[PATCH v2 00/16] hw/microblaze: Allow running cross-endian vCPUs

[PATCH v2 00/16] hw/microblaze: Allow running cross-endian vCPUs

[Philippe Mathieu-Daudé]

Since v1:
- Make device endianness configurable (Edgar)
- Convert more Xilinx devices
- Avoid preprocessor #if (Richard)
- Add R-b tags

Make machines endianness-agnostic, allowing to run a big-endian vCPU
on the little-endian 'qemu-system-microblazeel' binary, and a little
endian one on the big-endian 'qemu-system-microblaze' binary.

Tests added, following combinations covered:
- little-endian vCPU using little-endian binary (in-tree)
- little-endian vCPU using big-endian binary (new)
- big-endian vCPU using little-endian binary (new)
- big-endian vCPU using big-endian binary (in-tree)

To make a target endian-agnostic we need to remove the MO_TE uses.
In order to do that, we propagate the MemOp from earlier in the
call stack, or we extract it from the vCPU env (on MicroBlaze the
CPU endianness is exposed by the 'ENDI' bit).

Next step: Look at unifying binaries.

Please review,

Phil.

Philippe Mathieu-Daudé (16):
  hw/microblaze: Restrict MemoryRegionOps are implemented as 32-bit
  hw/microblaze: Propagate CPU endianness to microblaze_load_kernel()
  hw/intc/xilinx_intc: Make device endianness configurable
  RFC hw/net/xilinx_ethlite: Simplify by having configurable endianness
  RFC hw/timer/xilinx_timer: Allow down to 8-bit memory access
  hw/timer/xilinx_timer: Make device endianness configurable
  hw/char/xilinx_uartlite: Make device endianness configurable
  hw/ssi/xilinx_spi: Make device endianness configurable
  hw/ssi/xilinx_spips: Make device endianness configurable
  target/microblaze: Explode MO_TExx -> MO_TE | MO_xx
  target/microblaze: Set MO_TE once in do_load() / do_store()
  target/microblaze: Introduce mo_endian() helper
  target/microblaze: Consider endianness while translating code
  hw/microblaze: Support various endianness for s3adsp1800 machines
  tests/functional: Explicit endianness of microblaze assets
  tests/functional: Add microblaze cross-endianness tests

 hw/microblaze/boot.h                          |  4 +-
 include/hw/ssi/xilinx_spips.h                 |  1 +
 target/microblaze/cpu.h                       |  7 +++
 hw/arm/xilinx_zynq.c                          |  1 +
 hw/arm/xlnx-zynqmp.c                          |  4 ++
 hw/char/xilinx_uartlite.c                     | 40 +++++++++----
 hw/intc/xilinx_intc.c                         | 59 +++++++++++++++----
 hw/microblaze/boot.c                          |  8 +--
 hw/microblaze/petalogix_ml605_mmu.c           |  4 +-
 hw/microblaze/petalogix_s3adsp1800_mmu.c      | 59 ++++++++++++++++---
 hw/microblaze/xlnx-zynqmp-pmu.c               |  2 +-
 hw/net/xilinx_ethlite.c                       | 52 ++++++++++++----
 hw/ppc/virtex_ml507.c                         |  1 +
 hw/ssi/xilinx_spi.c                           | 29 ++++++---
 hw/ssi/xilinx_spips.c                         | 46 +++++++++++----
 hw/timer/xilinx_timer.c                       | 42 ++++++++++---
 target/microblaze/translate.c                 | 49 +++++++++------
 .../functional/test_microblaze_s3adsp1800.py  | 27 ++++++++-
 .../test_microblazeel_s3adsp1800.py           | 25 +++++++-
 19 files changed, 358 insertions(+), 102 deletions(-)

-- 
2.45.2

[PATCH v2 01/16] hw/microblaze: Restrict MemoryRegionOps are implemented as 32-bit

[Philippe Mathieu-Daudé]

All these MemoryRegionOps read() and write() handlers are
implemented expecting 32-bit accesses. Clarify that setting
.impl.min/max_access_size fields.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Message-Id: <20241105130431.22564-8-philmd@linaro.org>
---
 hw/char/xilinx_uartlite.c | 4 ++++
 hw/intc/xilinx_intc.c     | 4 ++++
 hw/net/xilinx_ethlite.c   | 4 ++++
 hw/timer/xilinx_timer.c   | 4 ++++
 4 files changed, 16 insertions(+)

diff --git a/hw/char/xilinx_uartlite.c b/hw/char/xilinx_uartlite.c
index f325084f8b..3022b3d8ef 100644
--- a/hw/char/xilinx_uartlite.c
+++ b/hw/char/xilinx_uartlite.c
@@ -170,6 +170,10 @@ static const MemoryRegionOps uart_ops = {
     .read = uart_read,
     .write = uart_write,
     .endianness = DEVICE_NATIVE_ENDIAN,
+    .impl = {
+        .min_access_size = 4,
+        .max_access_size = 4,
+    },
     .valid = {
         .min_access_size = 1,
         .max_access_size = 4
diff --git a/hw/intc/xilinx_intc.c b/hw/intc/xilinx_intc.c
index 6e5012e66e..8fb6b4f1a5 100644
--- a/hw/intc/xilinx_intc.c
+++ b/hw/intc/xilinx_intc.c
@@ -144,6 +144,10 @@ static const MemoryRegionOps pic_ops = {
     .read = pic_read,
     .write = pic_write,
     .endianness = DEVICE_NATIVE_ENDIAN,
+    .impl = {
+        .min_access_size = 4,
+        .max_access_size = 4,
+    },
     .valid = {
         .min_access_size = 4,
         .max_access_size = 4
diff --git a/hw/net/xilinx_ethlite.c b/hw/net/xilinx_ethlite.c
index bd81290808..e84b4cdd35 100644
--- a/hw/net/xilinx_ethlite.c
+++ b/hw/net/xilinx_ethlite.c
@@ -170,6 +170,10 @@ static const MemoryRegionOps eth_ops = {
     .read = eth_read,
     .write = eth_write,
     .endianness = DEVICE_NATIVE_ENDIAN,
+    .impl = {
+        .min_access_size = 4,
+        .max_access_size = 4,
+    },
     .valid = {
         .min_access_size = 4,
         .max_access_size = 4
diff --git a/hw/timer/xilinx_timer.c b/hw/timer/xilinx_timer.c
index 32a9df69e0..383fc8b3c8 100644
--- a/hw/timer/xilinx_timer.c
+++ b/hw/timer/xilinx_timer.c
@@ -193,6 +193,10 @@ static const MemoryRegionOps timer_ops = {
     .read = timer_read,
     .write = timer_write,
     .endianness = DEVICE_NATIVE_ENDIAN,
+    .impl = {
+        .min_access_size = 4,
+        .max_access_size = 4,
+    },
     .valid = {
         .min_access_size = 4,
         .max_access_size = 4
-- 
2.45.2

[PATCH v2 02/16] hw/microblaze: Propagate CPU endianness to microblaze_load_kernel()

[Philippe Mathieu-Daudé]

Pass vCPU endianness as argument so we can load kernels
with different endianness (different from the qemu-system-binary
builtin one).

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Message-Id: <20241105130431.22564-9-philmd@linaro.org>
---
 hw/microblaze/boot.h                     | 4 ++--
 hw/microblaze/boot.c                     | 8 ++++----
 hw/microblaze/petalogix_ml605_mmu.c      | 2 +-
 hw/microblaze/petalogix_s3adsp1800_mmu.c | 2 +-
 hw/microblaze/xlnx-zynqmp-pmu.c          | 2 +-
 5 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/hw/microblaze/boot.h b/hw/microblaze/boot.h
index 5a8c2f7975..d179a551a6 100644
--- a/hw/microblaze/boot.h
+++ b/hw/microblaze/boot.h
@@ -2,8 +2,8 @@
 #define MICROBLAZE_BOOT_H
 
 
-void microblaze_load_kernel(MicroBlazeCPU *cpu, hwaddr ddr_base,
-                            uint32_t ramsize,
+void microblaze_load_kernel(MicroBlazeCPU *cpu, bool is_little_endian,
+                            hwaddr ddr_base, uint32_t ramsize,
                             const char *initrd_filename,
                             const char *dtb_filename,
                             void (*machine_cpu_reset)(MicroBlazeCPU *));
diff --git a/hw/microblaze/boot.c b/hw/microblaze/boot.c
index ed61e483ee..3675489fa5 100644
--- a/hw/microblaze/boot.c
+++ b/hw/microblaze/boot.c
@@ -114,8 +114,8 @@ static uint64_t translate_kernel_address(void *opaque, uint64_t addr)
     return addr - 0x30000000LL;
 }
 
-void microblaze_load_kernel(MicroBlazeCPU *cpu, hwaddr ddr_base,
-                            uint32_t ramsize,
+void microblaze_load_kernel(MicroBlazeCPU *cpu, bool is_little_endian,
+                            hwaddr ddr_base, uint32_t ramsize,
                             const char *initrd_filename,
                             const char *dtb_filename,
                             void (*machine_cpu_reset)(MicroBlazeCPU *))
@@ -144,13 +144,13 @@ void microblaze_load_kernel(MicroBlazeCPU *cpu, hwaddr ddr_base,
         /* Boots a kernel elf binary.  */
         kernel_size = load_elf(kernel_filename, NULL, NULL, NULL,
                                &entry, NULL, &high, NULL,
-                               TARGET_BIG_ENDIAN, EM_MICROBLAZE, 0, 0);
+                               !is_little_endian, EM_MICROBLAZE, 0, 0);
         base32 = entry;
         if (base32 == 0xc0000000) {
             kernel_size = load_elf(kernel_filename, NULL,
                                    translate_kernel_address, NULL,
                                    &entry, NULL, NULL, NULL,
-                                   TARGET_BIG_ENDIAN, EM_MICROBLAZE, 0, 0);
+                                   !is_little_endian, EM_MICROBLAZE, 0, 0);
         }
         /* Always boot into physical ram.  */
         boot_info.bootstrap_pc = (uint32_t)entry;
diff --git a/hw/microblaze/petalogix_ml605_mmu.c b/hw/microblaze/petalogix_ml605_mmu.c
index 61e47d8398..d2b2109065 100644
--- a/hw/microblaze/petalogix_ml605_mmu.c
+++ b/hw/microblaze/petalogix_ml605_mmu.c
@@ -204,7 +204,7 @@ petalogix_ml605_init(MachineState *machine)
     cpu->cfg.pvr_regs[5] = 0xc56be000;
     cpu->cfg.pvr_regs[10] = 0x0e000000; /* virtex 6 */
 
