[qemu.git] / hw / vfio / ccw.c

/*
 * vfio based subchannel assignment support
 *
 * Copyright 2017 IBM Corp.
 * Copyright 2019 Red Hat, Inc.
 *
 * Author(s): Dong Jia Shi <[email protected]>
 *            Xiao Feng Ren <[email protected]>
 *            Pierre Morel <[email protected]>
 *            Cornelia Huck <[email protected]>
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or (at
 * your option) any later version. See the COPYING file in the top-level
 * directory.
 */

#include "qemu/osdep.h"
#include <linux/vfio.h>
#include <linux/vfio_ccw.h>
#include <sys/ioctl.h>

#include "qapi/error.h"
#include "hw/sysbus.h"
#include "hw/vfio/vfio.h"
#include "hw/vfio/vfio-common.h"
#include "hw/s390x/s390-ccw.h"
#include "hw/s390x/vfio-ccw.h"
#include "hw/qdev-properties.h"
#include "hw/s390x/ccw-device.h"
#include "exec/address-spaces.h"
#include "qemu/error-report.h"
#include "qemu/main-loop.h"
#include "qemu/module.h"

struct VFIOCCWDevice {
    S390CCWDevice cdev;
    VFIODevice vdev;
    uint64_t io_region_size;
    uint64_t io_region_offset;
    struct ccw_io_region *io_region;
    uint64_t async_cmd_region_size;
    uint64_t async_cmd_region_offset;
    struct ccw_cmd_region *async_cmd_region;
    EventNotifier io_notifier;
    bool force_orb_pfch;
    bool warned_orb_pfch;
};

static inline void warn_once_pfch(VFIOCCWDevice *vcdev, SubchDev *sch,
                                  const char *msg)
{
    warn_report_once_cond(&vcdev->warned_orb_pfch,
                          "vfio-ccw (devno %x.%x.%04x): %s",
                          sch->cssid, sch->ssid, sch->devno, msg);
}

static void vfio_ccw_compute_needs_reset(VFIODevice *vdev)
{
    vdev->needs_reset = false;
}

/*
 * We don't need vfio_hot_reset_multi and vfio_eoi operations for
 * vfio_ccw device now.
 */
struct VFIODeviceOps vfio_ccw_ops = {
    .vfio_compute_needs_reset = vfio_ccw_compute_needs_reset,
};

static IOInstEnding vfio_ccw_handle_request(SubchDev *sch)
{
    S390CCWDevice *cdev = sch->driver_data;
    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
    struct ccw_io_region *region = vcdev->io_region;
    int ret;

    if (!(sch->orb.ctrl0 & ORB_CTRL0_MASK_PFCH) && vcdev->force_orb_pfch) {
        sch->orb.ctrl0 |= ORB_CTRL0_MASK_PFCH;
        warn_once_pfch(vcdev, sch, "PFCH flag forced");
    }

    QEMU_BUILD_BUG_ON(sizeof(region->orb_area) != sizeof(ORB));
    QEMU_BUILD_BUG_ON(sizeof(region->scsw_area) != sizeof(SCSW));
    QEMU_BUILD_BUG_ON(sizeof(region->irb_area) != sizeof(IRB));

    memset(region, 0, sizeof(*region));

    memcpy(region->orb_area, &sch->orb, sizeof(ORB));
    memcpy(region->scsw_area, &sch->curr_status.scsw, sizeof(SCSW));

again:
    ret = pwrite(vcdev->vdev.fd, region,
                 vcdev->io_region_size, vcdev->io_region_offset);
    if (ret != vcdev->io_region_size) {
        if (errno == EAGAIN) {
            goto again;
        }
        error_report("vfio-ccw: write I/O region failed with errno=%d", errno);
        ret = -errno;
    } else {
        ret = region->ret_code;
    }
    switch (ret) {
    case 0:
        return IOINST_CC_EXPECTED;
    case -EBUSY:
        return IOINST_CC_BUSY;
    case -ENODEV:
    case -EACCES:
        return IOINST_CC_NOT_OPERATIONAL;
    case -EFAULT:
    default:
        sch_gen_unit_exception(sch);
        css_inject_io_interrupt(sch);
        return IOINST_CC_EXPECTED;
    }
}

static int vfio_ccw_handle_clear(SubchDev *sch)
{
    S390CCWDevice *cdev = sch->driver_data;
    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
    struct ccw_cmd_region *region = vcdev->async_cmd_region;
    int ret;

    if (!vcdev->async_cmd_region) {
        /* Async command region not available, fall back to emulation */
        return -ENOSYS;
    }

    memset(region, 0, sizeof(*region));
    region->command = VFIO_CCW_ASYNC_CMD_CSCH;

again:
    ret = pwrite(vcdev->vdev.fd, region,
                 vcdev->async_cmd_region_size, vcdev->async_cmd_region_offset);
    if (ret != vcdev->async_cmd_region_size) {
        if (errno == EAGAIN) {
            goto again;
        }
        error_report("vfio-ccw: write cmd region failed with errno=%d", errno);
        ret = -errno;
    } else {
        ret = region->ret_code;
    }
    switch (ret) {
    case 0:
    case -ENODEV:
    case -EACCES:
        return 0;
    case -EFAULT:
    default:
        sch_gen_unit_exception(sch);
        css_inject_io_interrupt(sch);
        return 0;
    }
}

static int vfio_ccw_handle_halt(SubchDev *sch)
{
    S390CCWDevice *cdev = sch->driver_data;
    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
    struct ccw_cmd_region *region = vcdev->async_cmd_region;
    int ret;

    if (!vcdev->async_cmd_region) {
        /* Async command region not available, fall back to emulation */
        return -ENOSYS;
    }

    memset(region, 0, sizeof(*region));
    region->command = VFIO_CCW_ASYNC_CMD_HSCH;

