[PULL 00/12] target-arm queue
[PULL v2 00/39] target-arm queue
1
One last arm pullreq before I stop work for the end of the year...
1
v2: fix format-string issue in a test case.
2
2
3
-- PMM
3
-- PMM
4
4
5
The following changes since commit 8e5943260a8f765216674ee87ce8588cc4e7463e:
5
The following changes since commit 6f34661b6c97a37a5efc27d31c037ddeda4547e2:
6
6
7
Merge remote-tracking branch 'remotes/vivier2/tags/trivial-branch-pull-request' into staging (2019-12-20 12:46:10 +0000)
7
Merge remote-tracking branch 'remotes/vivier2/tags/trivial-branch-for-6.0-pull-request' into staging (2021-03-11 18:55:27 +0000)
8
8
9
are available in the Git repository at:
9
are available in the Git repository at:
10
10
11
https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20191220
11
https://git.linaro.org/people/pmaydell/qemu-arm.git tags/pull-target-arm-20210314
12
12
13
for you to fetch changes up to c8fa6079eb35888587f1be27c1590da4edcc5098:
13
for you to fetch changes up to 6500ac13ff8e5c64ca69f5ef5d456028cfda6139:
14
14
15
arm/arm-powerctl: rebuild hflags after setting CP15 bits in arm_set_cpu_on() (2019-12-20 14:03:00 +0000)
15
hw/display/pxa2xx: Inline template header (2021-03-14 13:14:56 +0000)
16
16
17
----------------------------------------------------------------
17
----------------------------------------------------------------
18
target-arm queue:
18
target-arm queue:
19
* Support emulating the generic timers at frequencies other than 62.5MHz
19
* versal: Support XRAMs and XRAM controller
20
* Various fixes for SMMUv3 emulation bugs
20
* smmu: Various minor bug fixes
21
* Improve assert error message for hflags mismatches
21
* SVE emulation: fix bugs handling odd vector lengths
22
* arm-powerctl: rebuild hflags after setting CP15 bits in arm_set_cpu_on()
22
* allwinner-sun8i-emac: traverse transmit queue using TX_CUR_DESC register value
23
* tests/acceptance: fix orangepi-pc acceptance tests
24
* hw/timer/sse-timer: Propagate eventual error in sse_timer_realize()
25
* hw/arm/virt: KVM: The IPA lower bound is 32
26
* npcm7xx: support MFT module
27
* pl110, pxa2xx_lcd: tidy up template headers
23
28
24
----------------------------------------------------------------
29
----------------------------------------------------------------
25
Andrew Jeffery (4):
30
Andrew Jones (2):
26
target/arm: Remove redundant scaling of nexttick
31
accel: kvm: Fix kvm_type invocation
27
target/arm: Abstract the generic timer frequency
32
hw/arm/virt: KVM: The IPA lower bound is 32
28
target/arm: Prepare generic timer for per-platform CNTFRQ
29
ast2600: Configure CNTFRQ at 1125MHz
30
33
31
Niek Linnenbank (1):
34
Edgar E. Iglesias (2):
32
arm/arm-powerctl: rebuild hflags after setting CP15 bits in arm_set_cpu_on()
35
hw/misc: versal: Add a model of the XRAM controller
36
hw/arm: versal: Add support for the XRAMs
37
38
Eric Auger (7):
39
intel_iommu: Fix mask may be uninitialized in vtd_context_device_invalidate
40
dma: Introduce dma_aligned_pow2_mask()
41
virtio-iommu: Handle non power of 2 range invalidations
42
hw/arm/smmu-common: Fix smmu_iotlb_inv_iova when asid is not set
43
hw/arm/smmuv3: Enforce invalidation on a power of two range
44
hw/arm/smmuv3: Fix SMMU_CMD_CFGI_STE_RANGE handling
45
hw/arm/smmuv3: Uniformize sid traces
46
47
Hao Wu (5):
48
hw/misc: Add GPIOs for duty in NPCM7xx PWM
49
hw/misc: Add NPCM7XX MFT Module
50
hw/arm: Add MFT device to NPCM7xx Soc
51
hw/arm: Connect PWM fans in NPCM7XX boards
52
tests/qtest: Test PWM fan RPM using MFT in PWM test
53
54
Niek Linnenbank (5):
55
hw/net/allwinner-sun8i-emac: traverse transmit queue using TX_CUR_DESC register value
56
tests/acceptance/boot_linux_console: remove Armbian 19.11.3 bionic test for orangepi-pc machine
57
tests/acceptance/boot_linux_console: change URL for test_arm_orangepi_bionic_20_08
58
tests/acceptance: update sunxi kernel from armbian to 5.10.16
59
tests/acceptance: drop ARMBIAN_ARTIFACTS_CACHED condition for orangepi-pc, cubieboard tests
60
61
Peter Maydell (9):
62
hw/display/pl110: Remove dead code for non-32-bpp surfaces
63
hw/display/pl110: Pull included-once parts of template header into pl110.c
64
hw/display/pl110: Remove use of BITS from pl110_template.h
65
hw/display/pxa2xx_lcd: Remove dead code for non-32-bpp surfaces
66
hw/display/pxa2xx_lcd: Remove dest_width state field
67
hw/display/pxa2xx: Remove use of BITS in pxa2xx_template.h
68
hw/display/pxa2xx: Apply brace-related coding style fixes to template header
69
