* [PATCH v3 kvmtool 1/1] riscv: Move serial and rtc from IO port space to MMIO area.
@ 2023-02-02 19:13 Rajnesh Kanwal
2023-02-03 9:58 ` Alexandru Elisei
0 siblings, 1 reply; 3+ messages in thread
From: Rajnesh Kanwal @ 2023-02-02 19:13 UTC (permalink / raw)
To: apatel, atishp, andre.przywara, alexandru.elisei, kvm; +Cc: Rajnesh Kanwal
The default serial and rtc IO region overlaps with PCI IO bar
region leading bar 0 activation to fail. Moving these devices
to MMIO region similar to ARM.
Given serial has been moved from 0x3f8 to 0x10000000, this
requires us to now pass earlycon=uart8250,mmio,0x10000000
from cmdline rather than earlycon=uart8250,mmio,0x3f8.
To avoid the need to change the address every time the tool
is updated, we can also just pass "earlycon" from cmdline
and guest then finds the type and base address by following
the Device Tree's stdout-path property.
Signed-off-by: Rajnesh Kanwal <rkanwal@rivosinc.com>
Tested-by: Atish Patra <atishp@rivosinc.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>
---
v3: https://lore.kernel.org/all/20230201160137.486622-1-rkanwal@rivosinc.com/
Incorporated feedback from Andre Przywara and Alexandru Elisei.
Mainly updated the commit message to specify that we can simply pass
just "earlycon" from cmdline and avoid the need to specify uart address.
Also added Tested-by and Reviewed-by tags by Atish Patra.
v2: https://lore.kernel.org/all/20230201160137.486622-1-rkanwal@rivosinc.com/
Added further details in the commit message regarding the
UART address change required in kernel cmdline parameter.
v1: https://lore.kernel.org/all/20230124155251.1417682-1-rkanwal@rivosinc.com/
hw/rtc.c | 3 +++
hw/serial.c | 4 ++++
riscv/include/kvm/kvm-arch.h | 10 ++++++++--
3 files changed, 15 insertions(+), 2 deletions(-)
diff --git a/hw/rtc.c b/hw/rtc.c
index 9b8785a..da696e1 100644
--- a/hw/rtc.c
+++ b/hw/rtc.c
@@ -9,6 +9,9 @@
#if defined(CONFIG_ARM) || defined(CONFIG_ARM64)
#define RTC_BUS_TYPE DEVICE_BUS_MMIO
#define RTC_BASE_ADDRESS ARM_RTC_MMIO_BASE
+#elif defined(CONFIG_RISCV)
+#define RTC_BUS_TYPE DEVICE_BUS_MMIO
+#define RTC_BASE_ADDRESS RISCV_RTC_MMIO_BASE
#else
/* PORT 0070-007F - CMOS RAM/RTC (REAL TIME CLOCK) */
#define RTC_BUS_TYPE DEVICE_BUS_IOPORT
diff --git a/hw/serial.c b/hw/serial.c
index 3d53362..b6263a0 100644
--- a/hw/serial.c
+++ b/hw/serial.c
@@ -17,6 +17,10 @@
#define serial_iobase(nr) (ARM_UART_MMIO_BASE + (nr) * 0x1000)
#define serial_irq(nr) (32 + (nr))
#define SERIAL8250_BUS_TYPE DEVICE_BUS_MMIO
+#elif defined(CONFIG_RISCV)
+#define serial_iobase(nr) (RISCV_UART_MMIO_BASE + (nr) * 0x1000)
+#define serial_irq(nr) (1 + (nr))
+#define SERIAL8250_BUS_TYPE DEVICE_BUS_MMIO
#else
#define serial_iobase_0 (KVM_IOPORT_AREA + 0x3f8)
#define serial_iobase_1 (KVM_IOPORT_AREA + 0x2f8)
diff --git a/riscv/include/kvm/kvm-arch.h b/riscv/include/kvm/kvm-arch.h
