does not use
4
vanilla SDHCI implementation and it comes with a number of quirks.
5
6
Introduce i.MX SDHCI subtype of SDHCI block to add code necessary to
7
support unmodified Linux guest driver.
8
9
Cc: Peter Maydell <peter.maydell@linaro.org>
10
Cc: Jason Wang <jasowang@redhat.com>
11
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
12
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
13
Cc: Michael S. Tsirkin <mst@redhat.com>
14
Cc: qemu-devel@nongnu.org
15
Cc: qemu-arm@nongnu.org
16
Cc: yurovsky@gmail.com
17
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
18
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
19
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
20
[PMM: define and use ESDHC_UNDOCUMENTED_REG27]
21
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
22
---
23
hw/sd/sdhci-internal.h | 23 +++++
24
include/hw/sd/sdhci.h | 13 +++
25
hw/sd/sdhci.c | 230 ++++++++++++++++++++++++++++++++++++++++++++++++-
26
3 files changed, 265 insertions(+), 1 deletion(-)
27
28
diff --git a/hw/sd/sdhci-internal.h b/hw/sd/sdhci-internal.h
29
index XXXXXXX..XXXXXXX 100644
30
--- a/hw/sd/sdhci-internal.h
31
+++ b/hw/sd/sdhci-internal.h
32
@@ -XXX,XX +XXX,XX @@
33
34
/* R/W Host control Register 0x0 */
35
#define SDHC_HOSTCTL 0x28
36
+#define SDHC_CTRL_LED 0x01
37
#define SDHC_CTRL_DMA_CHECK_MASK 0x18
38
#define SDHC_CTRL_SDMA 0x00
39
#define SDHC_CTRL_ADMA1_32 0x08
40
#define SDHC_CTRL_ADMA2_32 0x10
41
#define SDHC_CTRL_ADMA2_64 0x18
42
#define SDHC_DMA_TYPE(x) ((x) & SDHC_CTRL_DMA_CHECK_MASK)
43
+#define SDHC_CTRL_4BITBUS 0x02
44
+#define SDHC_CTRL_8BITBUS 0x20
45
+#define SDHC_CTRL_CDTEST_INS 0x40
46
+#define SDHC_CTRL_CDTEST_EN 0x80
47
+
48
49
/* R/W Power Control Register 0x0 */
50
#define SDHC_PWRCON 0x29
51
@@ -XXX,XX +XXX,XX @@ enum {
52
sdhc_gap_write = 2 /* SDHC stopped at block gap during write operation */
53
};
54
55
+extern const VMStateDescription sdhci_vmstate;
56
+
57
+
58
+#define ESDHC_MIX_CTRL 0x48
59
+#define ESDHC_VENDOR_SPEC 0xc0
60
+#define ESDHC_DLL_CTRL 0x60
61
+
62
+#define ESDHC_TUNING_CTRL 0xcc
63
+#define ESDHC_TUNE_CTRL_STATUS 0x68
64
+#define ESDHC_WTMK_LVL 0x44
65
+
66
+/* Undocumented register used by guests working around erratum ERR004536 */
67
+#define ESDHC_UNDOCUMENTED_REG27 0x6c
68
+
69
+#define ESDHC_CTRL_4BITBUS (0x1 << 1)
70
+#define ESDHC_CTRL_8BITBUS (0x2 << 1)
71
+
72
#endif
73
diff --git a/include/hw/sd/sdhci.h b/include/hw/sd/sdhci.h
74
index XXXXXXX..XXXXXXX 100644
75
--- a/include/hw/sd/sdhci.h
76
+++ b/include/hw/sd/sdhci.h
77
@@ -XXX,XX +XXX,XX @@ typedef struct SDHCIState {
78
AddressSpace sysbus_dma_as;
79
AddressSpace *dma_as;
80
MemoryRegion *dma_mr;
81
+ const MemoryRegionOps *io_ops;
82
83
QEMUTimer *insert_timer; /* timer for 'changing' sd card. */
84
QEMUTimer *transfer_timer;
85
@@ -XXX,XX +XXX,XX @@ typedef struct SDHCIState {
86
87
/* Configurable properties */
88
bool pending_insert_quirk; /* Quirk for Raspberry Pi card insert int */
89
+ uint32_t quirks;
90
} SDHCIState;
91
92
+/*
93
+ * Controller does not provide transfer-complete interrupt when not
94
+ * busy.
95
+ *
96
+ * NOTE: This definition is taken out of Linux kernel and so the
97
+ * original bit number is preserved
98
+ */
99
+#define SDHCI_QUIRK_NO_BUSY_IRQ BIT(14)
100
+
101
#define TYPE_PCI_SDHCI "sdhci-pci"
102
#define PCI_SDHCI(obj) OBJECT_CHECK(SDHCIState, (obj), TYPE_PCI_SDHCI)
103
104
@@ -XXX,XX +XXX,XX @@ typedef struct SDHCIState {
105
#define SYSBUS_SDHCI(obj) \
106
OBJECT_CHECK(SDHCIState, (obj), TYPE_SYSBUS_SDHCI)
107
108
+#define TYPE_IMX_USDHC "imx-usdhc"
109
+
110
#endif /* SDHCI_H */
111
diff --git a/hw/sd/sdhci.c b/hw/sd/sdhci.c
112
index XXXXXXX..XXXXXXX 100644
113
--- a/hw/sd/sdhci.c
114
+++ b/hw/sd/sdhci.c
115
@@ -XXX,XX +XXX,XX @@ static void sdhci_send_command(SDHCIState *s)
116
}
117
}
118
119
- if ((s->norintstsen & SDHC_NISEN_TRSCMP) &&
120
+ if (!(s->quirks & SDHCI_QUIRK_NO_BUSY_IRQ) &&
121
+ (s->norintstsen & SDHC_NISEN_TRSCMP) &&
122
(s->cmdreg & SDHC_CMD_RESPONSE) == SDHC_CMD_RSP_WITH_BUSY) {
123
s->norintsts |= SDHC_NIS_TRSCMP;
124
}
125
@@ -XXX,XX +XXX,XX @@ static void sdhci_initfn(SDHCIState *s)
126
127
s->insert_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, sdhci_raise_insertion_irq, s);
128
s->transfer_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, sdhci_data_transfer, s);
129
+
130
+ s->io_ops = &sdhci_mmio_ops;
131
}
132
133
static void sdhci_uninitfn(SDHCIState *s)
134
@@ -XXX,XX +XXX,XX @@ static void sdhci_sysbus_realize(DeviceState *dev, Error ** errp)
135
}
136
137
sysbus_init_irq(sbd, &s->irq);
138
+
139
+ memory_region_init_io(&s->iomem, OBJECT(s), s->io_ops, s, "sdhci",
140
+ SDHC_REGISTERS_MAP_SIZE);
141
+
142
sysbus_init_mmio(sbd, &s->iomem);
143
}
144
145
@@ -XXX,XX +XXX,XX @@ static const TypeInfo sdhci_bus_info = {
146
.class_init = sdhci_bus_class_init,
147
};
148
149
+static uint64_t usdhc_read(void *opaque, hwaddr offset, unsigned size)
150
+{
151
+ SDHCIState *s = SYSBUS_SDHCI(opaque);
152
+ uint32_t ret;
153
+ uint16_t hostctl;
154
+
155
+ switch (offset) {
156
+ default:
157
+ return sdhci_read(opaque, offset, size);
158
+
159
+ case SDHC_HOSTCTL:
160
+ /*
161
+ * For a detailed explanation on the following bit
162
+ * manipulation code see comments in a similar part of
163
+ * usdhc_write()
164
+ */
165
+ hostctl = SDHC_DMA_TYPE(s->hostctl) << (8 - 3);
166
+
167
+ if (s->hostctl & SDHC_CTRL_8BITBUS) {
168
+ hostctl |= ESDHC_CTRL_8BITBUS;
169
+ }
170
+
171
+ if (s->hostctl & SDHC_CTRL_4BITBUS) {
172
+ hostctl |= ESDHC_CTRL_4BITBUS;
173
+ }
174
+
175
+ ret = hostctl;
176
+ ret |= (uint32_t)s->blkgap << 16;
177
+ ret |= (uint32_t)s->wakcon << 24;
178
+
179
+ break;
180
+
181
+ case ESDHC_DLL_CTRL:
182
+ case ESDHC_TUNE_CTRL_STATUS:
183
+ case ESDHC_UNDOCUMENTED_REG27:
184
+ case ESDHC_TUNING_CTRL:
185
+ case ESDHC_VENDOR_SPEC:
186
+ case ESDHC_MIX_CTRL:
187
+ case ESDHC_WTMK_LVL:
188
+ ret = 0;
189
+ break;
190
+ }
191
+
192
+ return ret;
193
+}
194
+
195
+static void
196
+usdhc_write(void *opaque, hwaddr offset, uint64_t val, unsigned size)
197
+{
198
+ SDHCIState *s = SYSBUS_SDHCI(opaque);
199
+ uint8_t hostctl;
200
+ uint32_t value = (uint32_t)val;
201
+
202
+ switch (offset) {
203
+ case ESDHC_DLL_CTRL:
204
+ case ESDHC_TUNE_CTRL_STATUS:
205
+ case ESDHC_UNDOCUMENTED_REG27:
206
+ case ESDHC_TUNING_CTRL:
207
+ case ESDHC_WTMK_LVL:
208
+ case ESDHC_VENDOR_SPEC:
209
+ break;
210
+
211
+ case SDHC_HOSTCTL:
212
+ /*
213
+ * Here's What ESDHCI has at offset 0x28 (SDHC_HOSTCTL)
214
+ *
215
+ * 7 6 5 4 3 2 1 0
216
+ * |-----------+--------+--------+-----------+----------+---------|
217
+ * | Card | Card | Endian | DATA3 | Data | Led |
218
+ * | Detect | Detect | Mode | as Card | Transfer | Control |
219
+ * | Signal | Test | | Detection | Width | |
220
+ * | Selection | Level | | Pin | | |
221
+ * |-----------+--------+--------+-----------+----------+---------|
222
+ *
223
+ * and 0x29
224
+ *
225
+ * 15 10 9 8
226
+ * |----------+------|
227
+ * | Reserved | DMA |
228
+ * | | Sel. |
229
+ * | | |
230
+ * |----------+------|
231
+ *
232
+ * and here's what SDCHI spec expects those offsets to be:
233
+ *
234
+ * 0x28 (Host Control Register)
235
+ *
236
+ * 7 6 5 4 3 2 1 0
237
+ * |--------+--------+----------+------+--------+----------+---------|
238
+ * | Card | Card | Extended | DMA | High | Data | LED |
239
+ * | Detect | Detect | Data | Sel. | Speed | Transfer | Control |
240
+ * | Signal | Test | Transfer | | Enable | Width | |
241
+ * | Sel. | Level | Width | | | | |
242
+ * |--------+--------+----------+------+--------+----------+---------|
243
+ *
244
+ * and 0x29 (Power Control Register)
245
+ *
246
+ * |----------------------------------|
247
+ * | Power Control Register |
248
+ * | |
249
+ * | Description omitted, |
250
+ * | since it has no analog in ESDHCI |
251
+ * | |
252
+ * |----------------------------------|
253
+ *
254
+ * Since offsets 0x2A and 0x2B should be compatible between
255
+ * both IP specs we only need to reconcile least 16-bit of the
256
+ * word we've been given.
