All of lore.kernel.org
 help / color / mirror / Atom feed
From: Stafford Horne <shorne@gmail.com>
To: QEMU Development <qemu-devel@nongnu.org>
Cc: Laurent Vivier <lvivier@redhat.com>,
	Openrisc <openrisc@lists.librecores.org>,
	Arnd Bergmann <arnd@arndb.de>
Subject: [PATCH] goldfish_rtc: Add endianness property
Date: Sun, 12 Jun 2022 22:03:33 +0900	[thread overview]
Message-ID: <20220612130333.257213-1-shorne@gmail.com> (raw)

Add an endianness property to allow configuring the RTC as either
native, little or big endian.

Cc: Laurent Vivier <lvivier@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Stafford Horne <shorne@gmail.com>
---
In a recent mail thread[0] it was discussed that future users of goldfish devices
should use little-endian rather than native endian.  This is to avoid having to
define special ioread/write routines in the Linux kernel as the default of the
Linux kernel is to use little-endian.

I have defined a property to allow us to make this change the endianness of the
device at initialization time.  For OpenRISC we can now set the qemu device
endianness property to little-endian and the device works.

For now I kept the default as DEVICE_NATIVE_ENDIAN for illustration purposes I
could also do:

 - Add the "endianness" property to all goldfish devices.
 - Switch the default to LITTLE_ENDIAN and just set BIG_ENDIAN for m68k only.
 - Just the bare minimum, using this patch and also set little-endian in
   OpenRISC.
 - Just drop this patch and use BIG_ENDIAN access for goldfish in the kernel for
   OpenRISC like m68k does.

I have tested this with the OpenRISC virt platform[1] available here (now with pci
support).

[0] https://lore.kernel.org/all/CAK8P3a13PTsMExyXZm5sZM3WBK-8hoNFjYoi19pHmKKCJQmHeA@mail.gmail.com/
[1] https://github.com/stffrdhrn/qemu/commits/or1k-virt-2

 hw/rtc/goldfish_rtc.c         | 45 ++++++++++++++++++++++++++++-------
 include/hw/rtc/goldfish_rtc.h |  2 ++
 2 files changed, 38 insertions(+), 9 deletions(-)

diff --git a/hw/rtc/goldfish_rtc.c b/hw/rtc/goldfish_rtc.c
index 35e493be31..ab1e15a3e2 100644
--- a/hw/rtc/goldfish_rtc.c
+++ b/hw/rtc/goldfish_rtc.c
@@ -216,14 +216,34 @@ static int goldfish_rtc_post_load(void *opaque, int version_id)
     return 0;
 }
 
-static const MemoryRegionOps goldfish_rtc_ops = {
-    .read = goldfish_rtc_read,
-    .write = goldfish_rtc_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .valid = {
-        .min_access_size = 4,
-        .max_access_size = 4
-    }
+static const MemoryRegionOps goldfish_rtc_ops[3] = {
+    [DEVICE_NATIVE_ENDIAN] = {
+        .read = goldfish_rtc_read,
+        .write = goldfish_rtc_write,
+        .endianness = DEVICE_NATIVE_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4
+        }
+    },
+    [DEVICE_LITTLE_ENDIAN] = {
+        .read = goldfish_rtc_read,
+        .write = goldfish_rtc_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4
+        }
+    },
+    [DEVICE_BIG_ENDIAN] = {
+        .read = goldfish_rtc_read,
+        .write = goldfish_rtc_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4
+        }
+    },
 };
 
 static const VMStateDescription goldfish_rtc_vmstate = {
@@ -265,7 +285,8 @@ static void goldfish_rtc_realize(DeviceState *d, Error **errp)
     SysBusDevice *dev = SYS_BUS_DEVICE(d);
     GoldfishRTCState *s = GOLDFISH_RTC(d);
 