index 3f96d00..620c796 100644
--- a/riscv/include/kvm/kvm-arch.h
+++ b/riscv/include/kvm/kvm-arch.h
@@ -11,7 +11,7 @@
#define RISCV_IOPORT 0x00000000ULL
#define RISCV_IOPORT_SIZE SZ_64K
#define RISCV_IRQCHIP 0x08000000ULL
-#define RISCV_IRQCHIP_SIZE SZ_128M
+#define RISCV_IRQCHIP_SIZE SZ_128M
#define RISCV_MMIO 0x10000000ULL
#define RISCV_MMIO_SIZE SZ_512M
#define RISCV_PCI 0x30000000ULL
@@ -35,10 +35,16 @@
#define RISCV_MAX_MEMORY(kvm) RISCV_LOMAP_MAX_MEMORY
#endif
+#define RISCV_UART_MMIO_BASE RISCV_MMIO
+#define RISCV_UART_MMIO_SIZE 0x10000
+
+#define RISCV_RTC_MMIO_BASE (RISCV_UART_MMIO_BASE + RISCV_UART_MMIO_SIZE)
+#define RISCV_RTC_MMIO_SIZE 0x10000
+
#define KVM_IOPORT_AREA RISCV_IOPORT
#define KVM_PCI_CFG_AREA RISCV_PCI
#define KVM_PCI_MMIO_AREA (KVM_PCI_CFG_AREA + RISCV_PCI_CFG_SIZE)
-#define KVM_VIRTIO_MMIO_AREA RISCV_MMIO
+#define KVM_VIRTIO_MMIO_AREA (RISCV_RTC_MMIO_BASE + RISCV_UART_MMIO_SIZE)
#define KVM_IOEVENTFD_HAS_PIO 0
--
2.25.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH v3 kvmtool 1/1] riscv: Move serial and rtc from IO port space to MMIO area.
2023-02-02 19:13 [PATCH v3 kvmtool 1/1] riscv: Move serial and rtc from IO port space to MMIO area Rajnesh Kanwal
@ 2023-02-03 9:58 ` Alexandru Elisei
2023-02-03 12:41 ` Rajnesh Kanwal
0 siblings, 1 reply; 3+ messages in thread
From: Alexandru Elisei @ 2023-02-03 9:58 UTC (permalink / raw)
To: Rajnesh Kanwal; +Cc: apatel, atishp, andre.przywara, kvm
Hi,
On Thu, Feb 02, 2023 at 07:13:01PM +0000, Rajnesh Kanwal wrote:
> The default serial and rtc IO region overlaps with PCI IO bar
> region leading bar 0 activation to fail. Moving these devices
> to MMIO region similar to ARM.
>
> Given serial has been moved from 0x3f8 to 0x10000000, this
> requires us to now pass earlycon=uart8250,mmio,0x10000000
> from cmdline rather than earlycon=uart8250,mmio,0x3f8.
>
> To avoid the need to change the address every time the tool
> is updated, we can also just pass "earlycon" from cmdline
> and guest then finds the type and base address by following
> the Device Tree's stdout-path property.
>
> Signed-off-by: Rajnesh Kanwal <rkanwal@rivosinc.com>
> Tested-by: Atish Patra <atishp@rivosinc.com>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> ---
> v3: https://lore.kernel.org/all/20230201160137.486622-1-rkanwal@rivosinc.com/
> Incorporated feedback from Andre Przywara and Alexandru Elisei.
> Mainly updated the commit message to specify that we can simply pass
> just "earlycon" from cmdline and avoid the need to specify uart address.
> Also added Tested-by and Reviewed-by tags by Atish Patra.
>
> v2: https://lore.kernel.org/all/20230201160137.486622-1-rkanwal@rivosinc.com/
> Added further details in the commit message regarding the
> UART address change required in kernel cmdline parameter.
