Re: [Qemu-devel] [PATCH V9 07/16] xen: Add the Xen platform pci device

[Anthony Liguori <anthony@codemonkey.ws>]

On 01/25/2011 08:29 AM, anthony.perard@citrix.com wrote:
> From: Steven Smith<ssmith@xensource.com>
>
> Introduce a new emulated PCI device, specific to fully virtualized Xen
> guests.  The device is necessary for PV on HVM drivers to work.
>
> Signed-off-by: Steven Smith<ssmith@xensource.com>
> Signed-off-by: Anthony PERARD<anthony.perard@citrix.com>
> Signed-off-by: Stefano Stabellini<stefano.stabellini@eu.citrix.com>
> ---
>   Makefile.target     |    1 +
>   hw/hw.h             |    3 +
>   hw/pci_ids.h        |    2 +
>   hw/xen_machine_fv.c |    3 +
>   hw/xen_platform.c   |  427 +++++++++++++++++++++++++++++++++++++++++++++++++++
>   hw/xen_platform.h   |    8 +
>   6 files changed, 444 insertions(+), 0 deletions(-)
>   create mode 100644 hw/xen_platform.c
>   create mode 100644 hw/xen_platform.h
>
> diff --git a/Makefile.target b/Makefile.target
> index 8b5fe36..8126da9 100644
> --- a/Makefile.target
> +++ b/Makefile.target
> @@ -218,6 +218,7 @@ obj-i386-$(CONFIG_XEN) += xen_machine_pv.o xen_domainbuild.o
>
>   # xen full virtualized machine
>   obj-i386-$(CONFIG_XEN) += xen_machine_fv.o
> +obj-i386-$(CONFIG_XEN) += xen_platform.o
>
>   # Inter-VM PCI shared memory
>   obj-$(CONFIG_KVM) += ivshmem.o
> diff --git a/hw/hw.h b/hw/hw.h
> index dd993de..298df31 100644
> --- a/hw/hw.h
> +++ b/hw/hw.h
> @@ -672,6 +672,9 @@ extern const VMStateDescription vmstate_i2c_slave;
>   #define VMSTATE_INT32_LE(_f, _s)                                   \
>       VMSTATE_SINGLE(_f, _s, 0, vmstate_info_int32_le, int32_t)
>
> +#define VMSTATE_UINT8_TEST(_f, _s, _t)                               \
> +    VMSTATE_SINGLE_TEST(_f, _s, _t, 0, vmstate_info_uint8, uint8_t)
> +
>   #define VMSTATE_UINT16_TEST(_f, _s, _t)                               \
>       VMSTATE_SINGLE_TEST(_f, _s, _t, 0, vmstate_info_uint16, uint16_t)
>
> diff --git a/hw/pci_ids.h b/hw/pci_ids.h
> index ea3418c..6e9eabc 100644
> --- a/hw/pci_ids.h
> +++ b/hw/pci_ids.h
> @@ -108,3 +108,5 @@
>   #define PCI_DEVICE_ID_INTEL_82371AB      0x7111
>   #define PCI_DEVICE_ID_INTEL_82371AB_2    0x7112
>   #define PCI_DEVICE_ID_INTEL_82371AB_3    0x7113
> +
> +#define PCI_VENDOR_ID_XENSOURCE          0x5853
> diff --git a/hw/xen_machine_fv.c b/hw/xen_machine_fv.c
> index 657c1e8..2f51625 100644
> --- a/hw/xen_machine_fv.c
> +++ b/hw/xen_machine_fv.c
> @@ -36,6 +36,7 @@
>
>   #include "xen_common.h"
>   #include "xen/hvm/hvm_info_table.h"
> +#include "xen_platform.h"
>
>   #define MAX_IDE_BUS 2
>
> @@ -89,6 +90,8 @@ static void xen_init_fv(ram_addr_t ram_size,
>
>       pc_vga_init(pci_bus);
>
> +    pci_xen_platform_init(pci_bus);
> +
>       /* init basic PC hardware */
>       pc_basic_device_init(isa_irq,&floppy_controller,&rtc_state);
>
> diff --git a/hw/xen_platform.c b/hw/xen_platform.c
> new file mode 100644
> index 0000000..abfcdb7
> --- /dev/null
> +++ b/hw/xen_platform.c
> @@ -0,0 +1,427 @@
> +/*
> + * XEN platform pci device, formerly known as the event channel device
> + *
> + * Copyright (c) 2003-2004 Intel Corp.
> + * Copyright (c) 2006 XenSource
> + *
> + * 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
> + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
> + * copies of the Software, and to permit persons to whom the Software is
> + * furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice shall be included in
> + * all copies or substantial portions of the Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
> + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
> + * THE SOFTWARE.
> + */
> +
> +#include "hw.h"
> +#include "pc.h"
> +#include "pci.h"
> +#include "irq.h"
> +#include "xen_common.h"