again:
    ret = pwrite(vcdev->vdev.fd, region,
                 vcdev->async_cmd_region_size, vcdev->async_cmd_region_offset);
    if (ret != vcdev->async_cmd_region_size) {
        if (errno == EAGAIN) {
            goto again;
        }
        error_report("vfio-ccw: write cmd region failed with errno=%d", errno);
        ret = -errno;
    } else {
        ret = region->ret_code;
    }
    switch (ret) {
    case 0:
    case -EBUSY:
    case -ENODEV:
    case -EACCES:
        return 0;
    case -EFAULT:
    default:
        sch_gen_unit_exception(sch);
        css_inject_io_interrupt(sch);
        return 0;
    }
}

static void vfio_ccw_reset(DeviceState *dev)
{
    CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
    S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);

    ioctl(vcdev->vdev.fd, VFIO_DEVICE_RESET);
}

static void vfio_ccw_io_notifier_handler(void *opaque)
{
    VFIOCCWDevice *vcdev = opaque;
    struct ccw_io_region *region = vcdev->io_region;
    S390CCWDevice *cdev = S390_CCW_DEVICE(vcdev);
    CcwDevice *ccw_dev = CCW_DEVICE(cdev);
    SubchDev *sch = ccw_dev->sch;
    SCHIB *schib = &sch->curr_status;
    SCSW s;
    IRB irb;
    int size;

    if (!event_notifier_test_and_clear(&vcdev->io_notifier)) {
        return;
    }

    size = pread(vcdev->vdev.fd, region, vcdev->io_region_size,
                 vcdev->io_region_offset);
    if (size == -1) {
        switch (errno) {
        case ENODEV:
            /* Generate a deferred cc 3 condition. */
            schib->scsw.flags |= SCSW_FLAGS_MASK_CC;
            schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
            schib->scsw.ctrl |= (SCSW_STCTL_ALERT | SCSW_STCTL_STATUS_PEND);
            goto read_err;
        case EFAULT:
            /* Memory problem, generate channel data check. */
            schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
            schib->scsw.cstat = SCSW_CSTAT_DATA_CHECK;
            schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
            schib->scsw.ctrl |= SCSW_STCTL_PRIMARY | SCSW_STCTL_SECONDARY |
                       SCSW_STCTL_ALERT | SCSW_STCTL_STATUS_PEND;
            goto read_err;
        default:
            /* Error, generate channel program check. */
            schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
            schib->scsw.cstat = SCSW_CSTAT_PROG_CHECK;
            schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
            schib->scsw.ctrl |= SCSW_STCTL_PRIMARY | SCSW_STCTL_SECONDARY |
                       SCSW_STCTL_ALERT | SCSW_STCTL_STATUS_PEND;
            goto read_err;
        }
    } else if (size != vcdev->io_region_size) {
        /* Information transfer error, generate channel-control check. */
        schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
        schib->scsw.cstat = SCSW_CSTAT_CHN_CTRL_CHK;
        schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
        schib->scsw.ctrl |= SCSW_STCTL_PRIMARY | SCSW_STCTL_SECONDARY |
                   SCSW_STCTL_ALERT | SCSW_STCTL_STATUS_PEND;
        goto read_err;
    }

    memcpy(&irb, region->irb_area, sizeof(IRB));

    /* Update control block via irb. */
    s = schib->scsw;
    copy_scsw_to_guest(&s, &irb.scsw);
    schib->scsw = s;

    /* If a uint check is pending, copy sense data. */
    if ((schib->scsw.dstat & SCSW_DSTAT_UNIT_CHECK) &&
        (schib->pmcw.chars & PMCW_CHARS_MASK_CSENSE)) {
        memcpy(sch->sense_data, irb.ecw, sizeof(irb.ecw));
    }

read_err:
    css_inject_io_interrupt(sch);
}

static void vfio_ccw_register_io_notifier(VFIOCCWDevice *vcdev, Error **errp)
{
    VFIODevice *vdev = &vcdev->vdev;
    struct vfio_irq_info *irq_info;
    size_t argsz;
    int fd;

    if (vdev->num_irqs < VFIO_CCW_IO_IRQ_INDEX + 1) {
        error_setg(errp, "vfio: unexpected number of io irqs %u",
                   vdev->num_irqs);
        return;
    }

    argsz = sizeof(*irq_info);
    irq_info = g_malloc0(argsz);
    irq_info->index = VFIO_CCW_IO_IRQ_INDEX;
    irq_info->argsz = argsz;
    if (ioctl(vdev->fd, VFIO_DEVICE_GET_IRQ_INFO,
              irq_info) < 0 || irq_info->count < 1) {
        error_setg_errno(errp, errno, "vfio: Error getting irq info");
        goto out_free_info;
    }

    if (event_notifier_init(&vcdev->io_notifier, 0)) {
        error_setg_errno(errp, errno,
                         "vfio: Unable to init event notifier for IO");
        goto out_free_info;
    }

    fd = event_notifier_get_fd(&vcdev->io_notifier);
    qemu_set_fd_handler(fd, vfio_ccw_io_notifier_handler, NULL, vcdev);

    if (vfio_set_irq_signaling(vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
                               VFIO_IRQ_SET_ACTION_TRIGGER, fd, errp)) {
        qemu_set_fd_handler(fd, NULL, NULL, vcdev);
        event_notifier_cleanup(&vcdev->io_notifier);
    }

out_free_info:
    g_free(irq_info);
}

static void vfio_ccw_unregister_io_notifier(VFIOCCWDevice *vcdev)
{
    Error *err = NULL;

    if (vfio_set_irq_signaling(&vcdev->vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
                               VFIO_IRQ_SET_ACTION_TRIGGER, -1, &err)) {
        error_reportf_err(err, VFIO_MSG_PREFIX, vcdev->vdev.name);
    }

    qemu_set_fd_handler(event_notifier_get_fd(&vcdev->io_notifier),
                        NULL, NULL, vcdev);
    event_notifier_cleanup(&vcdev->io_notifier);
}

static void vfio_ccw_get_region(VFIOCCWDevice *vcdev, Error **errp)
{
    VFIODevice *vdev = &vcdev->vdev;
    struct vfio_region_info *info;
    int ret;