257
+ */
258
+
259
+ /*
260
+ * First, save bits 7 6 and 0 since they are identical
261
+ */
262
+ hostctl = value & (SDHC_CTRL_LED |
263
+ SDHC_CTRL_CDTEST_INS |
264
+ SDHC_CTRL_CDTEST_EN);
265
+ /*
266
+ * Second, split "Data Transfer Width" from bits 2 and 1 in to
267
+ * bits 5 and 1
268
+ */
269
+ if (value & ESDHC_CTRL_8BITBUS) {
270
+ hostctl |= SDHC_CTRL_8BITBUS;
271
+ }
272
+
273
+ if (value & ESDHC_CTRL_4BITBUS) {
274
+ hostctl |= ESDHC_CTRL_4BITBUS;
275
+ }
276
+
277
+ /*
278
+ * Third, move DMA select from bits 9 and 8 to bits 4 and 3
279
+ */
280
+ hostctl |= SDHC_DMA_TYPE(value >> (8 - 3));
281
+
282
+ /*
283
+ * Now place the corrected value into low 16-bit of the value
284
+ * we are going to give standard SDHCI write function
285
+ *
286
+ * NOTE: This transformation should be the inverse of what can
287
+ * be found in drivers/mmc/host/sdhci-esdhc-imx.c in Linux
288
+ * kernel
289
+ */
290
+ value &= ~UINT16_MAX;
291
+ value |= hostctl;
292
+ value |= (uint16_t)s->pwrcon << 8;
293
+
294
+ sdhci_write(opaque, offset, value, size);
295
+ break;
296
+
297
+ case ESDHC_MIX_CTRL:
298
+ /*
299
+ * So, when SD/MMC stack in Linux tries to write to "Transfer
300
+ * Mode Register", ESDHC i.MX quirk code will translate it
301
+ * into a write to ESDHC_MIX_CTRL, so we do the opposite in
302
+ * order to get where we started
303
+ *
304
+ * Note that Auto CMD23 Enable bit is located in a wrong place
305
+ * on i.MX, but since it is not used by QEMU we do not care.
306
+ *
307
+ * We don't want to call sdhci_write(.., SDHC_TRNMOD, ...)
308
+ * here becuase it will result in a call to
309
+ * sdhci_send_command(s) which we don't want.
310
+ *
311
+ */
312
+ s->trnmod = value & UINT16_MAX;
313
+ break;
314
+ case SDHC_TRNMOD:
315
+ /*
316
+ * Similar to above, but this time a write to "Command
317
+ * Register" will be translated into a 4-byte write to
318
+ * "Transfer Mode register" where lower 16-bit of value would
319
+ * be set to zero. So what we do is fill those bits with
320
+ * cached value from s->trnmod and let the SDHCI
321
+ * infrastructure handle the rest
322
+ */
323
+ sdhci_write(opaque, offset, val | s->trnmod, size);
324
+ break;
325
+ case SDHC_BLKSIZE:
326
+ /*
327
+ * ESDHCI does not implement "Host SDMA Buffer Boundary", and
328
+ * Linux driver will try to zero this field out which will
329
+ * break the rest of SDHCI emulation.
330
+ *
331
+ * Linux defaults to maximum possible setting (512K boundary)
332
+ * and it seems to be the only option that i.MX IP implements,
333
+ * so we artificially set it to that value.
334
+ */
335
+ val |= 0x7 << 12;
336
+ /* FALLTHROUGH */
337
+ default:
338
+ sdhci_write(opaque, offset, val, size);
339
+ break;
340
+ }
341
+}
342
+
343
+
344
+static const MemoryRegionOps usdhc_mmio_ops = {
345
+ .read = usdhc_read,
346
+ .write = usdhc_write,
347
+ .valid = {
348
+ .min_access_size = 1,
349
+ .max_access_size = 4,
350
+ .unaligned = false
351
+ },
352
+ .endianness = DEVICE_LITTLE_ENDIAN,
353
+};
354
+
355
+static void imx_usdhc_init(Object *obj)
356
+{
357
+ SDHCIState *s = SYSBUS_SDHCI(obj);
358
+
359
+ s->io_ops = &usdhc_mmio_ops;
360
+ s->quirks = SDHCI_QUIRK_NO_BUSY_IRQ;
361
+}
362
+
363
+static const TypeInfo imx_usdhc_info = {
364
+ .name = TYPE_IMX_USDHC,
365
+ .parent = TYPE_SYSBUS_SDHCI,
366
+ .instance_init = imx_usdhc_init,
367
+};
368
+
369
static void sdhci_register_types(void)
370
{
371
type_register_static(&sdhci_pci_info);
372
type_register_static(&sdhci_sysbus_info);
373
type_register_static(&sdhci_bus_info);
374
+ type_register_static(&imx_usdhc_info);
375
}
376
377
type_init(sdhci_register_types)
378
--
379
2.16.1
380
381
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Convert i.MX6 to use TYPE_IMX_USDHC since that's what real HW comes
4
with.
5
6
Cc: Peter Maydell <peter.maydell@linaro.org>
7
Cc: Jason Wang <jasowang@redhat.com>
8
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
9
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
10
Cc: Michael S. Tsirkin <mst@redhat.com>
11
Cc: qemu-devel@nongnu.org
12
Cc: qemu-arm@nongnu.org
13
Cc: yurovsky@gmail.com
14
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
15
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
18
hw/arm/fsl-imx6.c | 2 +-
19
1 file changed, 1 insertion(+), 1 deletion(-)
20
21
diff --git a/hw/arm/fsl-imx6.c b/hw/arm/fsl-imx6.c
22
index XXXXXXX..XXXXXXX 100644
23
--- a/hw/arm/fsl-imx6.c
24
+++ b/hw/arm/fsl-imx6.c
25
@@ -XXX,XX +XXX,XX @@ static void fsl_imx6_init(Object *obj)
26
}
27
28
for (i = 0; i < FSL_IMX6_NUM_ESDHCS; i++) {
29
- object_initialize(&s->esdhc[i], sizeof(s->esdhc[i]), TYPE_SYSBUS_SDHCI);
30
+ object_initialize(&s->esdhc[i], sizeof(s->esdhc[i]), TYPE_IMX_USDHC);
31
qdev_set_parent_bus(DEVICE(&s->esdhc[i]), sysbus_get_default());
32
snprintf(name, NAME_SIZE, "sdhc%d", i + 1);
33
object_property_add_child(obj, name, OBJECT(&s->esdhc[i]), NULL);
34
--
35
2.16.1
36
37
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Add minimal code needed to allow upstream Linux guest to boot.
4
5
Cc: Peter Maydell <peter.maydell@linaro.org>
6
Cc: Jason Wang <jasowang@redhat.com>
7
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
9
Cc: Michael S. Tsirkin <mst@redhat.com>
10
Cc: qemu-devel@nongnu.org
11
Cc: qemu-arm@nongnu.org
12
Cc: yurovsky@gmail.com
13
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
14
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
15
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
---
17
hw/misc/Makefile.objs | 1 +
18
include/hw/misc/imx7_ccm.h | 139 +++++++++++++++++++++++
19
hw/misc/imx7_ccm.c | 277 +++++++++++++++++++++++++++++++++++++++++++++
20
3 files changed, 417 insertions(+)
21
create mode 100644 include/hw/misc/imx7_ccm.h
22
create mode 100644 hw/misc/imx7_ccm.c
23
24
diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs
25
index XXXXXXX..XXXXXXX 100644
26
--- a/hw/misc/Makefile.objs
27
+++ b/hw/misc/Makefile.objs
28
@@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_IMX) += imx31_ccm.o
29
obj-$(CONFIG_IMX) += imx25_ccm.o
30
obj-$(CONFIG_IMX) += imx6_ccm.o
31
obj-$(CONFIG_IMX) += imx6_src.o
32
+obj-$(CONFIG_IMX) += imx7_ccm.o
33
obj-$(CONFIG_MILKYMIST) += milkymist-hpdmc.o
34
obj-$(CONFIG_MILKYMIST) += milkymist-pfpu.o
35
obj-$(CONFIG_MAINSTONE) += mst_fpga.o
36
diff --git a/include/hw/misc/imx7_ccm.h b/include/hw/misc/imx7_ccm.h
37
new file mode 100644
38
index XXXXXXX..XXXXXXX
39
--- /dev/null
40
+++ b/include/hw/misc/imx7_ccm.h
41
@@ -XXX,XX +XXX,XX @@
42
+/*
43
+ * Copyright (c) 2017, Impinj, Inc.
44
+ *
45
+ * i.MX7 CCM, PMU and ANALOG IP blocks emulation code
46
+ *
47
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
48
+ *
49
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
50
+ * See the COPYING file in the top-level directory.
