qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 1/1 V6] VMWare PVSCSI paravirtual device impleme


From: Paolo Bonzini
Subject: Re: [Qemu-devel] [PATCH 1/1 V6] VMWare PVSCSI paravirtual device implementation
Date: Wed, 10 Apr 2013 11:33:56 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20130311 Thunderbird/17.0.4

Just a few comments, many of them aesthetic.

Il 08/04/2013 20:39, Dmitry Fleytman ha scritto:
> Signed-off-by: Dmitry Fleytman <address@hidden>
> Signed-off-by: Yan Vugenfirer <address@hidden>
> ---
>  default-configs/pci.mak    |    1 +
>  docs/specs/pvscsi-spec.txt |   92 ++++
>  hw/Makefile.objs           |    1 +
>  hw/pci/pci.h               |    1 +
>  hw/pvscsi.c                | 1194 
> ++++++++++++++++++++++++++++++++++++++++++++
>  hw/vmw_pvscsi.h            |  434 ++++++++++++++++
>  trace-events               |   36 ++
>  7 files changed, 1759 insertions(+)
>  create mode 100644 docs/specs/pvscsi-spec.txt
>  create mode 100644 hw/pvscsi.c
>  create mode 100644 hw/vmw_pvscsi.h
> 
> diff --git a/default-configs/pci.mak b/default-configs/pci.mak
> index ce56d58..3f8375c 100644
> --- a/default-configs/pci.mak
> +++ b/default-configs/pci.mak
> @@ -10,6 +10,7 @@ CONFIG_EEPRO100_PCI=y
>  CONFIG_PCNET_PCI=y
>  CONFIG_PCNET_COMMON=y
>  CONFIG_LSI_SCSI_PCI=y
> +CONFIG_PVSCSI_SCSI_PCI=y
>  CONFIG_MEGASAS_SCSI_PCI=y
>  CONFIG_RTL8139_PCI=y
>  CONFIG_E1000_PCI=y
> diff --git a/docs/specs/pvscsi-spec.txt b/docs/specs/pvscsi-spec.txt
> new file mode 100644
> index 0000000..b2c3a55
> --- /dev/null
> +++ b/docs/specs/pvscsi-spec.txt
> @@ -0,0 +1,92 @@
> +General Description
> +===================
> +
> +This document describes VMWare PVSCSI device interface specification.
> +Created by Dmitry Fleytman (address@hidden), Daynix Computing LTD.
> +Based on source code of PVSCSI Linux driver from kernel 3.0.4
> +
> +PVSCSI Device Interface Overview
> +================================
> +
> +The interface is based on memory area shared between hypervisor and VM.
> +Memory area is obtained by driver as device IO memory resource of
> +PVSCSI_MEM_SPACE_SIZE length.
> +The shared memory consists of registers area and rings area.
> +The registers area is used to raise hypervisor interrupts and issue device
> +commands. The rings area is used to transfer data descriptors and SCSI
> +commands from VM to hypervisor and to transfer messages produced by
> +hypervisor to VM. Data itself is transferred via virtual scatter-gather DMA.
> +
> +PVSCSI Device Registers
> +=======================
> +
> +Registers area length is 1 page (PVSCSI_MEM_SPACE_COMMAND_NUM_PAGES).
> +Registers area structure is described by PVSCSIRegOffset enumeration.

The length of the registers area is ...  The structure of the registers
area is described by the PVSCSIRegOffset enum.

> +There are registers to issue device command (with optional short data),
> +issue device interrupt, control interrupts masking.
> +
> +PVSCSI Device Rings
> +===================
> +
> +There are three rings in shared memory:
> +
> +    1. Request ring (struct PVSCSIRingReqDesc *req_ring)
> +        - ring for OS to device requests
> +    2. Completion ring (struct PVSCSIRingCmpDesc *cmp_ring)
> +        - ring for device request completions
> +    3. Message ring (struct PVSCSIRingMsgDesc *msg_ring)
> +        - ring for messages from device.
> +       This ring is optional and may be not configured.

and the guest might not configure it.

> +There is a control area (struct PVSCSIRingsState *rings_state) used to 
> control
> +rings operation.
> +
> +PVSCSI Device to Host Interrupts
> +================================
> +There are following interrupt types supported by PVSCSI device:
> +    1. Completion interrupts (completion ring notifications):
> +        PVSCSI_INTR_CMPL_0
> +        PVSCSI_INTR_CMPL_1
> +    2. Message interrupts (message ring notifications):
> +        PVSCSI_INTR_MSG_0
> +        PVSCSI_INTR_MSG_1
> +
> +Interrupts are controlled via PVSCSI_REG_OFFSET_INTR_MASK register
> +Bit set means interrupt enabled, bit cleared - disabled
> +
> +Interrupt modes supported are legacy, MSI and MSI-X
> +In case of legacy interrupts register PVSCSI_REG_OFFSET_INTR_STATUS
> +used to verify interrupt arrival and to clear interrupt state
> +Interrupts are cleared by writing processed bits back
> +to interrupt status register.

In case of legacy interrupts, register PVSCSI_REG_OFFSET_INTR_STATUS
is used to check which interrupt has arrived.  Interrupts are
acknowledged when the corresponding bit is written to the interrupt
status register.

> +
> +PVSCSI Device Operation Sequences
> +=================================
> +
> +1. Startup sequence:
> +    a. Issue PVSCSI_CMD_ADAPTER_RESET command;
> +    aa. Windows driver reads interrupt status register here;
> +    b. Issue PVSCSI_CMD_SETUP_MSG_RING command with no additional data,
> +       check status and disable device messages if error returned;
> +       (Omitted if device messages disabled by driver configuration)

Can you add a boolean property to enable/disable the message ring?