-    microblaze_load_kernel(cpu, MEMORY_BASEADDR, ram_size,
+    microblaze_load_kernel(cpu, true, MEMORY_BASEADDR, ram_size,
                            machine->initrd_filename,
                            BINARY_DEVICE_TREE_FILE,
                            NULL);
diff --git a/hw/microblaze/petalogix_s3adsp1800_mmu.c b/hw/microblaze/petalogix_s3adsp1800_mmu.c
index 6c0f5c6c65..8110be8371 100644
--- a/hw/microblaze/petalogix_s3adsp1800_mmu.c
+++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
@@ -129,7 +129,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
 
     create_unimplemented_device("xps_gpio", GPIO_BASEADDR, 0x10000);
 
-    microblaze_load_kernel(cpu, ddr_base, ram_size,
+    microblaze_load_kernel(cpu, !TARGET_BIG_ENDIAN, ddr_base, ram_size,
                            machine->initrd_filename,
                            BINARY_DEVICE_TREE_FILE,
                            NULL);
diff --git a/hw/microblaze/xlnx-zynqmp-pmu.c b/hw/microblaze/xlnx-zynqmp-pmu.c
index 567aad47bf..bdbf7328bf 100644
--- a/hw/microblaze/xlnx-zynqmp-pmu.c
+++ b/hw/microblaze/xlnx-zynqmp-pmu.c
@@ -172,7 +172,7 @@ static void xlnx_zynqmp_pmu_init(MachineState *machine)
     qdev_realize(DEVICE(pmu), NULL, &error_fatal);
 
     /* Load the kernel */
-    microblaze_load_kernel(&pmu->cpu, XLNX_ZYNQMP_PMU_RAM_ADDR,
+    microblaze_load_kernel(&pmu->cpu, true, XLNX_ZYNQMP_PMU_RAM_ADDR,
                            machine->ram_size,
                            machine->initrd_filename,
                            machine->dtb,
-- 
2.45.2

[PATCH v2 03/16] hw/intc/xilinx_intc: Make device endianness configurable

[Philippe Mathieu-Daudé]

Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
Add the "little-endian" property to select the device
endianness, defaulting to little endian.
Set the proper endianness for each machine using the device.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 hw/intc/xilinx_intc.c                    | 61 +++++++++++++++++++-----
 hw/microblaze/petalogix_ml605_mmu.c      |  1 +
 hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
 3 files changed, 50 insertions(+), 13 deletions(-)

diff --git a/hw/intc/xilinx_intc.c b/hw/intc/xilinx_intc.c
index 8fb6b4f1a5..4fffde961a 100644
--- a/hw/intc/xilinx_intc.c
+++ b/hw/intc/xilinx_intc.c
@@ -3,6 +3,9 @@
  *
  * Copyright (c) 2009 Edgar E. Iglesias.
  *
+ * https://docs.amd.com/v/u/en-US/xps_intc
+ * DS572: LogiCORE IP XPS Interrupt Controller (v2.01a)
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
  * in the Software without restriction, including without limitation the rights
@@ -49,6 +52,7 @@ struct XpsIntc
 {
     SysBusDevice parent_obj;
 
+    bool little_endian_model;
     MemoryRegion mmio;
     qemu_irq parent_irq;
 
@@ -140,17 +144,39 @@ static void pic_write(void *opaque, hwaddr addr,
     update_irq(p);
 }
 
-static const MemoryRegionOps pic_ops = {
-    .read = pic_read,
-    .write = pic_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .impl = {
-        .min_access_size = 4,
-        .max_access_size = 4,
+static const MemoryRegionOps pic_ops[2] = {
+    {
+        .read = pic_read,
+        .write = pic_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            /*
+             * All XPS INTC registers are accessed through the PLB interface.
+             * The base address for these registers is provided by the
+             * configuration parameter, C_BASEADDR. Each register is 32 bits
+             * although some bits may be unused and is accessed on a 4-byte
+             * boundary offset from the base address.
+             */
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
     },
-    .valid = {
-        .min_access_size = 4,
-        .max_access_size = 4
+    {
+        .read = pic_read,
+        .write = pic_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
     }
 };
 
@@ -174,13 +200,21 @@ static void xilinx_intc_init(Object *obj)
 
     qdev_init_gpio_in(DEVICE(obj), irq_handler, 32);
     sysbus_init_irq(SYS_BUS_DEVICE(obj), &p->parent_irq);
-
-    memory_region_init_io(&p->mmio, obj, &pic_ops, p, "xlnx.xps-intc",
-                          R_MAX * 4);
     sysbus_init_mmio(SYS_BUS_DEVICE(obj), &p->mmio);
 }
 
+static void xilinx_intc_realize(DeviceState *dev, Error **errp)
+{
+    XpsIntc *p = XILINX_INTC(dev);
+
+    memory_region_init_io(&p->mmio, OBJECT(dev),
+                          &pic_ops[p->little_endian_model],
+                          p, "xlnx.xps-intc",
+                          R_MAX * 4);
+}
+
 static Property xilinx_intc_properties[] = {
+    DEFINE_PROP_BOOL("little-endian", XpsIntc, little_endian_model, true),
     DEFINE_PROP_UINT32("kind-of-intr", XpsIntc, c_kind_of_intr, 0),
     DEFINE_PROP_END_OF_LIST(),
 };
@@ -189,6 +223,7 @@ static void xilinx_intc_class_init(ObjectClass *klass, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(klass);
 
+    dc->realize = xilinx_intc_realize;
     device_class_set_props(dc, xilinx_intc_properties);
 }
 
diff --git a/hw/microblaze/petalogix_ml605_mmu.c b/hw/microblaze/petalogix_ml605_mmu.c
index d2b2109065..64e8cadbee 100644
--- a/hw/microblaze/petalogix_ml605_mmu.c
+++ b/hw/microblaze/petalogix_ml605_mmu.c
@@ -111,6 +111,7 @@ petalogix_ml605_init(MachineState *machine)
 
 
     dev = qdev_new("xlnx.xps-intc");
+    qdev_prop_set_bit(dev, "little-endian", true);
     qdev_prop_set_uint32(dev, "kind-of-intr", 1 << TIMER_IRQ);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
     sysbus_mmio_map(SYS_BUS_DEVICE(dev), 0, INTC_BASEADDR);
diff --git a/hw/microblaze/petalogix_s3adsp1800_mmu.c b/hw/microblaze/petalogix_s3adsp1800_mmu.c
index 8110be8371..af949196d3 100644
--- a/hw/microblaze/petalogix_s3adsp1800_mmu.c
+++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
@@ -95,6 +95,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
                           64 * KiB, 1, 0x89, 0x18, 0x0000, 0x0, 1);
 
     dev = qdev_new("xlnx.xps-intc");
+    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
     qdev_prop_set_uint32(dev, "kind-of-intr",
                          1 << ETHLITE_IRQ | 1 << UARTLITE_IRQ);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
-- 
2.45.2

[RFC PATCH v2 04/16] hw/net/xilinx_ethlite: Simplify by having configurable endianness

[Philippe Mathieu-Daudé]

The Xilinx 'ethlite' device was added in commit b43848a100
("xilinx: Add ethlite emulation"), being only built back
then for a big-endian MicroBlaze target (see commit 72b675caac
"microblaze: Hook into the build-system").

I/O endianness access was then clarified in commit d48751ed4f
("xilinx-ethlite: Simplify byteswapping to/from brams"). Here
the 'fix' was to use tswap32(). Since the machine was built as
big-endian target, tswap32() use means the fix was for a little
endian host. While the datasheet (reference added in file header)
is not precise about it, we interpret such change as the device
expects accesses in big-endian order.

Instead of having a double swapping, one in the core memory layer
due to DEVICE_NATIVE_ENDIAN and a second one with the tswap calls,
allow the machine code to select the proper endianness desired,
removing the need of tswap().

Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair of
DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
Add the "little-endian" property to select the device endianness,
defaulting to little endian.
Set the proper endianness on the single machine using the device.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
RFC until I digest Paolo's review from v1:
https://lore.kernel.org/qemu-devel/34f6fe2f-06e0-4e2a-a361-2d662f6814b5@redhat.com/
---
 hw/microblaze/petalogix_s3adsp1800_mmu.c |  2 +
 hw/net/xilinx_ethlite.c                  | 54 +++++++++++++++++-------
 2 files changed, 40 insertions(+), 16 deletions(-)

diff --git a/hw/microblaze/petalogix_s3adsp1800_mmu.c b/hw/microblaze/petalogix_s3adsp1800_mmu.c
index af949196d3..f2e2dc2fd7 100644
--- a/hw/microblaze/petalogix_s3adsp1800_mmu.c
+++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
@@ -121,9 +121,11 @@ petalogix_s3adsp1800_init(MachineState *machine)
     sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, irq[TIMER_IRQ]);
 
     dev = qdev_new("xlnx.xps-ethernetlite");
+    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
     qemu_configure_nic_device(dev, true, NULL);
     qdev_prop_set_uint32(dev, "tx-ping-pong", 0);
     qdev_prop_set_uint32(dev, "rx-ping-pong", 0);
+    qdev_prop_set_bit(dev, "little-endian-model", !TARGET_BIG_ENDIAN);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
     sysbus_mmio_map(SYS_BUS_DEVICE(dev), 0, ETHLITE_BASEADDR);
     sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, irq[ETHLITE_IRQ]);
diff --git a/hw/net/xilinx_ethlite.c b/hw/net/xilinx_ethlite.c
index e84b4cdd35..d2e7939569 100644
--- a/hw/net/xilinx_ethlite.c
+++ b/hw/net/xilinx_ethlite.c
@@ -3,6 +3,9 @@
  *
  * Copyright (c) 2009 Edgar E. Iglesias.
  *
+ * DS580: https://docs.amd.com/v/u/en-US/xps_ethernetlite
+ * LogiCORE IP XPS Ethernet Lite Media Access Controller
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
  * in the Software without restriction, including without limitation the rights
@@ -25,7 +28,6 @@
 #include "qemu/osdep.h"
 #include "qemu/module.h"
 #include "qom/object.h"
-#include "exec/tswap.h"
 #include "hw/sysbus.h"
 #include "hw/irq.h"
 #include "hw/qdev-properties.h"
@@ -60,6 +62,7 @@ struct xlx_ethlite
 {
     SysBusDevice parent_obj;
 