51
+ */
52
+
53
+#ifndef IMX7_CCM_H
54
+#define IMX7_CCM_H
55
+
56
+#include "hw/misc/imx_ccm.h"
57
+#include "qemu/bitops.h"
58
+
59
+enum IMX7AnalogRegisters {
60
+ ANALOG_PLL_ARM,
61
+ ANALOG_PLL_ARM_SET,
62
+ ANALOG_PLL_ARM_CLR,
63
+ ANALOG_PLL_ARM_TOG,
64
+ ANALOG_PLL_DDR,
65
+ ANALOG_PLL_DDR_SET,
66
+ ANALOG_PLL_DDR_CLR,
67
+ ANALOG_PLL_DDR_TOG,
68
+ ANALOG_PLL_DDR_SS,
69
+ ANALOG_PLL_DDR_SS_SET,
70
+ ANALOG_PLL_DDR_SS_CLR,
71
+ ANALOG_PLL_DDR_SS_TOG,
72
+ ANALOG_PLL_DDR_NUM,
73
+ ANALOG_PLL_DDR_NUM_SET,
74
+ ANALOG_PLL_DDR_NUM_CLR,
75
+ ANALOG_PLL_DDR_NUM_TOG,
76
+ ANALOG_PLL_DDR_DENOM,
77
+ ANALOG_PLL_DDR_DENOM_SET,
78
+ ANALOG_PLL_DDR_DENOM_CLR,
79
+ ANALOG_PLL_DDR_DENOM_TOG,
80
+ ANALOG_PLL_480,
81
+ ANALOG_PLL_480_SET,
82
+ ANALOG_PLL_480_CLR,
83
+ ANALOG_PLL_480_TOG,
84
+ ANALOG_PLL_480A,
85
+ ANALOG_PLL_480A_SET,
86
+ ANALOG_PLL_480A_CLR,
87
+ ANALOG_PLL_480A_TOG,
88
+ ANALOG_PLL_480B,
89
+ ANALOG_PLL_480B_SET,
90
+ ANALOG_PLL_480B_CLR,
91
+ ANALOG_PLL_480B_TOG,
92
+ ANALOG_PLL_ENET,
93
+ ANALOG_PLL_ENET_SET,
94
+ ANALOG_PLL_ENET_CLR,
95
+ ANALOG_PLL_ENET_TOG,
96
+ ANALOG_PLL_AUDIO,
97
+ ANALOG_PLL_AUDIO_SET,
98
+ ANALOG_PLL_AUDIO_CLR,
99
+ ANALOG_PLL_AUDIO_TOG,
100
+ ANALOG_PLL_AUDIO_SS,
101
+ ANALOG_PLL_AUDIO_SS_SET,
102
+ ANALOG_PLL_AUDIO_SS_CLR,
103
+ ANALOG_PLL_AUDIO_SS_TOG,
104
+ ANALOG_PLL_AUDIO_NUM,
105
+ ANALOG_PLL_AUDIO_NUM_SET,
106
+ ANALOG_PLL_AUDIO_NUM_CLR,
107
+ ANALOG_PLL_AUDIO_NUM_TOG,
108
+ ANALOG_PLL_AUDIO_DENOM,
109
+ ANALOG_PLL_AUDIO_DENOM_SET,
110
+ ANALOG_PLL_AUDIO_DENOM_CLR,
111
+ ANALOG_PLL_AUDIO_DENOM_TOG,
112
+ ANALOG_PLL_VIDEO,
113
+ ANALOG_PLL_VIDEO_SET,
114
+ ANALOG_PLL_VIDEO_CLR,
115
+ ANALOG_PLL_VIDEO_TOG,
116
+ ANALOG_PLL_VIDEO_SS,
117
+ ANALOG_PLL_VIDEO_SS_SET,
118
+ ANALOG_PLL_VIDEO_SS_CLR,
119
+ ANALOG_PLL_VIDEO_SS_TOG,
120
+ ANALOG_PLL_VIDEO_NUM,
121
+ ANALOG_PLL_VIDEO_NUM_SET,
122
+ ANALOG_PLL_VIDEO_NUM_CLR,
123
+ ANALOG_PLL_VIDEO_NUM_TOG,
124
+ ANALOG_PLL_VIDEO_DENOM,
125
+ ANALOG_PLL_VIDEO_DENOM_SET,
126
+ ANALOG_PLL_VIDEO_DENOM_CLR,
127
+ ANALOG_PLL_VIDEO_DENOM_TOG,
128
+ ANALOG_PLL_MISC0,
129
+ ANALOG_PLL_MISC0_SET,
130
+ ANALOG_PLL_MISC0_CLR,
131
+ ANALOG_PLL_MISC0_TOG,
132
+
133
+ ANALOG_DIGPROG = 0x800 / sizeof(uint32_t),
134
+ ANALOG_MAX,
135
+
136
+ ANALOG_PLL_LOCK = BIT(31)
137
+};
138
+
139
+enum IMX7CCMRegisters {
140
+ CCM_MAX = 0xBE00 / sizeof(uint32_t) + 1,
141
+};
142
+
143
+enum IMX7PMURegisters {
144
+ PMU_MAX = 0x140 / sizeof(uint32_t),
145
+};
146
+
147
+#define TYPE_IMX7_CCM "imx7.ccm"
148
+#define IMX7_CCM(obj) OBJECT_CHECK(IMX7CCMState, (obj), TYPE_IMX7_CCM)
149
+
150
+typedef struct IMX7CCMState {
151
+ /* <private> */
152
+ IMXCCMState parent_obj;
153
+
154
+ /* <public> */
155
+ MemoryRegion iomem;
156
+
157
+ uint32_t ccm[CCM_MAX];
158
+} IMX7CCMState;
159
+
160
+
161
+#define TYPE_IMX7_ANALOG "imx7.analog"
162
+#define IMX7_ANALOG(obj) OBJECT_CHECK(IMX7AnalogState, (obj), TYPE_IMX7_ANALOG)
163
+
164
+typedef struct IMX7AnalogState {
165
+ /* <private> */
166
+ IMXCCMState parent_obj;
167
+
168
+ /* <public> */
169
+ struct {
170
+ MemoryRegion container;
171
+ MemoryRegion analog;
172
+ MemoryRegion digprog;
173
+ MemoryRegion pmu;
174
+ } mmio;
175
+
176
+ uint32_t analog[ANALOG_MAX];
177
+ uint32_t pmu[PMU_MAX];
178
+} IMX7AnalogState;
179
+
180
+#endif /* IMX7_CCM_H */
181
diff --git a/hw/misc/imx7_ccm.c b/hw/misc/imx7_ccm.c
182
new file mode 100644
183
index XXXXXXX..XXXXXXX
184
--- /dev/null
185
+++ b/hw/misc/imx7_ccm.c
186
@@ -XXX,XX +XXX,XX @@
187
+/*
188
+ * Copyright (c) 2018, Impinj, Inc.
189
+ *
190
+ * i.MX7 CCM, PMU and ANALOG IP blocks emulation code
191
+ *
192
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
193
+ *
194
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
195
+ * See the COPYING file in the top-level directory.
196
+ */
197
+
198
+#include "qemu/osdep.h"
199
+#include "qemu/log.h"
200
+
201
+#include "hw/misc/imx7_ccm.h"
202
+
203
+static void imx7_analog_reset(DeviceState *dev)
204
+{
205
+ IMX7AnalogState *s = IMX7_ANALOG(dev);
206
+
207
+ memset(s->pmu, 0, sizeof(s->pmu));
208
+ memset(s->analog, 0, sizeof(s->analog));
209
+
210
+ s->analog[ANALOG_PLL_ARM] = 0x00002042;
211
+ s->analog[ANALOG_PLL_DDR] = 0x0060302c;
212
+ s->analog[ANALOG_PLL_DDR_SS] = 0x00000000;
213
+ s->analog[ANALOG_PLL_DDR_NUM] = 0x06aaac4d;
214
+ s->analog[ANALOG_PLL_DDR_DENOM] = 0x100003ec;
215
+ s->analog[ANALOG_PLL_480] = 0x00002000;
216
+ s->analog[ANALOG_PLL_480A] = 0x52605a56;
217
+ s->analog[ANALOG_PLL_480B] = 0x52525216;
218
+ s->analog[ANALOG_PLL_ENET] = 0x00001fc0;
219
+ s->analog[ANALOG_PLL_AUDIO] = 0x0001301b;
220
+ s->analog[ANALOG_PLL_AUDIO_SS] = 0x00000000;
221
+ s->analog[ANALOG_PLL_AUDIO_NUM] = 0x05f5e100;
222
+ s->analog[ANALOG_PLL_AUDIO_DENOM] = 0x2964619c;
223
+ s->analog[ANALOG_PLL_VIDEO] = 0x0008201b;
224
+ s->analog[ANALOG_PLL_VIDEO_SS] = 0x00000000;
225
+ s->analog[ANALOG_PLL_VIDEO_NUM] = 0x0000f699;
226
+ s->analog[ANALOG_PLL_VIDEO_DENOM] = 0x000f4240;
227
+ s->analog[ANALOG_PLL_MISC0] = 0x00000000;
228
+
229
+ /* all PLLs need to be locked */
230
+ s->analog[ANALOG_PLL_ARM] |= ANALOG_PLL_LOCK;
231
+ s->analog[ANALOG_PLL_DDR] |= ANALOG_PLL_LOCK;
232
+ s->analog[ANALOG_PLL_480] |= ANALOG_PLL_LOCK;
233
+ s->analog[ANALOG_PLL_480A] |= ANALOG_PLL_LOCK;
234
+ s->analog[ANALOG_PLL_480B] |= ANALOG_PLL_LOCK;
235
+ s->analog[ANALOG_PLL_ENET] |= ANALOG_PLL_LOCK;
236
+ s->analog[ANALOG_PLL_AUDIO] |= ANALOG_PLL_LOCK;
237
+ s->analog[ANALOG_PLL_VIDEO] |= ANALOG_PLL_LOCK;
238
+ s->analog[ANALOG_PLL_MISC0] |= ANALOG_PLL_LOCK;
239
+
240
+ /*
241
+ * Since I couldn't find any info about this in the reference
242
+ * manual the value of this register is based strictly on matching
243
+ * what Linux kernel expects it to be.
244
+ */
245
+ s->analog[ANALOG_DIGPROG] = 0x720000;
246
+ /*
247
+ * Set revision to be 1.0 (Arbitrary choice, no particular
248
+ * reason).
