qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Gustavo Romero <gustavo.romero@linaro.org>
To: "Philippe Mathieu-Daudé" <philmd@linaro.org>, qemu-devel@nongnu.org
Cc: Thomas Huth <thuth@redhat.com>,
	Laurent Vivier <lvivier@redhat.com>,
	qemu-arm@nongnu.org, alex.bennee@linaro.org,
	Paolo Bonzini <pbonzini@redhat.com>,
	Anton Kochkov <anton.kochkov@proton.me>,
	richard.henderson@linaro.org,
	Peter Maydell <peter.maydell@linaro.org>
Subject: Re: [PATCH v2 2/6] hw/misc/ivshmem-flat: Allow device to wire itself on sysbus
Date: Tue, 20 Feb 2024 13:02:41 -0300	[thread overview]
Message-ID: <b03a659f-fef6-bb4e-b258-3e5a82b7edb4@linaro.org> (raw)
In-Reply-To: <20240216144456.34992-3-philmd@linaro.org>



On 2/16/24 11:44 AM, Philippe Mathieu-Daudé wrote:
> From: Gustavo Romero <gustavo.romero@linaro.org>
> 
> The following is an example on how to create the ivshmem-flat device on
> a Stellaris machine:
> 
> $ qemu-system-arm -cpu cortex-m3 -machine lm3s6965evb -nographic
>                    -net none -chardev stdio,id=con,mux=on
>                    -serial chardev:con -mon chardev=con,mode=readline
>                    -chardev socket,path=/tmp/ivshmem_socket,id=ivf
>                    -device ivshmem-flat,x-irq-qompath=/machine/unattached/device[1]/nvic/unnamed-gpio-in[0],x-bus-qompath="/sysbus",chardev=ivf
>                    -kernel zephyr_qemu.elf
> 
> The IRQ QOM path for the target machine can be determined by creating
> the VM without the ivshmem-flat device, going to the QEMU console and
> listing the QOM nodes with 'info qom-tree'. In the Stellaris example
> above the input IRQ is in the NVIC IC.
> 
> The MMRs for status and control (notification) are mapped to the MMIO
> region at 0x400FF000 (default), whilst the shared memory region start
> is mapped at addr. 0x40100000 (default), but both addresses can be set
> when creating the device by using 'x-bus-address-{mmr,shmem}' options,
> respectively.
> 
> Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
> Message-ID: <20231127052024.435743-2-gustavo.romero@linaro.org>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   docs/system/devices/ivshmem-flat.rst | 56 +++++++++++++++++++++++++
>   include/hw/misc/ivshmem-flat.h       |  9 ++++
>   hw/core/sysbus-fdt.c                 |  2 +
>   hw/misc/ivshmem-flat.c               | 63 ++++++++++++++++++++++++++--
>   4 files changed, 127 insertions(+), 3 deletions(-)
> 
> diff --git a/docs/system/devices/ivshmem-flat.rst b/docs/system/devices/ivshmem-flat.rst
> index 1f97052804..a10446a18f 100644
> --- a/docs/system/devices/ivshmem-flat.rst
> +++ b/docs/system/devices/ivshmem-flat.rst
> @@ -31,3 +31,59 @@ the IRQ mechanism is disabled). The shared memory region is always present.
>   The MMRs (INTRMASK, INTRSTATUS, IVPOSITION, and DOORBELL registers) offsets at
>   the MMR region, and their functions, follow the ivshmem spec, so they work
>   exactly as in the ivshmem PCI device (see ./specs/ivshmem-spec.txt).
> +
> +
> +Device Options
> +--------------
> +
> +The only required options to create an ivshmem-flat device are: (a) the UNIX
> +socket where the ivshmem server is listening, usually ``/tmp/ivshmem_socket``;
> +and (b) the bus type to be used by the device, which currently only supports
> +"/sysbus" bus type.
> +
> +Example:
> +
> +.. parsed-literal::
> +
> +    |qemu-system-arm| -chardev socket,path=/tmp/ivshmem_socket,id=ivshmem_flat -device ivshmem-flat,x-bus-qompath="/sysbus",chardev=ivshmem_flat

The doc need to be tweak now as well.