    /* Sanity check device */
    if (!(vdev->flags & VFIO_DEVICE_FLAGS_CCW)) {
        error_setg(errp, "vfio: Um, this isn't a vfio-ccw device");
        return;
    }

    /*
     * We always expect at least the I/O region to be present. We also
     * may have a variable number of regions governed by capabilities.
     */
    if (vdev->num_regions < VFIO_CCW_CONFIG_REGION_INDEX + 1) {
        error_setg(errp, "vfio: too few regions (%u), expected at least %u",
                   vdev->num_regions, VFIO_CCW_CONFIG_REGION_INDEX + 1);
        return;
    }

    ret = vfio_get_region_info(vdev, VFIO_CCW_CONFIG_REGION_INDEX, &info);
    if (ret) {
        error_setg_errno(errp, -ret, "vfio: Error getting config info");
        return;
    }

    vcdev->io_region_size = info->size;
    if (sizeof(*vcdev->io_region) != vcdev->io_region_size) {
        error_setg(errp, "vfio: Unexpected size of the I/O region");
        g_free(info);
        return;
    }

    vcdev->io_region_offset = info->offset;
    vcdev->io_region = g_malloc0(info->size);

    /* check for the optional async command region */
    ret = vfio_get_dev_region_info(vdev, VFIO_REGION_TYPE_CCW,
                                   VFIO_REGION_SUBTYPE_CCW_ASYNC_CMD, &info);
    if (!ret) {
        vcdev->async_cmd_region_size = info->size;
        if (sizeof(*vcdev->async_cmd_region) != vcdev->async_cmd_region_size) {
            error_setg(errp, "vfio: Unexpected size of the async cmd region");
            g_free(vcdev->io_region);
            g_free(info);
            return;
        }
        vcdev->async_cmd_region_offset = info->offset;
        vcdev->async_cmd_region = g_malloc0(info->size);
    }

    g_free(info);
}

static void vfio_ccw_put_region(VFIOCCWDevice *vcdev)
{
    g_free(vcdev->async_cmd_region);
    g_free(vcdev->io_region);
}

static void vfio_ccw_put_device(VFIOCCWDevice *vcdev)
{
    g_free(vcdev->vdev.name);
    vfio_put_base_device(&vcdev->vdev);
}

static void vfio_ccw_get_device(VFIOGroup *group, VFIOCCWDevice *vcdev,
                                Error **errp)
{
    char *name = g_strdup_printf("%x.%x.%04x", vcdev->cdev.hostid.cssid,
                                 vcdev->cdev.hostid.ssid,
                                 vcdev->cdev.hostid.devid);
    VFIODevice *vbasedev;

    QLIST_FOREACH(vbasedev, &group->device_list, next) {
        if (strcmp(vbasedev->name, name) == 0) {
            error_setg(errp, "vfio: subchannel %s has already been attached",
                       name);
            goto out_err;
        }
    }

    /*
     * All vfio-ccw devices are believed to operate in a way compatible with
     * memory ballooning, ie. pages pinned in the host are in the current
     * working set of the guest driver and therefore never overlap with pages
     * available to the guest balloon driver.  This needs to be set before
     * vfio_get_device() for vfio common to handle the balloon inhibitor.
     */
    vcdev->vdev.balloon_allowed = true;

    if (vfio_get_device(group, vcdev->cdev.mdevid, &vcdev->vdev, errp)) {
        goto out_err;
    }

    vcdev->vdev.ops = &vfio_ccw_ops;
    vcdev->vdev.type = VFIO_DEVICE_TYPE_CCW;
    vcdev->vdev.name = name;
    vcdev->vdev.dev = &vcdev->cdev.parent_obj.parent_obj;

    return;

out_err:
    g_free(name);
}

static VFIOGroup *vfio_ccw_get_group(S390CCWDevice *cdev, Error **errp)
{
    char *tmp, group_path[PATH_MAX];
    ssize_t len;
    int groupid;

    tmp = g_strdup_printf("/sys/bus/css/devices/%x.%x.%04x/%s/iommu_group",
                          cdev->hostid.cssid, cdev->hostid.ssid,
                          cdev->hostid.devid, cdev->mdevid);
    len = readlink(tmp, group_path, sizeof(group_path));
    g_free(tmp);

    if (len <= 0 || len >= sizeof(group_path)) {
        error_setg(errp, "vfio: no iommu_group found");
        return NULL;
    }

    group_path[len] = 0;

    if (sscanf(basename(group_path), "%d", &groupid) != 1) {
        error_setg(errp, "vfio: failed to read %s", group_path);
        return NULL;
    }

    return vfio_get_group(groupid, &address_space_memory, errp);
}

static void vfio_ccw_realize(DeviceState *dev, Error **errp)
{
    VFIOGroup *group;
    CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
    S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
    S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
    Error *err = NULL;