249
+ */
250
+ s->analog[ANALOG_DIGPROG] |= 0x000010;
251
+}
252
+
253
+static void imx7_ccm_reset(DeviceState *dev)
254
+{
255
+ IMX7CCMState *s = IMX7_CCM(dev);
256
+
257
+ memset(s->ccm, 0, sizeof(s->ccm));
258
+}
259
+
260
+#define CCM_INDEX(offset) (((offset) & ~(hwaddr)0xF) / sizeof(uint32_t))
261
+#define CCM_BITOP(offset) ((offset) & (hwaddr)0xF)
262
+
263
+enum {
264
+ CCM_BITOP_NONE = 0x00,
265
+ CCM_BITOP_SET = 0x04,
266
+ CCM_BITOP_CLR = 0x08,
267
+ CCM_BITOP_TOG = 0x0C,
268
+};
269
+
270
+static uint64_t imx7_set_clr_tog_read(void *opaque, hwaddr offset,
271
+ unsigned size)
272
+{
273
+ const uint32_t *mmio = opaque;
274
+
275
+ return mmio[CCM_INDEX(offset)];
276
+}
277
+
278
+static void imx7_set_clr_tog_write(void *opaque, hwaddr offset,
279
+ uint64_t value, unsigned size)
280
+{
281
+ const uint8_t bitop = CCM_BITOP(offset);
282
+ const uint32_t index = CCM_INDEX(offset);
283
+ uint32_t *mmio = opaque;
284
+
285
+ switch (bitop) {
286
+ case CCM_BITOP_NONE:
287
+ mmio[index] = value;
288
+ break;
289
+ case CCM_BITOP_SET:
290
+ mmio[index] |= value;
291
+ break;
292
+ case CCM_BITOP_CLR:
293
+ mmio[index] &= ~value;
294
+ break;
295
+ case CCM_BITOP_TOG:
296
+ mmio[index] ^= value;
297
+ break;
298
+ };
299
+}
300
+
301
+static const struct MemoryRegionOps imx7_set_clr_tog_ops = {
302
+ .read = imx7_set_clr_tog_read,
303
+ .write = imx7_set_clr_tog_write,
304
+ .endianness = DEVICE_NATIVE_ENDIAN,
305
+ .impl = {
306
+ /*
307
+ * Our device would not work correctly if the guest was doing
308
+ * unaligned access. This might not be a limitation on the real
309
+ * device but in practice there is no reason for a guest to access
310
+ * this device unaligned.
311
+ */
312
+ .min_access_size = 4,
313
+ .max_access_size = 4,
314
+ .unaligned = false,
315
+ },
316
+};
317
+
318
+static const struct MemoryRegionOps imx7_digprog_ops = {
319
+ .read = imx7_set_clr_tog_read,
320
+ .endianness = DEVICE_NATIVE_ENDIAN,
321
+ .impl = {
322
+ .min_access_size = 4,
323
+ .max_access_size = 4,
324
+ .unaligned = false,
325
+ },
326
+};
327
+
328
+static void imx7_ccm_init(Object *obj)
329
+{
330
+ SysBusDevice *sd = SYS_BUS_DEVICE(obj);
331
+ IMX7CCMState *s = IMX7_CCM(obj);
332
+
333
+ memory_region_init_io(&s->iomem,
334
+ obj,
335
+ &imx7_set_clr_tog_ops,
336
+ s->ccm,
337
+ TYPE_IMX7_CCM ".ccm",
338
+ sizeof(s->ccm));
339
+
340
+ sysbus_init_mmio(sd, &s->iomem);
341
+}
342
+
343
+static void imx7_analog_init(Object *obj)
344
+{
345
+ SysBusDevice *sd = SYS_BUS_DEVICE(obj);
346
+ IMX7AnalogState *s = IMX7_ANALOG(obj);
347
+
348
+ memory_region_init(&s->mmio.container, obj, TYPE_IMX7_ANALOG,
349
+ 0x10000);
350
+
351
+ memory_region_init_io(&s->mmio.analog,
352
+ obj,
353
+ &imx7_set_clr_tog_ops,
354
+ s->analog,
355
+ TYPE_IMX7_ANALOG,
356
+ sizeof(s->analog));
357
+
358
+ memory_region_add_subregion(&s->mmio.container,
359
+ 0x60, &s->mmio.analog);
360
+
361
+ memory_region_init_io(&s->mmio.pmu,
362
+ obj,
363
+ &imx7_set_clr_tog_ops,
364
+ s->pmu,
365
+ TYPE_IMX7_ANALOG ".pmu",
366
+ sizeof(s->pmu));
367
+
368
+ memory_region_add_subregion(&s->mmio.container,
369
+ 0x200, &s->mmio.pmu);
370
+
371
+ memory_region_init_io(&s->mmio.digprog,
372
+ obj,
373
+ &imx7_digprog_ops,
374
+ &s->analog[ANALOG_DIGPROG],
375
+ TYPE_IMX7_ANALOG ".digprog",
376
+ sizeof(uint32_t));
377
+
378
+ memory_region_add_subregion_overlap(&s->mmio.container,
379
+ 0x800, &s->mmio.digprog, 10);
380
+
381
+
382
+ sysbus_init_mmio(sd, &s->mmio.container);
383
+}
384
+
385
+static const VMStateDescription vmstate_imx7_ccm = {
386
+ .name = TYPE_IMX7_CCM,
387
+ .version_id = 1,
388
+ .minimum_version_id = 1,
389
+ .fields = (VMStateField[]) {
390
+ VMSTATE_UINT32_ARRAY(ccm, IMX7CCMState, CCM_MAX),
391
+ VMSTATE_END_OF_LIST()
392
+ },
393
+};
394
+
395
+static uint32_t imx7_ccm_get_clock_frequency(IMXCCMState *dev, IMXClk clock)
396
+{
397
+ /*
398
+ * This function is "consumed" by GPT emulation code, however on
399
+ * i.MX7 each GPT block can have their own clock root. This means
400
+ * that this functions needs somehow to know requester's identity
401
+ * and the way to pass it: be it via additional IMXClk constants
402
+ * or by adding another argument to this method needs to be
403
+ * figured out
404
+ */
405
+ qemu_log_mask(LOG_GUEST_ERROR, "[%s]%s: Not implemented\n",
406
+ TYPE_IMX7_CCM, __func__);
407
+ return 0;
408
+}
409
+
410
+static void imx7_ccm_class_init(ObjectClass *klass, void *data)
411
+{
412
+ DeviceClass *dc = DEVICE_CLASS(klass);
413
+ IMXCCMClass *ccm = IMX_CCM_CLASS(klass);
414
+
415
+ dc->reset = imx7_ccm_reset;
416
+ dc->vmsd = &vmstate_imx7_ccm;
417
+ dc->desc = "i.MX7 Clock Control Module";
418
+
419
+ ccm->get_clock_frequency = imx7_ccm_get_clock_frequency;
420
+}
421
+
422
+static const TypeInfo imx7_ccm_info = {
423
+ .name = TYPE_IMX7_CCM,
424
+ .parent = TYPE_IMX_CCM,
425
+ .instance_size = sizeof(IMX7CCMState),
426
+ .instance_init = imx7_ccm_init,
427
+ .class_init = imx7_ccm_class_init,
428
+};
429
+
430
+static const VMStateDescription vmstate_imx7_analog = {
431
+ .name = TYPE_IMX7_ANALOG,
432
+ .version_id = 1,
433
+ .minimum_version_id = 1,
434
+ .fields = (VMStateField[]) {
435
+ VMSTATE_UINT32_ARRAY(analog, IMX7AnalogState, ANALOG_MAX),
436
+ VMSTATE_UINT32_ARRAY(pmu, IMX7AnalogState, PMU_MAX),
437
+ VMSTATE_END_OF_LIST()
438
+ },
439
+};
440
+
441
+static void imx7_analog_class_init(ObjectClass *klass, void *data)
442
+{
443
+ DeviceClass *dc = DEVICE_CLASS(klass);
444
+
445
+ dc->reset = imx7_analog_reset;
446
+ dc->vmsd = &vmstate_imx7_analog;
447
+ dc->desc = "i.MX7 Analog Module";
448
+}
449
+
450
+static const TypeInfo imx7_analog_info = {
451
+ .name = TYPE_IMX7_ANALOG,
452
+ .parent = TYPE_SYS_BUS_DEVICE,
453
+ .instance_size = sizeof(IMX7AnalogState),
454
+ .instance_init = imx7_analog_init,
455
+ .class_init = imx7_analog_class_init,
456
+};
457
+
458
+static void imx7_ccm_register_type(void)
459
+{
460
+ type_register_static(&imx7_ccm_info);
461
+ type_register_static(&imx7_analog_info);
462
+}
463
+type_init(imx7_ccm_register_type)
464
--
465
2.16.1
466
467
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Add enough code to emulate i.MX2 watchdog IP block so it would be
4
possible to reboot the machine running Linux Guest.
5
6
Cc: Peter Maydell <peter.maydell@linaro.org>
7
Cc: Jason Wang <jasowang@redhat.com>
8
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
9
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
10
Cc: Michael S. Tsirkin <mst@redhat.com>
11
Cc: qemu-devel@nongnu.org
12
Cc: qemu-arm@nongnu.org
13
Cc: yurovsky@gmail.com
14
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
16
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
17
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
18
---
19
hw/misc/Makefile.objs | 1 +
20
include/hw/misc/imx2_wdt.h | 33 +++++++++++++++++
21
hw/misc/imx2_wdt.c | 89 ++++++++++++++++++++++++++++++++++++++++++++++
22
3 files changed, 123 insertions(+)
23
create mode 100644 include/hw/misc/imx2_wdt.h
24
create mode 100644 hw/misc/imx2_wdt.c
25
26
diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs
27
index XXXXXXX..XXXXXXX 100644
28
--- a/hw/misc/Makefile.objs
29
+++ b/hw/misc/Makefile.objs
30
@@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_IMX) += imx25_ccm.o
31
obj-$(CONFIG_IMX) += imx6_ccm.o
32
obj-$(CONFIG_IMX) += imx6_src.o
33
obj-$(CONFIG_IMX) += imx7_ccm.o
34
+obj-$(CONFIG_IMX) += imx2_wdt.o
35
obj-$(CONFIG_MILKYMIST) += milkymist-hpdmc.o
36
obj-$(CONFIG_MILKYMIST) += milkymist-pfpu.o
37
obj-$(CONFIG_MAINSTONE) += mst_fpga.o
38
diff --git a/include/hw/misc/imx2_wdt.h b/include/hw/misc/imx2_wdt.h
39
new file mode 100644
40
index XXXXXXX..XXXXXXX
41
--- /dev/null
42
+++ b/include/hw/misc/imx2_wdt.h
43
@@ -XXX,XX +XXX,XX @@
44
+/*
45
+ * Copyright (c) 2017, Impinj, Inc.
46
+ *
47
+ * i.MX2 Watchdog IP block
48
+ *
49
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
50
+ *
51
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
52
+ * See the COPYING file in the top-level directory.