>
> v1: https://lore.kernel.org/all/20230124155251.1417682-1-rkanwal@rivosinc.com/
>
> hw/rtc.c | 3 +++
> hw/serial.c | 4 ++++
> riscv/include/kvm/kvm-arch.h | 10 ++++++++--
> 3 files changed, 15 insertions(+), 2 deletions(-)
>
> diff --git a/hw/rtc.c b/hw/rtc.c
> index 9b8785a..da696e1 100644
> --- a/hw/rtc.c
> +++ b/hw/rtc.c
> @@ -9,6 +9,9 @@
> #if defined(CONFIG_ARM) || defined(CONFIG_ARM64)
> #define RTC_BUS_TYPE DEVICE_BUS_MMIO
> #define RTC_BASE_ADDRESS ARM_RTC_MMIO_BASE
> +#elif defined(CONFIG_RISCV)
> +#define RTC_BUS_TYPE DEVICE_BUS_MMIO
> +#define RTC_BASE_ADDRESS RISCV_RTC_MMIO_BASE
> #else
> /* PORT 0070-007F - CMOS RAM/RTC (REAL TIME CLOCK) */
> #define RTC_BUS_TYPE DEVICE_BUS_IOPORT
> diff --git a/hw/serial.c b/hw/serial.c
> index 3d53362..b6263a0 100644
> --- a/hw/serial.c
> +++ b/hw/serial.c
> @@ -17,6 +17,10 @@
> #define serial_iobase(nr) (ARM_UART_MMIO_BASE + (nr) * 0x1000)
> #define serial_irq(nr) (32 + (nr))
> #define SERIAL8250_BUS_TYPE DEVICE_BUS_MMIO
> +#elif defined(CONFIG_RISCV)
> +#define serial_iobase(nr) (RISCV_UART_MMIO_BASE + (nr) * 0x1000)
> +#define serial_irq(nr) (1 + (nr))
> +#define SERIAL8250_BUS_TYPE DEVICE_BUS_MMIO
> #else
> #define serial_iobase_0 (KVM_IOPORT_AREA + 0x3f8)
> #define serial_iobase_1 (KVM_IOPORT_AREA + 0x2f8)
> diff --git a/riscv/include/kvm/kvm-arch.h b/riscv/include/kvm/kvm-arch.h
> index 3f96d00..620c796 100644
> --- a/riscv/include/kvm/kvm-arch.h
> +++ b/riscv/include/kvm/kvm-arch.h
> @@ -11,7 +11,7 @@
> #define RISCV_IOPORT 0x00000000ULL
> #define RISCV_IOPORT_SIZE SZ_64K
> #define RISCV_IRQCHIP 0x08000000ULL
> -#define RISCV_IRQCHIP_SIZE SZ_128M
> +#define RISCV_IRQCHIP_SIZE SZ_128M
That's strange, for me the latest upstream commit is e17d182ad3f7 ("riscv: Add
--disable-<xyz> options to allow user disable extensions"), and I can't seem to
find those defines in the file. In fact, grep -r IRQCHIP riscv doesn't find
anything. Does this patch depend on another patch which hasn't been merged yet?
> #define RISCV_MMIO 0x10000000ULL
> #define RISCV_MMIO_SIZE SZ_512M
> #define RISCV_PCI 0x30000000ULL
> @@ -35,10 +35,16 @@
> #define RISCV_MAX_MEMORY(kvm) RISCV_LOMAP_MAX_MEMORY
> #endif
>
> +#define RISCV_UART_MMIO_BASE RISCV_MMIO
> +#define RISCV_UART_MMIO_SIZE 0x10000
> +
> +#define RISCV_RTC_MMIO_BASE (RISCV_UART_MMIO_BASE + RISCV_UART_MMIO_SIZE)
> +#define RISCV_RTC_MMIO_SIZE 0x10000
> +
> #define KVM_IOPORT_AREA RISCV_IOPORT
> #define KVM_PCI_CFG_AREA RISCV_PCI
> #define KVM_PCI_MMIO_AREA (KVM_PCI_CFG_AREA + RISCV_PCI_CFG_SIZE)
> -#define KVM_VIRTIO_MMIO_AREA RISCV_MMIO
> +#define KVM_VIRTIO_MMIO_AREA (RISCV_RTC_MMIO_BASE + RISCV_UART_MMIO_SIZE)
That should be RISCV_RTC_MMIO_SIZE, not RISCV_**UART**_MMIO_SIZE.