> +    c. Issue PVSCSI_CMD_SETUP_RINGS command, provide rings configuration
> +       as struct PVSCSICmdDescSetupRings;
> +    d. Issue PVSCSI_CMD_SETUP_MSG_RING command again, provide
> +       rings configuration as struct PVSCSICmdDescSetupMsgRing;
> +    e. Unmask completion and message (if device messages enabled) interrupts.
> +
> +2. Shutdown sequences
> +    a. Mask interrupts;
> +    b. Flush request ring using PVSCSI_REG_OFFSET_KICK_NON_RW_IO;
> +    c. Issue PVSCSI_CMD_ADAPTER_RESET command.
> +
> +3. Send request
> +    a. Fill next free request ring descriptor;
> +    b. Issue PVSCSI_REG_OFFSET_KICK_RW_IO for R/W operations;
> +       or PVSCSI_REG_OFFSET_KICK_NON_RW_IO for other operations.
> +
> +4. Abort command
> +    a. Issue PVSCSI_CMD_ABORT_CMD command;
> +
> +5. Request completion processing
> +    a. Upon completion interrupt arrival process completion
> +       and message (if enabled) rings.
> diff --git a/hw/Makefile.objs b/hw/Makefile.objs
> index d0b2ecb..6e43763 100644
> --- a/hw/Makefile.objs
> +++ b/hw/Makefile.objs
> @@ -130,6 +130,7 @@ common-obj-$(CONFIG_OPENCORES_ETH) += opencores_eth.o
>  # SCSI layer
>  common-obj-$(CONFIG_LSI_SCSI_PCI) += lsi53c895a.o
>  common-obj-$(CONFIG_MEGASAS_SCSI_PCI) += megasas.o
> +common-obj-$(CONFIG_PVSCSI_SCSI_PCI) += pvscsi.o
>  common-obj-$(CONFIG_ESP) += esp.o
>  common-obj-$(CONFIG_ESP_PCI) += esp-pci.o
>  
> diff --git a/hw/pci/pci.h b/hw/pci/pci.h
> index 9ea67a3..1767fe5 100644
> --- a/hw/pci/pci.h
> +++ b/hw/pci/pci.h
> @@ -59,6 +59,7 @@
>  #define PCI_DEVICE_ID_VMWARE_SVGA        0x0710
>  #define PCI_DEVICE_ID_VMWARE_NET         0x0720
>  #define PCI_DEVICE_ID_VMWARE_SCSI        0x0730
> +#define PCI_DEVICE_ID_VMWARE_PVSCSI      0x07C0
>  #define PCI_DEVICE_ID_VMWARE_IDE         0x1729
>  #define PCI_DEVICE_ID_VMWARE_VMXNET3     0x07B0
>  
> diff --git a/hw/pvscsi.c b/hw/pvscsi.c
> new file mode 100644
> index 0000000..4c66671
> --- /dev/null
> +++ b/hw/pvscsi.c
> @@ -0,0 +1,1194 @@
> +/*
> + * QEMU VMWARE PVSCSI paravirtual SCSI bus
> + *
> + * Copyright (c) 2012 Ravello Systems LTD (http://ravellosystems.com)
> + *
> + * Developed by Daynix Computing LTD (http://www.daynix.com)
> + *
> + * Based on implementation by Paolo Bonzini
> + * http://lists.gnu.org/archive/html/qemu-devel/2011-08/msg00729.html
> + *
> + * Authors:
> + * Paolo Bonzini <address@hidden>
> + * Dmitry Fleytman <address@hidden>
> + * Yan Vugenfirer <address@hidden>
> + *
> + * This work is licensed under the terms of the GNU GPL, version 2.
> + * See the COPYING file in the top-level directory.
> + *
> + * NOTE about MSI-X:
> + * MSI-X support has been removed for the moment because it leads Windows OS
> + * to crash on startup. The crash happens because Windows driver requires
> + * MSI-X shared memory to be part of the same BAR used for rings state
> + * registers, etc. This is not supported by QEMU infrastructure so separate
> + * BAR created from MSI-X purposes. Windows driver fails to deal with 2 BARs.
> + *
> + */
> +
> +#include "scsi-defs.h"
> +#include "hw/scsi.h"
> +#include "hw/pci/msi.h"
> +#include "vmw_pvscsi.h"
> +#include "trace.h"
> +
> +
> +#define PVSCSI_MSI_OFFSET        (0x50)
> +#define PVSCSI_USE_64BIT         (true)
> +#define PVSCSI_PER_VECTOR_MASK   (false)
> +
> +#define PVSCSI_MAX_DEVS                   (64)
> +#define PVSCSI_MSIX_NUM_VECTORS           (1)
> +
> +#define PVSCSI_MAX_CMD_DATA_WORDS \
> +    (sizeof(PVSCSICmdDescSetupRings)/sizeof(uint32_t))
> +
> +#define RS_GET_FIELD(rs_pa, field) \
> +    (ldl_le_phys(rs_pa + offsetof(struct PVSCSIRingsState, field)))
> +#define RS_SET_FIELD(rs_pa, field, val) \
> +    (stl_le_phys(rs_pa + offsetof(struct PVSCSIRingsState, field), val))
> +
> +#define TYPE_PVSCSI "pvscsi"
> +#define PVSCSI(obj) OBJECT_CHECK(PVSCSIState, (obj), TYPE_PVSCSI)
> +
> +typedef struct PVSCSIRingsMgr {

PVSCSIRingInfo, or just put it in PVSCSIState.

> +    uint64_t            rs_pa;
> +    uint32_t            txr_len_mask;
> +    uint32_t            rxr_len_mask;
> +    uint32_t            msg_len_mask;
> +    uint64_t            req_ring_pages_pa[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES];
> +    uint64_t            cmp_ring_pages_pa[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES];
> +    uint64_t            
> msg_ring_pages_pa[PVSCSI_SETUP_MSG_RING_MAX_NUM_PAGES];
> +    uint64_t            consumed_ptr;
> +    uint64_t            filled_cmp_ptr;
> +    uint64_t            filled_msg_ptr;
> +} PVSCSIRingsMgr;
> +
> +typedef struct PVSCSISGState {
> +    hwaddr elemAddr;
> +    hwaddr dataAddr;
> +    uint32_t resid;
> +} PVSCSISGState;
> +
> +typedef QTAILQ_HEAD(, PVSCSIRequest) PVSCSIRequestList;
> +
> +typedef struct {
> +    PCIDevice parent_obj;
> +    MemoryRegion io_space;
> +    SCSIBus bus;
> +    QEMUBH *completion_worker;
> +    PVSCSIRequestList pending_queue;
> +    PVSCSIRequestList completion_queue;
> +
> +    uint64_t reg_interrupt_status;        /* Interrupt status register value 
> */
> +    uint64_t reg_interrupt_enabled;       /* Interrupt mask register value   
> */
> +    uint64_t reg_command_status;          /* Command status register value   
> */
> +
> +    /* Command data adoption mechanism */
> +    uint64_t curr_cmd;                   /* Last command arrived             
> */
> +    uint32_t curr_cmd_data_cntr;         /* Amount of data for last command  
> */
> +
> +    /* Collector for current command data */
> +    uint32_t curr_cmd_data[PVSCSI_MAX_CMD_DATA_WORDS];
> +
> +    uint8_t rings_info_valid;            /* Whether data rings initialized   
> */
> +    uint8_t msg_ring_info_valid;         /* Whether message ring initialized 
> */
> +
> +    uint8_t msi_used;    /* Whether MSI support was installed successfully   
> */
> +
> +    PVSCSIRingsMgr rings;                /* Data transfer rings manager      
> */
> +} PVSCSIState;
> +
> +typedef struct PVSCSIRequest {
> +    SCSIRequest *sreq;
> +    PVSCSIState *dev;
> +    uint8_t sense_key;
> +    uint8_t completed;
> +    int lun;
> +    QEMUSGList sgl;
> +    PVSCSISGState sg;
> +    struct PVSCSIRingReqDesc req;
> +    struct PVSCSIRingCmpDesc cmp;
> +    QTAILQ_ENTRY(PVSCSIRequest) next;
> +} PVSCSIRequest;

This needs to be serialized and loaded back if you want to support
migration with rerror=stop/werror=stop.  See how it's done in
virtio-scsi.  To test it, you can make a disk readonly with blockdev
while QEMU is running, migrate to file, make it read-write again and
migrate back from the file.