53
+ */
54
+
55
+#ifndef IMX2_WDT_H
56
+#define IMX2_WDT_H
57
+
58
+#include "hw/sysbus.h"
59
+
60
+#define TYPE_IMX2_WDT "imx2.wdt"
61
+#define IMX2_WDT(obj) OBJECT_CHECK(IMX2WdtState, (obj), TYPE_IMX2_WDT)
62
+
63
+enum IMX2WdtRegisters {
64
+ IMX2_WDT_WCR = 0x0000,
65
+ IMX2_WDT_REG_NUM = 0x0008 / sizeof(uint16_t) + 1,
66
+};
67
+
68
+
69
+typedef struct IMX2WdtState {
70
+ /* <private> */
71
+ SysBusDevice parent_obj;
72
+
73
+ MemoryRegion mmio;
74
+} IMX2WdtState;
75
+
76
+#endif /* IMX7_SNVS_H */
77
diff --git a/hw/misc/imx2_wdt.c b/hw/misc/imx2_wdt.c
78
new file mode 100644
79
index XXXXXXX..XXXXXXX
80
--- /dev/null
81
+++ b/hw/misc/imx2_wdt.c
82
@@ -XXX,XX +XXX,XX @@
83
+/*
84
+ * Copyright (c) 2018, Impinj, Inc.
85
+ *
86
+ * i.MX2 Watchdog IP block
87
+ *
88
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
89
+ *
90
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
91
+ * See the COPYING file in the top-level directory.
92
+ */
93
+
94
+#include "qemu/osdep.h"
95
+#include "qemu/bitops.h"
96
+#include "sysemu/watchdog.h"
97
+
98
+#include "hw/misc/imx2_wdt.h"
99
+
100
+#define IMX2_WDT_WCR_WDA BIT(5) /* -> External Reset WDOG_B */
101
+#define IMX2_WDT_WCR_SRS BIT(4) /* -> Software Reset Signal */
102
+
103
+static uint64_t imx2_wdt_read(void *opaque, hwaddr addr,
104
+ unsigned int size)
105
+{
106
+ return 0;
107
+}
108
+
109
+static void imx2_wdt_write(void *opaque, hwaddr addr,
110
+ uint64_t value, unsigned int size)
111
+{
112
+ if (addr == IMX2_WDT_WCR &&
113
+ (value & (IMX2_WDT_WCR_WDA | IMX2_WDT_WCR_SRS))) {
114
+ watchdog_perform_action();
115
+ }
116
+}
117
+
118
+static const MemoryRegionOps imx2_wdt_ops = {
119
+ .read = imx2_wdt_read,
120
+ .write = imx2_wdt_write,
121
+ .endianness = DEVICE_NATIVE_ENDIAN,
122
+ .impl = {
123
+ /*
124
+ * Our device would not work correctly if the guest was doing
125
+ * unaligned access. This might not be a limitation on the
126
+ * real device but in practice there is no reason for a guest
127
+ * to access this device unaligned.
128
+ */
129
+ .min_access_size = 4,
130
+ .max_access_size = 4,
131
+ .unaligned = false,
132
+ },
133
+};
134
+
135
+static void imx2_wdt_realize(DeviceState *dev, Error **errp)
136
+{
137
+ IMX2WdtState *s = IMX2_WDT(dev);
138
+
139
+ memory_region_init_io(&s->mmio, OBJECT(dev),
140
+ &imx2_wdt_ops, s,
141
+ TYPE_IMX2_WDT".mmio",
142
+ IMX2_WDT_REG_NUM * sizeof(uint16_t));
143
+ sysbus_init_mmio(SYS_BUS_DEVICE(dev), &s->mmio);
144
+}
145
+
146
+static void imx2_wdt_class_init(ObjectClass *klass, void *data)
147
+{
148
+ DeviceClass *dc = DEVICE_CLASS(klass);
149
+
150
+ dc->realize = imx2_wdt_realize;
151
+ set_bit(DEVICE_CATEGORY_MISC, dc->categories);
152
+}
153
+
154
+static const TypeInfo imx2_wdt_info = {
155
+ .name = TYPE_IMX2_WDT,
156
+ .parent = TYPE_SYS_BUS_DEVICE,
157
+ .instance_size = sizeof(IMX2WdtState),
158
+ .class_init = imx2_wdt_class_init,
159
+};
160
+
161
+static WatchdogTimerModel model = {
162
+ .wdt_name = "imx2-watchdog",
163
+ .wdt_description = "i.MX2 Watchdog",
164
+};
165
+
166
+static void imx2_wdt_register_type(void)
167
+{
168
+ watchdog_add_model(&model);
169
+ type_register_static(&imx2_wdt_info);
170
+}
171
+type_init(imx2_wdt_register_type)
172
--
173
2.16.1
174
175
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Add code to emulate SNVS IP-block. Currently only the bits needed to
4
be able to emulate machine shutdown are implemented.
5
6
Cc: Peter Maydell <peter.maydell@linaro.org>
7
Cc: Jason Wang <jasowang@redhat.com>
8
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
9
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
10
Cc: Michael S. Tsirkin <mst@redhat.com>
11
Cc: qemu-devel@nongnu.org
12
Cc: qemu-arm@nongnu.org
13
Cc: yurovsky@gmail.com
14
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
18
hw/misc/Makefile.objs | 1 +
19
include/hw/misc/imx7_snvs.h | 35 +++++++++++++++++++
20
hw/misc/imx7_snvs.c | 83 +++++++++++++++++++++++++++++++++++++++++++++
21
3 files changed, 119 insertions(+)
22
create mode 100644 include/hw/misc/imx7_snvs.h
23
create mode 100644 hw/misc/imx7_snvs.c
24
25
diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs
26
index XXXXXXX..XXXXXXX 100644
27
--- a/hw/misc/Makefile.objs
28
+++ b/hw/misc/Makefile.objs
29
@@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_IMX) += imx6_ccm.o
30
obj-$(CONFIG_IMX) += imx6_src.o
31
obj-$(CONFIG_IMX) += imx7_ccm.o
32
obj-$(CONFIG_IMX) += imx2_wdt.o
33
+obj-$(CONFIG_IMX) += imx7_snvs.o
34
obj-$(CONFIG_MILKYMIST) += milkymist-hpdmc.o
35
obj-$(CONFIG_MILKYMIST) += milkymist-pfpu.o
36
obj-$(CONFIG_MAINSTONE) += mst_fpga.o
37
diff --git a/include/hw/misc/imx7_snvs.h b/include/hw/misc/imx7_snvs.h
38
new file mode 100644
39
index XXXXXXX..XXXXXXX
40
--- /dev/null
41
+++ b/include/hw/misc/imx7_snvs.h
42
@@ -XXX,XX +XXX,XX @@
43
+/*
44
+ * Copyright (c) 2017, Impinj, Inc.
45
+ *
46
+ * i.MX7 SNVS block emulation code
47
+ *
48
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
49
+ *
50
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
51
+ * See the COPYING file in the top-level directory.
52
+ */
53
+
54
+#ifndef IMX7_SNVS_H
55
+#define IMX7_SNVS_H
56
+
57
+#include "qemu/bitops.h"
58
+#include "hw/sysbus.h"
59
+
60
+
61
+enum IMX7SNVSRegisters {
62
+ SNVS_LPCR = 0x38,
63
+ SNVS_LPCR_TOP = BIT(6),
64
+ SNVS_LPCR_DP_EN = BIT(5)
65
+};
66
+
67
+#define TYPE_IMX7_SNVS "imx7.snvs"
68
+#define IMX7_SNVS(obj) OBJECT_CHECK(IMX7SNVSState, (obj), TYPE_IMX7_SNVS)
69
+
70
+typedef struct IMX7SNVSState {
71
+ /* <private> */
72
+ SysBusDevice parent_obj;
73
+
74
+ MemoryRegion mmio;
75
+} IMX7SNVSState;
76
+
77
+#endif /* IMX7_SNVS_H */
78
diff --git a/hw/misc/imx7_snvs.c b/hw/misc/imx7_snvs.c
79
new file mode 100644
80
index XXXXXXX..XXXXXXX
81
--- /dev/null
82
+++ b/hw/misc/imx7_snvs.c
83
@@ -XXX,XX +XXX,XX @@
84
+/*
85
+ * IMX7 Secure Non-Volatile Storage
86
+ *
87
+ * Copyright (c) 2018, Impinj, Inc.
88
+ *
89
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
90
+ *
91
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
92
+ * See the COPYING file in the top-level directory.
93
+ *
94
+ * Bare minimum emulation code needed to support being able to shut
95
+ * down linux guest gracefully.
96
+ */
97
+
98
+#include "qemu/osdep.h"
99
+#include "hw/misc/imx7_snvs.h"
100
+#include "qemu/log.h"
101
+#include "sysemu/sysemu.h"
102
+
103
+static uint64_t imx7_snvs_read(void *opaque, hwaddr offset, unsigned size)
104
+{
105
+ return 0;
106
+}
107
+
108
+static void imx7_snvs_write(void *opaque, hwaddr offset,
109
+ uint64_t v, unsigned size)
110
+{
111
+ const uint32_t value = v;
112
+ const uint32_t mask = SNVS_LPCR_TOP | SNVS_LPCR_DP_EN;
113
+
114
+ if (offset == SNVS_LPCR && ((value & mask) == mask)) {
115
+ qemu_system_shutdown_request(SHUTDOWN_CAUSE_GUEST_SHUTDOWN);
116
+ }
117
+}
118
+
119
+static const struct MemoryRegionOps imx7_snvs_ops = {
120
+ .read = imx7_snvs_read,
121
+ .write = imx7_snvs_write,
122
+ .endianness = DEVICE_NATIVE_ENDIAN,
123
+ .impl = {
124
+ /*
125
+ * Our device would not work correctly if the guest was doing
126
+ * unaligned access. This might not be a limitation on the real
127
+ * device but in practice there is no reason for a guest to access
128
+ * this device unaligned.
129
+ */
130
+ .min_access_size = 4,
131
+ .max_access_size = 4,
132
+ .unaligned = false,
133
+ },
134
+};
135
+
136
+static void imx7_snvs_init(Object *obj)
137
+{
138
+ SysBusDevice *sd = SYS_BUS_DEVICE(obj);
139
+ IMX7SNVSState *s = IMX7_SNVS(obj);
140
+
141
+ memory_region_init_io(&s->mmio, obj, &imx7_snvs_ops, s,
142
+ TYPE_IMX7_SNVS, 0x1000);
143
+
144
+ sysbus_init_mmio(sd, &s->mmio);
145
+}
146
+
147
+static void imx7_snvs_class_init(ObjectClass *klass, void *data)
148
+{
149
+ DeviceClass *dc = DEVICE_CLASS(klass);
150
+
151
+ dc->desc = "i.MX7 Secure Non-Volatile Storage Module";
152
+}
153
+
154
+static const TypeInfo imx7_snvs_info = {
155
+ .name = TYPE_IMX7_SNVS,
156
+ .parent = TYPE_SYS_BUS_DEVICE,
157
+ .instance_size = sizeof(IMX7SNVSState),
158
+ .instance_init = imx7_snvs_init,
159
+ .class_init = imx7_snvs_class_init,
160
+};
161
+
162
+static void imx7_snvs_register_type(void)
163
+{
164
+ type_register_static(&imx7_snvs_info);
165
+}
166
+type_init(imx7_snvs_register_type)
167
--
168
2.16.1
169
170
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Add minimal code needed to allow upstream Linux guest to boot.