Thanks,
Alex
>
> #define KVM_IOEVENTFD_HAS_PIO 0
>
> --
> 2.25.1
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v3 kvmtool 1/1] riscv: Move serial and rtc from IO port space to MMIO area.
2023-02-03 9:58 ` Alexandru Elisei
@ 2023-02-03 12:41 ` Rajnesh Kanwal
0 siblings, 0 replies; 3+ messages in thread
From: Rajnesh Kanwal @ 2023-02-03 12:41 UTC (permalink / raw)
To: Alexandru Elisei; +Cc: apatel, atishp, andre.przywara, kvm
On Fri, Feb 3, 2023 at 9:58 AM Alexandru Elisei
<alexandru.elisei@arm.com> wrote:
>
> Hi,
>
> On Thu, Feb 02, 2023 at 07:13:01PM +0000, Rajnesh Kanwal wrote:
> > The default serial and rtc IO region overlaps with PCI IO bar
> > region leading bar 0 activation to fail. Moving these devices
> > to MMIO region similar to ARM.
> >
> > Given serial has been moved from 0x3f8 to 0x10000000, this
> > requires us to now pass earlycon=uart8250,mmio,0x10000000
> > from cmdline rather than earlycon=uart8250,mmio,0x3f8.
> >
> > To avoid the need to change the address every time the tool
> > is updated, we can also just pass "earlycon" from cmdline
> > and guest then finds the type and base address by following
> > the Device Tree's stdout-path property.
> >
> > Signed-off-by: Rajnesh Kanwal <rkanwal@rivosinc.com>
> > Tested-by: Atish Patra <atishp@rivosinc.com>
> > Reviewed-by: Atish Patra <atishp@rivosinc.com>
> > ---
> > v3: https://lore.kernel.org/all/20230201160137.486622-1-rkanwal@rivosinc.com/
> > Incorporated feedback from Andre Przywara and Alexandru Elisei.
> > Mainly updated the commit message to specify that we can simply pass
> > just "earlycon" from cmdline and avoid the need to specify uart address.
> > Also added Tested-by and Reviewed-by tags by Atish Patra.
> >
> > v2: https://lore.kernel.org/all/20230201160137.486622-1-rkanwal@rivosinc.com/
> > Added further details in the commit message regarding the
> > UART address change required in kernel cmdline parameter.
> >
> > v1: https://lore.kernel.org/all/20230124155251.1417682-1-rkanwal@rivosinc.com/
> >
> > hw/rtc.c | 3 +++
> > hw/serial.c | 4 ++++
> > riscv/include/kvm/kvm-arch.h | 10 ++++++++--
> > 3 files changed, 15 insertions(+), 2 deletions(-)
> >
> > diff --git a/hw/rtc.c b/hw/rtc.c
> > index 9b8785a..da696e1 100644
> > --- a/hw/rtc.c
> > +++ b/hw/rtc.c
> > @@ -9,6 +9,9 @@
> > #if defined(CONFIG_ARM) || defined(CONFIG_ARM64)
> > #define RTC_BUS_TYPE DEVICE_BUS_MMIO
> > #define RTC_BASE_ADDRESS ARM_RTC_MMIO_BASE
> > +#elif defined(CONFIG_RISCV)
> > +#define RTC_BUS_TYPE DEVICE_BUS_MMIO
> > +#define RTC_BASE_ADDRESS RISCV_RTC_MMIO_BASE
> > #else
> > /* PORT 0070-007F - CMOS RAM/RTC (REAL TIME CLOCK) */
> > #define RTC_BUS_TYPE DEVICE_BUS_IOPORT
> > diff --git a/hw/serial.c b/hw/serial.c
> > index 3d53362..b6263a0 100644
> > --- a/hw/serial.c
> > +++ b/hw/serial.c
> > @@ -17,6 +17,10 @@