+    bool little_endian_model;
     MemoryRegion mmio;
     qemu_irq irq;
     NICState *nic;
@@ -103,9 +106,10 @@ eth_read(void *opaque, hwaddr addr, unsigned int size)
             break;
 
         default:
-            r = tswap32(s->regs[addr]);
+            r = s->regs[addr];
             break;
     }
+
     return r;
 }
 
@@ -161,22 +165,37 @@ eth_write(void *opaque, hwaddr addr,
             break;
 
         default:
-            s->regs[addr] = tswap32(value);
+            s->regs[addr] = value;
             break;
     }
 }
 
-static const MemoryRegionOps eth_ops = {
-    .read = eth_read,
-    .write = eth_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .impl = {
-        .min_access_size = 4,
-        .max_access_size = 4,
+static const MemoryRegionOps eth_ops[2] = {
+    {
+        .read = eth_read,
+        .write = eth_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
     },
-    .valid = {
-        .min_access_size = 4,
-        .max_access_size = 4
+    {
+        .read = eth_read,
+        .write = eth_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
     }
 };
 
@@ -237,6 +256,10 @@ static void xilinx_ethlite_realize(DeviceState *dev, Error **errp)
 {
     struct xlx_ethlite *s = XILINX_ETHLITE(dev);
 
+    memory_region_init_io(&s->mmio, OBJECT(dev),
+                          &eth_ops[s->little_endian_model], s,
+                          "xlnx.xps-ethernetlite", R_MAX * 4);
+
     qemu_macaddr_default_if_unset(&s->conf.macaddr);
     s->nic = qemu_new_nic(&net_xilinx_ethlite_info, &s->conf,
                           object_get_typename(OBJECT(dev)), dev->id,
@@ -249,13 +272,12 @@ static void xilinx_ethlite_init(Object *obj)
     struct xlx_ethlite *s = XILINX_ETHLITE(obj);
 
     sysbus_init_irq(SYS_BUS_DEVICE(obj), &s->irq);
-
-    memory_region_init_io(&s->mmio, obj, &eth_ops, s,
-                          "xlnx.xps-ethernetlite", R_MAX * 4);
     sysbus_init_mmio(SYS_BUS_DEVICE(obj), &s->mmio);
 }
 
 static Property xilinx_ethlite_properties[] = {
+    DEFINE_PROP_BOOL("little-endian", struct xlx_ethlite,
+                     little_endian_model, true),
     DEFINE_PROP_UINT32("tx-ping-pong", struct xlx_ethlite, c_tx_pingpong, 1),
     DEFINE_PROP_UINT32("rx-ping-pong", struct xlx_ethlite, c_rx_pingpong, 1),
     DEFINE_NIC_PROPERTIES(struct xlx_ethlite, conf),
-- 
2.45.2

[RFC PATCH v2 05/16] hw/timer/xilinx_timer: Allow down to 8-bit memory access

[Philippe Mathieu-Daudé]

Allow down to 8-bit access, per the datasheet (reference added
in previous commit):

 "Timer Counter registers are accessed as one of the following types:
  • Byte (8 bits)
  • Half word (2 bytes)
  • Word (4 bytes)"

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
---
RFC: This breaks the UART qtest, instead of having TX register
receiving 'T' = 0x54, it receives 0x54000000, converted to '\0'
char. It works if we use SWI instead of SBI (storing 32-bit).
---
 hw/timer/xilinx_timer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/hw/timer/xilinx_timer.c b/hw/timer/xilinx_timer.c
index 383fc8b3c8..c117bff225 100644
--- a/hw/timer/xilinx_timer.c
+++ b/hw/timer/xilinx_timer.c
@@ -198,7 +198,7 @@ static const MemoryRegionOps timer_ops = {
         .max_access_size = 4,
     },
     .valid = {
-        .min_access_size = 4,
+        .min_access_size = 1,
         .max_access_size = 4
     }
 };
-- 
2.45.2

[PATCH v2 06/16] hw/timer/xilinx_timer: Make device endianness configurable

[Philippe Mathieu-Daudé]

Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
Add the "little-endian" property to select the device
endianness, defaulting to little endian.
Set the proper endianness for each machine using the device.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 hw/microblaze/petalogix_ml605_mmu.c      |  1 +
 hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
 hw/ppc/virtex_ml507.c                    |  1 +
 hw/timer/xilinx_timer.c                  | 46 +++++++++++++++++-------
 4 files changed, 36 insertions(+), 13 deletions(-)

diff --git a/hw/microblaze/petalogix_ml605_mmu.c b/hw/microblaze/petalogix_ml605_mmu.c
index 64e8cadbee..f4ec983fee 100644
--- a/hw/microblaze/petalogix_ml605_mmu.c
+++ b/hw/microblaze/petalogix_ml605_mmu.c
@@ -127,6 +127,7 @@ petalogix_ml605_init(MachineState *machine)
 
     /* 2 timers at irq 2 @ 100 Mhz.  */
     dev = qdev_new("xlnx.xps-timer");
+    qdev_prop_set_bit(dev, "little-endian", true);
     qdev_prop_set_uint32(dev, "one-timer-only", 0);
     qdev_prop_set_uint32(dev, "clock-frequency", 100 * 1000000);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
diff --git a/hw/microblaze/petalogix_s3adsp1800_mmu.c b/hw/microblaze/petalogix_s3adsp1800_mmu.c
index f2e2dc2fd7..c0136d84c3 100644
--- a/hw/microblaze/petalogix_s3adsp1800_mmu.c
+++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
@@ -114,6 +114,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
 
     /* 2 timers at irq 2 @ 62 Mhz.  */
     dev = qdev_new("xlnx.xps-timer");
+    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
     qdev_prop_set_uint32(dev, "one-timer-only", 0);
     qdev_prop_set_uint32(dev, "clock-frequency", 62 * 1000000);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
diff --git a/hw/ppc/virtex_ml507.c b/hw/ppc/virtex_ml507.c
index f378e5c4a9..ea0b3a56fe 100644
--- a/hw/ppc/virtex_ml507.c
+++ b/hw/ppc/virtex_ml507.c
@@ -230,6 +230,7 @@ static void virtex_init(MachineState *machine)
 
     /* 2 timers at irq 2 @ 62 Mhz.  */
     dev = qdev_new("xlnx.xps-timer");
+    qdev_prop_set_bit(dev, "little-endian", false);
     qdev_prop_set_uint32(dev, "one-timer-only", 0);
     qdev_prop_set_uint32(dev, "clock-frequency", 62 * 1000000);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
diff --git a/hw/timer/xilinx_timer.c b/hw/timer/xilinx_timer.c
index c117bff225..d356807d7c 100644
--- a/hw/timer/xilinx_timer.c
+++ b/hw/timer/xilinx_timer.c
@@ -3,6 +3,9 @@
  *
  * Copyright (c) 2009 Edgar E. Iglesias.
  *
+ * DS573: https://docs.amd.com/v/u/en-US/xps_timer
+ * LogiCORE IP XPS Timer/Counter (v1.02a)
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a copy
  * of this software and associated documentation files (the "Software"), to deal
  * in the Software without restriction, including without limitation the rights
@@ -69,6 +72,7 @@ struct XpsTimerState
 {
     SysBusDevice parent_obj;
 
+    bool little_endian_model;
     MemoryRegion mmio;
     qemu_irq irq;
     uint8_t one_timer_only;
@@ -189,17 +193,31 @@ timer_write(void *opaque, hwaddr addr,
     timer_update_irq(t);
 }
 
-static const MemoryRegionOps timer_ops = {
-    .read = timer_read,
-    .write = timer_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .impl = {
-        .min_access_size = 4,
-        .max_access_size = 4,
-    },
-    .valid = {
-        .min_access_size = 1,
-        .max_access_size = 4
+static const MemoryRegionOps timer_ops[2] = {
+    {
+        .read = timer_read,
+        .write = timer_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 1,
+            .max_access_size = 4,
+        },
+    }, {
+        .read = timer_read,
+        .write = timer_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 1,
+            .max_access_size = 4,
+        },
     }
 };
 
@@ -233,8 +251,9 @@ static void xilinx_timer_realize(DeviceState *dev, Error **errp)
         ptimer_transaction_commit(xt->ptimer);
     }
 
-    memory_region_init_io(&t->mmio, OBJECT(t), &timer_ops, t, "xlnx.xps-timer",
-                          R_MAX * 4 * num_timers(t));
+    memory_region_init_io(&t->mmio, OBJECT(t),
+                          &timer_ops[t->little_endian_model], t,
+                          "xlnx.xps-timer", R_MAX * 4 * num_timers(t));
     sysbus_init_mmio(SYS_BUS_DEVICE(dev), &t->mmio);
 }
 
@@ -247,6 +266,7 @@ static void xilinx_timer_init(Object *obj)
 }
 
 static Property xilinx_timer_properties[] = {
+    DEFINE_PROP_BOOL("little-endian", XpsTimerState, little_endian_model, true),
     DEFINE_PROP_UINT32("clock-frequency", XpsTimerState, freq_hz, 62 * 1000000),
     DEFINE_PROP_UINT8("one-timer-only", XpsTimerState, one_timer_only, 0),
     DEFINE_PROP_END_OF_LIST(),
-- 
2.45.2