4
5
Cc: Peter Maydell <peter.maydell@linaro.org>
6
Cc: Jason Wang <jasowang@redhat.com>
7
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
9
Cc: Michael S. Tsirkin <mst@redhat.com>
10
Cc: qemu-devel@nongnu.org
11
Cc: qemu-arm@nongnu.org
12
Cc: yurovsky@gmail.com
13
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
14
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
15
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
16
---
17
hw/intc/Makefile.objs | 2 +-
18
include/hw/intc/imx_gpcv2.h | 22 ++++++++
19
hw/intc/imx_gpcv2.c | 125 ++++++++++++++++++++++++++++++++++++++++++++
20
3 files changed, 148 insertions(+), 1 deletion(-)
21
create mode 100644 include/hw/intc/imx_gpcv2.h
22
create mode 100644 hw/intc/imx_gpcv2.c
23
24
diff --git a/hw/intc/Makefile.objs b/hw/intc/Makefile.objs
25
index XXXXXXX..XXXXXXX 100644
26
--- a/hw/intc/Makefile.objs
27
+++ b/hw/intc/Makefile.objs
28
@@ -XXX,XX +XXX,XX @@ common-obj-$(CONFIG_XILINX) += xilinx_intc.o
29
common-obj-$(CONFIG_XLNX_ZYNQMP) += xlnx-pmu-iomod-intc.o
30
common-obj-$(CONFIG_XLNX_ZYNQMP) += xlnx-zynqmp-ipi.o
31
common-obj-$(CONFIG_ETRAXFS) += etraxfs_pic.o
32
-common-obj-$(CONFIG_IMX) += imx_avic.o
33
+common-obj-$(CONFIG_IMX) += imx_avic.o imx_gpcv2.o
34
common-obj-$(CONFIG_LM32) += lm32_pic.o
35
common-obj-$(CONFIG_REALVIEW) += realview_gic.o
36
common-obj-$(CONFIG_SLAVIO) += slavio_intctl.o
37
diff --git a/include/hw/intc/imx_gpcv2.h b/include/hw/intc/imx_gpcv2.h
38
new file mode 100644
39
index XXXXXXX..XXXXXXX
40
--- /dev/null
41
+++ b/include/hw/intc/imx_gpcv2.h
42
@@ -XXX,XX +XXX,XX @@
43
+#ifndef IMX_GPCV2_H
44
+#define IMX_GPCV2_H
45
+
46
+#include "hw/sysbus.h"
47
+
48
+enum IMXGPCv2Registers {
49
+ GPC_NUM = 0xE00 / sizeof(uint32_t),
50
+};
51
+
52
+typedef struct IMXGPCv2State {
53
+ /*< private >*/
54
+ SysBusDevice parent_obj;
55
+
56
+ /*< public >*/
57
+ MemoryRegion iomem;
58
+ uint32_t regs[GPC_NUM];
59
+} IMXGPCv2State;
60
+
61
+#define TYPE_IMX_GPCV2 "imx-gpcv2"
62
+#define IMX_GPCV2(obj) OBJECT_CHECK(IMXGPCv2State, (obj), TYPE_IMX_GPCV2)
63
+
64
+#endif /* IMX_GPCV2_H */
65
diff --git a/hw/intc/imx_gpcv2.c b/hw/intc/imx_gpcv2.c
66
new file mode 100644
67
index XXXXXXX..XXXXXXX
68
--- /dev/null
69
+++ b/hw/intc/imx_gpcv2.c
70
@@ -XXX,XX +XXX,XX @@
71
+/*
72
+ * Copyright (c) 2018, Impinj, Inc.
73
+ *
74
+ * i.MX7 GPCv2 block emulation code
75
+ *
76
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
77
+ *
78
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
79
+ * See the COPYING file in the top-level directory.
80
+ */
81
+
82
+#include "qemu/osdep.h"
83
+#include "hw/intc/imx_gpcv2.h"
84
+#include "qemu/log.h"
85
+
86
+#define GPC_PU_PGC_SW_PUP_REQ 0x0f8
87
+#define GPC_PU_PGC_SW_PDN_REQ 0x104
88
+
89
+#define USB_HSIC_PHY_SW_Pxx_REQ BIT(4)
90
+#define USB_OTG2_PHY_SW_Pxx_REQ BIT(3)
91
+#define USB_OTG1_PHY_SW_Pxx_REQ BIT(2)
92
+#define PCIE_PHY_SW_Pxx_REQ BIT(1)
93
+#define MIPI_PHY_SW_Pxx_REQ BIT(0)
94
+
95
+
96
+static void imx_gpcv2_reset(DeviceState *dev)
97
+{
98
+ IMXGPCv2State *s = IMX_GPCV2(dev);
99
+
100
+ memset(s->regs, 0, sizeof(s->regs));
101
+}
102
+
103
+static uint64_t imx_gpcv2_read(void *opaque, hwaddr offset,
104
+ unsigned size)
105
+{
106
+ IMXGPCv2State *s = opaque;
107
+
108
+ return s->regs[offset / sizeof(uint32_t)];
109
+}
110
+
111
+static void imx_gpcv2_write(void *opaque, hwaddr offset,
112
+ uint64_t value, unsigned size)
113
+{
114
+ IMXGPCv2State *s = opaque;
115
+ const size_t idx = offset / sizeof(uint32_t);
116
+
117
+ s->regs[idx] = value;
118
+
119
+ /*
120
+ * Real HW will clear those bits once as a way to indicate that
121
+ * power up request is complete
122
+ */
123
+ if (offset == GPC_PU_PGC_SW_PUP_REQ ||
124
+ offset == GPC_PU_PGC_SW_PDN_REQ) {
125
+ s->regs[idx] &= ~(USB_HSIC_PHY_SW_Pxx_REQ |
126
+ USB_OTG2_PHY_SW_Pxx_REQ |
127
+ USB_OTG1_PHY_SW_Pxx_REQ |
128
+ PCIE_PHY_SW_Pxx_REQ |
129
+ MIPI_PHY_SW_Pxx_REQ);
130
+ }
131
+}
132
+
133
+static const struct MemoryRegionOps imx_gpcv2_ops = {
134
+ .read = imx_gpcv2_read,
135
+ .write = imx_gpcv2_write,
136
+ .endianness = DEVICE_NATIVE_ENDIAN,
137
+ .impl = {
138
+ /*
139
+ * Our device would not work correctly if the guest was doing
140
+ * unaligned access. This might not be a limitation on the real
141
+ * device but in practice there is no reason for a guest to access
142
+ * this device unaligned.
143
+ */
144
+ .min_access_size = 4,
145
+ .max_access_size = 4,
146
+ .unaligned = false,
147
+ },
148
+};
149
+
150
+static void imx_gpcv2_init(Object *obj)
151
+{
152
+ SysBusDevice *sd = SYS_BUS_DEVICE(obj);
153
+ IMXGPCv2State *s = IMX_GPCV2(obj);
154
+
155
+ memory_region_init_io(&s->iomem,
156
+ obj,
157
+ &imx_gpcv2_ops,
158
+ s,
159
+ TYPE_IMX_GPCV2 ".iomem",
160
+ sizeof(s->regs));
161
+ sysbus_init_mmio(sd, &s->iomem);
162
+}
163
+
164
+static const VMStateDescription vmstate_imx_gpcv2 = {
165
+ .name = TYPE_IMX_GPCV2,
166
+ .version_id = 1,
167
+ .minimum_version_id = 1,
168
+ .fields = (VMStateField[]) {
169
+ VMSTATE_UINT32_ARRAY(regs, IMXGPCv2State, GPC_NUM),
170
+ VMSTATE_END_OF_LIST()
171
+ },
172
+};
173
+
174
+static void imx_gpcv2_class_init(ObjectClass *klass, void *data)
175
+{
176
+ DeviceClass *dc = DEVICE_CLASS(klass);
177
+
178
+ dc->reset = imx_gpcv2_reset;
179
+ dc->vmsd = &vmstate_imx_gpcv2;
180
+ dc->desc = "i.MX GPCv2 Module";
181
+}
182
+
183
+static const TypeInfo imx_gpcv2_info = {
184
+ .name = TYPE_IMX_GPCV2,
185
+ .parent = TYPE_SYS_BUS_DEVICE,
186
+ .instance_size = sizeof(IMXGPCv2State),
187
+ .instance_init = imx_gpcv2_init,
188
+ .class_init = imx_gpcv2_class_init,
189
+};
190
+
191
+static void imx_gpcv2_register_type(void)
192
+{
193
+ type_register_static(&imx_gpcv2_info);
194
+}
195
+type_init(imx_gpcv2_register_type)
196
--
197
2.16.1
198
199
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Add minimal code needed to allow upstream Linux guest to boot.
4
5
Cc: Peter Maydell <peter.maydell@linaro.org>
6
Cc: Jason Wang <jasowang@redhat.com>
7
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
9
Cc: Michael S. Tsirkin <mst@redhat.com>
10
Cc: qemu-devel@nongnu.org
11
Cc: qemu-arm@nongnu.org
12
Cc: yurovsky@gmail.com
13
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
14
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
15
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
16
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
17
---
18
hw/misc/Makefile.objs | 1 +
19
include/hw/misc/imx7_gpr.h | 28 ++++++++++
20
hw/misc/imx7_gpr.c | 124 +++++++++++++++++++++++++++++++++++++++++++++
21
hw/misc/trace-events | 4 ++
22
4 files changed, 157 insertions(+)
23
create mode 100644 include/hw/misc/imx7_gpr.h
24
create mode 100644 hw/misc/imx7_gpr.c
25
26
diff --git a/hw/misc/Makefile.objs b/hw/misc/Makefile.objs
27
index XXXXXXX..XXXXXXX 100644
28
--- a/hw/misc/Makefile.objs
29
+++ b/hw/misc/Makefile.objs
30
@@ -XXX,XX +XXX,XX @@ obj-$(CONFIG_IMX) += imx6_src.o
31
obj-$(CONFIG_IMX) += imx7_ccm.o
32
obj-$(CONFIG_IMX) += imx2_wdt.o
33
obj-$(CONFIG_IMX) += imx7_snvs.o
34
+obj-$(CONFIG_IMX) += imx7_gpr.o
35
obj-$(CONFIG_MILKYMIST) += milkymist-hpdmc.o
36
obj-$(CONFIG_MILKYMIST) += milkymist-pfpu.o
37
obj-$(CONFIG_MAINSTONE) += mst_fpga.o
38
diff --git a/include/hw/misc/imx7_gpr.h b/include/hw/misc/imx7_gpr.h
39
new file mode 100644
40
index XXXXXXX..XXXXXXX
41
--- /dev/null
42
+++ b/include/hw/misc/imx7_gpr.h
43
@@ -XXX,XX +XXX,XX @@
44
+/*
45
+ * Copyright (c) 2017, Impinj, Inc.