hw/display/pxa2xx: Apply whitespace-only coding style fixes to template header
70
hw/display/pxa2xx: Inline template header
33
71
34
Philippe Mathieu-Daudé (1):
72
Philippe Mathieu-Daudé (1):
35
target/arm: Display helpful message when hflags mismatch
73
hw/timer/sse-timer: Propagate eventual error in sse_timer_realize()
36
74
37
Simon Veith (6):
75
Richard Henderson (8):
38
hw/arm/smmuv3: Apply address mask to linear strtab base address
76
target/arm: Fix sve_uzp_p vs odd vector lengths
39
hw/arm/smmuv3: Correct SMMU_BASE_ADDR_MASK value
77
target/arm: Fix sve_zip_p vs odd vector lengths
40
hw/arm/smmuv3: Check stream IDs against actual table LOG2SIZE
78
target/arm: Fix sve_punpk_p vs odd vector lengths
41
hw/arm/smmuv3: Align stream table base address to table size
79
target/arm: Update find_last_active for PREDDESC
42
hw/arm/smmuv3: Use correct bit positions in EVT_SET_ADDR2 macro
80
target/arm: Update BRKA, BRKB, BRKN for PREDDESC
43
hw/arm/smmuv3: Report F_STE_FETCH fault address in correct word position
81
target/arm: Update CNTP for PREDDESC
82
target/arm: Update WHILE for PREDDESC
83
target/arm: Update sve reduction vs simd_desc
44
84
45
hw/arm/smmuv3-internal.h | 6 ++---
85
docs/system/arm/nuvoton.rst | 2 +-
46
target/arm/cpu.h | 5 ++++
86
docs/system/arm/xlnx-versal-virt.rst | 1 +
47
hw/arm/aspeed_ast2600.c | 3 +++
87
hw/arm/smmu-internal.h | 5 +
48
hw/arm/smmuv3.c | 28 +++++++++++++++-----
88
hw/display/pl110_template.h | 120 +-------
49
target/arm/arm-powerctl.c | 3 +++
89
hw/display/pxa2xx_template.h | 447 ---------------------------
50
target/arm/cpu.c | 65 +++++++++++++++++++++++++++++++++++++++++------
90
include/hw/arm/npcm7xx.h | 13 +-
51
target/arm/helper.c | 42 +++++++++++++++++++++++-------
91
include/hw/arm/xlnx-versal.h | 13 +
52
7 files changed, 125 insertions(+), 27 deletions(-)
92
include/hw/boards.h | 1 +
93
include/hw/misc/npcm7xx_mft.h | 70 +++++
94
include/hw/misc/npcm7xx_pwm.h | 4 +-
95
include/hw/misc/xlnx-versal-xramc.h | 97 ++++++
96
include/sysemu/dma.h | 12 +
97
target/arm/kvm_arm.h | 6 +-
98
accel/kvm/kvm-all.c | 2 +
99
hw/arm/npcm7xx.c | 45 ++-
100
hw/arm/npcm7xx_boards.c | 99 ++++++
101
hw/arm/smmu-common.c | 32 +-
102
hw/arm/smmuv3.c | 58 ++--
103
hw/arm/virt.c | 23 +-
104
hw/arm/xlnx-versal.c | 36 +++
105
hw/display/pl110.c | 123 +++++---
106
hw/display/pxa2xx_lcd.c | 520 ++++++++++++++++++++++++++-----
107
hw/i386/intel_iommu.c | 32 +-
108
hw/misc/npcm7xx_mft.c | 540 +++++++++++++++++++++++++++++++++
109
hw/misc/npcm7xx_pwm.c | 4 +
110
hw/misc/xlnx-versal-xramc.c | 253 +++++++++++++++
111
hw/net/allwinner-sun8i-emac.c | 62 ++--
112
hw/timer/sse-timer.c | 1 +
113
hw/virtio/virtio-iommu.c | 19 +-
114
softmmu/dma-helpers.c | 26 ++
115
target/arm/kvm.c | 4 +-
116
target/arm/sve_helper.c | 107 ++++---
117
target/arm/translate-sve.c | 26 +-
118
tests/qtest/npcm7xx_pwm-test.c | 205 ++++++++++++-
119
hw/arm/trace-events | 24 +-
120
hw/misc/meson.build | 2 +
121
hw/misc/trace-events | 8 +
122
tests/acceptance/boot_linux_console.py | 120 +++-----
123
tests/acceptance/replay_kernel.py | 10 +-
124
39 files changed, 2235 insertions(+), 937 deletions(-)
125
delete mode 100644 hw/display/pxa2xx_template.h
126
create mode 100644 include/hw/misc/npcm7xx_mft.h
127
create mode 100644 include/hw/misc/xlnx-versal-xramc.h
128
create mode 100644 hw/misc/npcm7xx_mft.c
129
create mode 100644 hw/misc/xlnx-versal-xramc.c
53
130
diff view generated by jsdifflib
[PULL 01/12] target/arm: Remove redundant scaling of nexttick
Deleted patch
1
From: Andrew Jeffery <andrew@aj.id.au>
2
1
3
The corner-case codepath was adjusting nexttick such that overflow
4
wouldn't occur when timer_mod() scaled the value back up. Remove a use
5
of GTIMER_SCALE and avoid unnecessary operations by calling
6
timer_mod_ns() directly.
7
8
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
9
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
10
Reviewed-by: Cédric Le Goater <clg@kaod.org>
11
Message-id: f8c680720e3abe55476e6d9cb604ad27fdbeb2e0.1576215453.git-series.andrew@aj.id.au
12
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
13
---
14
target/arm/helper.c | 5 +++--
15
1 file changed, 3 insertions(+), 2 deletions(-)
16
17
diff --git a/target/arm/helper.c b/target/arm/helper.c
18
index XXXXXXX..XXXXXXX 100644
19
--- a/target/arm/helper.c
20
+++ b/target/arm/helper.c
21
@@ -XXX,XX +XXX,XX @@ static void gt_recalc_timer(ARMCPU *cpu, int timeridx)
22
* timer expires we will reset the timer for any remaining period.