> +
> +The other options are for fine tuning the device.
> +
> +``x-irq-qompath``. Used to inform the device which IRQ input line it can attach
> +to enable the notification mechanism (IRQ). The ivshmem-flat device currently
> +only supports notification via vector 0, ignoring other vectors.
> +
> +Two examples for different machines follow.
> +
> +Stellaris machine (``- machine lm3s6965evb``):
> +
> +::
> +
> +    x-irq-qompath=/machine/unattached/device[1]/nvic/unnamed-gpio-in[0]
> +
> +Arm mps2-an385 machine (``-machine mps2-an385``):
> +
> +::
> +
> +    x-irq-qompath=/machine/armv7m/nvic/unnamed-gpio-in[0]
> +
> +The available IRQ input lines on a given VM that the ivshmem-flat device can be
> +attached to can be inspected from the QEMU monitor (Ctrl-a + c) with:
> +
> +(qemu) info qom-tree
> +
> +``x-bus-address-mmr``. Allows changing the address where the MMRs are mapped
> +into the VM memory layout. Default is 0x400FF000, but this address might be
> +already taken on some VMs, hence it's  necessary to adjust the MMR location on
> +some VMs.
> +
> + ``x-bus-address-shmem``. Allows changing the address where the shared memory
> +region is mapped into the VM memory layout. Default is 0x40100000, but this
> +address might be already taken on some VMs, hence it's necessary to adjust the
> +shared memory location.
> +
> +``shmem-size``. Allows changing the size (in bytes) of shared memroy region.
> +Default is 4 MiB, which is the same default value used by the ivshmem server, so
> +usually it's not necessary to change it. The  size must match the size of the
> +shared memory reserverd and informed by the ivshmem server, otherwise device
> +creation fails.
> diff --git a/include/hw/misc/ivshmem-flat.h b/include/hw/misc/ivshmem-flat.h
> index 97ca0ddce6..d5b6d99ae4 100644
> --- a/include/hw/misc/ivshmem-flat.h
> +++ b/include/hw/misc/ivshmem-flat.h
> @@ -25,6 +25,12 @@
>    *    socket
>    *  + QOM property "shmem-size" sets the size of the RAM region shared between
>    *    the device and the ivshmem server
> + *  + QOM property "x-bus-address-iomem" is the base address of the I/O region
> + *    on the main system bus
> + *  + QOM property "x-bus-address-shmem" is the base address of the shared RAM
> + *    region on the main system bus
> + *  + QOM property "x-irq-qompath" is the QOM path of the interrupt being
> + *    notified
>    *  + sysbus MMIO region 0: device I/O mapped registers
>    *  + sysbus MMIO region 1: shared memory with ivshmem server
>    *  + sysbus IRQ 0: single output interrupt
> @@ -68,9 +74,11 @@ struct IvshmemFTState {
>   
>       /* IRQ */
>       qemu_irq irq;
> +    char *x_sysbus_irq_qompath;
>   
>       /* I/O registers */
>       MemoryRegion iomem;
> +    uint64_t x_sysbus_mmio_addr;
>       uint32_t intmask;
>       uint32_t intstatus;
>       uint32_t ivposition;
> @@ -80,6 +88,7 @@ struct IvshmemFTState {
>       MemoryRegion shmem;
>       int shmem_fd;
>       uint32_t shmem_size;
> +    uint64_t x_sysbus_shmem_addr;
>   };
>   
>   #endif /* IVSHMEM_FLAT_H */
> diff --git a/hw/core/sysbus-fdt.c b/hw/core/sysbus-fdt.c
> index eebcd28f9a..40d7356cae 100644
> --- a/hw/core/sysbus-fdt.c
> +++ b/hw/core/sysbus-fdt.c
> @@ -31,6 +31,7 @@
>   #include "qemu/error-report.h"
>   #include "sysemu/device_tree.h"
>   #include "sysemu/tpm.h"
> +#include "hw/misc/ivshmem-flat.h"
>   #include "hw/platform-bus.h"
>   #include "hw/vfio/vfio-platform.h"
>   #include "hw/vfio/vfio-calxeda-xgmac.h"
> @@ -495,6 +496,7 @@ static const BindingEntry bindings[] = {
>       TYPE_BINDING(TYPE_TPM_TIS_SYSBUS, add_tpm_tis_fdt_node),
>   #endif
>       TYPE_BINDING(TYPE_RAMFB_DEVICE, no_fdt_node),
> +    TYPE_BINDING(TYPE_IVSHMEM_FLAT, no_fdt_node),
>       TYPE_BINDING("", NULL), /* last element */
>   };
>   
> diff --git a/hw/misc/ivshmem-flat.c b/hw/misc/ivshmem-flat.c
> index 998c78a5ee..433a47c8d3 100644
> --- a/hw/misc/ivshmem-flat.c
> +++ b/hw/misc/ivshmem-flat.c
> @@ -421,16 +421,75 @@ static bool ivshmem_flat_connect_server(DeviceState *dev, Error **errp)
>       return true;
>   }
>   
> +static bool ivshmem_flat_sysbus_wire(DeviceState *dev, Error **errp)
> +{
> +    IvshmemFTState *s = IVSHMEM_FLAT(dev);
> +    SysBusDevice *sbd = SYS_BUS_DEVICE(dev);
> +
> +    if (s->x_sysbus_mmio_addr != UINT64_MAX) {
> +        trace_ivshmem_flat_mmio_map(s->x_sysbus_mmio_addr);
> +        sysbus_mmio_map(sbd, 0, s->x_sysbus_mmio_addr);
> +    }
> +    if (s->x_sysbus_shmem_addr != UINT64_MAX) {
> +        trace_ivshmem_flat_shmem_map(s->x_sysbus_shmem_addr);
> +        sysbus_mmio_map(sbd, 1, s->x_sysbus_shmem_addr);
> +    }
> +
> +    /* Check for input IRQ line, if it's provided, connect it. */
> +    if (s->x_sysbus_irq_qompath) {
> +        Object *oirq;
> +        bool ambiguous = false;
> +        qemu_irq irq;
> +
> +        oirq = object_resolve_path_type(s->x_sysbus_irq_qompath, TYPE_IRQ,
> +                                        &ambiguous);
> +        if (ambiguous) {
> +            error_setg(errp, "Specified IRQ is ambiguous. Can't create"
> +                             " ivshmem-flat device.");
> +            return false;
> +        }
> +
> +        if (!oirq) {
> +            error_setg(errp, "Can't resolve IRQ QOM path.");
> +            return false;
> +        }
> +        irq = (qemu_irq)oirq;
> +        trace_ivshmem_flat_irq_resolved(s->x_sysbus_irq_qompath);
> +
> +        /*
> +         * Connect device out irq line to interrupt controller input irq line.
> +         */
> +        sysbus_connect_irq(sbd, 0, irq);
> +    } else {
> +       /*
> +        * If input IRQ is not provided, warn user the device won't be able
> +        * to trigger any interrupts.
> +        */
> +        warn_report("Input IRQ not specified, device won't be able"
> +                    " to handle IRQs!");
> +    }
> +
> +    return true;
> +}
> +
>   static void ivshmem_flat_realize(DeviceState *dev, Error **errp)
>   {
>       if (!ivshmem_flat_connect_server(dev, errp)) {
>           return;
>       }
> +    if (!ivshmem_flat_sysbus_wire(dev, errp)) {
> +        return;
> +    }
>   }
>   
>   static Property ivshmem_flat_props[] = {
>       DEFINE_PROP_CHR("chardev", IvshmemFTState, server_chr),
>       DEFINE_PROP_UINT32("shmem-size", IvshmemFTState, shmem_size, 4 * MiB),
> +    DEFINE_PROP_STRING("x-irq-qompath", IvshmemFTState, x_sysbus_irq_qompath),
> +    DEFINE_PROP_UINT64("x-bus-address-iomem", IvshmemFTState,
> +                       x_sysbus_mmio_addr, UINT64_MAX),
> +    DEFINE_PROP_UINT64("x-bus-address-shmem", IvshmemFTState,
> +                       x_sysbus_shmem_addr, UINT64_MAX),
>       DEFINE_PROP_END_OF_LIST(),
>   };
>   
> @@ -439,13 +498,11 @@ static void ivshmem_flat_class_init(ObjectClass *klass, void *data)
>       DeviceClass *dc = DEVICE_CLASS(klass);
>   
>       dc->hotpluggable = true;
> +    dc->user_creatable = true;
>       dc->realize = ivshmem_flat_realize;
>   
>       set_bit(DEVICE_CATEGORY_MISC, dc->categories);
>       device_class_set_props(dc, ivshmem_flat_props);
> -
> -    /* Reason: Must be wired up in code (sysbus MRs and IRQ) */
> -    dc->user_creatable = false;
>   }
>   
>   static const TypeInfo ivshmem_flat_info = {
> 