[PATCH v2 07/16] hw/char/xilinx_uartlite: Make device endianness configurable

[Philippe Mathieu-Daudé]

Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
Add the "little-endian" property to select the device
endianness, defaulting to little endian.
Set the proper endianness on the single machine using the
device.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 hw/char/xilinx_uartlite.c                | 44 ++++++++++++++++--------
 hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
 2 files changed, 31 insertions(+), 14 deletions(-)

diff --git a/hw/char/xilinx_uartlite.c b/hw/char/xilinx_uartlite.c
index 3022b3d8ef..c43bf1a030 100644
--- a/hw/char/xilinx_uartlite.c
+++ b/hw/char/xilinx_uartlite.c
@@ -57,6 +57,7 @@
 struct XilinxUARTLite {
     SysBusDevice parent_obj;
 
+    bool little_endian_model;
     MemoryRegion mmio;
     CharBackend chr;
     qemu_irq irq;
@@ -166,21 +167,36 @@ uart_write(void *opaque, hwaddr addr,
     uart_update_irq(s);
 }
 
-static const MemoryRegionOps uart_ops = {
-    .read = uart_read,
-    .write = uart_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .impl = {
-        .min_access_size = 4,
-        .max_access_size = 4,
-    },
-    .valid = {
-        .min_access_size = 1,
-        .max_access_size = 4
+static const MemoryRegionOps uart_ops[2] = {
+    {
+        .read = uart_read,
+        .write = uart_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 1,
+            .max_access_size = 4,
+        },
+    }, {
+        .read = uart_read,
+        .write = uart_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 1,
+            .max_access_size = 4,
+        },
     }
 };
 
 static Property xilinx_uartlite_properties[] = {
+    DEFINE_PROP_BOOL("little-endian", XilinxUARTLite, little_endian_model, true),
     DEFINE_PROP_CHR("chardev", XilinxUARTLite, chr),
     DEFINE_PROP_END_OF_LIST(),
 };
@@ -219,6 +235,9 @@ static void xilinx_uartlite_realize(DeviceState *dev, Error **errp)
 {
     XilinxUARTLite *s = XILINX_UARTLITE(dev);
 
+    memory_region_init_io(&s->mmio, OBJECT(dev),
+                          &uart_ops[s->little_endian_model],
+                          s, "xlnx.xps-uartlite", R_MAX * 4);
     qemu_chr_fe_set_handlers(&s->chr, uart_can_rx, uart_rx,
                              uart_event, NULL, s, NULL, true);
 }
@@ -228,9 +247,6 @@ static void xilinx_uartlite_init(Object *obj)
     XilinxUARTLite *s = XILINX_UARTLITE(obj);
 
     sysbus_init_irq(SYS_BUS_DEVICE(obj), &s->irq);
-
-    memory_region_init_io(&s->mmio, obj, &uart_ops, s,
-                          "xlnx.xps-uartlite", R_MAX * 4);
     sysbus_init_mmio(SYS_BUS_DEVICE(obj), &s->mmio);
 }
 
diff --git a/hw/microblaze/petalogix_s3adsp1800_mmu.c b/hw/microblaze/petalogix_s3adsp1800_mmu.c
index c0136d84c3..bd8b85fa54 100644
--- a/hw/microblaze/petalogix_s3adsp1800_mmu.c
+++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
@@ -107,6 +107,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
     }
 
     dev = qdev_new(TYPE_XILINX_UARTLITE);
+    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
     qdev_prop_set_chr(dev, "chardev", serial_hd(0));
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
     sysbus_mmio_map(SYS_BUS_DEVICE(dev), 0, UARTLITE_BASEADDR);
-- 
2.45.2

[PATCH v2 08/16] hw/ssi/xilinx_spi: Make device endianness configurable

[Philippe Mathieu-Daudé]

Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
Add the "little-endian" property to select the device
endianness, defaulting to little endian.
Set the proper endianness on the single machine using the
device.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 hw/arm/xlnx-zynqmp.c |  4 ++++
 hw/ssi/xilinx_spi.c  | 29 +++++++++++++++++++++--------
 2 files changed, 25 insertions(+), 8 deletions(-)

diff --git a/hw/arm/xlnx-zynqmp.c b/hw/arm/xlnx-zynqmp.c
index ab2d50e31b..e735dbdf82 100644
--- a/hw/arm/xlnx-zynqmp.c
+++ b/hw/arm/xlnx-zynqmp.c
@@ -714,6 +714,10 @@ static void xlnx_zynqmp_realize(DeviceState *dev, Error **errp)
     for (i = 0; i < XLNX_ZYNQMP_NUM_SPIS; i++) {
         gchar *bus_name;
 
+        if (!object_property_set_bool(OBJECT(&s->spi[i])), "little-endian",
+                                      true, errp)) {
+            return;
+        }
         if (!sysbus_realize(SYS_BUS_DEVICE(&s->spi[i]), errp)) {
             return;
         }
diff --git a/hw/ssi/xilinx_spi.c b/hw/ssi/xilinx_spi.c
index 7f1e1808c5..2a0c9bca05 100644
--- a/hw/ssi/xilinx_spi.c
+++ b/hw/ssi/xilinx_spi.c
@@ -83,6 +83,7 @@ OBJECT_DECLARE_SIMPLE_TYPE(XilinxSPI, XILINX_SPI)
 struct XilinxSPI {
     SysBusDevice parent_obj;
 
+    bool little_endian_model;
     MemoryRegion mmio;
 
     qemu_irq irq;
@@ -313,13 +314,23 @@ done:
     xlx_spi_update_irq(s);
 }
 
-static const MemoryRegionOps spi_ops = {
-    .read = spi_read,
-    .write = spi_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .valid = {
-        .min_access_size = 4,
-        .max_access_size = 4
+static const MemoryRegionOps spi_ops[2] = {
+    {
+        .read = spi_read,
+        .write = spi_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+    }, {
+        .read = spi_read,
+        .write = spi_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
     }
 };
 
@@ -339,7 +350,8 @@ static void xilinx_spi_realize(DeviceState *dev, Error **errp)
         sysbus_init_irq(sbd, &s->cs_lines[i]);
     }
 
-    memory_region_init_io(&s->mmio, OBJECT(s), &spi_ops, s,
+    memory_region_init_io(&s->mmio, OBJECT(s),
+                          &spi_ops[s->little_endian_model], s,
                           "xilinx-spi", R_MAX * 4);
     sysbus_init_mmio(sbd, &s->mmio);
 
@@ -362,6 +374,7 @@ static const VMStateDescription vmstate_xilinx_spi = {
 };
 
 static Property xilinx_spi_properties[] = {
+    DEFINE_PROP_BOOL("little-endian", XilinxSPI, little_endian_model, true),
     DEFINE_PROP_UINT8("num-ss-bits", XilinxSPI, num_cs, 1),
     DEFINE_PROP_END_OF_LIST(),
 };
-- 
2.45.2

[PATCH v2 09/16] hw/ssi/xilinx_spips: Make device endianness configurable

[Philippe Mathieu-Daudé]

Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
Add the "little-endian" property to select the device
endianness, defaulting to little endian.
Set the proper endianness on the single machine using the
device.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 include/hw/ssi/xilinx_spips.h |  1 +
 hw/arm/xilinx_zynq.c          |  1 +
 hw/ssi/xilinx_spips.c         | 46 ++++++++++++++++++++++++++---------
 3 files changed, 36 insertions(+), 12 deletions(-)

diff --git a/include/hw/ssi/xilinx_spips.h b/include/hw/ssi/xilinx_spips.h
index 7a754bf67a..451c3758b3 100644
--- a/include/hw/ssi/xilinx_spips.h
+++ b/include/hw/ssi/xilinx_spips.h
@@ -101,6 +101,7 @@ typedef struct XilinxQSPIPS XilinxQSPIPS;
 struct XlnxZynqMPQSPIPS {
     XilinxQSPIPS parent_obj;
 
+    bool little_endian_model;
     StreamSink *dma;
     int gqspi_irqline;
 
diff --git a/hw/arm/xilinx_zynq.c b/hw/arm/xilinx_zynq.c
index fde4d946b7..bcc0022c17 100644
--- a/hw/arm/xilinx_zynq.c
+++ b/hw/arm/xilinx_zynq.c
@@ -142,6 +142,7 @@ static inline int zynq_init_spi_flashes(uint32_t base_addr, qemu_irq irq,
     int num_ss = is_qspi ? NUM_QSPI_FLASHES : NUM_SPI_FLASHES;
 
     dev = qdev_new(is_qspi ? "xlnx.ps7-qspi" : "xlnx.ps7-spi");
+    qdev_prop_set_bit(dev, "little-endian", true);
     qdev_prop_set_uint8(dev, "num-txrx-bytes", is_qspi ? 4 : 1);
     qdev_prop_set_uint8(dev, "num-ss-bits", num_ss);
     qdev_prop_set_uint8(dev, "num-busses", num_busses);
diff --git a/hw/ssi/xilinx_spips.c b/hw/ssi/xilinx_spips.c
index aeb462c3ce..5c6f0dd079 100644
--- a/hw/ssi/xilinx_spips.c
+++ b/hw/ssi/xilinx_spips.c
@@ -1251,17 +1251,32 @@ static MemTxResult lqspi_write(void *opaque, hwaddr offset, uint64_t value,
     return MEMTX_ERROR;
 }
 
-static const MemoryRegionOps lqspi_ops = {
-    .read_with_attrs = lqspi_read,
-    .write_with_attrs = lqspi_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .impl = {
-        .min_access_size = 4,
-        .max_access_size = 4,
+static const MemoryRegionOps lqspi_ops[2] = {
+    {
+        .read_with_attrs = lqspi_read,
+        .write_with_attrs = lqspi_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 1,
+            .max_access_size = 4
+        },
     },
-    .valid = {
-        .min_access_size = 1,
-        .max_access_size = 4
+    {
+        .read_with_attrs = lqspi_read,
+        .write_with_attrs = lqspi_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .impl = {
+            .min_access_size = 4,
+            .max_access_size = 4,
+        },
+        .valid = {
+            .min_access_size = 1,
+            .max_access_size = 4,
+        },
     }
 };
 