23
*/
24
if (nexttick > INT64_MAX / GTIMER_SCALE) {
25
- nexttick = INT64_MAX / GTIMER_SCALE;
26
+ timer_mod_ns(cpu->gt_timer[timeridx], INT64_MAX);
27
+ } else {
28
+ timer_mod(cpu->gt_timer[timeridx], nexttick);
29
}
30
- timer_mod(cpu->gt_timer[timeridx], nexttick);
31
trace_arm_gt_recalc(timeridx, irqstate, nexttick);
32
} else {
33
/* Timer disabled: ISTATUS and timer output always clear */
34
--
35
2.20.1
36
37
diff view generated by jsdifflib
[PULL 02/12] target/arm: Abstract the generic timer frequency
Deleted patch
1
From: Andrew Jeffery <andrew@aj.id.au>
2
1
3
Prepare for SoCs such as the ASPEED AST2600 whose firmware configures
4
CNTFRQ to values significantly larger than the static 62.5MHz value
5
currently derived from GTIMER_SCALE. As the OS potentially derives its
6
timer periods from the CNTFRQ value the lack of support for running
7
QEMUTimers at the appropriate rate leads to sticky behaviour in the
8
guest.
9
10
Substitute the GTIMER_SCALE constant with use of a helper to derive the
11
period from gt_cntfrq_hz stored in struct ARMCPU. Initially set
12
gt_cntfrq_hz to the frequency associated with GTIMER_SCALE so current
13
behaviour is maintained.
14
15
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
16
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
17
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
18
Message-id: 40bd8df043f66e1ccfb3e9482999d099ac72bb2e.1576215453.git-series.andrew@aj.id.au
19
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
20
---
21
target/arm/cpu.h | 5 +++++
22
target/arm/cpu.c | 8 ++++++++
23
target/arm/helper.c | 10 +++++++---
24
3 files changed, 20 insertions(+), 3 deletions(-)
25
26
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
27
index XXXXXXX..XXXXXXX 100644
28
--- a/target/arm/cpu.h
29
+++ b/target/arm/cpu.h
30
@@ -XXX,XX +XXX,XX @@ struct ARMCPU {
31
*/
32
DECLARE_BITMAP(sve_vq_map, ARM_MAX_VQ);
33
DECLARE_BITMAP(sve_vq_init, ARM_MAX_VQ);
34
+
35
+ /* Generic timer counter frequency, in Hz */
36
+ uint64_t gt_cntfrq_hz;
37
};
38
39
+unsigned int gt_cntfrq_period_ns(ARMCPU *cpu);
40
+
41
void arm_cpu_post_init(Object *obj);
42
43
uint64_t arm_cpu_mp_affinity(int idx, uint8_t clustersz);
44
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
45
index XXXXXXX..XXXXXXX 100644
46
--- a/target/arm/cpu.c
47
+++ b/target/arm/cpu.c
48
@@ -XXX,XX +XXX,XX @@ static void arm_cpu_initfn(Object *obj)
49
if (tcg_enabled()) {
50
cpu->psci_version = 2; /* TCG implements PSCI 0.2 */
51
}
52
+
53
+ cpu->gt_cntfrq_hz = NANOSECONDS_PER_SECOND / GTIMER_SCALE;
54
}
55
56
static Property arm_cpu_reset_cbar_property =
57
@@ -XXX,XX +XXX,XX @@ static void arm_set_init_svtor(Object *obj, Visitor *v, const char *name,
58
visit_type_uint32(v, name, &cpu->init_svtor, errp);
59
}
60
61
+unsigned int gt_cntfrq_period_ns(ARMCPU *cpu)
62
+{
63
+ return NANOSECONDS_PER_SECOND > cpu->gt_cntfrq_hz ?
64
+ NANOSECONDS_PER_SECOND / cpu->gt_cntfrq_hz : 1;
65
+}
66
+
67
void arm_cpu_post_init(Object *obj)
68
{
69
ARMCPU *cpu = ARM_CPU(obj);
70
diff --git a/target/arm/helper.c b/target/arm/helper.c
71
index XXXXXXX..XXXXXXX 100644
72
--- a/target/arm/helper.c
73
+++ b/target/arm/helper.c
74
@@ -XXX,XX +XXX,XX @@ static CPAccessResult gt_stimer_access(CPUARMState *env,
75
76
static uint64_t gt_get_countervalue(CPUARMState *env)
77
{
78
- return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) / GTIMER_SCALE;
79
+ ARMCPU *cpu = env_archcpu(env);
80
+
81
+ return qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) / gt_cntfrq_period_ns(cpu);
82
}
83
84
static void gt_recalc_timer(ARMCPU *cpu, int timeridx)
85
@@ -XXX,XX +XXX,XX @@ static void gt_recalc_timer(ARMCPU *cpu, int timeridx)
86
* set the timer for as far in the future as possible. When the
87
* timer expires we will reset the timer for any remaining period.
88
*/
89
- if (nexttick > INT64_MAX / GTIMER_SCALE) {
90
+ if (nexttick > INT64_MAX / gt_cntfrq_period_ns(cpu)) {
91
timer_mod_ns(cpu->gt_timer[timeridx], INT64_MAX);
92
} else {
93
timer_mod(cpu->gt_timer[timeridx], nexttick);
94
@@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo generic_timer_cp_reginfo[] = {
95
96
static uint64_t gt_virt_cnt_read(CPUARMState *env, const ARMCPRegInfo *ri)
97
{
98
+ ARMCPU *cpu = env_archcpu(env);
99
+
100
/* Currently we have no support for QEMUTimer in linux-user so we
101
* can't call gt_get_countervalue(env), instead we directly
102
* call the lower level functions.
103
*/
104
- return cpu_get_clock() / GTIMER_SCALE;
105
+ return cpu_get_clock() / gt_cntfrq_period_ns(cpu);
106
}
107
108
static const ARMCPRegInfo generic_timer_cp_reginfo[] = {
109
--
110
2.20.1
111
112
diff view generated by jsdifflib
[PULL 03/12] target/arm: Prepare generic timer for per-platform CNTFRQ
Deleted patch
1
From: Andrew Jeffery <andrew@aj.id.au>
2
1
3
The ASPEED AST2600 clocks the generic timer at the rate of HPLL. On
4
recent firmwares this is at 1125MHz, which is considerably quicker than
5
the assumed 62.5MHz of the current generic timer implementation. The
6
delta between the value as read from CNTFRQ and the true rate of the
7
underlying QEMUTimer leads to sticky behaviour in AST2600 guests.