This is not blocking the patch though.


> +/* Integer binary logarithm */
> +static int
> +pvscsi_log2(uint32_t input)
> +{
> +    int log = 0;
> +    assert(input > 0);
> +    while (input >> ++log) {
> +    }
> +    return log;
> +}
> +
> +static void
> +pvscsi_rings_mgr_init_data(PVSCSIRingsMgr *m, PVSCSICmdDescSetupRings *ri)

s/pvscsi_rings_mgr/pvscsi_ring/g

> +{
> +    int i;
> +    uint32_t txr_len_log2, rxr_len_log2;
> +    uint32_t req_ring_size, cmp_ring_size;
> +    m->rs_pa = ri->ringsStatePPN << VMW_PAGE_SHIFT;
> +
> +    req_ring_size = ri->reqRingNumPages * 
> PVSCSI_MAX_NUM_REQ_ENTRIES_PER_PAGE;
> +    cmp_ring_size = ri->cmpRingNumPages * 
> PVSCSI_MAX_NUM_CMP_ENTRIES_PER_PAGE;
> +    txr_len_log2 = pvscsi_log2(req_ring_size - 1);
> +    rxr_len_log2 = pvscsi_log2(cmp_ring_size - 1);
> +
> +    m->txr_len_mask = MASK(txr_len_log2);
> +    m->rxr_len_mask = MASK(rxr_len_log2);
> +
> +    m->consumed_ptr = 0;
> +    m->filled_cmp_ptr = 0;
> +
> +    for (i = 0; i < ri->reqRingNumPages; i++) {
> +        m->req_ring_pages_pa[i] = ri->reqRingPPNs[i] << VMW_PAGE_SHIFT;
> +    }
> +
> +    for (i = 0; i < ri->cmpRingNumPages; i++) {
> +        m->cmp_ring_pages_pa[i] = ri->cmpRingPPNs[i] << VMW_PAGE_SHIFT;
> +    }
> +
> +    RS_SET_FIELD(m->rs_pa, reqProdIdx, 0);
> +    RS_SET_FIELD(m->rs_pa, reqConsIdx, 0);
> +    RS_SET_FIELD(m->rs_pa, reqNumEntriesLog2, txr_len_log2);
> +
> +    RS_SET_FIELD(m->rs_pa, cmpProdIdx, 0);
> +    RS_SET_FIELD(m->rs_pa, cmpConsIdx, 0);
> +    RS_SET_FIELD(m->rs_pa, cmpNumEntriesLog2, rxr_len_log2);
> +
> +    trace_pvscsi_rings_mgr_init_data(txr_len_log2, rxr_len_log2);
> +
> +    /* Flush ring state page changes */
> +    smp_wmb();
> +}
> +
> +static void
> +pvscsi_rings_mgr_init_msg(PVSCSIRingsMgr *m, PVSCSICmdDescSetupMsgRing *ri)
> +{
> +    int i;
> +    uint32_t len_log2;
> +    uint32_t ring_size;
> +
> +    ring_size = ri->numPages * PVSCSI_MAX_NUM_MSG_ENTRIES_PER_PAGE;
> +    len_log2 = pvscsi_log2(ring_size - 1);
> +
> +    m->msg_len_mask = MASK(len_log2);
> +
> +    m->filled_msg_ptr = 0;
> +
> +    for (i = 0; i < ri->numPages; i++) {
> +        m->msg_ring_pages_pa[i] = ri->ringPPNs[i] << VMW_PAGE_SHIFT;
> +    }
> +
> +    RS_SET_FIELD(m->rs_pa, msgProdIdx, 0);
> +    RS_SET_FIELD(m->rs_pa, msgConsIdx, 0);
> +    RS_SET_FIELD(m->rs_pa, msgNumEntriesLog2, len_log2);
> +
> +    trace_pvscsi_rings_mgr_init_msg(len_log2);
> +
> +    /* Flush ring state page changes */
> +    smp_wmb();
> +}
> +
> +static void
> +pvscsi_rings_mgr_cleanup(PVSCSIRingsMgr *mgr)
> +{
> +    mgr->rs_pa = 0;
> +    mgr->txr_len_mask = 0;
> +    mgr->rxr_len_mask = 0;
> +    mgr->msg_len_mask = 0;
> +    mgr->consumed_ptr = 0;
> +    mgr->filled_cmp_ptr = 0;
> +    mgr->filled_msg_ptr = 0;
> +    memset(mgr->req_ring_pages_pa, 0, sizeof(mgr->req_ring_pages_pa));
> +    memset(mgr->cmp_ring_pages_pa, 0, sizeof(mgr->cmp_ring_pages_pa));
> +    memset(mgr->msg_ring_pages_pa, 0, sizeof(mgr->msg_ring_pages_pa));
> +}
> +
> +static hwaddr
> +pvscsi_rings_mgr_pop_req_descr(PVSCSIRingsMgr *mgr)
> +{
> +    uint32_t ready_ptr = RS_GET_FIELD(mgr->rs_pa, reqProdIdx);
> +
> +    if (ready_ptr != mgr->consumed_ptr) {
> +        uint32_t next_ready_ptr =
> +            mgr->consumed_ptr++ & mgr->txr_len_mask;
> +        uint32_t next_ready_page =
> +            next_ready_ptr / PVSCSI_MAX_NUM_REQ_ENTRIES_PER_PAGE;
> +        uint32_t inpage_idx =
> +            next_ready_ptr % PVSCSI_MAX_NUM_REQ_ENTRIES_PER_PAGE;
> +
> +        return mgr->req_ring_pages_pa[next_ready_page] +
> +               inpage_idx * sizeof(PVSCSIRingReqDesc);
> +    } else {
> +        return 0;
> +    }
> +}
> +
> +static void
> +pvscsi_rings_mgr_flush_req_ring(PVSCSIRingsMgr *mgr)

pvscsi_ring_flush_req, and likewise for other functions that have _ring
at the end.