@@ -1325,8 +1340,9 @@ static void xilinx_qspips_realize(DeviceState *dev, Error **errp)
     s->num_txrx_bytes = 4;
 
     xilinx_spips_realize(dev, errp);
-    memory_region_init_io(&s->mmlqspi, OBJECT(s), &lqspi_ops, s, "lqspi",
-                          (1 << LQSPI_ADDRESS_BITS) * 2);
+    memory_region_init_io(&s->mmlqspi, OBJECT(s),
+                          &lqspi_ops[s->little_endian_model],
+                          s, "lqspi", (1 << LQSPI_ADDRESS_BITS) * 2);
     sysbus_init_mmio(sbd, &s->mmlqspi);
 
     q->lqspi_cached_addr = ~0ULL;
@@ -1432,12 +1448,18 @@ static Property xilinx_spips_properties[] = {
     DEFINE_PROP_END_OF_LIST(),
 };
 
+static Property xilinx_qspips_properties[] = {
+    DEFINE_PROP_BOOL("little-endian", XilinxQSPIPS, little_endian_model, true),
+    DEFINE_PROP_END_OF_LIST(),
+};
+
 static void xilinx_qspips_class_init(ObjectClass *klass, void * data)
 {
     DeviceClass *dc = DEVICE_CLASS(klass);
     XilinxSPIPSClass *xsc = XILINX_SPIPS_CLASS(klass);
 
     dc->realize = xilinx_qspips_realize;
+    device_class_set_props(dc, xilinx_qspips_properties);
     xsc->reg_ops = &qspips_ops;
     xsc->reg_size = XLNX_SPIPS_R_MAX * 4;
     xsc->rx_fifo_size = RXFF_A_Q;
-- 
2.45.2

[PATCH v2 10/16] target/microblaze: Explode MO_TExx -> MO_TE | MO_xx

[Philippe Mathieu-Daudé]

Extract the implicit MO_TE definition in order to replace
it by runtime variable in the next commit.

Mechanical change using:

  $ for n in UW UL UQ UO SW SL SQ; do \
      sed -i -e "s/MO_TE$n/MO_TE | MO_$n/" \
           $(git grep -l MO_TE$n target/microblaze); \
    done

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
---
 target/microblaze/translate.c | 36 +++++++++++++++++------------------
 1 file changed, 18 insertions(+), 18 deletions(-)

diff --git a/target/microblaze/translate.c b/target/microblaze/translate.c
index 4beaf69e76..4c25b1e438 100644
--- a/target/microblaze/translate.c
+++ b/target/microblaze/translate.c
@@ -779,13 +779,13 @@ static bool trans_lbui(DisasContext *dc, arg_typeb *arg)
 static bool trans_lhu(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TEUW, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
 }
 
 static bool trans_lhur(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TEUW, dc->mem_index, true);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, true);
 }
 
 static bool trans_lhuea(DisasContext *dc, arg_typea *arg)
@@ -797,26 +797,26 @@ static bool trans_lhuea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TEUW, MMU_NOMMU_IDX, false);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_lhui(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_load(dc, arg->rd, addr, MO_TEUW, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
 }
 
 static bool trans_lw(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TEUL, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
 }
 
 static bool trans_lwr(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TEUL, dc->mem_index, true);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, true);
 }
 
 static bool trans_lwea(DisasContext *dc, arg_typea *arg)
@@ -828,14 +828,14 @@ static bool trans_lwea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TEUL, MMU_NOMMU_IDX, false);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_lwi(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_load(dc, arg->rd, addr, MO_TEUL, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
 }
 
 static bool trans_lwx(DisasContext *dc, arg_typea *arg)
@@ -845,7 +845,7 @@ static bool trans_lwx(DisasContext *dc, arg_typea *arg)
     /* lwx does not throw unaligned access errors, so force alignment */
     tcg_gen_andi_tl(addr, addr, ~3);
 
-    tcg_gen_qemu_ld_i32(cpu_res_val, addr, dc->mem_index, MO_TEUL);
+    tcg_gen_qemu_ld_i32(cpu_res_val, addr, dc->mem_index, MO_TE | MO_UL);
     tcg_gen_mov_tl(cpu_res_addr, addr);
 
     if (arg->rd) {
@@ -929,13 +929,13 @@ static bool trans_sbi(DisasContext *dc, arg_typeb *arg)
 static bool trans_sh(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TEUW, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
 }
 
 static bool trans_shr(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TEUW, dc->mem_index, true);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, true);
 }
 
 static bool trans_shea(DisasContext *dc, arg_typea *arg)
@@ -947,26 +947,26 @@ static bool trans_shea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TEUW, MMU_NOMMU_IDX, false);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_shi(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_store(dc, arg->rd, addr, MO_TEUW, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
 }
 
 static bool trans_sw(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TEUL, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
 }
 
 static bool trans_swr(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TEUL, dc->mem_index, true);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, true);
 }
 
 static bool trans_swea(DisasContext *dc, arg_typea *arg)
@@ -978,14 +978,14 @@ static bool trans_swea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TEUL, MMU_NOMMU_IDX, false);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_swi(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_store(dc, arg->rd, addr, MO_TEUL, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
 }
 
 static bool trans_swx(DisasContext *dc, arg_typea *arg)
@@ -1014,7 +1014,7 @@ static bool trans_swx(DisasContext *dc, arg_typea *arg)
 
     tcg_gen_atomic_cmpxchg_i32(tval, cpu_res_addr, cpu_res_val,
                                reg_for_write(dc, arg->rd),
-                               dc->mem_index, MO_TEUL);
+                               dc->mem_index, MO_TE | MO_UL);
 
     tcg_gen_brcond_i32(TCG_COND_NE, cpu_res_val, tval, swx_fail);
 
-- 
2.45.2

[PATCH v2 11/16] target/microblaze: Set MO_TE once in do_load() / do_store()

[Philippe Mathieu-Daudé]

All callers of do_load() / do_store() set MO_TE flag.
Set it once in the callees.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/microblaze/translate.c | 36 +++++++++++++++++++----------------
 1 file changed, 20 insertions(+), 16 deletions(-)

diff --git a/target/microblaze/translate.c b/target/microblaze/translate.c
index 4c25b1e438..86f3c19618 100644
--- a/target/microblaze/translate.c
+++ b/target/microblaze/translate.c
@@ -712,6 +712,8 @@ static bool do_load(DisasContext *dc, int rd, TCGv addr, MemOp mop,
 {
     MemOp size = mop & MO_SIZE;
 
+    mop |= MO_TE;
+
     /*
      * When doing reverse accesses we need to do two things.
      *
@@ -779,13 +781,13 @@ static bool trans_lbui(DisasContext *dc, arg_typeb *arg)
 static bool trans_lhu(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_UW, dc->mem_index, false);
 }
 
 static bool trans_lhur(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, true);
+    return do_load(dc, arg->rd, addr, MO_UW, dc->mem_index, true);
 }
 
 static bool trans_lhuea(DisasContext *dc, arg_typea *arg)
@@ -797,26 +799,26 @@ static bool trans_lhuea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, MMU_NOMMU_IDX, false);
+    return do_load(dc, arg->rd, addr, MO_UW, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_lhui(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_UW, dc->mem_index, false);
 }
 
 static bool trans_lw(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_UL, dc->mem_index, false);
 }
 
 static bool trans_lwr(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, true);
+    return do_load(dc, arg->rd, addr, MO_UL, dc->mem_index, true);
 }
 
 static bool trans_lwea(DisasContext *dc, arg_typea *arg)
@@ -828,14 +830,14 @@ static bool trans_lwea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, MMU_NOMMU_IDX, false);
+    return do_load(dc, arg->rd, addr, MO_UL, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_lwi(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_load(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
+    return do_load(dc, arg->rd, addr, MO_UL, dc->mem_index, false);
 }
 
 static bool trans_lwx(DisasContext *dc, arg_typea *arg)
@@ -862,6 +864,8 @@ static bool do_store(DisasContext *dc, int rd, TCGv addr, MemOp mop,
 {
     MemOp size = mop & MO_SIZE;
 
+    mop |= MO_TE;
+
     /*
      * When doing reverse accesses we need to do two things.
      *
@@ -929,13 +933,13 @@ static bool trans_sbi(DisasContext *dc, arg_typeb *arg)
 static bool trans_sh(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_UW, dc->mem_index, false);
 }
 
 static bool trans_shr(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, true);
+    return do_store(dc, arg->rd, addr, MO_UW, dc->mem_index, true);
 }
 
 static bool trans_shea(DisasContext *dc, arg_typea *arg)
@@ -947,26 +951,26 @@ static bool trans_shea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, MMU_NOMMU_IDX, false);
+    return do_store(dc, arg->rd, addr, MO_UW, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_shi(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UW, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_UW, dc->mem_index, false);
 }
 
 static bool trans_sw(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_UL, dc->mem_index, false);
 }
 
 static bool trans_swr(DisasContext *dc, arg_typea *arg)
 {
     TCGv addr = compute_ldst_addr_typea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, true);
+    return do_store(dc, arg->rd, addr, MO_UL, dc->mem_index, true);
 }
 
 static bool trans_swea(DisasContext *dc, arg_typea *arg)
@@ -978,14 +982,14 @@ static bool trans_swea(DisasContext *dc, arg_typea *arg)
     return true;
 #else
     TCGv addr = compute_ldst_addr_ea(dc, arg->ra, arg->rb);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, MMU_NOMMU_IDX, false);
+    return do_store(dc, arg->rd, addr, MO_UL, MMU_NOMMU_IDX, false);
 #endif
 }
 
 static bool trans_swi(DisasContext *dc, arg_typeb *arg)
 {
     TCGv addr = compute_ldst_addr_typeb(dc, arg->ra, arg->imm);
-    return do_store(dc, arg->rd, addr, MO_TE | MO_UL, dc->mem_index, false);
+    return do_store(dc, arg->rd, addr, MO_UL, dc->mem_index, false);
 }
 
 static bool trans_swx(DisasContext *dc, arg_typea *arg)
-- 
2.45.2

[PATCH v2 12/16] target/microblaze: Introduce mo_endian() helper

[Philippe Mathieu-Daudé]

mo_endian() returns the target endianness, currently static.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/microblaze/translate.c | 14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

diff --git a/target/microblaze/translate.c b/target/microblaze/translate.c
index 86f3c19618..0b466db694 100644
--- a/target/microblaze/translate.c
+++ b/target/microblaze/translate.c
@@ -707,12 +707,17 @@ static void record_unaligned_ess(DisasContext *dc, int rd,
 }
 #endif
 
