[mirror_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 <bjsdjshi@linux.vnet.ibm.com>
 *            Xiao Feng Ren <renxiaof@linux.vnet.ibm.com>
 *            Pierre Morel <pmorel@linux.vnet.ibm.com>
 *            Cornelia Huck <cohuck@redhat.com>
 *
 * 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/s390x/ccw-device.h"
#include "exec/address-spaces.h"
#include "qemu/error-report.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)) {
        if (!(vcdev->force_orb_pfch)) {
            warn_once_pfch(vcdev, sch, "requires PFCH flag set");
            sch_gen_unit_exception(sch);
            css_inject_io_interrupt(sch);
            return IOINST_CC_EXPECTED;
        } else {
            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: wirte 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;

    vfio_set_irq_signaling(&vcdev->vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
                           VFIO_IRQ_SET_ACTION_TRIGGER, -1, &err);
    if (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, NULL);
    }
out_err_propagate:
    error_propagate(errp, err);
}

static void vfio_ccw_unrealize(DeviceState *dev, Error **errp)
{
    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, errp);
    }
}

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);

    dc->props = 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 <bjsdjshi@linux.vnet.ibm.com>
	8	* Xiao Feng Ren <renxiaof@linux.vnet.ibm.com>
	9	* Pierre Morel <pmorel@linux.vnet.ibm.com>
8fadea24	10	* Cornelia Huck <cohuck@redhat.com>
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"
1dcac3e1	28	#include "hw/s390x/ccw-device.h"
d791937f	29	#include "exec/address-spaces.h"
4886b3e9	30	#include "qemu/error-report.h"
0b8fa32f	31	#include "qemu/module.h"
1dcac3e1	32
44445d86	33	struct VFIOCCWDevice {
1dcac3e1 XFR	34	S390CCWDevice cdev;
1dcac3e1 XFR	35	VFIODevice vdev;
c14e706c DJS	36	uint64_t io_region_size;
	37	uint64_t io_region_offset;
	38	struct ccw_io_region *io_region;
8fadea24 CH	39	uint64_t async_cmd_region_size;
	40	uint64_t async_cmd_region_offset;
	41	struct ccw_cmd_region *async_cmd_region;
4886b3e9	42	EventNotifier io_notifier;
9a51c9ee HP	43	bool force_orb_pfch;
9a51c9ee HP	44	bool warned_orb_pfch;
44445d86	45	};
1dcac3e1	46
9a51c9ee HP	47	static inline void warn_once_pfch(VFIOCCWDevice vcdev, SubchDev sch,
	48	const char *msg)
	49	{
c55510b7 CH	50	warn_report_once_cond(&vcdev->warned_orb_pfch,
	51	"vfio-ccw (devno %x.%x.%04x): %s",
	52	sch->cssid, sch->ssid, sch->devno, msg);
9a51c9ee HP	53	}
9a51c9ee HP	54
1dcac3e1 XFR	55	static void vfio_ccw_compute_needs_reset(VFIODevice *vdev)
	56	{
	57	vdev->needs_reset = false;
	58	}
	59
	60	/*
	61	* We don't need vfio_hot_reset_multi and vfio_eoi operations for
	62	* vfio_ccw device now.