46
+ *
47
+ * i.MX7 GPR IP block emulation code
48
+ *
49
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
50
+ *
51
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
52
+ * See the COPYING file in the top-level directory.
53
+ */
54
+
55
+#ifndef IMX7_GPR_H
56
+#define IMX7_GPR_H
57
+
58
+#include "qemu/bitops.h"
59
+#include "hw/sysbus.h"
60
+
61
+#define TYPE_IMX7_GPR "imx7.gpr"
62
+#define IMX7_GPR(obj) OBJECT_CHECK(IMX7GPRState, (obj), TYPE_IMX7_GPR)
63
+
64
+typedef struct IMX7GPRState {
65
+ /* <private> */
66
+ SysBusDevice parent_obj;
67
+
68
+ MemoryRegion mmio;
69
+} IMX7GPRState;
70
+
71
+#endif /* IMX7_GPR_H */
72
diff --git a/hw/misc/imx7_gpr.c b/hw/misc/imx7_gpr.c
73
new file mode 100644
74
index XXXXXXX..XXXXXXX
75
--- /dev/null
76
+++ b/hw/misc/imx7_gpr.c
77
@@ -XXX,XX +XXX,XX @@
78
+/*
79
+ * Copyright (c) 2018, Impinj, Inc.
80
+ *
81
+ * i.MX7 GPR IP block emulation code
82
+ *
83
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
84
+ *
85
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
86
+ * See the COPYING file in the top-level directory.
87
+ *
88
+ * Bare minimum emulation code needed to support being able to shut
89
+ * down linux guest gracefully.
90
+ */
91
+
92
+#include "qemu/osdep.h"
93
+#include "hw/misc/imx7_gpr.h"
94
+#include "qemu/log.h"
95
+#include "sysemu/sysemu.h"
96
+
97
+#include "trace.h"
98
+
99
+enum IMX7GPRRegisters {
100
+ IOMUXC_GPR0 = 0x00,
101
+ IOMUXC_GPR1 = 0x04,
102
+ IOMUXC_GPR2 = 0x08,
103
+ IOMUXC_GPR3 = 0x0c,
104
+ IOMUXC_GPR4 = 0x10,
105
+ IOMUXC_GPR5 = 0x14,
106
+ IOMUXC_GPR6 = 0x18,
107
+ IOMUXC_GPR7 = 0x1c,
108
+ IOMUXC_GPR8 = 0x20,
109
+ IOMUXC_GPR9 = 0x24,
110
+ IOMUXC_GPR10 = 0x28,
111
+ IOMUXC_GPR11 = 0x2c,
112
+ IOMUXC_GPR12 = 0x30,
113
+ IOMUXC_GPR13 = 0x34,
114
+ IOMUXC_GPR14 = 0x38,
115
+ IOMUXC_GPR15 = 0x3c,
116
+ IOMUXC_GPR16 = 0x40,
117
+ IOMUXC_GPR17 = 0x44,
118
+ IOMUXC_GPR18 = 0x48,
119
+ IOMUXC_GPR19 = 0x4c,
120
+ IOMUXC_GPR20 = 0x50,
121
+ IOMUXC_GPR21 = 0x54,
122
+ IOMUXC_GPR22 = 0x58,
123
+};
124
+
125
+#define IMX7D_GPR1_IRQ_MASK BIT(12)
126
+#define IMX7D_GPR1_ENET1_TX_CLK_SEL_MASK BIT(13)
127
+#define IMX7D_GPR1_ENET2_TX_CLK_SEL_MASK BIT(14)
128
+#define IMX7D_GPR1_ENET_TX_CLK_SEL_MASK (0x3 << 13)
129
+#define IMX7D_GPR1_ENET1_CLK_DIR_MASK BIT(17)
130
+#define IMX7D_GPR1_ENET2_CLK_DIR_MASK BIT(18)
131
+#define IMX7D_GPR1_ENET_CLK_DIR_MASK (0x3 << 17)
132
+
133
+#define IMX7D_GPR5_CSI_MUX_CONTROL_MIPI BIT(4)
134
+#define IMX7D_GPR12_PCIE_PHY_REFCLK_SEL BIT(5)
135
+#define IMX7D_GPR22_PCIE_PHY_PLL_LOCKED BIT(31)
136
+
137
+
138
+static uint64_t imx7_gpr_read(void *opaque, hwaddr offset, unsigned size)
139
+{
140
+ trace_imx7_gpr_read(offset);
141
+
142
+ if (offset == IOMUXC_GPR22) {
143
+ return IMX7D_GPR22_PCIE_PHY_PLL_LOCKED;
144
+ }
145
+
146
+ return 0;
147
+}
148
+
149
+static void imx7_gpr_write(void *opaque, hwaddr offset,
150
+ uint64_t v, unsigned size)
151
+{
152
+ trace_imx7_gpr_write(offset, v);
153
+}
154
+
155
+static const struct MemoryRegionOps imx7_gpr_ops = {
156
+ .read = imx7_gpr_read,
157
+ .write = imx7_gpr_write,
158
+ .endianness = DEVICE_NATIVE_ENDIAN,
159
+ .impl = {
160
+ /*
161
+ * Our device would not work correctly if the guest was doing
162
+ * unaligned access. This might not be a limitation on the
163
+ * real device but in practice there is no reason for a guest
164
+ * to access this device unaligned.
165
+ */
166
+ .min_access_size = 4,
167
+ .max_access_size = 4,
168
+ .unaligned = false,
169
+ },
170
+};
171
+
172
+static void imx7_gpr_init(Object *obj)
173
+{
174
+ SysBusDevice *sd = SYS_BUS_DEVICE(obj);
175
+ IMX7GPRState *s = IMX7_GPR(obj);
176
+
177
+ memory_region_init_io(&s->mmio, obj, &imx7_gpr_ops, s,
178
+ TYPE_IMX7_GPR, 64 * 1024);
179
+ sysbus_init_mmio(sd, &s->mmio);
180
+}
181
+
182
+static void imx7_gpr_class_init(ObjectClass *klass, void *data)
183
+{
184
+ DeviceClass *dc = DEVICE_CLASS(klass);
185
+
186
+ dc->desc = "i.MX7 General Purpose Registers Module";
187
+}
188
+
189
+static const TypeInfo imx7_gpr_info = {
190
+ .name = TYPE_IMX7_GPR,
191
+ .parent = TYPE_SYS_BUS_DEVICE,
192
+ .instance_size = sizeof(IMX7GPRState),
193
+ .instance_init = imx7_gpr_init,
194
+ .class_init = imx7_gpr_class_init,
195
+};
196
+
197
+static void imx7_gpr_register_type(void)
198
+{
199
+ type_register_static(&imx7_gpr_info);
200
+}
201
+type_init(imx7_gpr_register_type)
202
diff --git a/hw/misc/trace-events b/hw/misc/trace-events
203
index XXXXXXX..XXXXXXX 100644
204
--- a/hw/misc/trace-events
205
+++ b/hw/misc/trace-events
206
@@ -XXX,XX +XXX,XX @@ mps2_scc_cfg_read(unsigned function, unsigned device, uint32_t value) "MPS2 SCC
207
msf2_sysreg_write(uint64_t offset, uint32_t val, uint32_t prev) "msf2-sysreg write: addr 0x%08" HWADDR_PRIx " data 0x%" PRIx32 " prev 0x%" PRIx32
208
msf2_sysreg_read(uint64_t offset, uint32_t val) "msf2-sysreg read: addr 0x%08" HWADDR_PRIx " data 0x%08" PRIx32
209
msf2_sysreg_write_pll_status(void) "Invalid write to read only PLL status register"
210
+
211
+#hw/misc/imx7_gpr.c
212
+imx7_gpr_read(uint64_t offset) "addr 0x%08" HWADDR_PRIx
213
+imx7_gpr_write(uint64_t offset, uint64_t value) "addr 0x%08" HWADDR_PRIx "value 0x%08" HWADDR_PRIx
214
--
215
2.16.1
216
217
diff view generated by jsdifflib
Deleted patch
1
From: Andrey Smirnov <andrew.smirnov@gmail.com>
2
1
3
Add code to emulate Chipidea USB IP (used in i.MX SoCs). Tested to
4
work against:
5
6
-usb -drive if=none,id=stick,file=usb.img,format=raw -device \
7
usb-storage,bus=usb-bus.0,drive=stick
8
9
Cc: Peter Maydell <peter.maydell@linaro.org>
10
Cc: Jason Wang <jasowang@redhat.com>
11
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
12
Cc: Marcel Apfelbaum <marcel.apfelbaum@zoho.com>
13
Cc: Michael S. Tsirkin <mst@redhat.com>
14
Cc: qemu-devel@nongnu.org
15
Cc: qemu-arm@nongnu.org
16
Cc: yurovsky@gmail.com
17
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
18
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
19
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
20
---
21
hw/usb/Makefile.objs | 1 +
22
include/hw/usb/chipidea.h | 16 +++++
23
hw/usb/chipidea.c | 176 ++++++++++++++++++++++++++++++++++++++++++++++
24
3 files changed, 193 insertions(+)
25
create mode 100644 include/hw/usb/chipidea.h
26
create mode 100644 hw/usb/chipidea.c
27
28
diff --git a/hw/usb/Makefile.objs b/hw/usb/Makefile.objs
29
index XXXXXXX..XXXXXXX 100644
30
--- a/hw/usb/Makefile.objs
31
+++ b/hw/usb/Makefile.objs
32
@@ -XXX,XX +XXX,XX @@ common-obj-$(CONFIG_USB_XHCI_NEC) += hcd-xhci-nec.o
33
common-obj-$(CONFIG_USB_MUSB) += hcd-musb.o
34
35
obj-$(CONFIG_TUSB6010) += tusb6010.o
36
+obj-$(CONFIG_IMX) += chipidea.o
37
38
# emulated usb devices
39
common-obj-$(CONFIG_USB) += dev-hub.o
40
diff --git a/include/hw/usb/chipidea.h b/include/hw/usb/chipidea.h
41
new file mode 100644
42
index XXXXXXX..XXXXXXX
43
--- /dev/null
44
+++ b/include/hw/usb/chipidea.h
45
@@ -XXX,XX +XXX,XX @@
46
+#ifndef CHIPIDEA_H
47
+#define CHIPIDEA_H
48
+
49
+#include "hw/usb/hcd-ehci.h"
50
+
51
+typedef struct ChipideaState {
52
+ /*< private >*/
53
+ EHCISysBusState parent_obj;
54
+
55
+ MemoryRegion iomem[3];
56
+} ChipideaState;
57
+
58
+#define TYPE_CHIPIDEA "usb-chipidea"
59
+#define CHIPIDEA(obj) OBJECT_CHECK(ChipideaState, (obj), TYPE_CHIPIDEA)
60
+
61
+#endif /* CHIPIDEA_H */
62
diff --git a/hw/usb/chipidea.c b/hw/usb/chipidea.c
63
new file mode 100644
64
index XXXXXXX..XXXXXXX
65
--- /dev/null
66
+++ b/hw/usb/chipidea.c
67
@@ -XXX,XX +XXX,XX @@
68
+/*
69
+ * Copyright (c) 2018, Impinj, Inc.