8
9
Add a feature-gated property exposing CNTFRQ for ARM CPUs providing the
10
generic timer. This allows platforms to configure CNTFRQ (and the
11
associated QEMUTimer) to the appropriate frequency prior to starting the
12
guest.
13
14
As the platform can now determine the rate of CNTFRQ we're exposed to
15
limitations of QEMUTimer that didn't previously materialise: In the
16
course of emulation we need to arbitrarily and accurately convert
17
between guest ticks and time, but we're constrained by QEMUTimer's use
18
of an integer scaling factor. The effect is QEMUTimer cannot exactly
19
capture the period of frequencies that do not cleanly divide
20
NANOSECONDS_PER_SECOND for scaling ticks to time. As such, provide an
21
equally inaccurate scaling factor for scaling time to ticks so at least
22
a self-consistent inverse relationship holds.
23
24
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
25
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
26
Message-id: a22db9325f96e39f76e3c2baddcb712149f46bf2.1576215453.git-series.andrew@aj.id.au
27
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
28
---
29
target/arm/cpu.c | 61 +++++++++++++++++++++++++++++++++++++--------
30
target/arm/helper.c | 9 ++++++-
31
2 files changed, 59 insertions(+), 11 deletions(-)
32
33
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
34
index XXXXXXX..XXXXXXX 100644
35
--- a/target/arm/cpu.c
36
+++ b/target/arm/cpu.c
37
@@ -XXX,XX +XXX,XX @@ static void arm_cpu_initfn(Object *obj)
38
if (tcg_enabled()) {
39
cpu->psci_version = 2; /* TCG implements PSCI 0.2 */
40
}
41
-
42
- cpu->gt_cntfrq_hz = NANOSECONDS_PER_SECOND / GTIMER_SCALE;
43
}
44
45
+static Property arm_cpu_gt_cntfrq_property =
46
+ DEFINE_PROP_UINT64("cntfrq", ARMCPU, gt_cntfrq_hz,
47
+ NANOSECONDS_PER_SECOND / GTIMER_SCALE);
48
+
49
static Property arm_cpu_reset_cbar_property =
50
DEFINE_PROP_UINT64("reset-cbar", ARMCPU, reset_cbar, 0);
51
52
@@ -XXX,XX +XXX,XX @@ static void arm_set_init_svtor(Object *obj, Visitor *v, const char *name,
53
54
unsigned int gt_cntfrq_period_ns(ARMCPU *cpu)
55
{
56
+ /*
57
+ * The exact approach to calculating guest ticks is:
58
+ *
59
+ * muldiv64(qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL), cpu->gt_cntfrq_hz,
60
+ * NANOSECONDS_PER_SECOND);
61
+ *
62
+ * We don't do that. Rather we intentionally use integer division
63
+ * truncation below and in the caller for the conversion of host monotonic
64
+ * time to guest ticks to provide the exact inverse for the semantics of
65
+ * the QEMUTimer scale factor. QEMUTimer's scale facter is an integer, so
66
+ * it loses precision when representing frequencies where
67
+ * `(NANOSECONDS_PER_SECOND % cpu->gt_cntfrq) > 0` holds. Failing to
68
+ * provide an exact inverse leads to scheduling timers with negative
69
+ * periods, which in turn leads to sticky behaviour in the guest.
70
+ *
71
+ * Finally, CNTFRQ is effectively capped at 1GHz to ensure our scale factor
72
+ * cannot become zero.
73
+ */
74
return NANOSECONDS_PER_SECOND > cpu->gt_cntfrq_hz ?
75
NANOSECONDS_PER_SECOND / cpu->gt_cntfrq_hz : 1;
76
}
77
@@ -XXX,XX +XXX,XX @@ void arm_cpu_post_init(Object *obj)
78
79
qdev_property_add_static(DEVICE(obj), &arm_cpu_cfgend_property,
80
&error_abort);
81
+
82
+ if (arm_feature(&cpu->env, ARM_FEATURE_GENERIC_TIMER)) {
83
+ qdev_property_add_static(DEVICE(cpu), &arm_cpu_gt_cntfrq_property,
84
+ &error_abort);
85
+ }
86
}
87
88
static void arm_cpu_finalizefn(Object *obj)
89
@@ -XXX,XX +XXX,XX @@ static void arm_cpu_realizefn(DeviceState *dev, Error **errp)
90
}
91
}
92
93
- cpu->gt_timer[GTIMER_PHYS] = timer_new(QEMU_CLOCK_VIRTUAL, GTIMER_SCALE,
94
- arm_gt_ptimer_cb, cpu);
95
- cpu->gt_timer[GTIMER_VIRT] = timer_new(QEMU_CLOCK_VIRTUAL, GTIMER_SCALE,
96
- arm_gt_vtimer_cb, cpu);
97
- cpu->gt_timer[GTIMER_HYP] = timer_new(QEMU_CLOCK_VIRTUAL, GTIMER_SCALE,
98
- arm_gt_htimer_cb, cpu);
99
- cpu->gt_timer[GTIMER_SEC] = timer_new(QEMU_CLOCK_VIRTUAL, GTIMER_SCALE,
100
- arm_gt_stimer_cb, cpu);
101
+
102
+ {
103
+ uint64_t scale;
104
+
105
+ if (arm_feature(env, ARM_FEATURE_GENERIC_TIMER)) {
106
+ if (!cpu->gt_cntfrq_hz) {
107
+ error_setg(errp, "Invalid CNTFRQ: %"PRId64"Hz",
108
+ cpu->gt_cntfrq_hz);
109
+ return;
110
+ }
111
+ scale = gt_cntfrq_period_ns(cpu);
112
+ } else {
113
+ scale = GTIMER_SCALE;
114
+ }
115
+
116
+ cpu->gt_timer[GTIMER_PHYS] = timer_new(QEMU_CLOCK_VIRTUAL, scale,
117
+ arm_gt_ptimer_cb, cpu);
118
+ cpu->gt_timer[GTIMER_VIRT] = timer_new(QEMU_CLOCK_VIRTUAL, scale,
119
+ arm_gt_vtimer_cb, cpu);
120
+ cpu->gt_timer[GTIMER_HYP] = timer_new(QEMU_CLOCK_VIRTUAL, scale,
121
+ arm_gt_htimer_cb, cpu);
122
+ cpu->gt_timer[GTIMER_SEC] = timer_new(QEMU_CLOCK_VIRTUAL, scale,
123
+ arm_gt_stimer_cb, cpu);
124
+ }
125
#endif
126
127
cpu_exec_realizefn(cs, &local_err);
128
diff --git a/target/arm/helper.c b/target/arm/helper.c
129
index XXXXXXX..XXXXXXX 100644
130
--- a/target/arm/helper.c
131
+++ b/target/arm/helper.c
132
@@ -XXX,XX +XXX,XX @@ void arm_gt_stimer_cb(void *opaque)
133
gt_recalc_timer(cpu, GTIMER_SEC);
134
}
135
136
+static void arm_gt_cntfrq_reset(CPUARMState *env, const ARMCPRegInfo *opaque)
137
+{
138
+ ARMCPU *cpu = env_archcpu(env);
139
+
140
+ cpu->env.cp15.c14_cntfrq = cpu->gt_cntfrq_hz;
141
+}
142
+
143
static const ARMCPRegInfo generic_timer_cp_reginfo[] = {
144
/* Note that CNTFRQ is purely reads-as-written for the benefit
145
* of software; writing it doesn't actually change the timer frequency.