	63	*/
	64	struct VFIODeviceOps vfio_ccw_ops = {
	65	.vfio_compute_needs_reset = vfio_ccw_compute_needs_reset,
	66	};
	67
66dc50f7	68	static IOInstEnding vfio_ccw_handle_request(SubchDev *sch)
8ca2b376	69	{
66dc50f7	70	S390CCWDevice *cdev = sch->driver_data;
8ca2b376 XFR	71	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	72	struct ccw_io_region *region = vcdev->io_region;
	73	int ret;
	74
9a51c9ee HP	75	if (!(sch->orb.ctrl0 & ORB_CTRL0_MASK_PFCH)) {
	76	if (!(vcdev->force_orb_pfch)) {
	77	warn_once_pfch(vcdev, sch, "requires PFCH flag set");
	78	sch_gen_unit_exception(sch);
	79	css_inject_io_interrupt(sch);
	80	return IOINST_CC_EXPECTED;
	81	} else {
	82	sch->orb.ctrl0 \|= ORB_CTRL0_MASK_PFCH;
	83	warn_once_pfch(vcdev, sch, "PFCH flag forced");
	84	}
	85	}
	86
8ca2b376 XFR	87	QEMU_BUILD_BUG_ON(sizeof(region->orb_area) != sizeof(ORB));
	88	QEMU_BUILD_BUG_ON(sizeof(region->scsw_area) != sizeof(SCSW));
	89	QEMU_BUILD_BUG_ON(sizeof(region->irb_area) != sizeof(IRB));
	90
	91	memset(region, 0, sizeof(*region));
	92
66dc50f7 HP	93	memcpy(region->orb_area, &sch->orb, sizeof(ORB));
66dc50f7 HP	94	memcpy(region->scsw_area, &sch->curr_status.scsw, sizeof(SCSW));
8ca2b376 XFR	95
	96	again:
	97	ret = pwrite(vcdev->vdev.fd, region,
	98	vcdev->io_region_size, vcdev->io_region_offset);
	99	if (ret != vcdev->io_region_size) {
	100	if (errno == EAGAIN) {
	101	goto again;
	102	}
	103	error_report("vfio-ccw: wirte I/O region failed with errno=%d", errno);
66dc50f7 HP	104	ret = -errno;
	105	} else {
	106	ret = region->ret_code;
	107	}
	108	switch (ret) {
	109	case 0:
	110	return IOINST_CC_EXPECTED;
	111	case -EBUSY:
	112	return IOINST_CC_BUSY;
	113	case -ENODEV:
	114	case -EACCES:
	115	return IOINST_CC_NOT_OPERATIONAL;
	116	case -EFAULT:
	117	default:
	118	sch_gen_unit_exception(sch);
	119	css_inject_io_interrupt(sch);
	120	return IOINST_CC_EXPECTED;
8ca2b376	121	}
8ca2b376 XFR	122	}
8ca2b376 XFR	123
8fadea24 CH	124	static int vfio_ccw_handle_clear(SubchDev *sch)
	125	{
	126	S390CCWDevice *cdev = sch->driver_data;
	127	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	128	struct ccw_cmd_region *region = vcdev->async_cmd_region;
	129	int ret;
	130
	131	if (!vcdev->async_cmd_region) {
	132	/* Async command region not available, fall back to emulation */
	133	return -ENOSYS;
	134	}
	135
	136	memset(region, 0, sizeof(*region));
	137	region->command = VFIO_CCW_ASYNC_CMD_CSCH;
	138
	139	again:
	140	ret = pwrite(vcdev->vdev.fd, region,
	141	vcdev->async_cmd_region_size, vcdev->async_cmd_region_offset);
	142	if (ret != vcdev->async_cmd_region_size) {
	143	if (errno == EAGAIN) {
	144	goto again;
	145	}
	146	error_report("vfio-ccw: write cmd region failed with errno=%d", errno);
	147	ret = -errno;
	148	} else {
	149	ret = region->ret_code;
	150	}
	151	switch (ret) {
	152	case 0:
	153	case -ENODEV:
	154	case -EACCES:
	155	return 0;
	156	case -EFAULT:
	157	default:
	158	sch_gen_unit_exception(sch);
	159	css_inject_io_interrupt(sch);
	160	return 0;
	161	}
	162	}
	163
	164	static int vfio_ccw_handle_halt(SubchDev *sch)
	165	{
	166	S390CCWDevice *cdev = sch->driver_data;
	167	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	168	struct ccw_cmd_region *region = vcdev->async_cmd_region;
	169	int ret;
	170
	171	if (!vcdev->async_cmd_region) {
	172	/* Async command region not available, fall back to emulation */
	173	return -ENOSYS;
	174	}
	175
	176	memset(region, 0, sizeof(*region));