> +#include "net.h"
> +#include "xen_platform.h"
> +#include "xen_backend.h"
> +#include "qemu-log.h"
> +#include "rwhandler.h"
> +
> +#include<assert.h>
> +#include<xenguest.h>
> +
> +//#define DEBUG_PLATFORM
> +
> +#ifdef DEBUG_PLATFORM
> +#define DPRINTF(fmt, ...) do { \
> +    fprintf(stderr, "xen_platform: " fmt, ## __VA_ARGS__); \
> +} while (0)
> +#else
> +#define DPRINTF(fmt, ...) do { } while (0)
> +#endif
> +
> +#define PFFLAG_ROM_LOCK 1 /* Sets whether ROM memory area is RW or RO */
> +
> +typedef struct PCIXenPlatformState {
> +    PCIDevice  pci_dev;
> +    uint8_t flags; /* used only for version_id == 2 */
> +    int drivers_blacklisted;
> +    uint16_t driver_product_version;
> +
> +    /* Log from guest drivers */
> +    int throttling_disabled;
> +    char log_buffer[4096];
> +    int log_buffer_off;
> +} PCIXenPlatformState;
> +
> +#define XEN_PLATFORM_IOPORT 0x10
> +
> +/* We throttle access to dom0 syslog, to avoid DOS attacks.  This is
> +   modelled as a token bucket, with one token for every byte of log.
> +   The bucket size is 128KB (->1024 lines of 128 bytes each) and
> +   refills at 256B/s.  It starts full.  The guest is blocked if no
> +   tokens are available when it tries to generate a log message. */
> +#define BUCKET_MAX_SIZE (128*1024)
> +#define BUCKET_FILL_RATE 256
> +
> +static void throttle(PCIXenPlatformState *s, unsigned count)
> +{
> +    static unsigned available;
> +    static int64_t last_refill;
> +    static int started;
> +    static int warned;
> +
> +    int64_t waiting_for, now;
> +    int64_t delay;
> +
> +    if (s->throttling_disabled) {
> +        return;
> +    }
> +
> +    if (!started) {
> +        last_refill = qemu_get_clock_ns(rt_clock);
> +        available = BUCKET_MAX_SIZE;
> +        started = 1;
> +    }
> +
> +    if (count>  BUCKET_MAX_SIZE) {
> +        DPRINTF("tried to get %u tokens, but bucket size is %u\n",
> +                BUCKET_MAX_SIZE, count);
> +        exit(1);
> +    }
> +
> +    if (available<  count) {
> +        /* The bucket is empty.  Refill it */
> +
> +        /* When will it be full enough to handle this request? */
> +        delay = muldiv64(count - available, 1000000000, BUCKET_FILL_RATE);
> +
> +        waiting_for = last_refill + delay;
> +
> +        /* How long do we have to wait? (might be negative) */
> +        waiting_for = waiting_for - qemu_get_clock_ns(rt_clock);
> +
> +        /* Wait for it. */
> +        if (waiting_for>  0) {
> +            struct timespec ts;
> +            if (!warned) {
> +                DPRINTF("throttling guest access to syslog");
> +                warned = 1;
> +            }
> +            ts.tv_sec = waiting_for / 1000000000;
> +            ts.tv_nsec = waiting_for % 1000000000;
> +            while (nanosleep(&ts,&ts)<  0&&  errno == EINTR) {
> +            }
> +        }
> +
> +        /* Refill */
> +        now = qemu_get_clock_ns(rt_clock);
> +        available += muldiv64(now - last_refill,
> +                              BUCKET_FILL_RATE,
> +                              1000000000);
> +        if (available>  BUCKET_MAX_SIZE) {
> +            available = BUCKET_MAX_SIZE;
> +        }
> +        last_refill = now;
> +    }
> +
> +    assert(available>= count);
> +
> +    available -= count;
> +}
> +
> +/* Send bytes to syslog */
> +static void log_writeb(PCIXenPlatformState *s, char val)
> +{
> +    if (val == '\n' || s->log_buffer_off == sizeof(s->log_buffer) - 1) {
> +        /* Flush buffer */
> +        s->log_buffer[s->log_buffer_off] = 0;
> +        throttle(s, s->log_buffer_off);
> +        DPRINTF("%s\n", s->log_buffer);
> +        s->log_buffer_off = 0;
> +    } else {
> +        s->log_buffer[s->log_buffer_off++] = val;
> +    }
> +}
>    