146
@@ -XXX,XX +XXX,XX @@ static const ARMCPRegInfo generic_timer_cp_reginfo[] = {
147
.opc0 = 3, .opc1 = 3, .crn = 14, .crm = 0, .opc2 = 0,
148
.access = PL1_RW | PL0_R, .accessfn = gt_cntfrq_access,
149
.fieldoffset = offsetof(CPUARMState, cp15.c14_cntfrq),
150
- .resetvalue = (1000 * 1000 * 1000) / GTIMER_SCALE,
151
+ .resetfn = arm_gt_cntfrq_reset,
152
},
153
/* overall control: mostly access permissions */
154
{ .name = "CNTKCTL", .state = ARM_CP_STATE_BOTH,
155
--
156
2.20.1
157
158
diff view generated by jsdifflib
[PULL 04/12] ast2600: Configure CNTFRQ at 1125MHz
Deleted patch
1
From: Andrew Jeffery <andrew@aj.id.au>
2
1
3
This matches the configuration set by u-boot on the AST2600.
4
5
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
6
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
7
Reviewed-by: Cédric Le Goater <clg@kaod.org>
8
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
9
Message-id: 080ca1267a09381c43cf3c50d434fb6c186f2b6e.1576215453.git-series.andrew@aj.id.au
10
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
11
---
12
hw/arm/aspeed_ast2600.c | 3 +++
13
1 file changed, 3 insertions(+)
14
15
diff --git a/hw/arm/aspeed_ast2600.c b/hw/arm/aspeed_ast2600.c
16
index XXXXXXX..XXXXXXX 100644
17
--- a/hw/arm/aspeed_ast2600.c
18
+++ b/hw/arm/aspeed_ast2600.c
19
@@ -XXX,XX +XXX,XX @@ static void aspeed_soc_ast2600_realize(DeviceState *dev, Error **errp)
20
object_property_set_int(OBJECT(&s->cpu[i]), aspeed_calc_affinity(i),
21
"mp-affinity", &error_abort);
22
23
+ object_property_set_int(OBJECT(&s->cpu[i]), 1125000000, "cntfrq",
24
+ &error_abort);
25
+
26
/*
27
* TODO: the secondary CPUs are started and a boot helper
28
* is needed when using -kernel
29
--
30
2.20.1
31
32
diff view generated by jsdifflib
[PULL 05/12] hw/arm/smmuv3: Apply address mask to linear strtab base address
Deleted patch
1
From: Simon Veith <sveith@amazon.de>
2
1
3
In the SMMU_STRTAB_BASE register, the stream table base address only
4
occupies bits [51:6]. Other bits, such as RA (bit [62]), must be masked
5
out to obtain the base address.
6
7
The branch for 2-level stream tables correctly applies this mask by way
8
of SMMU_BASE_ADDR_MASK, but the one for linear stream tables does not.
9
10
Apply the missing mask in that case as well so that the correct stream
11
base address is used by guests which configure a linear stream table.
12
13
Linux guests are unaffected by this change because they choose a 2-level
14
stream table layout for the QEMU SMMUv3, based on the size of its stream
15
ID space.
16
17
ref. ARM IHI 0070C, section 6.3.23.
18
19
Signed-off-by: Simon Veith <sveith@amazon.de>
20
Acked-by: Eric Auger <eric.auger@redhat.com>
21
Tested-by: Eric Auger <eric.auger@redhat.com>
22
Message-id: 1576509312-13083-2-git-send-email-sveith@amazon.de
23
Cc: Eric Auger <eric.auger@redhat.com>
24
Cc: qemu-devel@nongnu.org
25
Cc: qemu-arm@nongnu.org
26
Acked-by: Eric Auger <eric.auger@redhat.com>
27
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
28
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
29
---
30
hw/arm/smmuv3.c | 2 +-
31
1 file changed, 1 insertion(+), 1 deletion(-)
32
33
diff --git a/hw/arm/smmuv3.c b/hw/arm/smmuv3.c
34
index XXXXXXX..XXXXXXX 100644
35
--- a/hw/arm/smmuv3.c
36
+++ b/hw/arm/smmuv3.c
37
@@ -XXX,XX +XXX,XX @@ static int smmu_find_ste(SMMUv3State *s, uint32_t sid, STE *ste,
38
}
39
addr = l2ptr + l2_ste_offset * sizeof(*ste);
40
} else {
41
- addr = s->strtab_base + sid * sizeof(*ste);
42
+ addr = (s->strtab_base & SMMU_BASE_ADDR_MASK) + sid * sizeof(*ste);
43
}
44
45
if (smmu_get_ste(s, addr, ste, event)) {
46
--
47
2.20.1
48
49
diff view generated by jsdifflib
[PULL 06/12] hw/arm/smmuv3: Correct SMMU_BASE_ADDR_MASK value
Deleted patch
1
From: Simon Veith <sveith@amazon.de>
2
1
3
There are two issues with the current value of SMMU_BASE_ADDR_MASK:
4
5
- At the lower end, we are clearing bits [4:0]. Per the SMMUv3 spec,
6
we should also be treating bit 5 as zero in the base address.