    /* Call the class init function for subchannel. */
    if (cdc->realize) {
        cdc->realize(cdev, vcdev->vdev.sysfsdev, &err);
        if (err) {
            goto out_err_propagate;
        }
    }

    group = vfio_ccw_get_group(cdev, &err);
    if (!group) {
        goto out_group_err;
    }

    vfio_ccw_get_device(group, vcdev, &err);
    if (err) {
        goto out_device_err;
    }

    vfio_ccw_get_region(vcdev, &err);
    if (err) {
        goto out_region_err;
    }

    vfio_ccw_register_io_notifier(vcdev, &err);
    if (err) {
        goto out_notifier_err;
    }

    return;

out_notifier_err:
    vfio_ccw_put_region(vcdev);
out_region_err:
    vfio_ccw_put_device(vcdev);
out_device_err:
    vfio_put_group(group);
out_group_err:
    if (cdc->unrealize) {
        cdc->unrealize(cdev);
    }
out_err_propagate:
    error_propagate(errp, err);
}

static void vfio_ccw_unrealize(DeviceState *dev)
{
    CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
    S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
    VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
    S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
    VFIOGroup *group = vcdev->vdev.group;

    vfio_ccw_unregister_io_notifier(vcdev);
    vfio_ccw_put_region(vcdev);
    vfio_ccw_put_device(vcdev);
    vfio_put_group(group);

    if (cdc->unrealize) {
        cdc->unrealize(cdev);
    }
}

static Property vfio_ccw_properties[] = {
    DEFINE_PROP_STRING("sysfsdev", VFIOCCWDevice, vdev.sysfsdev),
    DEFINE_PROP_BOOL("force-orb-pfch", VFIOCCWDevice, force_orb_pfch, false),
    DEFINE_PROP_END_OF_LIST(),
};

static const VMStateDescription vfio_ccw_vmstate = {
    .name = "vfio-ccw",
    .unmigratable = 1,
};

static void vfio_ccw_class_init(ObjectClass *klass, void *data)
{
    DeviceClass *dc = DEVICE_CLASS(klass);
    S390CCWDeviceClass *cdc = S390_CCW_DEVICE_CLASS(klass);

    device_class_set_props(dc, vfio_ccw_properties);
    dc->vmsd = &vfio_ccw_vmstate;
    dc->desc = "VFIO-based subchannel assignment";
    set_bit(DEVICE_CATEGORY_MISC, dc->categories);
    dc->realize = vfio_ccw_realize;
    dc->unrealize = vfio_ccw_unrealize;
    dc->reset = vfio_ccw_reset;

    cdc->handle_request = vfio_ccw_handle_request;
    cdc->handle_halt = vfio_ccw_handle_halt;
    cdc->handle_clear = vfio_ccw_handle_clear;
}

static const TypeInfo vfio_ccw_info = {
    .name = TYPE_VFIO_CCW,
    .parent = TYPE_S390_CCW,
    .instance_size = sizeof(VFIOCCWDevice),
    .class_init = vfio_ccw_class_init,
};

static void register_vfio_ccw_type(void)
{
    type_register_static(&vfio_ccw_info);
}