This doesn't belong in qemu.  Just pipe it out to a character device and 
you can direct that to whatever you want with whatever type of 
throttling you see fit.

Hard coding a policy like this in qemu is wrong.

> +
> +/* Xen Platform, Fixed IOPort */
> +
> +static void platform_fixed_ioport_writew(void *opaque, uint32_t addr, uint32_t val)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    switch (addr - XEN_PLATFORM_IOPORT) {
> +    case 0:
> +        /* TODO: */
> +        /* Unplug devices.  Value is a bitmask of which devices to
> +           unplug, with bit 0 the IDE devices, bit 1 the network
> +           devices, and bit 2 the non-primary-master IDE devices. */
> +        break;
> +    case 2:
> +        switch (val) {
> +        case 1:
> +            DPRINTF("Citrix Windows PV drivers loaded in guest\n");
> +            break;
> +        case 0:
> +            DPRINTF("Guest claimed to be running PV product 0?\n");
> +            break;
> +        default:
> +            DPRINTF("Unknown PV product %d loaded in guest\n", val);
> +            break;
> +        }
> +        s->driver_product_version = val;
> +        break;
> +    }
> +}
> +
> +static void platform_fixed_ioport_writel(void *opaque, uint32_t addr,
> +                                         uint32_t val)
> +{
> +    switch (addr - XEN_PLATFORM_IOPORT) {
> +    case 0:
> +        /* PV driver version */
> +        break;
> +    }
> +}
> +
> +static void platform_fixed_ioport_writeb(void *opaque, uint32_t addr, uint32_t val)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    switch (addr - XEN_PLATFORM_IOPORT) {
> +    case 0: /* Platform flags */ {
> +        hvmmem_type_t mem_type = (val&  PFFLAG_ROM_LOCK) ?
> +            HVMMEM_ram_ro : HVMMEM_ram_rw;
> +        if (xc_hvm_set_mem_type(xen_xc, xen_domid, mem_type, 0xc0, 0x40)) {
> +            DPRINTF("unable to change ro/rw state of ROM memory area!\n");
> +        } else {
> +            s->flags = val&  PFFLAG_ROM_LOCK;
> +            DPRINTF("changed ro/rw state of ROM memory area. now is %s state.\n",
> +                    (mem_type == HVMMEM_ram_ro ? "ro":"rw"));
> +        }
> +        break;
> +    }
> +    case 2:
> +        log_writeb(s, val);
> +        break;
> +    }
> +}
> +
> +static uint32_t platform_fixed_ioport_readw(void *opaque, uint32_t addr)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    switch (addr - XEN_PLATFORM_IOPORT) {
> +    case 0:
> +        if (s->drivers_blacklisted) {
> +            /* The drivers will recognise this magic number and refuse
> +             * to do anything. */
> +            return 0xd249;
> +        } else {
> +            /* Magic value so that you can identify the interface. */
> +            return 0x49d2;
> +        }
> +    default:
> +        return 0xffff;
> +    }
> +}
> +
> +static uint32_t platform_fixed_ioport_readb(void *opaque, uint32_t addr)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    switch (addr - XEN_PLATFORM_IOPORT) {
> +    case 0:
> +        /* Platform flags */
> +        return s->flags;
> +    case 2:
> +        /* Version number */
> +        return 1;
> +    default:
> +        return 0xff;
> +    }
> +}
> +
> +static void platform_fixed_ioport_reset(void *opaque)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    platform_fixed_ioport_writeb(s, XEN_PLATFORM_IOPORT, 0);
> +}
> +
> +static void platform_fixed_ioport_init(PCIXenPlatformState* s)
> +{
> +    register_ioport_write(XEN_PLATFORM_IOPORT, 16, 4, platform_fixed_ioport_writel, s);
> +    register_ioport_write(XEN_PLATFORM_IOPORT, 16, 2, platform_fixed_ioport_writew, s);
> +    register_ioport_write(XEN_PLATFORM_IOPORT, 16, 1, platform_fixed_ioport_writeb, s);
> +    register_ioport_read(XEN_PLATFORM_IOPORT, 16, 2, platform_fixed_ioport_readw, s);