+static inline MemOp mo_endian(DisasContext *dc)
+{
+    return MO_TE;
+}
+
 static bool do_load(DisasContext *dc, int rd, TCGv addr, MemOp mop,
                     int mem_index, bool rev)
 {
     MemOp size = mop & MO_SIZE;
 
-    mop |= MO_TE;
+    mop |= mo_endian(dc);
 
     /*
      * When doing reverse accesses we need to do two things.
@@ -847,7 +852,8 @@ static bool trans_lwx(DisasContext *dc, arg_typea *arg)
     /* lwx does not throw unaligned access errors, so force alignment */
     tcg_gen_andi_tl(addr, addr, ~3);
 
-    tcg_gen_qemu_ld_i32(cpu_res_val, addr, dc->mem_index, MO_TE | MO_UL);
+    tcg_gen_qemu_ld_i32(cpu_res_val, addr, dc->mem_index,
+                        mo_endian(dc) | MO_UL);
     tcg_gen_mov_tl(cpu_res_addr, addr);
 
     if (arg->rd) {
@@ -864,7 +870,7 @@ static bool do_store(DisasContext *dc, int rd, TCGv addr, MemOp mop,
 {
     MemOp size = mop & MO_SIZE;
 
-    mop |= MO_TE;
+    mop |= mo_endian(dc);
 
     /*
      * When doing reverse accesses we need to do two things.
@@ -1018,7 +1024,7 @@ static bool trans_swx(DisasContext *dc, arg_typea *arg)
 
     tcg_gen_atomic_cmpxchg_i32(tval, cpu_res_addr, cpu_res_val,
                                reg_for_write(dc, arg->rd),
-                               dc->mem_index, MO_TE | MO_UL);
+                               dc->mem_index, mo_endian(dc) | MO_UL);
 
     tcg_gen_brcond_i32(TCG_COND_NE, cpu_res_val, tval, swx_fail);
 
-- 
2.45.2

[PATCH v2 13/16] target/microblaze: Consider endianness while translating code

[Philippe Mathieu-Daudé]

Consider the CPU ENDI bit, swap instructions when the CPU
endianness doesn't match the binary one.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/microblaze/cpu.h       | 7 +++++++
 target/microblaze/translate.c | 5 +++--
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/target/microblaze/cpu.h b/target/microblaze/cpu.h
index 3e5a3e5c60..6d540713eb 100644
--- a/target/microblaze/cpu.h
+++ b/target/microblaze/cpu.h
@@ -412,6 +412,13 @@ void mb_tcg_init(void);
 /* Ensure there is no overlap between the two masks. */
 QEMU_BUILD_BUG_ON(MSR_TB_MASK & IFLAGS_TB_MASK);
 
+static inline bool mb_cpu_is_big_endian(CPUState *cs)
+{
+    MicroBlazeCPU *cpu = MICROBLAZE_CPU(cs);
+
+    return !cpu->cfg.endi;
+}
+
 static inline void cpu_get_tb_cpu_state(CPUMBState *env, vaddr *pc,
                                         uint64_t *cs_base, uint32_t *flags)
 {
diff --git a/target/microblaze/translate.c b/target/microblaze/translate.c
index 0b466db694..5595ae4fad 100644
--- a/target/microblaze/translate.c
+++ b/target/microblaze/translate.c
@@ -709,7 +709,7 @@ static void record_unaligned_ess(DisasContext *dc, int rd,
 
 static inline MemOp mo_endian(DisasContext *dc)
 {
-    return MO_TE;
+    return dc->cfg->endi ? MO_LE : MO_BE;
 }
 
 static bool do_load(DisasContext *dc, int rd, TCGv addr, MemOp mop,
@@ -1646,7 +1646,8 @@ static void mb_tr_translate_insn(DisasContextBase *dcb, CPUState *cs)
 
     dc->tb_flags_to_set = 0;
 
-    ir = translator_ldl(cpu_env(cs), &dc->base, dc->base.pc_next);
+    ir = translator_ldl_swap(cpu_env(cs), &dc->base, dc->base.pc_next,
+                             mb_cpu_is_big_endian(cs) != TARGET_BIG_ENDIAN);
     if (!decode(dc, ir)) {
         trap_illegal(dc, true);
     }
-- 
2.45.2

[PATCH v2 14/16] hw/microblaze: Support various endianness for s3adsp1800 machines

[Philippe Mathieu-Daudé]

Introduce an abstract machine parent class which defines
the 'little_endian' property. Duplicate the current machine,
which endian is tied to the binary endianness, to one big
endian and a little endian machine; updating the machine
description. Keep the current default machine for each binary.

'petalogix-s3adsp1800' machine is aliased as:
- 'petalogix-s3adsp1800-be' on big-endian binary,
- 'petalogix-s3adsp1800-le' on little-endian one.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
---
 hw/microblaze/petalogix_s3adsp1800_mmu.c | 62 +++++++++++++++++++-----
 1 file changed, 51 insertions(+), 11 deletions(-)

diff --git a/hw/microblaze/petalogix_s3adsp1800_mmu.c b/hw/microblaze/petalogix_s3adsp1800_mmu.c
index bd8b85fa54..533a833b41 100644
--- a/hw/microblaze/petalogix_s3adsp1800_mmu.c
+++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
@@ -55,8 +55,17 @@
 #define ETHLITE_IRQ         1
 #define UARTLITE_IRQ        3
 
+typedef struct PetalogixS3adsp1800MachineClass {
+    MachineClass parent_obj;
+
+    bool little_endian;
+} PetalogixS3adsp1800MachineClass;
+
 #define TYPE_PETALOGIX_S3ADSP1800_MACHINE \
-            MACHINE_TYPE_NAME("petalogix-s3adsp1800")
+            MACHINE_TYPE_NAME("petalogix-s3adsp1800-common")
+DECLARE_CLASS_CHECKERS(PetalogixS3adsp1800MachineClass,
+                       PETALOGIX_S3ADSP1800_MACHINE,
+                       TYPE_PETALOGIX_S3ADSP1800_MACHINE)
 
 static void
 petalogix_s3adsp1800_init(MachineState *machine)
@@ -71,11 +80,13 @@ petalogix_s3adsp1800_init(MachineState *machine)
     MemoryRegion *phys_ram = g_new(MemoryRegion, 1);
     qemu_irq irq[32];
     MemoryRegion *sysmem = get_system_memory();
+    PetalogixS3adsp1800MachineClass *pmc;
 
+    pmc = PETALOGIX_S3ADSP1800_MACHINE_GET_CLASS(machine);
     cpu = MICROBLAZE_CPU(object_new(TYPE_MICROBLAZE_CPU));
     object_property_set_str(OBJECT(cpu), "version", "7.10.d", &error_abort);
     object_property_set_bool(OBJECT(cpu), "little-endian",
-                             !TARGET_BIG_ENDIAN, &error_abort);
+                             pmc->little_endian, &error_abort);
     qdev_realize(DEVICE(cpu), NULL, &error_abort);
 
     /* Attach emulated BRAM through the LMB.  */
@@ -95,7 +106,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
                           64 * KiB, 1, 0x89, 0x18, 0x0000, 0x0, 1);
 
     dev = qdev_new("xlnx.xps-intc");
-    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
+    qdev_prop_set_bit(dev, "little-endian", pmc->little_endian);
     qdev_prop_set_uint32(dev, "kind-of-intr",
                          1 << ETHLITE_IRQ | 1 << UARTLITE_IRQ);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
@@ -107,7 +118,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
     }
 
     dev = qdev_new(TYPE_XILINX_UARTLITE);
-    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
+    qdev_prop_set_bit(dev, "little-endian", pmc->little_endian);
     qdev_prop_set_chr(dev, "chardev", serial_hd(0));
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
     sysbus_mmio_map(SYS_BUS_DEVICE(dev), 0, UARTLITE_BASEADDR);
@@ -115,7 +126,7 @@ petalogix_s3adsp1800_init(MachineState *machine)
 
     /* 2 timers at irq 2 @ 62 Mhz.  */
     dev = qdev_new("xlnx.xps-timer");
-    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
+    qdev_prop_set_bit(dev, "little-endian", pmc->little_endian);
     qdev_prop_set_uint32(dev, "one-timer-only", 0);
     qdev_prop_set_uint32(dev, "clock-frequency", 62 * 1000000);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
@@ -127,33 +138,62 @@ petalogix_s3adsp1800_init(MachineState *machine)
     qemu_configure_nic_device(dev, true, NULL);
     qdev_prop_set_uint32(dev, "tx-ping-pong", 0);
     qdev_prop_set_uint32(dev, "rx-ping-pong", 0);
-    qdev_prop_set_bit(dev, "little-endian-model", !TARGET_BIG_ENDIAN);
+    qdev_prop_set_bit(dev, "little-endian", pmc->little_endian);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
     sysbus_mmio_map(SYS_BUS_DEVICE(dev), 0, ETHLITE_BASEADDR);
     sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, irq[ETHLITE_IRQ]);
 
     create_unimplemented_device("xps_gpio", GPIO_BASEADDR, 0x10000);
 
-    microblaze_load_kernel(cpu, !TARGET_BIG_ENDIAN, ddr_base, ram_size,
+    microblaze_load_kernel(cpu, pmc->little_endian, ddr_base, ram_size,
                            machine->initrd_filename,
                            BINARY_DEVICE_TREE_FILE,
                            NULL);
 }
 