70
+ *
71
+ * Chipidea USB block emulation code
72
+ *
73
+ * Author: Andrey Smirnov <andrew.smirnov@gmail.com>
74
+ *
75
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
76
+ * See the COPYING file in the top-level directory.
77
+ */
78
+
79
+#include "qemu/osdep.h"
80
+#include "hw/usb/hcd-ehci.h"
81
+#include "hw/usb/chipidea.h"
82
+#include "qemu/log.h"
83
+
84
+enum {
85
+ CHIPIDEA_USBx_DCIVERSION = 0x000,
86
+ CHIPIDEA_USBx_DCCPARAMS = 0x004,
87
+ CHIPIDEA_USBx_DCCPARAMS_HC = BIT(8),
88
+};
89
+
90
+static uint64_t chipidea_read(void *opaque, hwaddr offset,
91
+ unsigned size)
92
+{
93
+ return 0;
94
+}
95
+
96
+static void chipidea_write(void *opaque, hwaddr offset,
97
+ uint64_t value, unsigned size)
98
+{
99
+}
100
+
101
+static const struct MemoryRegionOps chipidea_ops = {
102
+ .read = chipidea_read,
103
+ .write = chipidea_write,
104
+ .endianness = DEVICE_NATIVE_ENDIAN,
105
+ .impl = {
106
+ /*
107
+ * Our device would not work correctly if the guest was doing
108
+ * unaligned access. This might not be a limitation on the
109
+ * real device but in practice there is no reason for a guest
110
+ * to access this device unaligned.
111
+ */
112
+ .min_access_size = 4,
113
+ .max_access_size = 4,
114
+ .unaligned = false,
115
+ },
116
+};
117
+
118
+static uint64_t chipidea_dc_read(void *opaque, hwaddr offset,
119
+ unsigned size)
120
+{
121
+ switch (offset) {
122
+ case CHIPIDEA_USBx_DCIVERSION:
123
+ return 0x1;
124
+ case CHIPIDEA_USBx_DCCPARAMS:
125
+ /*
126
+ * Real hardware (at least i.MX7) will also report the
127
+ * controller as "Device Capable" (and 8 supported endpoints),
128
+ * but there doesn't seem to be much point in doing so, since
129
+ * we don't emulate that part.
130
+ */
131
+ return CHIPIDEA_USBx_DCCPARAMS_HC;
132
+ }
133
+
134
+ return 0;
135
+}
136
+
137
+static void chipidea_dc_write(void *opaque, hwaddr offset,
138
+ uint64_t value, unsigned size)
139
+{
140
+}
141
+
142
+static const struct MemoryRegionOps chipidea_dc_ops = {
143
+ .read = chipidea_dc_read,
144
+ .write = chipidea_dc_write,
145
+ .endianness = DEVICE_NATIVE_ENDIAN,
146
+ .impl = {
147
+ /*
148
+ * Our device would not work correctly if the guest was doing
149
+ * unaligned access. This might not be a limitation on the real
150
+ * device but in practice there is no reason for a guest to access
151
+ * this device unaligned.
152
+ */
153
+ .min_access_size = 4,
154
+ .max_access_size = 4,
155
+ .unaligned = false,
156
+ },
157
+};
158
+
159
+static void chipidea_init(Object *obj)
160
+{
161
+ EHCIState *ehci = &SYS_BUS_EHCI(obj)->ehci;
162
+ ChipideaState *ci = CHIPIDEA(obj);
163
+ int i;
164
+
165
+ for (i = 0; i < ARRAY_SIZE(ci->iomem); i++) {
166
+ const struct {
167
+ const char *name;
168
+ hwaddr offset;
169
+ uint64_t size;
170
+ const struct MemoryRegionOps *ops;
171
+ } regions[ARRAY_SIZE(ci->iomem)] = {
172
+ /*
173
+ * Registers located between offsets 0x000 and 0xFC
174
+ */
175
+ {
176
+ .name = TYPE_CHIPIDEA ".misc",
177
+ .offset = 0x000,
178
+ .size = 0x100,
179
+ .ops = &chipidea_ops,
180
+ },
181
+ /*
182
+ * Registers located between offsets 0x1A4 and 0x1DC
183
+ */
184
+ {
185
+ .name = TYPE_CHIPIDEA ".endpoints",
186
+ .offset = 0x1A4,
187
+ .size = 0x1DC - 0x1A4 + 4,
188
+ .ops = &chipidea_ops,
189
+ },
190
+ /*
191
+ * USB_x_DCIVERSION and USB_x_DCCPARAMS
192
+ */
193
+ {
194
+ .name = TYPE_CHIPIDEA ".dc",
195
+ .offset = 0x120,
196
+ .size = 8,
197
+ .ops = &chipidea_dc_ops,
198
+ },
199
+ };
200
+
201
+ memory_region_init_io(&ci->iomem[i],
202
+ obj,
203
+ regions[i].ops,
204
+ ci,
205
+ regions[i].name,
206
+ regions[i].size);
207
+
208
+ memory_region_add_subregion(&ehci->mem,
209
+ regions[i].offset,
210
+ &ci->iomem[i]);
211
+ }
212
+}
213
+
214
+static void chipidea_class_init(ObjectClass *klass, void *data)
215
+{
216
+ DeviceClass *dc = DEVICE_CLASS(klass);
217
+ SysBusEHCIClass *sec = SYS_BUS_EHCI_CLASS(klass);
218
+
219
+ /*
220
+ * Offsets used were taken from i.MX7Dual Applications Processor
221
+ * Reference Manual, Rev 0.1, p. 3177, Table 11-59
222
+ */
223
+ sec->capsbase = 0x100;
224
+ sec->opregbase = 0x140;
225
+ sec->portnr = 1;
226
+
227
+ set_bit(DEVICE_CATEGORY_USB, dc->categories);
228
+ dc->desc = "Chipidea USB Module";
229
+}
230
+
231
+static const TypeInfo chipidea_info = {
232
+ .name = TYPE_CHIPIDEA,
233
+ .parent = TYPE_SYS_BUS_EHCI,
234
+ .instance_size = sizeof(ChipideaState),
235
+ .instance_init = chipidea_init,
236
+ .class_init = chipidea_class_init,
237
+};
238
+
239
+static void chipidea_register_type(void)
240
+{
241
+ type_register_static(&chipidea_info);
242
+}
243
+type_init(chipidea_register_type)
244
--
245
2.16.1
246
247
diff view generated by jsdifflib
Deleted patch
1
From: Richard Henderson <richard.henderson@linaro.org>
2
1
3
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
4
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
5
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
6
Message-id: 20180123035349.24538-3-richard.henderson@linaro.org
7
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
8
---
9
target/arm/cpu.h | 12 ++++++++++++
10
1 file changed, 12 insertions(+)
11
12
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
13
index XXXXXXX..XXXXXXX 100644
14
--- a/target/arm/cpu.h
15
+++ b/target/arm/cpu.h
16
@@ -XXX,XX +XXX,XX @@ typedef struct ARMVectorReg {
17
uint64_t d[2 * ARM_MAX_VQ] QEMU_ALIGNED(16);
18
} ARMVectorReg;
19
20
+/* In AArch32 mode, predicate registers do not exist at all. */
21
+#ifdef TARGET_AARCH64
22
+typedef struct ARMPredicateReg {
23
+ uint64_t p[2 * ARM_MAX_VQ / 8] QEMU_ALIGNED(16);
24
+} ARMPredicateReg;
25
+#endif
26
+
27
28
typedef struct CPUARMState {
29
/* Regs for current mode. */
30
@@ -XXX,XX +XXX,XX @@ typedef struct CPUARMState {
31
struct {
32
ARMVectorReg zregs[32];
33
34
+#ifdef TARGET_AARCH64
35
+ /* Store FFR as pregs[16] to make it easier to treat as any other. */
36
+ ARMPredicateReg pregs[17];
37
+#endif
38
+
39
uint32_t xregs[16];
40
/* We store these fpcsr fields separately for convenience. */
41
int vec_len;
42
--
43
2.16.1
44
45
diff view generated by jsdifflib
Deleted patch
1
The code where we added the TT instruction was accidentally
2
missing a 'break', which meant that after generating the code
3
to execute the TT we would fall through to 'goto illegal_op'
4
and generate code to take an UNDEF insn.
5
1
6
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
7
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
8
Message-id: 20180206103941.13985-1-peter.maydell@linaro.org
9
---
10
target/arm/translate.c | 1 +
11
1 file changed, 1 insertion(+)
12
13
diff --git a/target/arm/translate.c b/target/arm/translate.c
14
index XXXXXXX..XXXXXXX 100644
15
--- a/target/arm/translate.c
16
+++ b/target/arm/translate.c
17
@@ -XXX,XX +XXX,XX @@ static void disas_thumb2_insn(DisasContext *s, uint32_t insn)
18
tcg_temp_free_i32(addr);
19
tcg_temp_free_i32(op);
20
store_reg(s, rd, ttresp);
21
+ break;
22
}
23
goto illegal_op;
24
}
25
--
26
2.16.1
27
28
diff view generated by jsdifflib