	177	region->command = VFIO_CCW_ASYNC_CMD_HSCH;
	178
	179	again:
	180	ret = pwrite(vcdev->vdev.fd, region,
	181	vcdev->async_cmd_region_size, vcdev->async_cmd_region_offset);
	182	if (ret != vcdev->async_cmd_region_size) {
	183	if (errno == EAGAIN) {
	184	goto again;
	185	}
	186	error_report("vfio-ccw: write cmd region failed with errno=%d", errno);
	187	ret = -errno;
188	} else {
189	ret = region->ret_code;
190	}
191	switch (ret) {
192	case 0:
193	case -EBUSY:
194	case -ENODEV:
195	case -EACCES:
196	return 0;
197	case -EFAULT:
198	default:
199	sch_gen_unit_exception(sch);
200	css_inject_io_interrupt(sch);
201	return 0;
202	}
203	}
204
1dcac3e1 XFR	205	static void vfio_ccw_reset(DeviceState *dev)
	206	{
	207	CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
	208	S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
	209	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	210
	211	ioctl(vcdev->vdev.fd, VFIO_DEVICE_RESET);
	212	}
	213
4886b3e9 DJS	214	static void vfio_ccw_io_notifier_handler(void *opaque)
	215	{
	216	VFIOCCWDevice *vcdev = opaque;
8ca2b376 XFR	217	struct ccw_io_region *region = vcdev->io_region;
	218	S390CCWDevice *cdev = S390_CCW_DEVICE(vcdev);
	219	CcwDevice *ccw_dev = CCW_DEVICE(cdev);
	220	SubchDev *sch = ccw_dev->sch;
e1d0b372 DB	221	SCHIB *schib = &sch->curr_status;
e1d0b372 DB	222	SCSW s;
8ca2b376 XFR	223	IRB irb;
8ca2b376 XFR	224	int size;
4886b3e9 DJS	225
	226	if (!event_notifier_test_and_clear(&vcdev->io_notifier)) {
	227	return;
	228	}
8ca2b376 XFR	229
	230	size = pread(vcdev->vdev.fd, region, vcdev->io_region_size,
	231	vcdev->io_region_offset);
	232	if (size == -1) {
	233	switch (errno) {
	234	case ENODEV:
	235	/* Generate a deferred cc 3 condition. */
e1d0b372 DB	236	schib->scsw.flags \|= SCSW_FLAGS_MASK_CC;
	237	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	238	schib->scsw.ctrl \|= (SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND);
8ca2b376 XFR	239	goto read_err;
	240	case EFAULT:
	241	/* Memory problem, generate channel data check. */
e1d0b372 DB	242	schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
	243	schib->scsw.cstat = SCSW_CSTAT_DATA_CHECK;
	244	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	245	schib->scsw.ctrl \|= SCSW_STCTL_PRIMARY \| SCSW_STCTL_SECONDARY \|
8ca2b376 XFR	246	SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND;
	247	goto read_err;
	248	default:
	249	/* Error, generate channel program check. */
e1d0b372 DB	250	schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
	251	schib->scsw.cstat = SCSW_CSTAT_PROG_CHECK;
	252	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	253	schib->scsw.ctrl \|= SCSW_STCTL_PRIMARY \| SCSW_STCTL_SECONDARY \|
8ca2b376 XFR	254	SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND;
	255	goto read_err;
	256	}
	257	} else if (size != vcdev->io_region_size) {
	258	/* Information transfer error, generate channel-control check. */
e1d0b372 DB	259	schib->scsw.ctrl &= ~SCSW_ACTL_START_PEND;
	260	schib->scsw.cstat = SCSW_CSTAT_CHN_CTRL_CHK;
	261	schib->scsw.ctrl &= ~SCSW_CTRL_MASK_STCTL;
	262	schib->scsw.ctrl \|= SCSW_STCTL_PRIMARY \| SCSW_STCTL_SECONDARY \|
8ca2b376 XFR	263	SCSW_STCTL_ALERT \| SCSW_STCTL_STATUS_PEND;
	264	goto read_err;
	265	}
	266
	267	memcpy(&irb, region->irb_area, sizeof(IRB));
	268
	269	/* Update control block via irb. */
e1d0b372 DB	270	s = schib->scsw;
	271	copy_scsw_to_guest(&s, &irb.scsw);
	272	schib->scsw = s;