-    memory_region_init_io(&s->iomem, OBJECT(s), &goldfish_rtc_ops, s,
+    memory_region_init_io(&s->iomem, OBJECT(s),
+                          &goldfish_rtc_ops[s->endianness], s,
                           "goldfish_rtc", 0x24);
     sysbus_init_mmio(dev, &s->iomem);
 
@@ -274,10 +295,16 @@ static void goldfish_rtc_realize(DeviceState *d, Error **errp)
     s->timer = timer_new_ns(rtc_clock, goldfish_rtc_interrupt, s);
 }
 
+static Property goldfish_rtc_properties[] = {
+    DEFINE_PROP_UINT8("endianness", GoldfishRTCState, endianness, DEVICE_NATIVE_ENDIAN),
+    DEFINE_PROP_END_OF_LIST(),
+};
+
 static void goldfish_rtc_class_init(ObjectClass *klass, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(klass);
 
+    device_class_set_props(dc, goldfish_rtc_properties);
     dc->realize = goldfish_rtc_realize;
     dc->reset = goldfish_rtc_reset;
     dc->vmsd = &goldfish_rtc_vmstate;
diff --git a/include/hw/rtc/goldfish_rtc.h b/include/hw/rtc/goldfish_rtc.h
index 79ca7daf5d..8e1aeb85e3 100644
--- a/include/hw/rtc/goldfish_rtc.h
+++ b/include/hw/rtc/goldfish_rtc.h
@@ -42,6 +42,8 @@ struct GoldfishRTCState {
     uint32_t irq_pending;
     uint32_t irq_enabled;
     uint32_t time_high;
+
+    uint8_t endianness;
 };
 
 #endif
-- 
2.36.1


WARNING: multiple messages have this Message-ID (diff)
From: Stafford Horne <shorne@gmail.com>
To: QEMU Development <qemu-devel@nongnu.org>
Cc: Openrisc <openrisc@lists.librecores.org>,
	Stafford Horne <shorne@gmail.com>,
	Laurent Vivier <lvivier@redhat.com>,
	Arnd Bergmann <arnd@arndb.de>
Subject: [PATCH] goldfish_rtc: Add endianness property
Date: Sun, 12 Jun 2022 22:03:33 +0900	[thread overview]
Message-ID: <20220612130333.257213-1-shorne@gmail.com> (raw)

Add an endianness property to allow configuring the RTC as either
native, little or big endian.

Cc: Laurent Vivier <lvivier@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Stafford Horne <shorne@gmail.com>
---
In a recent mail thread[0] it was discussed that future users of goldfish devices
should use little-endian rather than native endian.  This is to avoid having to
define special ioread/write routines in the Linux kernel as the default of the
Linux kernel is to use little-endian.

I have defined a property to allow us to make this change the endianness of the
device at initialization time.  For OpenRISC we can now set the qemu device
endianness property to little-endian and the device works.

For now I kept the default as DEVICE_NATIVE_ENDIAN for illustration purposes I
could also do:

 - Add the "endianness" property to all goldfish devices.
 - Switch the default to LITTLE_ENDIAN and just set BIG_ENDIAN for m68k only.
 - Just the bare minimum, using this patch and also set little-endian in
   OpenRISC.
 - Just drop this patch and use BIG_ENDIAN access for goldfish in the kernel for
   OpenRISC like m68k does.

I have tested this with the OpenRISC virt platform[1] available here (now with pci
support).

[0] https://lore.kernel.org/all/CAK8P3a13PTsMExyXZm5sZM3WBK-8hoNFjYoi19pHmKKCJQmHeA@mail.gmail.com/
[1] https://github.com/stffrdhrn/qemu/commits/or1k-virt-2

 hw/rtc/goldfish_rtc.c         | 45 ++++++++++++++++++++++++++++-------
 include/hw/rtc/goldfish_rtc.h |  2 ++
 2 files changed, 38 insertions(+), 9 deletions(-)

diff --git a/hw/rtc/goldfish_rtc.c b/hw/rtc/goldfish_rtc.c
index 35e493be31..ab1e15a3e2 100644
--- a/hw/rtc/goldfish_rtc.c
+++ b/hw/rtc/goldfish_rtc.c
@@ -216,14 +216,34 @@ static int goldfish_rtc_post_load(void *opaque, int version_id)
     return 0;
 }
 