-static void petalogix_s3adsp1800_machine_class_init(ObjectClass *oc, void *data)
+static void petalogix_s3adsp1800_machine_class_init(ObjectClass *oc,
+                                                    bool little_endian)
 {
     MachineClass *mc = MACHINE_CLASS(oc);
+    PetalogixS3adsp1800MachineClass *pmc = PETALOGIX_S3ADSP1800_MACHINE_CLASS(oc);
 
-    mc->desc = "PetaLogix linux refdesign for xilinx Spartan 3ADSP1800";
     mc->init = petalogix_s3adsp1800_init;
-    mc->is_default = true;
+    pmc->little_endian = little_endian;
+    mc->desc = little_endian
+        ? "PetaLogix linux refdesign for xilinx Spartan 3ADSP1800 (little endian)"
+        : "PetaLogix linux refdesign for xilinx Spartan 3ADSP1800 (big endian)";
+    if (little_endian == !TARGET_BIG_ENDIAN) {
+        mc->is_default = true;
+        mc->alias = "petalogix-s3adsp1800";
+    }
+}
+
+static void petalogix_s3adsp1800_machine_class_init_be(ObjectClass *oc, void *data)
+{
+    petalogix_s3adsp1800_machine_class_init(oc, false);
+}
+
+static void petalogix_s3adsp1800_machine_class_init_le(ObjectClass *oc, void *data)
+{
+    petalogix_s3adsp1800_machine_class_init(oc, true);
 }
 
 static const TypeInfo petalogix_s3adsp1800_machine_types[] = {
     {
         .name           = TYPE_PETALOGIX_S3ADSP1800_MACHINE,
         .parent         = TYPE_MACHINE,
-        .class_init     = petalogix_s3adsp1800_machine_class_init,
+        .abstract       = true,
+        .class_size     = sizeof(PetalogixS3adsp1800MachineClass),
+    },
+    {
+        .name           = MACHINE_TYPE_NAME("petalogix-s3adsp1800-be"),
+        .parent         = TYPE_PETALOGIX_S3ADSP1800_MACHINE,
+        .class_init     = petalogix_s3adsp1800_machine_class_init_be,
+    },
+    {
+        .name           = MACHINE_TYPE_NAME("petalogix-s3adsp1800-le"),
+        .parent         = TYPE_PETALOGIX_S3ADSP1800_MACHINE,
+        .class_init     = petalogix_s3adsp1800_machine_class_init_le,
     },
 };
 
-- 
2.45.2

[PATCH v2 15/16] tests/functional: Explicit endianness of microblaze assets

[Philippe Mathieu-Daudé]

The archive used in test_microblaze_s3adsp1800.py (testing a
big-endian target) contains a big-endian kernel. Rename using
the _BE suffix.

Similarly, the archive in test_microblazeel_s3adsp1800 (testing
a little-endian target) contains a little-endian kernel. Rename
using _LE suffix.

These changes will help when adding cross-endian kernel tests.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 tests/functional/test_microblaze_s3adsp1800.py   | 6 +++---
 tests/functional/test_microblazeel_s3adsp1800.py | 6 +++---
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/tests/functional/test_microblaze_s3adsp1800.py b/tests/functional/test_microblaze_s3adsp1800.py
index 4f692ffdb1..2b2f782270 100755
--- a/tests/functional/test_microblaze_s3adsp1800.py
+++ b/tests/functional/test_microblaze_s3adsp1800.py
@@ -17,14 +17,14 @@ class MicroblazeMachine(QemuSystemTest):
 
     timeout = 90
 
-    ASSET_IMAGE = Asset(
+    ASSET_IMAGE_BE = Asset(
         ('https://qemu-advcal.gitlab.io/qac-best-of-multiarch/download/'
          'day17.tar.xz'),
         '3ba7439dfbea7af4876662c97f8e1f0cdad9231fc166e4861d17042489270057')
 
-    def test_microblaze_s3adsp1800(self):
+    def test_microblaze_s3adsp1800_be(self):
         self.set_machine('petalogix-s3adsp1800')
-        file_path = self.ASSET_IMAGE.fetch()
+        file_path = self.ASSET_IMAGE_BE.fetch()
         archive_extract(file_path, self.workdir)
         self.vm.set_console()
         self.vm.add_args('-kernel', self.workdir + '/day17/ballerina.bin')
diff --git a/tests/functional/test_microblazeel_s3adsp1800.py b/tests/functional/test_microblazeel_s3adsp1800.py
index faa3927f2e..1aee5149fb 100755
--- a/tests/functional/test_microblazeel_s3adsp1800.py
+++ b/tests/functional/test_microblazeel_s3adsp1800.py
@@ -17,14 +17,14 @@ class MicroblazeelMachine(QemuSystemTest):
 
     timeout = 90
 
-    ASSET_IMAGE = Asset(
+    ASSET_IMAGE_LE = Asset(
         ('http://www.qemu-advent-calendar.org/2023/download/day13.tar.gz'),
         'b9b3d43c5dd79db88ada495cc6e0d1f591153fe41355e925d791fbf44de50c22')
 
-    def test_microblazeel_s3adsp1800(self):
+    def test_microblazeel_s3adsp1800_le(self):
         self.require_netdev('user')
         self.set_machine('petalogix-s3adsp1800')
-        file_path = self.ASSET_IMAGE.fetch()
+        file_path = self.ASSET_IMAGE_LE.fetch()
         archive_extract(file_path, self.workdir)
         self.vm.set_console()
         self.vm.add_args('-kernel', self.workdir + '/day13/xmaton.bin')
-- 
2.45.2

[PATCH v2 16/16] tests/functional: Add microblaze cross-endianness tests

[Philippe Mathieu-Daudé]

Copy/paste the current tests, but call the opposite endianness
machines, testing:
- petalogix-s3adsp1800-le machine (little-endian CPU) on the
  qemu-system-microblaze binary (big-endian)
- petalogix-s3adsp1800-be machine (big-endian CPU) on the
  qemu-system-microblazeel binary (little-endian).

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
---
 .../functional/test_microblaze_s3adsp1800.py  | 21 +++++++++++++++++++
 .../test_microblazeel_s3adsp1800.py           | 19 +++++++++++++++++
 2 files changed, 40 insertions(+)

diff --git a/tests/functional/test_microblaze_s3adsp1800.py b/tests/functional/test_microblaze_s3adsp1800.py
index 2b2f782270..7f5e8b6024 100755
--- a/tests/functional/test_microblaze_s3adsp1800.py
+++ b/tests/functional/test_microblaze_s3adsp1800.py
@@ -36,5 +36,26 @@ def test_microblaze_s3adsp1800_be(self):
         # message, that's why we don't test for a later string here. This
         # needs some investigation by a microblaze wizard one day...
 
+    ASSET_IMAGE_LE = Asset(
+        ('http://www.qemu-advent-calendar.org/2023/download/day13.tar.gz'),
+        'b9b3d43c5dd79db88ada495cc6e0d1f591153fe41355e925d791fbf44de50c22')
+
+    def test_microblaze_s3adsp1800_le(self):
+        self.require_netdev('user')
+        self.set_machine('petalogix-s3adsp1800-le')
+        file_path = self.ASSET_IMAGE_LE.fetch()
+        archive_extract(file_path, self.workdir)
+        self.vm.set_console()
+        self.vm.add_args('-kernel', self.workdir + '/day13/xmaton.bin')
+        self.vm.add_args('-nic', 'user,tftp=' + self.workdir + '/day13/')
+        self.vm.launch()
+        wait_for_console_pattern(self, 'QEMU Advent Calendar 2023')
+        time.sleep(0.1)
+        exec_command(self, 'root')
+        time.sleep(0.1)
+        exec_command_and_wait_for_pattern(self,
+                'tftp -g -r xmaton.png 10.0.2.2 ; md5sum xmaton.png',
+                '821cd3cab8efd16ad6ee5acc3642a8ea')
+
 if __name__ == '__main__':
     QemuSystemTest.main()
diff --git a/tests/functional/test_microblazeel_s3adsp1800.py b/tests/functional/test_microblazeel_s3adsp1800.py
index 1aee5149fb..60543009ba 100755
--- a/tests/functional/test_microblazeel_s3adsp1800.py
+++ b/tests/functional/test_microblazeel_s3adsp1800.py
@@ -38,5 +38,24 @@ def test_microblazeel_s3adsp1800_le(self):
                 'tftp -g -r xmaton.png 10.0.2.2 ; md5sum xmaton.png',
                 '821cd3cab8efd16ad6ee5acc3642a8ea')
 
+    ASSET_IMAGE_BE = Asset(
+        ('https://qemu-advcal.gitlab.io/qac-best-of-multiarch/download/'
+         'day17.tar.xz'),
+        '3ba7439dfbea7af4876662c97f8e1f0cdad9231fc166e4861d17042489270057')
+
+    def test_microblazeel_s3adsp1800_be(self):
+        self.set_machine('petalogix-s3adsp1800-be')
+        file_path = self.ASSET_IMAGE_BE.fetch()
+        archive_extract(file_path, self.workdir)
+        self.vm.set_console()
+        self.vm.add_args('-kernel', self.workdir + '/day17/ballerina.bin')
+        self.vm.launch()
+        wait_for_console_pattern(self, 'This architecture does not have '
+                                       'kernel memory protection')
+        # Note:
+        # The kernel sometimes gets stuck after the "This architecture ..."
+        # message, that's why we don't test for a later string here. This
+        # needs some investigation by a microblaze wizard one day...
+
 if __name__ == '__main__':
     QemuSystemTest.main()
-- 
2.45.2

Re: [PATCH v2 02/16] hw/microblaze: Propagate CPU endianness to microblaze_load_kernel()