7
- At the upper end, we are clearing bits [63:48]. Per the SMMUv3 spec,
8
only bits [63:52] must be explicitly treated as zero.
9
10
Update the SMMU_BASE_ADDR_MASK value to mask out bits [63:52] and [5:0].
11
12
ref. ARM IHI 0070C, section 6.3.23.
13
14
Signed-off-by: Simon Veith <sveith@amazon.de>
15
Acked-by: Eric Auger <eric.auger@redhat.com>
16
Tested-by: Eric Auger <eric.auger@redhat.com>
17
Message-id: 1576509312-13083-3-git-send-email-sveith@amazon.de
18
Cc: Eric Auger <eric.auger@redhat.com>
19
Cc: qemu-devel@nongnu.org
20
Cc: qemu-arm@nongnu.org
21
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
22
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
23
---
24
hw/arm/smmuv3-internal.h | 2 +-
25
1 file changed, 1 insertion(+), 1 deletion(-)
26
27
diff --git a/hw/arm/smmuv3-internal.h b/hw/arm/smmuv3-internal.h
28
index XXXXXXX..XXXXXXX 100644
29
--- a/hw/arm/smmuv3-internal.h
30
+++ b/hw/arm/smmuv3-internal.h
31
@@ -XXX,XX +XXX,XX @@ REG32(GERROR_IRQ_CFG2, 0x74)
32
33
#define A_STRTAB_BASE 0x80 /* 64b */
34
35
-#define SMMU_BASE_ADDR_MASK 0xffffffffffe0
36
+#define SMMU_BASE_ADDR_MASK 0xfffffffffffc0
37
38
REG32(STRTAB_BASE_CFG, 0x88)
39
FIELD(STRTAB_BASE_CFG, FMT, 16, 2)
40
--
41
2.20.1
42
43
diff view generated by jsdifflib
[PULL 07/12] hw/arm/smmuv3: Check stream IDs against actual table LOG2SIZE
Deleted patch
1
From: Simon Veith <sveith@amazon.de>
2
1
3
When checking whether a stream ID is in range of the stream table, we
4
have so far been only checking it against our implementation limit
5
(SMMU_IDR1_SIDSIZE). However, the guest can program the
6
STRTAB_BASE_CFG.LOG2SIZE field to a size that is smaller than this
7
limit.
8
9
Check the stream ID against this limit as well to match the hardware
10
behavior of raising C_BAD_STREAMID events in case the limit is exceeded.
11
Also, ensure that we do not go one entry beyond the end of the table by
12
checking that its index is strictly smaller than the table size.
13
14
ref. ARM IHI 0070C, section 6.3.24.
15
16
Signed-off-by: Simon Veith <sveith@amazon.de>
17
Acked-by: Eric Auger <eric.auger@redhat.com>
18
Tested-by: Eric Auger <eric.auger@redhat.com>
19
Message-id: 1576509312-13083-4-git-send-email-sveith@amazon.de
20
Cc: Eric Auger <eric.auger@redhat.com>
21
Cc: qemu-devel@nongnu.org
22
Cc: qemu-arm@nongnu.org
23
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
24
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
25
---
26
hw/arm/smmuv3.c | 8 ++++++--
27
1 file changed, 6 insertions(+), 2 deletions(-)
28
29
diff --git a/hw/arm/smmuv3.c b/hw/arm/smmuv3.c
30
index XXXXXXX..XXXXXXX 100644
31
--- a/hw/arm/smmuv3.c
32
+++ b/hw/arm/smmuv3.c
33
@@ -XXX,XX +XXX,XX @@ static int smmu_find_ste(SMMUv3State *s, uint32_t sid, STE *ste,
34
SMMUEventInfo *event)
35
{
36
dma_addr_t addr;
37
+ uint32_t log2size;
38
int ret;
39
40
trace_smmuv3_find_ste(sid, s->features, s->sid_split);
41
- /* Check SID range */
42
- if (sid > (1 << SMMU_IDR1_SIDSIZE)) {
43
+ log2size = FIELD_EX32(s->strtab_base_cfg, STRTAB_BASE_CFG, LOG2SIZE);
44
+ /*
45
+ * Check SID range against both guest-configured and implementation limits
46
+ */
47
+ if (sid >= (1 << MIN(log2size, SMMU_IDR1_SIDSIZE))) {
48
event->type = SMMU_EVT_C_BAD_STREAMID;
49
return -EINVAL;
50
}
51
--
52
2.20.1
53
54
diff view generated by jsdifflib
[PULL 08/12] hw/arm/smmuv3: Align stream table base address to table size
Deleted patch
1
From: Simon Veith <sveith@amazon.de>
2
1
3
Per the specification, and as observed in hardware, the SMMUv3 aligns
4
the SMMU_STRTAB_BASE address to the size of the table by masking out the
5
respective least significant bits in the ADDR field.
6
7
Apply this masking logic to our smmu_find_ste() lookup function per the
8
specification.
9
10
ref. ARM IHI 0070C, section 6.3.23.