> +    register_ioport_read(XEN_PLATFORM_IOPORT, 16, 1, platform_fixed_ioport_readb, s);
> +}
> +
> +/* Xen Platform PCI Device */
> +
> +static uint32_t xen_platform_ioport_readb(void *opaque, uint32_t addr)
> +{
> +    addr&= 0xff;
> +
> +    if (addr == 0) {
> +        return platform_fixed_ioport_readb(opaque, XEN_PLATFORM_IOPORT);
> +    } else {
> +        return ~0u;
> +    }
> +}
> +
> +static void xen_platform_ioport_writeb(void *opaque, uint32_t addr, uint32_t val)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    addr&= 0xff;
> +    val&= 0xff;
> +
> +    switch (addr) {
> +    case 0: /* Platform flags */
> +        platform_fixed_ioport_writeb(opaque, XEN_PLATFORM_IOPORT, val);
> +        break;
> +    case 8:
> +        log_writeb(s, val);
> +        break;
> +    default:
> +        break;
> +    }
> +}
> +
> +static void platform_ioport_map(PCIDevice *pci_dev, int region_num, pcibus_t addr, pcibus_t size, int type)
> +{
> +    PCIXenPlatformState *d = DO_UPCAST(PCIXenPlatformState, pci_dev, pci_dev);
> +
> +    register_ioport_write(addr, size, 1, xen_platform_ioport_writeb, d);
> +    register_ioport_read(addr, size, 1, xen_platform_ioport_readb, d);
> +}
> +
> +static uint32_t platform_mmio_read(ReadWriteHandler *handler, pcibus_t addr, int len)
> +{
> +    DPRINTF("Warning: attempted read from physical address "
> +            "0x" TARGET_FMT_plx " in xen platform mmio space\n", addr);
> +
> +    return 0;
> +}
> +
> +static void platform_mmio_write(ReadWriteHandler *handler, pcibus_t addr,
> +                                uint32_t val, int len)
> +{
> +    DPRINTF("Warning: attempted write of 0x%x to physical "
> +            "address 0x" TARGET_FMT_plx " in xen platform mmio space\n",
> +            val, addr);
> +}
> +
> +static ReadWriteHandler platform_mmio_handler = {
> +    .read =&platform_mmio_read,
> +    .write =&platform_mmio_write,
> +};
> +
> +static void platform_mmio_map(PCIDevice *d, int region_num,
> +                              pcibus_t addr, pcibus_t size, int type)
> +{
> +    int mmio_io_addr;
> +
> +    mmio_io_addr = cpu_register_io_memory_simple(&platform_mmio_handler,
> +                                                 DEVICE_NATIVE_ENDIAN);
> +
> +    cpu_register_physical_memory(addr, size, mmio_io_addr);
> +}
> +
> +static int xen_platform_post_load(void *opaque, int version_id)
> +{
> +    PCIXenPlatformState *s = opaque;
> +
> +    platform_fixed_ioport_writeb(s, XEN_PLATFORM_IOPORT, s->flags);
> +
> +    return 0;
> +}
> +
> +static const VMStateDescription vmstate_xen_platform = {
> +    .name = "platform",
> +    .version_id = 4,
> +    .minimum_version_id = 4,
> +    .minimum_version_id_old = 4,
> +    .post_load = xen_platform_post_load,
> +    .fields = (VMStateField []) {
> +        VMSTATE_PCI_DEVICE(pci_dev, PCIXenPlatformState),
> +        VMSTATE_UINT8(flags, PCIXenPlatformState),
> +        VMSTATE_END_OF_LIST()
> +    }
> +};
> +
> +static int xen_platform_initfn(PCIDevice *dev)
> +{
> +    PCIXenPlatformState *d = DO_UPCAST(PCIXenPlatformState, pci_dev, dev);
> +    uint8_t *pci_conf;
> +
> +    pci_conf = d->pci_dev.config;
> +
> +    pci_config_set_vendor_id(pci_conf, PCI_VENDOR_ID_XENSOURCE);
> +    pci_config_set_device_id(pci_conf, 0x0001);
> +    pci_set_word(pci_conf + PCI_COMMAND, PCI_COMMAND_IO | PCI_COMMAND_MEMORY);
> +
> +    pci_config_set_revision(pci_conf, 1);
> +    pci_config_set_prog_interface(pci_conf, 0);
> +
> +    pci_config_set_class(pci_conf, PCI_CLASS_OTHERS<<  8 | 0x80);
> +
> +    pci_conf[PCI_INTERRUPT_PIN] = 1;
> +
> +    /* Microsoft WHQL requires non-zero subsystem IDs. */
> +    /* http://www.pcisig.com/reflector/msg02205.html.  */
> +    pci_set_word(pci_conf + PCI_SUBSYSTEM_VENDOR_ID, pci_conf[PCI_VENDOR_ID]);
> +    pci_set_word(pci_conf + PCI_SUBSYSTEM_ID, 0x0001);
>    