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> Pass vCPU endianness as argument so we can load kernels
> with different endianness (different from the qemu-system-binary
> builtin one).
> 
> Signed-off-by: Philippe Mathieu-Daudé<philmd@linaro.org>
> Reviewed-by: Anton Johansson<anjo@rev.ng>
> Reviewed-by: Alistair Francis<alistair.francis@wdc.com>
> Reviewed-by: Edgar E. Iglesias<edgar.iglesias@amd.com>
> Message-Id:<20241105130431.22564-9-philmd@linaro.org>
> ---
>   hw/microblaze/boot.h                     | 4 ++--
>   hw/microblaze/boot.c                     | 8 ++++----
>   hw/microblaze/petalogix_ml605_mmu.c      | 2 +-
>   hw/microblaze/petalogix_s3adsp1800_mmu.c | 2 +-
>   hw/microblaze/xlnx-zynqmp-pmu.c          | 2 +-
>   5 files changed, 9 insertions(+), 9 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

Re: [PATCH v2 03/16] hw/intc/xilinx_intc: Make device endianness configurable

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
> Add the "little-endian" property to select the device
> endianness, defaulting to little endian.
> Set the proper endianness for each machine using the device.
> 
> Signed-off-by: Philippe Mathieu-Daudé<philmd@linaro.org>
> ---
>   hw/intc/xilinx_intc.c                    | 61 +++++++++++++++++++-----
>   hw/microblaze/petalogix_ml605_mmu.c      |  1 +
>   hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
>   3 files changed, 50 insertions(+), 13 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

Re: [RFC PATCH v2 04/16] hw/net/xilinx_ethlite: Simplify by having configurable endianness

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> +++ b/hw/microblaze/petalogix_s3adsp1800_mmu.c
> @@ -121,9 +121,11 @@ petalogix_s3adsp1800_init(MachineState *machine)
>       sysbus_connect_irq(SYS_BUS_DEVICE(dev), 0, irq[TIMER_IRQ]);
>   
>       dev = qdev_new("xlnx.xps-ethernetlite");
> +    qdev_prop_set_bit(dev, "little-endian", !TARGET_BIG_ENDIAN);
>       qemu_configure_nic_device(dev, true, NULL);
>       qdev_prop_set_uint32(dev, "tx-ping-pong", 0);
>       qdev_prop_set_uint32(dev, "rx-ping-pong", 0);
> +    qdev_prop_set_bit(dev, "little-endian-model", !TARGET_BIG_ENDIAN);

Surely only one of these.

r~

Re: [PATCH v2 06/16] hw/timer/xilinx_timer: Make device endianness configurable

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
> Add the "little-endian" property to select the device
> endianness, defaulting to little endian.
> Set the proper endianness for each machine using the device.
> 
> Signed-off-by: Philippe Mathieu-Daudé<philmd@linaro.org>
> ---
>   hw/microblaze/petalogix_ml605_mmu.c      |  1 +
>   hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
>   hw/ppc/virtex_ml507.c                    |  1 +
>   hw/timer/xilinx_timer.c                  | 46 +++++++++++++++++-------
>   4 files changed, 36 insertions(+), 13 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

Re: [PATCH v2 07/16] hw/char/xilinx_uartlite: Make device endianness configurable

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
> Add the "little-endian" property to select the device
> endianness, defaulting to little endian.
> Set the proper endianness on the single machine using the
> device.
> 
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   hw/char/xilinx_uartlite.c                | 44 ++++++++++++++++--------
>   hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
>   2 files changed, 31 insertions(+), 14 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

> +static const MemoryRegionOps uart_ops[2] = {
> +    {
> +        .read = uart_read,
> +        .write = uart_write,
> +        .endianness = DEVICE_BIG_ENDIAN,
> +        .impl = {
> +            .min_access_size = 4,
> +            .max_access_size = 4,
> +        },
> +        .valid = {
> +            .min_access_size = 1,
> +            .max_access_size = 4,
> +        },
> +    }, {
> +        .read = uart_read,
> +        .write = uart_write,
> +        .endianness = DEVICE_LITTLE_ENDIAN,
> +        .impl = {
> +            .min_access_size = 4,
> +            .max_access_size = 4,
> +        },
> +        .valid = {
> +            .min_access_size = 1,
> +            .max_access_size = 4,
> +        },
>       }
>   };

Having looked at several of these now, it occurs to me that you can avoid repetition:

static const MemoryRegionOps uart_ops[2] = {
     [0 ... 1] = {
         .read = uart_read,
         etc,
     },
     [0].endianness = DEVICE_BIG_ENDIAN,
     [1].endianness = DEVICE_LITTLE_ENDIAN,
};


r~

Re: [PATCH v2 08/16] hw/ssi/xilinx_spi: Make device endianness configurable

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
> Add the "little-endian" property to select the device
> endianness, defaulting to little endian.
> Set the proper endianness on the single machine using the
> device.
> 
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   hw/arm/xlnx-zynqmp.c |  4 ++++
>   hw/ssi/xilinx_spi.c  | 29 +++++++++++++++++++++--------
>   2 files changed, 25 insertions(+), 8 deletions(-)
> 
> diff --git a/hw/arm/xlnx-zynqmp.c b/hw/arm/xlnx-zynqmp.c
> index ab2d50e31b..e735dbdf82 100644
> --- a/hw/arm/xlnx-zynqmp.c
> +++ b/hw/arm/xlnx-zynqmp.c
> @@ -714,6 +714,10 @@ static void xlnx_zynqmp_realize(DeviceState *dev, Error **errp)
>       for (i = 0; i < XLNX_ZYNQMP_NUM_SPIS; i++) {
>           gchar *bus_name;
>   
> +        if (!object_property_set_bool(OBJECT(&s->spi[i])), "little-endian",
> +                                      true, errp)) {
> +            return;
> +        }

How might this ever fail?
Despite xlnx_zynqmp_realize having an errp argument, I'd be inclined to use error_abort.

Otherwise,
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~

Re: [PATCH v2 09/16] hw/ssi/xilinx_spips: Make device endianness configurable

[Richard Henderson]

On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
> Add the "little-endian" property to select the device
> endianness, defaulting to little endian.
> Set the proper endianness on the single machine using the
> device.
> 
> Signed-off-by: Philippe Mathieu-Daudé<philmd@linaro.org>
> ---
>   include/hw/ssi/xilinx_spips.h |  1 +
>   hw/arm/xilinx_zynq.c          |  1 +
>   hw/ssi/xilinx_spips.c         | 46 ++++++++++++++++++++++++++---------
>   3 files changed, 36 insertions(+), 12 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~

Re: [PATCH v2 07/16] hw/char/xilinx_uartlite: Make device endianness configurable

[Philippe Mathieu-Daudé]

On 7/11/24 10:27, Richard Henderson wrote:
> On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
>> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
>> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
>> Add the "little-endian" property to select the device
>> endianness, defaulting to little endian.
>> Set the proper endianness on the single machine using the
>> device.
>>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>> ---
>>   hw/char/xilinx_uartlite.c                | 44 ++++++++++++++++--------
>>   hw/microblaze/petalogix_s3adsp1800_mmu.c |  1 +
>>   2 files changed, 31 insertions(+), 14 deletions(-)
> 
> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
> 
>> +static const MemoryRegionOps uart_ops[2] = {
>> +    {
>> +        .read = uart_read,
>> +        .write = uart_write,
>> +        .endianness = DEVICE_BIG_ENDIAN,
>> +        .impl = {
>> +            .min_access_size = 4,
>> +            .max_access_size = 4,
>> +        },
>> +        .valid = {
>> +            .min_access_size = 1,
>> +            .max_access_size = 4,
>> +        },
>> +    }, {
>> +        .read = uart_read,
>> +        .write = uart_write,
>> +        .endianness = DEVICE_LITTLE_ENDIAN,
>> +        .impl = {
>> +            .min_access_size = 4,
>> +            .max_access_size = 4,
>> +        },
>> +        .valid = {
>> +            .min_access_size = 1,
>> +            .max_access_size = 4,
>> +        },
>>       }
>>   };
> 
> Having looked at several of these now, it occurs to me that you can 
> avoid repetition:
> 
> static const MemoryRegionOps uart_ops[2] = {
>      [0 ... 1] = {
>          .read = uart_read,
>          etc,
>      },
>      [0].endianness = DEVICE_BIG_ENDIAN,
>      [1].endianness = DEVICE_LITTLE_ENDIAN,
> };

Thank you :) I had the idea it was possible then forgot about it.

Re: [PATCH v2 08/16] hw/ssi/xilinx_spi: Make device endianness configurable

[Philippe Mathieu-Daudé]

On 7/11/24 11:01, Richard Henderson wrote:
> On 11/7/24 01:22, Philippe Mathieu-Daudé wrote:
>> Replace the DEVICE_NATIVE_ENDIAN MemoryRegionOps by a pair
>> of DEVICE_LITTLE_ENDIAN / DEVICE_BIG_ENDIAN.
>> Add the "little-endian" property to select the device
>> endianness, defaulting to little endian.
>> Set the proper endianness on the single machine using the
>> device.
>>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>> ---
>>   hw/arm/xlnx-zynqmp.c |  4 ++++
>>   hw/ssi/xilinx_spi.c  | 29 +++++++++++++++++++++--------
>>   2 files changed, 25 insertions(+), 8 deletions(-)
>>
>> diff --git a/hw/arm/xlnx-zynqmp.c b/hw/arm/xlnx-zynqmp.c
>> index ab2d50e31b..e735dbdf82 100644
>> --- a/hw/arm/xlnx-zynqmp.c
>> +++ b/hw/arm/xlnx-zynqmp.c
>> @@ -714,6 +714,10 @@ static void xlnx_zynqmp_realize(DeviceState *dev, 
>> Error **errp)
>>       for (i = 0; i < XLNX_ZYNQMP_NUM_SPIS; i++) {
>>           gchar *bus_name;
>> +        if (!object_property_set_bool(OBJECT(&s->spi[i])), 
>> "little-endian",
>> +                                      true, errp)) {
>> +            return;
>> +        }
> 
> How might this ever fail?
> Despite xlnx_zynqmp_realize having an errp argument, I'd be inclined to 
> use error_abort.

Yeah I also noticed that, but this is the pattern this file uses, so
I'd rather clean it up in a separate patch.

> Otherwise,
> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

Thanks!