11
12
Signed-off-by: Simon Veith <sveith@amazon.de>
13
Acked-by: Eric Auger <eric.auger@redhat.com>
14
Tested-by: Eric Auger <eric.auger@redhat.com>
15
Message-id: 1576509312-13083-5-git-send-email-sveith@amazon.de
16
Cc: Eric Auger <eric.auger@redhat.com>
17
Cc: qemu-devel@nongnu.org
18
Cc: qemu-arm@nongnu.org
19
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
20
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
21
---
22
hw/arm/smmuv3.c | 18 ++++++++++++++----
23
1 file changed, 14 insertions(+), 4 deletions(-)
24
25
diff --git a/hw/arm/smmuv3.c b/hw/arm/smmuv3.c
26
index XXXXXXX..XXXXXXX 100644
27
--- a/hw/arm/smmuv3.c
28
+++ b/hw/arm/smmuv3.c
29
@@ -XXX,XX +XXX,XX @@ bad_ste:
30
static int smmu_find_ste(SMMUv3State *s, uint32_t sid, STE *ste,
31
SMMUEventInfo *event)
32
{
33
- dma_addr_t addr;
34
+ dma_addr_t addr, strtab_base;
35
uint32_t log2size;
36
+ int strtab_size_shift;
37
int ret;
38
39
trace_smmuv3_find_ste(sid, s->features, s->sid_split);
40
@@ -XXX,XX +XXX,XX @@ static int smmu_find_ste(SMMUv3State *s, uint32_t sid, STE *ste,
41
}
42
if (s->features & SMMU_FEATURE_2LVL_STE) {
43
int l1_ste_offset, l2_ste_offset, max_l2_ste, span;
44
- dma_addr_t strtab_base, l1ptr, l2ptr;
45
+ dma_addr_t l1ptr, l2ptr;
46
STEDesc l1std;
47
48
- strtab_base = s->strtab_base & SMMU_BASE_ADDR_MASK;
49
+ /*
50
+ * Align strtab base address to table size. For this purpose, assume it
51
+ * is not bounded by SMMU_IDR1_SIDSIZE.
52
+ */
53
+ strtab_size_shift = MAX(5, (int)log2size - s->sid_split - 1 + 3);
54
+ strtab_base = s->strtab_base & SMMU_BASE_ADDR_MASK &
55
+ ~MAKE_64BIT_MASK(0, strtab_size_shift);
56
l1_ste_offset = sid >> s->sid_split;
57
l2_ste_offset = sid & ((1 << s->sid_split) - 1);
58
l1ptr = (dma_addr_t)(strtab_base + l1_ste_offset * sizeof(l1std));
59
@@ -XXX,XX +XXX,XX @@ static int smmu_find_ste(SMMUv3State *s, uint32_t sid, STE *ste,
60
}
61
addr = l2ptr + l2_ste_offset * sizeof(*ste);
62
} else {
63
- addr = (s->strtab_base & SMMU_BASE_ADDR_MASK) + sid * sizeof(*ste);
64
+ strtab_size_shift = log2size + 5;
65
+ strtab_base = s->strtab_base & SMMU_BASE_ADDR_MASK &
66
+ ~MAKE_64BIT_MASK(0, strtab_size_shift);
67
+ addr = strtab_base + sid * sizeof(*ste);
68
}
69
70
if (smmu_get_ste(s, addr, ste, event)) {
71
--
72
2.20.1
73
74
diff view generated by jsdifflib
[PULL 09/12] hw/arm/smmuv3: Use correct bit positions in EVT_SET_ADDR2 macro
Deleted patch
1
From: Simon Veith <sveith@amazon.de>
2
1
3
The bit offsets in the EVT_SET_ADDR2 macro do not match those specified
4
in the ARM SMMUv3 Architecture Specification. In all events that use
5
this macro, e.g. F_WALK_EABT, the faulting fetch address or IPA actually
6
occupies the 32-bit words 6 and 7 in the event record contiguously, with
7
the upper and lower unused bits clear due to alignment or maximum
8
supported address bits. How many bits are clear depends on the
9
individual event type.
10
11
Update the macro to write to the correct words in the event record so
12
that guest drivers can obtain accurate address information on events.
13
14
ref. ARM IHI 0070C, sections 7.3.12 through 7.3.16.
15
16
Signed-off-by: Simon Veith <sveith@amazon.de>
17
Acked-by: Eric Auger <eric.auger@redhat.com>
18
Tested-by: Eric Auger <eric.auger@redhat.com>
19
Message-id: 1576509312-13083-6-git-send-email-sveith@amazon.de
20
Cc: Eric Auger <eric.auger@redhat.com>
21
Cc: qemu-devel@nongnu.org
22
Cc: qemu-arm@nongnu.org
23
Acked-by: Eric Auger <eric.auger@redhat.com>
24
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
25
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
26
---
27
hw/arm/smmuv3-internal.h | 4 ++--
28
1 file changed, 2 insertions(+), 2 deletions(-)
29
30
diff --git a/hw/arm/smmuv3-internal.h b/hw/arm/smmuv3-internal.h
31
index XXXXXXX..XXXXXXX 100644
32
--- a/hw/arm/smmuv3-internal.h
33
+++ b/hw/arm/smmuv3-internal.h
34
@@ -XXX,XX +XXX,XX @@ typedef struct SMMUEventInfo {
35
} while (0)
36
#define EVT_SET_ADDR2(x, addr) \
37
do { \
38
- (x)->word[7] = deposit32((x)->word[7], 3, 29, addr >> 16); \
39
- (x)->word[7] = deposit32((x)->word[7], 0, 16, addr & 0xffff);\
40
+ (x)->word[7] = (uint32_t)(addr >> 32); \
41
+ (x)->word[6] = (uint32_t)(addr & 0xffffffff); \
42
} while (0)
43
44
void smmuv3_record_event(SMMUv3State *s, SMMUEventInfo *event);
45
--
46
2.20.1
47
48
diff view generated by jsdifflib
[PULL 10/12] hw/arm/smmuv3: Report F_STE_FETCH fault address in correct word position
Deleted patch
1
From: Simon Veith <sveith@amazon.de>
2
1
3
The smmuv3_record_event() function that generates the F_STE_FETCH error
4
uses the EVT_SET_ADDR macro to record the fetch address, placing it in
5
32-bit words 4 and 5.
6
7
The correct position for this address is in words 6 and 7, per the
8
SMMUv3 Architecture Specification.