type_init(register_vfio_ccw_type)
Commit	Line	Data
1dcac3e1 XFR	1	/*
	2	* vfio based subchannel assignment support
	3	*
	4	* Copyright 2017 IBM Corp.
8fadea24 CH	5	* Copyright 2019 Red Hat, Inc.
8fadea24 CH	6	*
1dcac3e1 XFR	7	* Author(s): Dong Jia Shi <[email protected]>
	8	* Xiao Feng Ren <[email protected]>
	9	* Pierre Morel <[email protected]>
8fadea24	10	* Cornelia Huck <[email protected]>
1dcac3e1	11	*
08b824aa CH	12	* This work is licensed under the terms of the GNU GPL, version 2 or (at
08b824aa CH	13	* your option) any later version. See the COPYING file in the top-level
1dcac3e1 XFR	14	* directory.
	15	*/
	16
e9808d09	17	#include "qemu/osdep.h"
1dcac3e1	18	#include <linux/vfio.h>
c14e706c	19	#include <linux/vfio_ccw.h>
1dcac3e1 XFR	20	#include <sys/ioctl.h>
1dcac3e1 XFR	21
1dcac3e1 XFR	22	#include "qapi/error.h"
	23	#include "hw/sysbus.h"
	24	#include "hw/vfio/vfio.h"
	25	#include "hw/vfio/vfio-common.h"
	26	#include "hw/s390x/s390-ccw.h"
44445d86	27	#include "hw/s390x/vfio-ccw.h"
a27bd6c7	28	#include "hw/qdev-properties.h"
1dcac3e1	29	#include "hw/s390x/ccw-device.h"
d791937f	30	#include "exec/address-spaces.h"
4886b3e9	31	#include "qemu/error-report.h"
db725815	32	#include "qemu/main-loop.h"
0b8fa32f	33	#include "qemu/module.h"
1dcac3e1	34
44445d86	35	struct VFIOCCWDevice {
1dcac3e1 XFR	36	S390CCWDevice cdev;
1dcac3e1 XFR	37	VFIODevice vdev;
c14e706c DJS	38	uint64_t io_region_size;
	39	uint64_t io_region_offset;
	40	struct ccw_io_region *io_region;
8fadea24 CH	41	uint64_t async_cmd_region_size;
	42	uint64_t async_cmd_region_offset;
	43	struct ccw_cmd_region *async_cmd_region;
4886b3e9	44	EventNotifier io_notifier;
9a51c9ee HP	45	bool force_orb_pfch;
9a51c9ee HP	46	bool warned_orb_pfch;
44445d86	47	};
1dcac3e1	48
9a51c9ee HP	49	static inline void warn_once_pfch(VFIOCCWDevice vcdev, SubchDev sch,
	50	const char *msg)
	51	{
c55510b7 CH	52	warn_report_once_cond(&vcdev->warned_orb_pfch,
	53	"vfio-ccw (devno %x.%x.%04x): %s",
	54	sch->cssid, sch->ssid, sch->devno, msg);
9a51c9ee HP	55	}
9a51c9ee HP	56
1dcac3e1 XFR	57	static void vfio_ccw_compute_needs_reset(VFIODevice *vdev)
	58	{
	59	vdev->needs_reset = false;
	60	}
	61
	62	/*
	63	* We don't need vfio_hot_reset_multi and vfio_eoi operations for
	64	* vfio_ccw device now.
	65	*/
	66	struct VFIODeviceOps vfio_ccw_ops = {
	67	.vfio_compute_needs_reset = vfio_ccw_compute_needs_reset,
	68	};
	69
66dc50f7	70	static IOInstEnding vfio_ccw_handle_request(SubchDev *sch)
8ca2b376	71	{
66dc50f7	72	S390CCWDevice *cdev = sch->driver_data;
8ca2b376 XFR	73	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	74	struct ccw_io_region *region = vcdev->io_region;
	75	int ret;
	76
24e58a7b JR	77	if (!(sch->orb.ctrl0 & ORB_CTRL0_MASK_PFCH) && vcdev->force_orb_pfch) {
	78	sch->orb.ctrl0 \|= ORB_CTRL0_MASK_PFCH;
	79	warn_once_pfch(vcdev, sch, "PFCH flag forced");
9a51c9ee HP	80	}
9a51c9ee HP	81
8ca2b376 XFR	82	QEMU_BUILD_BUG_ON(sizeof(region->orb_area) != sizeof(ORB));
	83	QEMU_BUILD_BUG_ON(sizeof(region->scsw_area) != sizeof(SCSW));
	84	QEMU_BUILD_BUG_ON(sizeof(region->irb_area) != sizeof(IRB));
	85
	86	memset(region, 0, sizeof(*region));
	87
66dc50f7 HP	88	memcpy(region->orb_area, &sch->orb, sizeof(ORB));
66dc50f7 HP	89	memcpy(region->scsw_area, &sch->curr_status.scsw, sizeof(SCSW));
8ca2b376 XFR	90
	91	again:
	92	ret = pwrite(vcdev->vdev.fd, region,
	93	vcdev->io_region_size, vcdev->io_region_offset);
	94	if (ret != vcdev->io_region_size) {
	95	if (errno == EAGAIN) {
	96	goto again;
	97	}
91f751dc	98	error_report("vfio-ccw: write I/O region failed with errno=%d", errno);
66dc50f7 HP	99	ret = -errno;
	100	} else {
	101	ret = region->ret_code;
	102	}
	103	switch (ret) {
	104	case 0:
	105	return IOINST_CC_EXPECTED;
	106	case -EBUSY:
	107	return IOINST_CC_BUSY;
	108	case -ENODEV:
	109	case -EACCES:
	110	return IOINST_CC_NOT_OPERATIONAL;
	111	case -EFAULT:
	112	default:
	113	sch_gen_unit_exception(sch);
	114	css_inject_io_interrupt(sch);
	115	return IOINST_CC_EXPECTED;
8ca2b376	116	}
8ca2b376 XFR	117	}
8ca2b376 XFR	118
8fadea24 CH	119	static int vfio_ccw_handle_clear(SubchDev *sch)
	120	{
	121	S390CCWDevice *cdev = sch->driver_data;
	122	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	123	struct ccw_cmd_region *region = vcdev->async_cmd_region;
	124	int ret;
	125
	126	if (!vcdev->async_cmd_region) {
	127	/* Async command region not available, fall back to emulation */
	128	return -ENOSYS;
	129	}
	130
	131	memset(region, 0, sizeof(*region));
	132	region->command = VFIO_CCW_ASYNC_CMD_CSCH;
	133
	134	again:
	135	ret = pwrite(vcdev->vdev.fd, region,
	136	vcdev->async_cmd_region_size, vcdev->async_cmd_region_offset);
	137	if (ret != vcdev->async_cmd_region_size) {
	138	if (errno == EAGAIN) {
	139	goto again;
	140	}
	141	error_report("vfio-ccw: write cmd region failed with errno=%d", errno);
	142	ret = -errno;