> +{
> +    RS_SET_FIELD(mgr->rs_pa, reqConsIdx, mgr->consumed_ptr);
> +}
> +
> +static hwaddr
> +pvscsi_rings_mgr_pop_cmp_descr(PVSCSIRingsMgr *mgr)
> +{
> +    /*
> +     * According to Linux driver code it explicitly verifies that number
> +     * of requests being processed by device is less then the size of
> +     * completion queue, so device may omit completion queue overflow
> +     * conditions check. We assume that this is true for other (Windows)
> +     * drivers as well.
> +     */
> +
> +    uint32_t free_cmp_ptr =
> +        mgr->filled_cmp_ptr++ & mgr->rxr_len_mask;
> +    uint32_t free_cmp_page =
> +        free_cmp_ptr / PVSCSI_MAX_NUM_CMP_ENTRIES_PER_PAGE;
> +    uint32_t inpage_idx =
> +        free_cmp_ptr % PVSCSI_MAX_NUM_CMP_ENTRIES_PER_PAGE;
> +    return mgr->cmp_ring_pages_pa[free_cmp_page] +
> +           inpage_idx * sizeof(PVSCSIRingCmpDesc);
> +}
> +
> +static hwaddr
> +pvscsi_rings_mgr_pop_msg_descr(PVSCSIRingsMgr *mgr)
> +{
> +    uint32_t free_msg_ptr =
> +        mgr->filled_msg_ptr++ & mgr->msg_len_mask;
> +    uint32_t free_msg_page =
> +        free_msg_ptr / PVSCSI_MAX_NUM_MSG_ENTRIES_PER_PAGE;
> +    uint32_t inpage_idx =
> +        free_msg_ptr % PVSCSI_MAX_NUM_MSG_ENTRIES_PER_PAGE;
> +    return mgr->msg_ring_pages_pa[free_msg_page] +
> +           inpage_idx * sizeof(PVSCSIRingMsgDesc);
> +}
> +
> +static void
> +pvscsi_rings_mgr_flush_cmp_ring(PVSCSIRingsMgr *mgr)
> +{
> +    /* Flush descriptor changes */
> +    smp_wmb();
> +
> +    trace_pvscsi_rings_mgr_flush_cmp_ring(mgr->filled_cmp_ptr);
> +
> +    RS_SET_FIELD(mgr->rs_pa, cmpProdIdx, mgr->filled_cmp_ptr);
> +}
> +
> +static bool
> +pvscsi_rings_mgr_msg_has_room(PVSCSIRingsMgr *mgr)
> +{
> +    uint32_t prodIdx = RS_GET_FIELD(mgr->rs_pa, msgProdIdx);
> +    uint32_t consIdx = RS_GET_FIELD(mgr->rs_pa, msgConsIdx);
> +
> +    return (prodIdx - consIdx) < (mgr->msg_len_mask + 1);
> +}
> +
> +static void
> +pvscsi_rings_mgr_flush_msg_ring(PVSCSIRingsMgr *mgr)
> +{
> +    /* Flush descriptor changes */
> +    smp_wmb();
> +
> +    trace_pvscsi_rings_mgr_flush_msg_ring(mgr->filled_msg_ptr);
> +
> +    RS_SET_FIELD(mgr->rs_pa, msgProdIdx, mgr->filled_msg_ptr);
> +}
> +
> +static void
> +pvscsi_reset_state(PVSCSIState *s)
> +{
> +    s->curr_cmd = PVSCSI_CMD_FIRST;
> +    s->curr_cmd_data_cntr = 0;
> +    s->reg_command_status = PVSCSI_COMMAND_PROCESSING_SUCCEEDED;
> +    s->reg_interrupt_status = 0;
> +    pvscsi_rings_mgr_cleanup(&s->rings);
> +    s->rings_info_valid = FALSE;
> +    s->msg_ring_info_valid = FALSE;
> +    QTAILQ_INIT(&s->pending_queue);
> +    QTAILQ_INIT(&s->completion_queue);
> +}
> +
> +static void
> +pvscsi_free_queue(PVSCSIRequestList *req_list)

This shouldn't be needed.

> +{
> +    PVSCSIRequest *pvscsi_req;
> +
> +    while (!QTAILQ_EMPTY(req_list)) {
> +        pvscsi_req = QTAILQ_FIRST(req_list);
> +        QTAILQ_REMOVE(req_list, pvscsi_req, next);
> +        g_free(pvscsi_req);
> +    }
> +}
> +
> +static void
> +pvscsi_reset_adapter(PVSCSIState *s)
> +{
> +    qbus_reset_all_fn(&s->bus);
> +    pvscsi_free_queue(&s->completion_queue);
> +    assert(QTAILQ_EMPTY(&s->pending_queue));
> +    pvscsi_reset_state(s);
> +}
> +
> +static void
> +pvscsi_update_irq_status(PVSCSIState *s)
> +{
> +    PCIDevice *d = PCI_DEVICE(s);
> +    bool should_raise = s->reg_interrupt_enabled & s->reg_interrupt_status;
> +
> +    trace_pvscsi_update_irq_level(should_raise, s->reg_interrupt_enabled,
> +                                  s->reg_interrupt_status);
> +
> +    if (s->msi_used && msi_enabled(d)) {
> +        if (should_raise) {
> +            trace_pvscsi_update_irq_msi();
> +            msi_notify(d, PVSCSI_VECTOR_COMPLETION);
> +        }
> +        return;
> +    }
> +
> +    qemu_set_irq(d->irq[0], !!should_raise);
> +}
> +
> +static void
> +pvscsi_raise_completion_interrupt(PVSCSIState *s)
> +{
> +    s->reg_interrupt_status |= PVSCSI_INTR_CMPL_0;

Did you find out how you're supposed to use PVSCSI_INTR_CMPL_1?