8ca2b376	273
334e7685	274	/* If a uint check is pending, copy sense data. */
e1d0b372 DB	275	if ((schib->scsw.dstat & SCSW_DSTAT_UNIT_CHECK) &&
e1d0b372 DB	276	(schib->pmcw.chars & PMCW_CHARS_MASK_CSENSE)) {
334e7685 DJS	277	memcpy(sch->sense_data, irb.ecw, sizeof(irb.ecw));
	278	}
	279
8ca2b376 XFR	280	read_err:
8ca2b376 XFR	281	css_inject_io_interrupt(sch);
4886b3e9 DJS	282	}
	283
	284	static void vfio_ccw_register_io_notifier(VFIOCCWDevice vcdev, Error *errp)
	285	{
	286	VFIODevice *vdev = &vcdev->vdev;
	287	struct vfio_irq_info *irq_info;
4886b3e9	288	size_t argsz;
b5e89f04	289	int fd;
4886b3e9 DJS	290
	291	if (vdev->num_irqs < VFIO_CCW_IO_IRQ_INDEX + 1) {
	292	error_setg(errp, "vfio: unexpected number of io irqs %u",
	293	vdev->num_irqs);
	294	return;
	295	}
	296
28e22d4b	297	argsz = sizeof(*irq_info);
4886b3e9 DJS	298	irq_info = g_malloc0(argsz);
	299	irq_info->index = VFIO_CCW_IO_IRQ_INDEX;
	300	irq_info->argsz = argsz;
	301	if (ioctl(vdev->fd, VFIO_DEVICE_GET_IRQ_INFO,
	302	irq_info) < 0 \|\| irq_info->count < 1) {
	303	error_setg_errno(errp, errno, "vfio: Error getting irq info");
	304	goto out_free_info;
	305	}
	306
	307	if (event_notifier_init(&vcdev->io_notifier, 0)) {
	308	error_setg_errno(errp, errno,
	309	"vfio: Unable to init event notifier for IO");
	310	goto out_free_info;
	311	}
	312
b5e89f04 CH	313	fd = event_notifier_get_fd(&vcdev->io_notifier);
	314	qemu_set_fd_handler(fd, vfio_ccw_io_notifier_handler, NULL, vcdev);
	315
	316	if (vfio_set_irq_signaling(vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
	317	VFIO_IRQ_SET_ACTION_TRIGGER, fd, errp)) {
	318	qemu_set_fd_handler(fd, NULL, NULL, vcdev);
4886b3e9 DJS	319	event_notifier_cleanup(&vcdev->io_notifier);
	320	}
	321
4886b3e9 DJS	322	out_free_info:
	323	g_free(irq_info);
	324	}
	325
	326	static void vfio_ccw_unregister_io_notifier(VFIOCCWDevice *vcdev)
	327	{
b5e89f04 CH	328	Error *err = NULL;
	329
	330	vfio_set_irq_signaling(&vcdev->vdev, VFIO_CCW_IO_IRQ_INDEX, 0,
	331	VFIO_IRQ_SET_ACTION_TRIGGER, -1, &err);
	332	if (err) {
	333	error_reportf_err(err, VFIO_MSG_PREFIX, vcdev->vdev.name);
4886b3e9 DJS	334	}
	335
	336	qemu_set_fd_handler(event_notifier_get_fd(&vcdev->io_notifier),
	337	NULL, NULL, vcdev);
	338	event_notifier_cleanup(&vcdev->io_notifier);
4886b3e9 DJS	339	}
4886b3e9 DJS	340
c14e706c DJS	341	static void vfio_ccw_get_region(VFIOCCWDevice vcdev, Error *errp)
	342	{
	343	VFIODevice *vdev = &vcdev->vdev;
	344	struct vfio_region_info *info;
	345	int ret;
	346
	347	/* Sanity check device */
	348	if (!(vdev->flags & VFIO_DEVICE_FLAGS_CCW)) {
	349	error_setg(errp, "vfio: Um, this isn't a vfio-ccw device");
	350	return;
	351	}
	352
8fadea24 CH	353	/*
	354	* We always expect at least the I/O region to be present. We also
	355	* may have a variable number of regions governed by capabilities.
	356	*/
c14e706c	357	if (vdev->num_regions < VFIO_CCW_CONFIG_REGION_INDEX + 1) {
8fadea24 CH	358	error_setg(errp, "vfio: too few regions (%u), expected at least %u",
8fadea24 CH	359	vdev->num_regions, VFIO_CCW_CONFIG_REGION_INDEX + 1);
c14e706c DJS	360	return;
	361	}
	362
	363	ret = vfio_get_region_info(vdev, VFIO_CCW_CONFIG_REGION_INDEX, &info);
	364	if (ret) {
	365	error_setg_errno(errp, -ret, "vfio: Error getting config info");
	366	return;
	367	}
	368
	369	vcdev->io_region_size = info->size;
	370	if (sizeof(*vcdev->io_region) != vcdev->io_region_size) {