The generic PCI layer handles this FWIW.

Regards,

Anthony Liguori

> +
> +    pci_register_bar(&d->pci_dev, 0, 0x100,
> +            PCI_BASE_ADDRESS_SPACE_IO, platform_ioport_map);
> +
> +    /* reserve 16MB mmio address for share memory*/
> +    pci_register_bar(&d->pci_dev, 1, 0x1000000,
> +            PCI_BASE_ADDRESS_MEM_PREFETCH, platform_mmio_map);
> +
> +    platform_fixed_ioport_init(d);
> +
> +    return 0;
> +}
> +
> +static void platform_reset(DeviceState *dev)
> +{
> +    PCIXenPlatformState *s = DO_UPCAST(PCIXenPlatformState, pci_dev.qdev, dev);
> +
> +    platform_fixed_ioport_reset(s);
> +}
> +
> +void pci_xen_platform_init(PCIBus *bus)
> +{
> +    PCIDevice *dev;
> +
> +    dev = pci_create(bus, -1, "xen-platform");
> +
> +    qdev_init_nofail(&dev->qdev);
> +}
> +
> +static PCIDeviceInfo xen_platform_info = {
> +    .init = xen_platform_initfn,
> +    .qdev.name = "xen-platform",
> +    .qdev.desc = "XEN platform pci device",
> +    .qdev.size = sizeof(PCIXenPlatformState),
> +    .qdev.vmsd =&vmstate_xen_platform,
> +    .qdev.reset = platform_reset,
> +};
> +
> +static void xen_platform_register(void)
> +{
> +    pci_qdev_register(&xen_platform_info);
> +}
> +
> +device_init(xen_platform_register);
> diff --git a/hw/xen_platform.h b/hw/xen_platform.h
> new file mode 100644
> index 0000000..574eecd
> --- /dev/null
> +++ b/hw/xen_platform.h
> @@ -0,0 +1,8 @@
> +#ifndef XEN_PLATFORM_H
> +#define XEN_PLATFORM_H
> +
> +#include "hw/pci.h"
> +
> +void pci_xen_platform_init(PCIBus *bus);
> +
> +#endif
>