> +    /* Memory barrier to flush interrupt status register changes*/
> +    smp_wmb();
> +
> +    pvscsi_update_irq_status(s);
> +}
> +
> +static void
> +pvscsi_raise_message_interrupt(PVSCSIState *s)
> +{
> +    s->reg_interrupt_status |= PVSCSI_INTR_MSG_0;
> +
> +    /* Memory barrier to flush interrupt status register changes*/
> +    smp_wmb();
> +
> +    pvscsi_update_irq_status(s);
> +}
> +
> +static void
> +pvscsi_cmp_ring_put(PVSCSIState *s, struct PVSCSIRingCmpDesc *cmp_desc)
> +{
> +    hwaddr cmp_descr_pa;
> +
> +    cmp_descr_pa = pvscsi_rings_mgr_pop_cmp_descr(&s->rings);
> +    trace_pvscsi_cmp_ring_put(cmp_descr_pa);
> +    cpu_physical_memory_write(cmp_descr_pa, (void *)cmp_desc,
> +                              sizeof(*cmp_desc));
> +}
> +
> +static void
> +pvscsi_msg_ring_put(PVSCSIState *s, struct PVSCSIRingMsgDesc *msg_desc)
> +{
> +    hwaddr msg_descr_pa;
> +
> +    msg_descr_pa = pvscsi_rings_mgr_pop_msg_descr(&s->rings);
> +    trace_pvscsi_msg_ring_put(msg_descr_pa);
> +    cpu_physical_memory_write(msg_descr_pa, (void *)msg_desc,
> +                              sizeof(*msg_desc));
> +}
> +
> +static void
> +pvscsi_process_completion_queue(void *opaque)
> +{
> +    PVSCSIState *s = opaque;
> +    PVSCSIRequest *pvscsi_req;
> +    bool has_completed = false;
> +
> +    while (!QTAILQ_EMPTY(&s->completion_queue)) {
> +        pvscsi_req = QTAILQ_FIRST(&s->completion_queue);
> +        QTAILQ_REMOVE(&s->completion_queue, pvscsi_req, next);
> +        pvscsi_cmp_ring_put(s, &pvscsi_req->cmp);
> +        g_free(pvscsi_req);
> +        has_completed++;
> +    }
> +
> +    if (has_completed) {
> +        pvscsi_rings_mgr_flush_cmp_ring(&s->rings);
> +        pvscsi_raise_completion_interrupt(s);
> +    }
> +}
> +
> +static void
> +pvscsi_schedule_completion_processing(PVSCSIState *s)
> +{
> +    /* Try putting more complete requests on the ring. */
> +    if (!QTAILQ_EMPTY(&s->completion_queue)) {
> +        qemu_bh_schedule(s->completion_worker);
> +    }
> +}
> +
> +static void
> +pvscsi_complete_request(PVSCSIState *s, PVSCSIRequest *r)
> +{
> +    assert(!r->completed);
> +
> +    trace_pvscsi_complete_request(r->cmp.context, r->cmp.dataLen,
> +                                  r->sense_key);
> +    if (r->sreq != NULL) {
> +        scsi_req_unref(r->sreq);
> +        r->sreq = NULL;
> +    }
> +    r->completed = 1;
> +    QTAILQ_REMOVE(&s->pending_queue, r, next);
> +    QTAILQ_INSERT_TAIL(&s->completion_queue, r, next);
> +    pvscsi_schedule_completion_processing(s);
> +}
> +
> +static QEMUSGList *pvscsi_get_sg_list(SCSIRequest *r)
> +{
> +    PVSCSIRequest *req = r->hba_private;
> +
> +    trace_pvscsi_get_sg_list(req->sgl.nsg, req->sgl.size);
> +
> +    return &req->sgl;
> +}
> +
> +static void
> +pvscsi_get_next_sg_elem(PVSCSISGState *sg)
> +{
> +    struct PVSCSISGElement elem;
> +
> +    for (;; sg->elemAddr = elem.addr) {

Please remove the for loop altogether.

> +        cpu_physical_memory_read(sg->elemAddr, (void *)&elem, sizeof(elem));
> +        if ((elem.flags & ~PVSCSI_KNOWN_FLAGS) != 0) {
> +            /*
> +             * There is PVSCSI_SGE_FLAG_CHAIN_ELEMENT flag described in
> +             * header file but its value is unknown. This flag requires
> +             * additional processing, so we put warning here to catch it
> +             * some day and make proper implementation
> +             */
> +            trace_pvscsi_get_next_sg_elem(elem.flags);
> +        }
> +        break;
> +    }
> +
> +    sg->elemAddr += sizeof(elem);
> +    sg->dataAddr = elem.addr;
> +    sg->resid = elem.length;
> +}
> +
> +static void
> +pvscsi_write_sense(PVSCSIRequest *r, uint8_t *sense, int len)
> +{
> +    r->cmp.senseLen = MIN(r->req.senseLen, len);
> +    r->sense_key = sense[2];

The key is in sense[1] if bit 1 of sense[0] is 1.  See scsi_build_sense
in hw/scsi/scsi-bus.c.

> +    cpu_physical_memory_write(r->req.senseAddr, sense, r->cmp.senseLen);
> +}
> +
> +static void
> +pvscsi_command_complete(SCSIRequest *req, uint32_t status, size_t resid)
> +{
> +    PVSCSIRequest *pvscsi_req = req->hba_private;
> +    PVSCSIState *s = pvscsi_req->dev;
> +
> +    if (!pvscsi_req) {
> +        trace_pvscsi_command_complete_not_found(req->tag);
> +        return;
> +    }
> +
> +    if (resid) {
> +        /* Short transfer.  */
> +        trace_pvscsi_command_complete_data_run();
> +        pvscsi_req->cmp.hostStatus = BTSTAT_DATARUN;
> +    }
> +
> +    pvscsi_req->cmp.scsiStatus = status;
> +    if (pvscsi_req->cmp.scsiStatus == CHECK_CONDITION) {
> +        uint8_t sense[SCSI_SENSE_BUF_SIZE];
> +        int sense_len =
> +            scsi_req_get_sense(pvscsi_req->sreq, sense, sizeof(sense));
> +
> +        trace_pvscsi_command_complete_sense_len(sense_len);
> +        pvscsi_write_sense(pvscsi_req, sense, sense_len);
> +    }
> +    qemu_sglist_destroy(&pvscsi_req->sgl);
> +    pvscsi_complete_request(s, pvscsi_req);
> +}
> +
> +static void
> +pvscsi_send_msg(PVSCSIState *s, SCSIDevice *dev, uint32_t msg_type)
> +{
> +    if (s->msg_ring_info_valid && pvscsi_rings_mgr_msg_has_room(&s->rings)) {
> +        PVSCSIMsgDescDevStatusChanged msg = {0};
> +
> +        msg.type = msg_type;
> +        msg.bus = dev->channel;
> +        msg.target = dev->id;
> +        msg.lun[1] = dev->lun;
> +
> +        pvscsi_msg_ring_put(s, (PVSCSIRingMsgDesc *)&msg);
> +        pvscsi_rings_mgr_flush_msg_ring(&s->rings);
> +        pvscsi_raise_message_interrupt(s);
> +    }
> +}
> +
> +static void
> +pvscsi_hotplug(SCSIBus *bus, SCSIDevice *dev)
> +{
> +    PVSCSIState *s = container_of(bus, PVSCSIState, bus);
> +    pvscsi_send_msg(s, dev, PVSCSI_MSG_DEV_ADDED);
> +}
> +
> +static void
> +pvscsi_hot_unplug(SCSIBus *bus, SCSIDevice *dev)
> +{
> +    PVSCSIState *s = container_of(bus, PVSCSIState, bus);
> +    pvscsi_send_msg(s, dev, PVSCSI_MSG_DEV_REMOVED);
> +}
> +
> +static void
> +pvscsi_request_cancelled(SCSIRequest *req)
> +{
> +    PVSCSIRequest *pvscsi_req = req->hba_private;
> +    PVSCSIState *s = pvscsi_req->dev;
> +
> +    if (pvscsi_req->cmp.hostStatus == BTSTAT_SUCCESS) {
> +        pvscsi_req->cmp.hostStatus = BTSTAT_ABORTQUEUE;
> +    }

virtio-scsi has a "resetting" field and sets BTSTAT_BUSRESET if it is
one.  Perhaps you can do the same.