	371	error_setg(errp, "vfio: Unexpected size of the I/O region");
	372	g_free(info);
	373	return;
	374	}
	375
	376	vcdev->io_region_offset = info->offset;
	377	vcdev->io_region = g_malloc0(info->size);
	378
8fadea24 CH	379	/* check for the optional async command region */
	380	ret = vfio_get_dev_region_info(vdev, VFIO_REGION_TYPE_CCW,
	381	VFIO_REGION_SUBTYPE_CCW_ASYNC_CMD, &info);
	382	if (!ret) {
	383	vcdev->async_cmd_region_size = info->size;
	384	if (sizeof(*vcdev->async_cmd_region) != vcdev->async_cmd_region_size) {
	385	error_setg(errp, "vfio: Unexpected size of the async cmd region");
	386	g_free(vcdev->io_region);
	387	g_free(info);
	388	return;
	389	}
	390	vcdev->async_cmd_region_offset = info->offset;
	391	vcdev->async_cmd_region = g_malloc0(info->size);
	392	}
	393
c14e706c DJS	394	g_free(info);
	395	}
	396
	397	static void vfio_ccw_put_region(VFIOCCWDevice *vcdev)
	398	{
8fadea24	399	g_free(vcdev->async_cmd_region);
c14e706c DJS	400	g_free(vcdev->io_region);
	401	}
	402
c96f2c2a	403	static void vfio_ccw_put_device(VFIOCCWDevice *vcdev)
1dcac3e1 XFR	404	{
	405	g_free(vcdev->vdev.name);
	406	vfio_put_base_device(&vcdev->vdev);
	407	}
	408
c96f2c2a GK	409	static void vfio_ccw_get_device(VFIOGroup group, VFIOCCWDevice vcdev,
	410	Error **errp)
	411	{
	412	char *name = g_strdup_printf("%x.%x.%04x", vcdev->cdev.hostid.cssid,
	413	vcdev->cdev.hostid.ssid,
	414	vcdev->cdev.hostid.devid);
	415	VFIODevice *vbasedev;
	416
	417	QLIST_FOREACH(vbasedev, &group->device_list, next) {
	418	if (strcmp(vbasedev->name, name) == 0) {
	419	error_setg(errp, "vfio: subchannel %s has already been attached",
	420	name);
	421	goto out_err;
	422	}
	423	}
238e9172 AW	424
	425	/*
	426	* All vfio-ccw devices are believed to operate in a way compatible with
	427	* memory ballooning, ie. pages pinned in the host are in the current
	428	* working set of the guest driver and therefore never overlap with pages
	429	* available to the guest balloon driver. This needs to be set before
	430	* vfio_get_device() for vfio common to handle the balloon inhibitor.
	431	*/
	432	vcdev->vdev.balloon_allowed = true;
c96f2c2a GK	433
	434	if (vfio_get_device(group, vcdev->cdev.mdevid, &vcdev->vdev, errp)) {
	435	goto out_err;
	436	}
	437
	438	vcdev->vdev.ops = &vfio_ccw_ops;
	439	vcdev->vdev.type = VFIO_DEVICE_TYPE_CCW;
	440	vcdev->vdev.name = name;
	441	vcdev->vdev.dev = &vcdev->cdev.parent_obj.parent_obj;
	442
	443	return;
	444
	445	out_err:
	446	g_free(name);
	447	}
	448
1dcac3e1 XFR	449	static VFIOGroup vfio_ccw_get_group(S390CCWDevice cdev, Error **errp)
	450	{
	451	char *tmp, group_path[PATH_MAX];
	452	ssize_t len;
	453	int groupid;
	454
	455	tmp = g_strdup_printf("/sys/bus/css/devices/%x.%x.%04x/%s/iommu_group",
	456	cdev->hostid.cssid, cdev->hostid.ssid,
	457	cdev->hostid.devid, cdev->mdevid);
	458	len = readlink(tmp, group_path, sizeof(group_path));
	459	g_free(tmp);
	460
	461	if (len <= 0 \|\| len >= sizeof(group_path)) {
	462	error_setg(errp, "vfio: no iommu_group found");
	463	return NULL;
	464	}
	465
	466	group_path[len] = 0;
	467
	468	if (sscanf(basename(group_path), "%d", &groupid) != 1) {
	469	error_setg(errp, "vfio: failed to read %s", group_path);
	470	return NULL;
	471	}
	472
	473	return vfio_get_group(groupid, &address_space_memory, errp);
	474	}
	475
	476	static void vfio_ccw_realize(DeviceState dev, Error *errp)
	477	{
1dcac3e1 XFR	478	VFIOGroup *group;
	479	CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