  parent reply	other threads:[~2024-02-20 16:03 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-16 14:44 [PATCH v2 0/6] Add ivshmem-flat device Philippe Mathieu-Daudé
2024-02-16 14:44 ` [PATCH v2 1/6] hw/misc/ivshmem: " Philippe Mathieu-Daudé
2024-02-16 14:44 ` [PATCH v2 2/6] hw/misc/ivshmem-flat: Allow device to wire itself on sysbus Philippe Mathieu-Daudé
2024-02-20 15:50   ` Gustavo Romero
2024-02-20 16:02   ` Gustavo Romero [this message]
2024-02-16 14:44 ` [PATCH v2 3/6] hw/arm: Allow some machines to use the ivshmem-flat device Philippe Mathieu-Daudé
2024-02-16 14:44 ` [PATCH v2 4/6] hw/misc/ivshmem: Rename ivshmem to ivshmem-pci Philippe Mathieu-Daudé
2024-02-16 14:44 ` [PATCH v2 5/6] tests/qtest: Reorganize common code in ivshmem-test Philippe Mathieu-Daudé
2024-02-19 13:54   ` Thomas Huth
2024-02-16 14:44 ` [PATCH v2 6/6] tests/qtest: Add ivshmem-flat test Philippe Mathieu-Daudé
2024-02-19 14:04   ` Thomas Huth
2024-02-20 16:01   ` Gustavo Romero
2024-02-20 15:42 ` [PATCH v2 0/6] Add ivshmem-flat device Gustavo Romero

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=b03a659f-fef6-bb4e-b258-3e5a82b7edb4@linaro.org \
    --to=gustavo.romero@linaro.org \
    --cc=alex.bennee@linaro.org \
    --cc=anton.kochkov@proton.me \
    --cc=lvivier@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=philmd@linaro.org \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=thuth@redhat.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).