	143	} else {
	144	ret = region->ret_code;
	145	}
	146	switch (ret) {
	147	case 0:
	148	case -ENODEV:
	149	case -EACCES:
	150	return 0;
	151	case -EFAULT:
	152	default:
	153	sch_gen_unit_exception(sch);
	154	css_inject_io_interrupt(sch);
	155	return 0;
	156	}
	157	}
	158
	159	static int vfio_ccw_handle_halt(SubchDev *sch)
	160	{
	161	S390CCWDevice *cdev = sch->driver_data;
	162	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	163	struct ccw_cmd_region *region = vcdev->async_cmd_region;
	164	int ret;
	165
	166	if (!vcdev->async_cmd_region) {
	167	/* Async command region not available, fall back to emulation */
	168	return -ENOSYS;
	169	}
	170
	171	memset(region, 0, sizeof(*region));
	172	region->command = VFIO_CCW_ASYNC_CMD_HSCH;
	173
	174	again:
	175	ret = pwrite(vcdev->vdev.fd, region,
	176	vcdev->async_cmd_region_size, vcdev->async_cmd_region_offset);
	177	if (ret != vcdev->async_cmd_region_size) {
	178	if (errno == EAGAIN) {
	179	goto again;
	180	}
	181	error_report("vfio-ccw: write cmd region failed with errno=%d", errno);
	182	ret = -errno;
183	} else {
184	ret = region->ret_code;
185	}
186	switch (ret) {
187	case 0:
188	case -EBUSY:
189	case -ENODEV:
190	case -EACCES:
191	return 0;
192	case -EFAULT:
193	default:
194	sch_gen_unit_exception(sch);
195	css_inject_io_interrupt(sch);
196	return 0;
197	}
198	}
199
1dcac3e1 XFR	200	static void vfio_ccw_reset(DeviceState *dev)
	201	{
	202	CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
	203	S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
	204	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	205
	206	ioctl(vcdev->vdev.fd, VFIO_DEVICE_RESET);
	207	}
	208
4886b3e9 DJS	209	static void vfio_ccw_io_notifier_handler(void *opaque)
	210	{
	211	VFIOCCWDevice *vcdev = opaque;
8ca2b376 XFR	212	struct ccw_io_region *region = vcdev->io_region;
	213	S390CCWDevice *cdev = S390_CCW_DEVICE(vcdev);
	214	CcwDevice *ccw_dev = CCW_DEVICE(cdev);
	215	SubchDev *sch = ccw_dev->sch;
e1d0b372 DB	216	SCHIB *schib = &sch->curr_status;
e1d0b372 DB	217	SCSW s;
8ca2b376 XFR	218	IRB irb;
8ca2b376 XFR	219	int size;
4886b3e9 DJS	220
	221	if (!event_notifier_test_and_clear(&vcdev->io_notifier)) {
	222	return;
	223	}
8ca2b376 XFR	224
	225	size = pread(vcdev->vdev.fd, region, vcdev->io_region_size,
	226	vcdev->io_region_offset);
	227	if (size == -1) {
	228	switch (errno) {
	229	case ENODEV:
	230	/* Generate a deferred cc 3 condition. */
e1d0b372 DB	231	schib->scsw.flags \|= SCSW_FLAGS_MASK_CC;
	232	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	233	schib->scsw.ctrl \|= (SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND);
8ca2b376 XFR	234	goto read_err;
	235	case EFAULT:
	236	/* Memory problem, generate channel data check. */
e1d0b372 DB	237	schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
	238	schib->scsw.cstat = SCSW_CSTAT_DATA_CHECK;
	239	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	240	schib->scsw.ctrl \|= SCSW_STCTL_PRIMARY \| SCSW_STCTL_SECONDARY \|
8ca2b376 XFR	241	SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND;
	242	goto read_err;
	243	default:
	244	/* Error, generate channel program check. */
e1d0b372 DB	245	schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
	246	schib->scsw.cstat = SCSW_CSTAT_PROG_CHECK;
	247	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	248	schib->scsw.ctrl \|= SCSW_STCTL_PRIMARY \| SCSW_STCTL_SECONDARY \|
8ca2b376 XFR	249	SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND;
	250	goto read_err;
	251	}
	252	} else if (size != vcdev->io_region_size) {
	253	/* Information transfer error, generate channel-control check. */
e1d0b372 DB	254	schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
	255	schib->scsw.cstat = SCSW_CSTAT_CHN_CTRL_CHK;
	256	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	257	schib->scsw.ctrl \|= SCSW_STCTL_PRIMARY \| SCSW_STCTL_SECONDARY \|
8ca2b376 XFR	258	SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND;
	259	goto read_err;
	260	}
	261
	262	memcpy(&irb, region->irb_area, sizeof(IRB));
	263
	264	/* Update control block via irb. */
e1d0b372 DB	265	s = schib->scsw;
	266	copy_scsw_to_guest(&s, &irb.scsw);
	267	schib->scsw = s;
8ca2b376	268
334e7685	269	/* If a uint check is pending, copy sense data. */
e1d0b372 DB	270	if ((schib->scsw.dstat & SCSW_DSTAT_UNIT_CHECK) &&
e1d0b372 DB	271	(schib->pmcw.chars & PMCW_CHARS_MASK_CSENSE)) {
334e7685 DJS	272	memcpy(sch->sense_data, irb.ecw, sizeof(irb.ecw));
	273	}
	274
8ca2b376 XFR	275	read_err:
8ca2b376 XFR	276	css_inject_io_interrupt(sch);
4886b3e9 DJS	277	}
	278
	279	static void vfio_ccw_register_io_notifier(VFIOCCWDevice vcdev, Error *errp)
	280	{
	281	VFIODevice *vdev = &vcdev->vdev;
	282	struct vfio_irq_info *irq_info;
4886b3e9	283	size_t argsz;
b5e89f04	284	int fd;
4886b3e9 DJS	285
	286	if (vdev->num_irqs < VFIO_CCW_IO_IRQ_INDEX + 1) {