9
10
Update the function to use the EVT_SET_ADDR2 macro instead, which is the
11
macro intended for writing to these words.
12
13
ref. ARM IHI 0070C, section 7.3.4.
14
15
Signed-off-by: Simon Veith <sveith@amazon.de>
16
Acked-by: Eric Auger <eric.auger@redhat.com>
17
Tested-by: Eric Auger <eric.auger@redhat.com>
18
Message-id: 1576509312-13083-7-git-send-email-sveith@amazon.de
19
Cc: Eric Auger <eric.auger@redhat.com>
20
Cc: qemu-devel@nongnu.org
21
Cc: qemu-arm@nongnu.org
22
Acked-by: Eric Auger <eric.auger@redhat.com>
23
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
24
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
25
---
26
hw/arm/smmuv3.c | 2 +-
27
1 file changed, 1 insertion(+), 1 deletion(-)
28
29
diff --git a/hw/arm/smmuv3.c b/hw/arm/smmuv3.c
30
index XXXXXXX..XXXXXXX 100644
31
--- a/hw/arm/smmuv3.c
32
+++ b/hw/arm/smmuv3.c
33
@@ -XXX,XX +XXX,XX @@ void smmuv3_record_event(SMMUv3State *s, SMMUEventInfo *info)
34
case SMMU_EVT_F_STE_FETCH:
35
EVT_SET_SSID(&evt, info->u.f_ste_fetch.ssid);
36
EVT_SET_SSV(&evt, info->u.f_ste_fetch.ssv);
37
- EVT_SET_ADDR(&evt, info->u.f_ste_fetch.addr);
38
+ EVT_SET_ADDR2(&evt, info->u.f_ste_fetch.addr);
39
break;
40
case SMMU_EVT_C_BAD_STE:
41
EVT_SET_SSID(&evt, info->u.c_bad_ste.ssid);
42
--
43
2.20.1
44
45
diff view generated by jsdifflib
[PULL 11/12] target/arm: Display helpful message when hflags mismatch
Deleted patch
1
From: Philippe Mathieu-Daudé <philmd@redhat.com>
2
1
3
Instead of crashing in a confuse way, give some hint to the user
4
about why we aborted. He might report the issue without having
5
to use a debugger.
6
7
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
8
Message-id: 20191209134552.27733-1-philmd@redhat.com
9
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
10
Tested-by: Niek Linnenbank <nieklinnenbank@gmail.com>
11
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
12
---
13
target/arm/helper.c | 18 +++++++++++++++---
14
1 file changed, 15 insertions(+), 3 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 @@ void HELPER(rebuild_hflags_a64)(CPUARMState *env, int el)
21
env->hflags = rebuild_hflags_a64(env, el, fp_el, mmu_idx);
22
}
23
24
+static inline void assert_hflags_rebuild_correctly(CPUARMState *env)
25
+{
26
+#ifdef CONFIG_DEBUG_TCG
27
+ uint32_t env_flags_current = env->hflags;
28
+ uint32_t env_flags_rebuilt = rebuild_hflags_internal(env);
29
+
30
+ if (unlikely(env_flags_current != env_flags_rebuilt)) {
31
+ fprintf(stderr, "TCG hflags mismatch (current:0x%08x rebuilt:0x%08x)\n",
32
+ env_flags_current, env_flags_rebuilt);
33
+ abort();
34
+ }
35
+#endif
36
+}
37
+
38
void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc,
39
target_ulong *cs_base, uint32_t *pflags)
40
{
41
@@ -XXX,XX +XXX,XX @@ void cpu_get_tb_cpu_state(CPUARMState *env, target_ulong *pc,
42
uint32_t pstate_for_ss;
43
44
*cs_base = 0;
45
-#ifdef CONFIG_DEBUG_TCG
46
- assert(flags == rebuild_hflags_internal(env));
47
-#endif
48
+ assert_hflags_rebuild_correctly(env);
49
50
if (FIELD_EX32(flags, TBFLAG_ANY, AARCH64_STATE)) {
51
*pc = env->pc;
52
--
53
2.20.1
54
55
diff view generated by jsdifflib
[PULL 12/12] arm/arm-powerctl: rebuild hflags after setting CP15 bits in arm_set_cpu_on()
Deleted patch
1
From: Niek Linnenbank <nieklinnenbank@gmail.com>
2
1
3
After setting CP15 bits in arm_set_cpu_on() the cached hflags must
4
be rebuild to reflect the changed processor state. Without rebuilding,
5
the cached hflags would be inconsistent until the next call to
6
arm_rebuild_hflags(). When QEMU is compiled with debugging enabled
7
(--enable-debug), this problem is captured shortly after the first
8
call to arm_set_cpu_on() for CPUs running in ARM 32-bit non-secure mode:
9
10
qemu-system-arm: target/arm/helper.c:11359: cpu_get_tb_cpu_state:
11
Assertion `flags == rebuild_hflags_internal(env)' failed.
12
Aborted (core dumped)
13
14
Fixes: 0c7f8c43daf65
15
Cc: qemu-stable@nongnu.org
16
Signed-off-by: Niek Linnenbank <nieklinnenbank@gmail.com>
17
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
18
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
19
---
20
target/arm/arm-powerctl.c | 3 +++
21
1 file changed, 3 insertions(+)
22
23
diff --git a/target/arm/arm-powerctl.c b/target/arm/arm-powerctl.c
24
index XXXXXXX..XXXXXXX 100644
25
--- a/target/arm/arm-powerctl.c
26
+++ b/target/arm/arm-powerctl.c
27
@@ -XXX,XX +XXX,XX @@ static void arm_set_cpu_on_async_work(CPUState *target_cpu_state,
28
target_cpu->env.regs[0] = info->context_id;
29
}
30
31
+ /* CP15 update requires rebuilding hflags */
32
+ arm_rebuild_hflags(&target_cpu->env);
33
+
34
/* Start the new CPU at the requested address */
35
cpu_set_pc(target_cpu_state, info->entry);
36
37
--
38
2.20.1
39
40
diff view generated by jsdifflib

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.