	480	S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
	481	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	482	S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
	483	Error *err = NULL;
	484
	485	/* Call the class init function for subchannel. */
	486	if (cdc->realize) {
	487	cdc->realize(cdev, vcdev->vdev.sysfsdev, &err);
	488	if (err) {
	489	goto out_err_propagate;
	490	}
	491	}
	492
	493	group = vfio_ccw_get_group(cdev, &err);
	494	if (!group) {
	495	goto out_group_err;
	496	}
	497
c96f2c2a GK	498	vfio_ccw_get_device(group, vcdev, &err);
c96f2c2a GK	499	if (err) {
1dcac3e1 XFR	500	goto out_device_err;
	501	}
	502
c14e706c DJS	503	vfio_ccw_get_region(vcdev, &err);
	504	if (err) {
	505	goto out_region_err;
	506	}
	507
4886b3e9 DJS	508	vfio_ccw_register_io_notifier(vcdev, &err);
	509	if (err) {
	510	goto out_notifier_err;
	511	}
	512
1dcac3e1 XFR	513	return;
1dcac3e1 XFR	514
4886b3e9 DJS	515	out_notifier_err:
4886b3e9 DJS	516	vfio_ccw_put_region(vcdev);
c14e706c	517	out_region_err:
c96f2c2a	518	vfio_ccw_put_device(vcdev);
1dcac3e1 XFR	519	out_device_err:
	520	vfio_put_group(group);
	521	out_group_err:
	522	if (cdc->unrealize) {
	523	cdc->unrealize(cdev, NULL);
	524	}
	525	out_err_propagate:
	526	error_propagate(errp, err);
	527	}
	528
	529	static void vfio_ccw_unrealize(DeviceState dev, Error *errp)
	530	{
	531	CcwDevice *ccw_dev = DO_UPCAST(CcwDevice, parent_obj, dev);
	532	S390CCWDevice *cdev = DO_UPCAST(S390CCWDevice, parent_obj, ccw_dev);
	533	VFIOCCWDevice *vcdev = DO_UPCAST(VFIOCCWDevice, cdev, cdev);
	534	S390CCWDeviceClass *cdc = S390_CCW_DEVICE_GET_CLASS(cdev);
	535	VFIOGroup *group = vcdev->vdev.group;
	536
4886b3e9	537	vfio_ccw_unregister_io_notifier(vcdev);
c14e706c	538	vfio_ccw_put_region(vcdev);
c96f2c2a	539	vfio_ccw_put_device(vcdev);
1dcac3e1 XFR	540	vfio_put_group(group);
	541
	542	if (cdc->unrealize) {
	543	cdc->unrealize(cdev, errp);
	544	}
	545	}
	546
	547	static Property vfio_ccw_properties[] = {
	548	DEFINE_PROP_STRING("sysfsdev", VFIOCCWDevice, vdev.sysfsdev),
9a51c9ee	549	DEFINE_PROP_BOOL("force-orb-pfch", VFIOCCWDevice, force_orb_pfch, false),
1dcac3e1 XFR	550	DEFINE_PROP_END_OF_LIST(),
	551	};
	552
	553	static const VMStateDescription vfio_ccw_vmstate = {
da56e330	554	.name = "vfio-ccw",
1dcac3e1 XFR	555	.unmigratable = 1,
	556	};
	557
	558	static void vfio_ccw_class_init(ObjectClass klass, void data)
	559	{
	560	DeviceClass *dc = DEVICE_CLASS(klass);
8ca2b376	561	S390CCWDeviceClass *cdc = S390_CCW_DEVICE_CLASS(klass);
1dcac3e1 XFR	562
	563	dc->props = vfio_ccw_properties;
	564	dc->vmsd = &vfio_ccw_vmstate;
	565	dc->desc = "VFIO-based subchannel assignment";
bd2aef10	566	set_bit(DEVICE_CATEGORY_MISC, dc->categories);
1dcac3e1 XFR	567	dc->realize = vfio_ccw_realize;
	568	dc->unrealize = vfio_ccw_unrealize;
	569	dc->reset = vfio_ccw_reset;
8ca2b376 XFR	570
8ca2b376 XFR	571	cdc->handle_request = vfio_ccw_handle_request;
8fadea24 CH	572	cdc->handle_halt = vfio_ccw_handle_halt;
8fadea24 CH	573	cdc->handle_clear = vfio_ccw_handle_clear;
1dcac3e1 XFR	574	}
	575
	576	static const TypeInfo vfio_ccw_info = {
	577	.name = TYPE_VFIO_CCW,
	578	.parent = TYPE_S390_CCW,
	579	.instance_size = sizeof(VFIOCCWDevice),
	580	.class_init = vfio_ccw_class_init,
	581	};
	582
	583	static void register_vfio_ccw_type(void)
	584	{
	585	type_register_static(&vfio_ccw_info);
	586	}
	587
	588	type_init(register_vfio_ccw_type)