	287	error_setg(errp, "vfio: unexpected number of io irqs %u",
	288	vdev->num_irqs);
	289	return;
	290	}
	291
28e22d4b	292	argsz = sizeof(*irq_info);
4886b3e9 DJS	293	irq_info = g_malloc0(argsz);
	294	irq_info->index = VFIO_CCW_IO_IRQ_INDEX;
	295	irq_info->argsz = argsz;
	296	if (ioctl(vdev->fd, VFIO_DEVICE_GET_IRQ_INFO,
	297	irq_info) < 0 \|\| irq_info->count < 1) {
	298	error_setg_errno(errp, errno, "vfio: Error getting irq info");
	299	goto out_free_info;
	300	}
	301
	302	if (event_notifier_init(&vcdev->io_notifier, 0)) {
	303	error_setg_errno(errp, errno,
	304	"vfio: Unable to init event notifier for IO");
	305	goto out_free_info;
	306	}
	307
b5e89f04 CH	308	fd = event_notifier_get_fd(&vcdev->io_notifier);
	309	qemu_set_fd_handler(fd, vfio_ccw_io_notifier_handler, NULL, vcdev);
	310
	311	if (vfio_set_irq_signaling(vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
	312	VFIO_IRQ_SET_ACTION_TRIGGER, fd, errp)) {
	313	qemu_set_fd_handler(fd, NULL, NULL, vcdev);
4886b3e9 DJS	314	event_notifier_cleanup(&vcdev->io_notifier);
	315	}
	316
4886b3e9 DJS	317	out_free_info:
	318	g_free(irq_info);
	319	}
	320
	321	static void vfio_ccw_unregister_io_notifier(VFIOCCWDevice *vcdev)
	322	{
b5e89f04 CH	323	Error *err = NULL;
b5e89f04 CH	324
f5cf94cd AW	325	if (vfio_set_irq_signaling(&vcdev->vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
f5cf94cd AW	326	VFIO_IRQ_SET_ACTION_TRIGGER, -1, &err)) {
b5e89f04	327	error_reportf_err(err, VFIO_MSG_PREFIX, vcdev->vdev.name);
4886b3e9 DJS	328	}
	329
	330	qemu_set_fd_handler(event_notifier_get_fd(&vcdev->io_notifier),
	331	NULL, NULL, vcdev);
	332	event_notifier_cleanup(&vcdev->io_notifier);
4886b3e9 DJS	333	}
4886b3e9 DJS	334
c14e706c DJS	335	static void vfio_ccw_get_region(VFIOCCWDevice vcdev, Error *errp)
	336	{
	337	VFIODevice *vdev = &vcdev->vdev;
	338	struct vfio_region_info *info;
	339	int ret;
	340
	341	/* Sanity check device */
	342	if (!(vdev->flags & VFIO_DEVICE_FLAGS_CCW)) {
	343	error_setg(errp, "vfio: Um, this isn't a vfio-ccw device");
	344	return;
	345	}
	346
8fadea24 CH	347	/*
	348	* We always expect at least the I/O region to be present. We also
	349	* may have a variable number of regions governed by capabilities.
	350	*/
c14e706c	351	if (vdev->num_regions < VFIO_CCW_CONFIG_REGION_INDEX + 1) {
8fadea24 CH	352	error_setg(errp, "vfio: too few regions (%u), expected at least %u",
8fadea24 CH	353	vdev->num_regions, VFIO_CCW_CONFIG_REGION_INDEX + 1);
c14e706c DJS	354	return;
	355	}
	356
	357	ret = vfio_get_region_info(vdev, VFIO_CCW_CONFIG_REGION_INDEX, &info);
	358	if (ret) {
	359	error_setg_errno(errp, -ret, "vfio: Error getting config info");
	360	return;
	361	}
	362
	363	vcdev->io_region_size = info->size;
	364	if (sizeof(*vcdev->io_region) != vcdev->io_region_size) {
	365	error_setg(errp, "vfio: Unexpected size of the I/O region");
	366	g_free(info);
	367	return;
	368	}
	369
	370	vcdev->io_region_offset = info->offset;
	371	vcdev->io_region = g_malloc0(info->size);
	372
8fadea24 CH	373	/* check for the optional async command region */
	374	ret = vfio_get_dev_region_info(vdev, VFIO_REGION_TYPE_CCW,
	375	VFIO_REGION_SUBTYPE_CCW_ASYNC_CMD, &info);
	376	if (!ret) {
	377	vcdev->async_cmd_region_size = info->size;
	378	if (sizeof(*vcdev->async_cmd_region) != vcdev->async_cmd_region_size) {
	379	error_setg(errp, "vfio: Unexpected size of the async cmd region");
	380	g_free(vcdev->io_region);
	381	g_free(info);
	382	return;
	383	}
	384	vcdev->async_cmd_region_offset = info->offset;
	385	vcdev->async_cmd_region = g_malloc0(info->size);
	386	}
	387
c14e706c DJS	388	g_free(info);
	389	}
	390
	391	static void vfio_ccw_put_region(VFIOCCWDevice *vcdev)
	392	{
8fadea24	393	g_free(vcdev->async_cmd_region);
c14e706c DJS	394	g_free(vcdev->io_region);
	395	}
	396
c96f2c2a	397	static void vfio_ccw_put_device(VFIOCCWDevice *vcdev)
1dcac3e1 XFR	398	{
	399	g_free(vcdev->vdev.name);
	400	vfio_put_base_device(&vcdev->vdev);
	401	}
	402
c96f2c2a GK	403	static void vfio_ccw_get_device(VFIOGroup group, VFIOCCWDevice vcdev,
	404	Error **errp)
	405	{
	406	char *name = g_strdup_printf("%x.%x.%04x", vcdev->cdev.hostid.cssid,
	407	vcdev->cdev.hostid.ssid,
	408	vcdev->cdev.hostid.devid);
	409	VFIODevice *vbasedev;
	410
	411	QLIST_FOREACH(vbasedev, &group->device_list, next) {
	412	if (strcmp(vbasedev->name, name) == 0) {
	413	error_setg(errp, "vfio: subchannel %s has already been attached",
	414	name);
	415	goto out_err;
	416	}
	417	}
238e9172 AW	418
	419	/*
	420	* All vfio-ccw devices are believed to operate in a way compatible with
	421	* memory ballooning, ie. pages pinned in the host are in the current
	422	* working set of the guest driver and therefore never overlap with pages
	423	* available to the guest balloon driver. This needs to be set before
	424	* vfio_get_device() for vfio common to handle the balloon inhibitor.
	425	*/