> > #define serial_iobase(nr) (ARM_UART_MMIO_BASE + (nr) * 0x1000)
> > #define serial_irq(nr) (32 + (nr))
> > #define SERIAL8250_BUS_TYPE DEVICE_BUS_MMIO
> > +#elif defined(CONFIG_RISCV)
> > +#define serial_iobase(nr) (RISCV_UART_MMIO_BASE + (nr) * 0x1000)
> > +#define serial_irq(nr) (1 + (nr))
> > +#define SERIAL8250_BUS_TYPE DEVICE_BUS_MMIO
> > #else
> > #define serial_iobase_0 (KVM_IOPORT_AREA + 0x3f8)
> > #define serial_iobase_1 (KVM_IOPORT_AREA + 0x2f8)
> > diff --git a/riscv/include/kvm/kvm-arch.h b/riscv/include/kvm/kvm-arch.h
> > index 3f96d00..620c796 100644
> > --- a/riscv/include/kvm/kvm-arch.h
> > +++ b/riscv/include/kvm/kvm-arch.h
> > @@ -11,7 +11,7 @@
> > #define RISCV_IOPORT 0x00000000ULL
> > #define RISCV_IOPORT_SIZE SZ_64K
> > #define RISCV_IRQCHIP 0x08000000ULL
> > -#define RISCV_IRQCHIP_SIZE SZ_128M
> > +#define RISCV_IRQCHIP_SIZE SZ_128M
>
> That's strange, for me the latest upstream commit is e17d182ad3f7 ("riscv: Add
> --disable-<xyz> options to allow user disable extensions"), and I can't seem to
> find those defines in the file. In fact, grep -r IRQCHIP riscv doesn't find
> anything. Does this patch depend on another patch which hasn't been merged yet?
Extremely sorry. My bad. The change was based on AIA IRQCHIP changes but those
haven't merged yet and this fix doesn't depend on it. I have rebased
the changes and
sent another patch. I have also fixed the problem you have mentioned below.
Thanks for the review.
https://lore.kernel.org/all/20230203122934.18714-1-rkanwal@rivosinc.com/
I also retested everything on a pristine setup using instruction in
https://github.com/kvm-riscv/howto/wiki/KVM-RISCV64-on-QEMU.
Thanks
Rajnesh
>
> > #define RISCV_MMIO 0x10000000ULL
> > #define RISCV_MMIO_SIZE SZ_512M
> > #define RISCV_PCI 0x30000000ULL
> > @@ -35,10 +35,16 @@
> > #define RISCV_MAX_MEMORY(kvm) RISCV_LOMAP_MAX_MEMORY
> > #endif
> >
> > +#define RISCV_UART_MMIO_BASE RISCV_MMIO
> > +#define RISCV_UART_MMIO_SIZE 0x10000
> > +
> > +#define RISCV_RTC_MMIO_BASE (RISCV_UART_MMIO_BASE + RISCV_UART_MMIO_SIZE)
> > +#define RISCV_RTC_MMIO_SIZE 0x10000
> > +
> > #define KVM_IOPORT_AREA RISCV_IOPORT
> > #define KVM_PCI_CFG_AREA RISCV_PCI
> > #define KVM_PCI_MMIO_AREA (KVM_PCI_CFG_AREA + RISCV_PCI_CFG_SIZE)
> > -#define KVM_VIRTIO_MMIO_AREA RISCV_MMIO
> > +#define KVM_VIRTIO_MMIO_AREA (RISCV_RTC_MMIO_BASE + RISCV_UART_MMIO_SIZE)
>
> That should be RISCV_RTC_MMIO_SIZE, not RISCV_**UART**_MMIO_SIZE.
>
> Thanks,
> Alex
>
> >
> > #define KVM_IOEVENTFD_HAS_PIO 0
> >
> > --
> > 2.25.1
> >
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2023-02-03 12:41 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-02-02 19:13 [PATCH v3 kvmtool 1/1] riscv: Move serial and rtc from IO port space to MMIO area Rajnesh Kanwal
2023-02-03 9:58 ` Alexandru Elisei
2023-02-03 12:41 ` Rajnesh Kanwal
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.