> +    pvscsi_complete_request(s, pvscsi_req);
> +}
> +
> +static SCSIDevice*
> +pvscsi_device_find(PVSCSIState *s, int channel, int target,
> +                   uint8_t *requested_lun, uint8_t *target_lun)
> +{
> +    if (requested_lun[0] || requested_lun[2] || requested_lun[3] ||
> +        requested_lun[4] || requested_lun[5] || requested_lun[6] ||
> +        requested_lun[7] || (target > PVSCSI_MAX_DEVS)) {
> +        return NULL;
> +    } else {
> +        *target_lun = requested_lun[1];
> +        return scsi_device_find(&s->bus, channel, target, *target_lun);
> +    }
> +}
> +
> +static PVSCSIRequest *
> +pvscsi_queue_pending_descriptor(PVSCSIState *s, SCSIDevice **d,
> +                                struct PVSCSIRingReqDesc *descr)
> +{
> +    PVSCSIRequest *pvscsi_req;
> +    uint8_t lun;
> +
> +    pvscsi_req = g_malloc0(sizeof(*pvscsi_req));
> +    pvscsi_req->dev = s;
> +    pvscsi_req->req = *descr;
> +    pvscsi_req->cmp.context = pvscsi_req->req.context;
> +    QTAILQ_INSERT_TAIL(&s->pending_queue, pvscsi_req, next);
> +
> +    *d = pvscsi_device_find(s, descr->bus, descr->target, descr->lun, &lun);
> +    if (!*d) {
> +        return pvscsi_req;
> +    }
> +
> +    pvscsi_req->lun = lun;
> +    return pvscsi_req;
> +}
> +
> +static void
> +pvscsi_convert_sglist(PVSCSIRequest *r)
> +{
> +    int chunk_size;
> +    uint64_t data_length = r->req.dataLen;
> +    PVSCSISGState sg = r->sg;
> +    while (data_length) {
> +        while (!sg.resid) {
> +            pvscsi_get_next_sg_elem(&sg);
> +            trace_pvscsi_convert_sglist(r->req.context, r->sg.dataAddr,
> +                                        r->sg.resid);
> +        }
> +        assert(data_length > 0);
> +        chunk_size = MIN((unsigned) data_length, sg.resid);
> +        if (chunk_size) {
> +            qemu_sglist_add(&r->sgl, sg.dataAddr, chunk_size);
> +        }
> +
> +        sg.dataAddr += chunk_size;
> +        data_length -= chunk_size;
> +        sg.resid -= chunk_size;
> +    }
> +}
> +
> +static void
> +pvscsi_build_sglist(PVSCSIState *s, PVSCSIRequest *r)
> +{
> +    PCIDevice *d = PCI_DEVICE(s);
> +
> +    qemu_sglist_init(&r->sgl, 1, pci_dma_context(d));
> +    if (r->req.flags & PVSCSI_FLAG_CMD_WITH_SG_LIST) {
> +        pvscsi_convert_sglist(r);
> +    } else {
> +        qemu_sglist_add(&r->sgl, r->req.dataAddr, r->req.dataLen);
> +    }
> +}
> +
> +static void
> +pvscsi_process_request_descriptor(PVSCSIState *s,
> +                                  struct PVSCSIRingReqDesc *descr)
> +{
> +    SCSIDevice *d;
> +    PVSCSIRequest *r = pvscsi_queue_pending_descriptor(s, &d, descr);
> +    int64_t n;
> +
> +    trace_pvscsi_process_req_descr(descr->cdb[0], descr->context);
> +
> +    if (!d) {
> +        r->cmp.hostStatus = BTSTAT_SELTIMEO;
> +        trace_pvscsi_process_req_descr_unknown_device();
> +        pvscsi_complete_request(s, r);
> +        return;
> +    }
> +
> +    if (descr->flags & PVSCSI_FLAG_CMD_WITH_SG_LIST) {
> +        r->sg.elemAddr = descr->dataAddr;
> +    }
> +
> +    r->sreq = scsi_req_new(d, descr->context, r->lun, descr->cdb, r);
> +    if (r->sreq->cmd.mode == SCSI_XFER_FROM_DEV &&
> +        (descr->flags & PVSCSI_FLAG_CMD_DIR_TODEVICE)) {
> +        r->cmp.hostStatus = BTSTAT_BADMSG;
> +        trace_pvscsi_process_req_descr_invalid_dir();
> +        scsi_req_cancel(r->sreq);
> +        return;
> +    }
> +    if (r->sreq->cmd.mode == SCSI_XFER_TO_DEV &&
> +        (descr->flags & PVSCSI_FLAG_CMD_DIR_TOHOST)) {
> +        r->cmp.hostStatus = BTSTAT_BADMSG;
> +        trace_pvscsi_process_req_descr_invalid_dir();
> +        scsi_req_cancel(r->sreq);
> +        return;
> +    }
> +
> +    pvscsi_build_sglist(s, r);
> +    n = scsi_req_enqueue(r->sreq);
> +
> +    if (n) {
> +        scsi_req_continue(r->sreq);
> +    }
> +}
> +
> +static void
> +pvscsi_process_io(PVSCSIState *s)
> +{
> +    PVSCSIRingReqDesc descr;
> +    hwaddr next_descr_pa;
> +
> +    assert(s->rings_info_valid);
> +    while ((next_descr_pa = pvscsi_rings_mgr_pop_req_descr(&s->rings)) != 0) 
> {
> +
> +        /* Only read after production index verification */
> +        smp_rmb();
> +
> +        trace_pvscsi_process_io(next_descr_pa);
> +        cpu_physical_memory_read(next_descr_pa, &descr, sizeof(descr));
> +        pvscsi_process_request_descriptor(s, &descr);
> +    }
> +
> +    pvscsi_rings_mgr_flush_req_ring(&s->rings);
> +}
> +
> +static void
> +pvscsi_dbg_dump_tx_rings_config(PVSCSICmdDescSetupRings *rc)
> +{
> +    int i;
> +    trace_pvscsi_tx_rings_ppn("Rings State", rc->ringsStatePPN);
> +
> +    trace_pvscsi_tx_rings_num_pages("Request Ring", rc->reqRingNumPages);
> +    for (i = 0; i < rc->reqRingNumPages; i++) {
> +        trace_pvscsi_tx_rings_ppn("Request Ring", rc->reqRingPPNs[i]);
> +    }
> +
> +    trace_pvscsi_tx_rings_num_pages("Confirm Ring", rc->cmpRingNumPages);
> +    for (i = 0; i < rc->cmpRingNumPages; i++) {
> +        trace_pvscsi_tx_rings_ppn("Confirm Ring", rc->reqRingPPNs[i]);
> +    }
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_config(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_noimpl("PVSCSI_CMD_CONFIG");
> +    return PVSCSI_COMMAND_PROCESSING_FAILED;
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_unplug(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_noimpl("PVSCSI_CMD_DEVICE_UNPLUG");
> +    return PVSCSI_COMMAND_PROCESSING_FAILED;
> +}
> +
> +static uint64_t
> +pvscsi_on_issue_scsi(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_noimpl("PVSCSI_CMD_ISSUE_SCSI");
> +    return PVSCSI_COMMAND_PROCESSING_FAILED;
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_setup_rings(PVSCSIState *s)
> +{
> +    PVSCSICmdDescSetupRings *rc =
> +        (PVSCSICmdDescSetupRings *) s->curr_cmd_data;
> +
> +    trace_pvscsi_on_cmd_arrived("PVSCSI_CMD_SETUP_RINGS");
> +
> +    pvscsi_dbg_dump_tx_rings_config(rc);
> +    pvscsi_rings_mgr_init_data(&s->rings, rc);
> +    s->rings_info_valid = TRUE;
> +    return PVSCSI_COMMAND_PROCESSING_SUCCEEDED;
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_abort(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_abort(
> +        ((struct PVSCSICmdDescAbortCmd *) s->curr_cmd_data)->context,
> +        ((struct PVSCSICmdDescAbortCmd *) s->curr_cmd_data)->target);
> +    return PVSCSI_COMMAND_PROCESSING_SUCCEEDED;

You need to call scsi_req_cancel here before returning.

> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_unknown(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_unknown_data(s->curr_cmd_data[0]);
> +    return PVSCSI_COMMAND_PROCESSING_FAILED;
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_reset_device(PVSCSIState *s)
> +{
> +    uint8_t target_lun = 0;
> +    struct PVSCSICmdDescResetDevice *cmd =
> +        (struct PVSCSICmdDescResetDevice *) s->curr_cmd_data;
> +    SCSIDevice *sdev;
> +
> +    sdev = pvscsi_device_find(s, 0, cmd->target, cmd->lun, &target_lun);
> +
> +    trace_pvscsi_on_cmd_reset_dev(cmd->target, (int) target_lun, sdev);
> +
> +    if (sdev != NULL) {
> +        device_reset(&sdev->qdev);
> +        return PVSCSI_COMMAND_PROCESSING_SUCCEEDED;
> +    }
> +
> +    return PVSCSI_COMMAND_PROCESSING_FAILED;
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_reset_bus(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_arrived("PVSCSI_CMD_RESET_BUS");
> +
> +    qbus_reset_all_fn(&s->bus);
> +    return PVSCSI_COMMAND_PROCESSING_SUCCEEDED;
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_setup_msg_ring(PVSCSIState *s)
> +{
> +    PVSCSICmdDescSetupMsgRing *rc =
> +        (PVSCSICmdDescSetupMsgRing *) s->curr_cmd_data;
> +
> +    trace_pvscsi_on_cmd_arrived("PVSCSI_CMD_SETUP_MSG_RING");
> +
> +    if (s->rings_info_valid) {
> +        pvscsi_rings_mgr_init_msg(&s->rings, rc);
> +        s->msg_ring_info_valid = TRUE;
> +    }
> +    return sizeof(PVSCSICmdDescSetupMsgRing) / sizeof(uint32_t);
> +}
> +
> +static uint64_t
> +pvscsi_on_cmd_adapter_reset(PVSCSIState *s)
> +{
> +    trace_pvscsi_on_cmd_arrived("PVSCSI_CMD_ADAPTER_RESET");
> +
> +    pvscsi_reset_adapter(s);
> +    return PVSCSI_COMMAND_PROCESSING_SUCCEEDED;
> +}
> +
> +static const struct {
> +    int       data_size;
> +    uint64_t  (*handler_fn)(PVSCSIState *s);
> +} pvscsi_commands[] = {
> +    [PVSCSI_CMD_FIRST] = {
> +        .data_size = 0,
> +        .handler_fn = pvscsi_on_cmd_unknown,
> +    },
> +
> +    /* Not implemented, data size defined based on what arrives on windows */
> +    [PVSCSI_CMD_CONFIG] = {
> +        .data_size = 6 * sizeof(uint32_t),
> +        .handler_fn = pvscsi_on_cmd_config,
> +    },
> +
> +    /* Command not implemented, data size is unknown */
> +    [PVSCSI_CMD_ISSUE_SCSI] = {
> +        .data_size = 0,
> +        .handler_fn = pvscsi_on_issue_scsi,
> +    },
> +
> +    /* Command not implemented, data size is unknown */
> +    [PVSCSI_CMD_DEVICE_UNPLUG] = {
> +        .data_size = 0,
> +        .handler_fn = pvscsi_on_cmd_unplug,
> +    },
> +
> +    [PVSCSI_CMD_SETUP_RINGS] = {
> +        .data_size = sizeof(PVSCSICmdDescSetupRings),
> +        .handler_fn = pvscsi_on_cmd_setup_rings,
> +    },
> +
> +    [PVSCSI_CMD_RESET_DEVICE] = {
> +        .data_size = sizeof(struct PVSCSICmdDescResetDevice),
> +        .handler_fn = pvscsi_on_cmd_reset_device,
> +    },
> +
> +    [PVSCSI_CMD_RESET_BUS] = {
> +        .data_size = 0,
> +        .handler_fn = pvscsi_on_cmd_reset_bus,
> +    },
> +
> +    [PVSCSI_CMD_SETUP_MSG_RING] = {
> +        .data_size = sizeof(PVSCSICmdDescSetupMsgRing),
> +        .handler_fn = pvscsi_on_cmd_setup_msg_ring,
> +    },
> +
> +    [PVSCSI_CMD_ADAPTER_RESET] = {
> +        .data_size = 0,
> +        .handler_fn = pvscsi_on_cmd_adapter_reset,
> +    },
> +
> +    [PVSCSI_CMD_ABORT_CMD] = {
> +        .data_size = sizeof(struct PVSCSICmdDescAbortCmd),
> +        .handler_fn = pvscsi_on_cmd_abort,
> +    },
> +};
> +
> +static void
> +pvscsi_do_command_processing(PVSCSIState *s)
> +{
> +    size_t bytes_arrived = s->curr_cmd_data_cntr * sizeof(uint32_t);
> +
> +    assert(s->curr_cmd < PVSCSI_CMD_LAST);
> +    if (bytes_arrived >= pvscsi_commands[s->curr_cmd].data_size) {
> +        s->reg_command_status = pvscsi_commands[s->curr_cmd].handler_fn(s);
> +        s->curr_cmd = PVSCSI_CMD_FIRST;
> +        s->curr_cmd_data_cntr   = 0;
> +    }
> +}
> +
> +static void
> +pvscsi_on_command_data(PVSCSIState *s, uint32_t value)
> +{
> +    size_t bytes_arrived = s->curr_cmd_data_cntr * sizeof(uint32_t);
> +
> +    assert(bytes_arrived < sizeof(s->curr_cmd_data));
> +    s->curr_cmd_data[s->curr_cmd_data_cntr++] = value;
> +
> +    pvscsi_do_command_processing(s);
> +}
> +
> +static void
> +pvscsi_on_command(PVSCSIState *s, uint64_t cmd_id)
> +{
> +    if ((cmd_id > PVSCSI_CMD_FIRST) && (cmd_id < PVSCSI_CMD_LAST)) {
> +        s->curr_cmd = cmd_id;
> +    } else {
> +        s->curr_cmd = PVSCSI_CMD_FIRST;
> +        trace_pvscsi_on_cmd_unknown(cmd_id);
> +    }
> +
> +    s->curr_cmd_data_cntr = 0;
> +    s->reg_command_status = PVSCSI_COMMAND_NOT_ENOUGH_DATA;
> +
> +    pvscsi_do_command_processing(s);
> +}
> +
> +static void
> +pvscsi_io_write(void *opaque, hwaddr addr,
> +                uint64_t val, unsigned size)