	426	vcdev->vdev.balloon_allowed = true;
c96f2c2a GK	427
	428	if (vfio_get_device(group, vcdev->cdev.mdevid, &vcdev->vdev, errp)) {
	429	goto out_err;
	430	}
	431
	432	vcdev->vdev.ops = &vfio_ccw_ops;
	433	vcdev->vdev.type = VFIO_DEVICE_TYPE_CCW;
	434	vcdev->vdev.name = name;
	435	vcdev->vdev.dev = &vcdev->cdev.parent_obj.parent_obj;
	436
	437	return;
	438
	439	out_err:
	440	g_free(name);
	441	}
	442
1dcac3e1 XFR	443	static VFIOGroup vfio_ccw_get_group(S390CCWDevice cdev, Error **errp)
	444	{
	445	char *tmp, group_path[PATH_MAX];
	446	ssize_t len;
	447	int groupid;
	448
	449	tmp = g_strdup_printf("/sys/bus/css/devices/%x.%x.%04x/%s/iommu_group",
	450	cdev->hostid.cssid, cdev->hostid.ssid,
	451	cdev->hostid.devid, cdev->mdevid);
	452	len = readlink(tmp, group_path, sizeof(group_path));
	453	g_free(tmp);
	454
	455	if (len <= 0 \|\| len >= sizeof(group_path)) {
	456	error_setg(errp, "vfio: no iommu_group found");
	457	return NULL;
	458	}
	459
	460	group_path[len] = 0;
	461
	462	if (sscanf(basename(group_path), "%d", &groupid) != 1) {
	463	error_setg(errp, "vfio: failed to read %s", group_path);
	464	return NULL;
	465	}
	466
	467	return vfio_get_group(groupid, &address_space_memory, errp);
	468	}
	469
	470	static void vfio_ccw_realize(DeviceState dev, Error *errp)
	471	{
1dcac3e1 XFR	472	VFIOGroup *group;
	473	CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
	474	S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
	475	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	476	S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
	477	Error *err = NULL;
	478
	479	/* Call the class init function for subchannel. */
	480	if (cdc->realize) {
	481	cdc->realize(cdev, vcdev->vdev.sysfsdev, &err);
	482	if (err) {
	483	goto out_err_propagate;
	484	}
	485	}
	486
	487	group = vfio_ccw_get_group(cdev, &err);
	488	if (!group) {
	489	goto out_group_err;
	490	}
	491
c96f2c2a GK	492	vfio_ccw_get_device(group, vcdev, &err);
c96f2c2a GK	493	if (err) {
1dcac3e1 XFR	494	goto out_device_err;
	495	}
	496
c14e706c DJS	497	vfio_ccw_get_region(vcdev, &err);
	498	if (err) {
	499	goto out_region_err;
	500	}
	501
4886b3e9 DJS	502	vfio_ccw_register_io_notifier(vcdev, &err);
	503	if (err) {
	504	goto out_notifier_err;
	505	}
	506
1dcac3e1 XFR	507	return;
1dcac3e1 XFR	508
4886b3e9 DJS	509	out_notifier_err:
4886b3e9 DJS	510	vfio_ccw_put_region(vcdev);
c14e706c	511	out_region_err:
c96f2c2a	512	vfio_ccw_put_device(vcdev);
1dcac3e1 XFR	513	out_device_err:
	514	vfio_put_group(group);
	515	out_group_err:
	516	if (cdc->unrealize) {
b69c3c21	517	cdc->unrealize(cdev);
1dcac3e1 XFR	518	}
	519	out_err_propagate:
	520	error_propagate(errp, err);
	521	}
	522
b69c3c21	523	static void vfio_ccw_unrealize(DeviceState *dev)
1dcac3e1 XFR	524	{
	525	CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
	526	S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
	527	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	528	S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
	529	VFIOGroup *group = vcdev->vdev.group;
	530
4886b3e9	531	vfio_ccw_unregister_io_notifier(vcdev);
c14e706c	532	vfio_ccw_put_region(vcdev);
c96f2c2a	533	vfio_ccw_put_device(vcdev);
1dcac3e1 XFR	534	vfio_put_group(group);
	535
	536	if (cdc->unrealize) {
b69c3c21	537	cdc->unrealize(cdev);
1dcac3e1 XFR	538	}
	539	}
	540
	541	static Property vfio_ccw_properties[] = {
	542	DEFINE_PROP_STRING("sysfsdev", VFIOCCWDevice, vdev.sysfsdev),
9a51c9ee	543	DEFINE_PROP_BOOL("force-orb-pfch", VFIOCCWDevice, force_orb_pfch, false),
1dcac3e1 XFR	544	DEFINE_PROP_END_OF_LIST(),
	545	};
	546
	547	static const VMStateDescription vfio_ccw_vmstate = {
da56e330	548	.name = "vfio-ccw",
1dcac3e1 XFR	549	.unmigratable = 1,
	550	};
	551
	552	static void vfio_ccw_class_init(ObjectClass klass, void data)
	553	{
	554	DeviceClass *dc = DEVICE_CLASS(klass);
8ca2b376	555	S390CCWDeviceClass *cdc = S390_CCW_DEVICE_CLASS(klass);
1dcac3e1	556
4f67d30b	557	device_class_set_props(dc, vfio_ccw_properties);
1dcac3e1 XFR	558	dc->vmsd = &vfio_ccw_vmstate;
1dcac3e1 XFR	559	dc->desc = "VFIO-based subchannel assignment";
bd2aef10	560	set_bit(DEVICE_CATEGORY_MISC, dc->categories);
1dcac3e1 XFR	561	dc->realize = vfio_ccw_realize;
	562	dc->unrealize = vfio_ccw_unrealize;
	563	dc->reset = vfio_ccw_reset;
8ca2b376 XFR	564
8ca2b376 XFR	565	cdc->handle_request = vfio_ccw_handle_request;
8fadea24 CH	566	cdc->handle_halt = vfio_ccw_handle_halt;
8fadea24 CH	567	cdc->handle_clear = vfio_ccw_handle_clear;
1dcac3e1 XFR	568	}
	569
	570	static const TypeInfo vfio_ccw_info = {
	571	.name = TYPE_VFIO_CCW,
	572	.parent = TYPE_S390_CCW,
	573	.instance_size = sizeof(VFIOCCWDevice),
	574	.class_init = vfio_ccw_class_init,
	575	};
	576
	577	static void register_vfio_ccw_type(void)
	578	{
	579	type_register_static(&vfio_ccw_info);
	580	}
	581
	582	type_init(register_vfio_ccw_type)