-static const MemoryRegionOps goldfish_rtc_ops = {
-    .read = goldfish_rtc_read,
-    .write = goldfish_rtc_write,
-    .endianness = DEVICE_NATIVE_ENDIAN,
-    .valid = {
-        .min_access_size = 4,
-        .max_access_size = 4
-    }
+static const MemoryRegionOps goldfish_rtc_ops[3] = {
+    [DEVICE_NATIVE_ENDIAN] = {
+        .read = goldfish_rtc_read,
+        .write = goldfish_rtc_write,
+        .endianness = DEVICE_NATIVE_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4
+        }
+    },
+    [DEVICE_LITTLE_ENDIAN] = {
+        .read = goldfish_rtc_read,
+        .write = goldfish_rtc_write,
+        .endianness = DEVICE_LITTLE_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4
+        }
+    },
+    [DEVICE_BIG_ENDIAN] = {
+        .read = goldfish_rtc_read,
+        .write = goldfish_rtc_write,
+        .endianness = DEVICE_BIG_ENDIAN,
+        .valid = {
+            .min_access_size = 4,
+            .max_access_size = 4
+        }
+    },
 };
 
 static const VMStateDescription goldfish_rtc_vmstate = {
@@ -265,7 +285,8 @@ static void goldfish_rtc_realize(DeviceState *d, Error **errp)
     SysBusDevice *dev = SYS_BUS_DEVICE(d);
     GoldfishRTCState *s = GOLDFISH_RTC(d);
 
-    memory_region_init_io(&s->iomem, OBJECT(s), &goldfish_rtc_ops, s,
+    memory_region_init_io(&s->iomem, OBJECT(s),
+                          &goldfish_rtc_ops[s->endianness], s,
                           "goldfish_rtc", 0x24);
     sysbus_init_mmio(dev, &s->iomem);
 
@@ -274,10 +295,16 @@ static void goldfish_rtc_realize(DeviceState *d, Error **errp)
     s->timer = timer_new_ns(rtc_clock, goldfish_rtc_interrupt, s);
 }
 
+static Property goldfish_rtc_properties[] = {
+    DEFINE_PROP_UINT8("endianness", GoldfishRTCState, endianness, DEVICE_NATIVE_ENDIAN),
+    DEFINE_PROP_END_OF_LIST(),
+};
+
 static void goldfish_rtc_class_init(ObjectClass *klass, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(klass);
 
+    device_class_set_props(dc, goldfish_rtc_properties);
     dc->realize = goldfish_rtc_realize;
     dc->reset = goldfish_rtc_reset;
     dc->vmsd = &goldfish_rtc_vmstate;
diff --git a/include/hw/rtc/goldfish_rtc.h b/include/hw/rtc/goldfish_rtc.h
index 79ca7daf5d..8e1aeb85e3 100644
--- a/include/hw/rtc/goldfish_rtc.h
+++ b/include/hw/rtc/goldfish_rtc.h
@@ -42,6 +42,8 @@ struct GoldfishRTCState {
     uint32_t irq_pending;
     uint32_t irq_enabled;
     uint32_t time_high;
+
+    uint8_t endianness;
 };
 
 #endif
-- 
2.36.1



             reply	other threads:[~2022-06-12 13:03 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-06-12 13:03 Stafford Horne [this message]
2022-06-12 13:03 ` [PATCH] goldfish_rtc: Add endianness property Stafford Horne

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220612130333.257213-1-shorne@gmail.com \
    --to=shorne@gmail.com \
    --cc=arnd@arndb.de \
    --cc=lvivier@redhat.com \
    --cc=openrisc@lists.librecores.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.