> +{
> +    PVSCSIState *s = opaque;
> +
> +    switch (addr) {
> +    case PVSCSI_REG_OFFSET_COMMAND:
> +        pvscsi_on_command(s, val);
> +        break;
> +
> +    case PVSCSI_REG_OFFSET_COMMAND_DATA:
> +        pvscsi_on_command_data(s, (uint32_t) val);
> +        break;
> +
> +    case PVSCSI_REG_OFFSET_INTR_STATUS:
> +        trace_pvscsi_io_write("PVSCSI_REG_OFFSET_INTR_STATUS", val);
> +        s->reg_interrupt_status &= ~val;
> +        pvscsi_update_irq_status(s);
> +        pvscsi_schedule_completion_processing(s);
> +        break;
> +
> +    case PVSCSI_REG_OFFSET_INTR_MASK:
> +        trace_pvscsi_io_write("PVSCSI_REG_OFFSET_INTR_MASK", val);
> +        s->reg_interrupt_enabled = val;
> +        pvscsi_update_irq_status(s);
> +        break;
> +
> +    case PVSCSI_REG_OFFSET_KICK_NON_RW_IO:
> +        trace_pvscsi_io_write("PVSCSI_REG_OFFSET_KICK_NON_RW_IO", val);
> +        pvscsi_process_io(s);
> +        break;
> +
> +    case PVSCSI_REG_OFFSET_KICK_RW_IO:
> +        trace_pvscsi_io_write("PVSCSI_REG_OFFSET_KICK_RW_IO", val);
> +        pvscsi_process_io(s);
> +        break;
> +
> +    case PVSCSI_REG_OFFSET_DEBUG:
> +        trace_pvscsi_io_write("PVSCSI_REG_OFFSET_DEBUG", val);
> +        break;
> +
> +    default:
> +        trace_pvscsi_io_write_unknown(addr, size, val);
> +        break;
> +    }
> +
> +}
> +
> +static uint64_t
> +pvscsi_io_read(void *opaque, hwaddr addr, unsigned size)
> +{
> +    PVSCSIState *s = opaque;
> +
> +    switch (addr) {
> +    case PVSCSI_REG_OFFSET_INTR_STATUS:
> +        trace_pvscsi_io_read("PVSCSI_REG_OFFSET_INTR_STATUS",
> +                             s->reg_interrupt_status);
> +        return s->reg_interrupt_status;
> +
> +    case PVSCSI_REG_OFFSET_INTR_MASK:
> +        trace_pvscsi_io_read("PVSCSI_REG_OFFSET_INTR_MASK",
> +                             s->reg_interrupt_status);
> +        return s->reg_interrupt_enabled;
> +
> +    case PVSCSI_REG_OFFSET_COMMAND_STATUS:
> +        trace_pvscsi_io_read("PVSCSI_REG_OFFSET_COMMAND_STATUS",
> +                             s->reg_interrupt_status);
> +        return s->reg_command_status;
> +
> +    default:
> +        trace_pvscsi_io_read_unknown(addr, size);
> +        return 0;
> +    }
> +}
> +
> +
> +static bool
> +pvscsi_init_msi(PVSCSIState *s)
> +{
> +    int res;
> +    PCIDevice *d = PCI_DEVICE(s);
> +
> +    res = msi_init(d, PVSCSI_MSI_OFFSET, PVSCSI_MSIX_NUM_VECTORS,
> +                   PVSCSI_USE_64BIT, PVSCSI_PER_VECTOR_MASK);
> +    if (res < 0) {
> +        trace_pvscsi_init_msi_fail(res);
> +        s->msi_used = false;
> +    } else {
> +        s->msi_used = true;
> +    }
> +
> +    return s->msi_used;
> +}
> +
> +static void
> +pvscsi_cleanup_msi(PVSCSIState *s)
> +{
> +    PCIDevice *d = PCI_DEVICE(s);
> +
> +    if (s->msi_used) {
> +        msi_uninit(d);
> +    }
> +}
> +
> +static const MemoryRegionOps pvscsi_ops = {
> +        .read = pvscsi_io_read,
> +        .write = pvscsi_io_write,
> +        .endianness = DEVICE_LITTLE_ENDIAN,
> +        .impl = {
> +                .min_access_size = 4,
> +                .max_access_size = 4,
> +        },
> +};
> +
> +static const struct SCSIBusInfo pvscsi_scsi_info = {
> +        .tcq = true,
> +        .max_target = PVSCSI_MAX_DEVS,
> +        .max_channel = 0,
> +        .max_lun = 0,
> +
> +        .get_sg_list = pvscsi_get_sg_list,
> +        .complete = pvscsi_command_complete,
> +        .cancel = pvscsi_request_cancelled,
> +        .hotplug = pvscsi_hotplug,
> +        .hot_unplug = pvscsi_hot_unplug,
> +};
> +
> +static int
> +pvscsi_init(PCIDevice *pci_dev)
> +{
> +    PVSCSIState *s = PVSCSI(pci_dev);
> +
> +    trace_pvscsi_state("init");
> +
> +    /* PCI subsystem ID */
> +    pci_dev->config[PCI_SUBSYSTEM_ID] = 0x00;
> +    pci_dev->config[PCI_SUBSYSTEM_ID + 1] = 0x10;
> +
> +    /* PCI latency timer = 255 */
> +    pci_dev->config[PCI_LATENCY_TIMER] = 0xff;
> +
> +    /* Interrupt pin A */
> +    pci_config_set_interrupt_pin(pci_dev->config, 1);
> +
> +    memory_region_init_io(&s->io_space, &pvscsi_ops, s,
> +                          "pvscsi-io", PVSCSI_MEM_SPACE_SIZE);
> +    pci_register_bar(pci_dev, 0, PCI_BASE_ADDRESS_SPACE_MEMORY, 
> &s->io_space);
> +
> +    pvscsi_init_msi(s);
> +
> +    s->completion_worker = qemu_bh_new(pvscsi_process_completion_queue, s);
> +    if (!s->completion_worker) {
> +        pvscsi_cleanup_msi(s);
> +        memory_region_destroy(&s->io_space);
> +        return -ENOMEM;
> +    }
> +
> +    scsi_bus_new(&s->bus, &pci_dev->qdev, &pvscsi_scsi_info);
> +    pvscsi_reset_state(s);
> +
> +    return 0;
> +}
> +
> +static void
> +pvscsi_uninit(PCIDevice *pci_dev)
> +{
> +    PVSCSIState *s = PVSCSI(pci_dev);
> +
> +    trace_pvscsi_state("uninit");
> +    qemu_bh_delete(s->completion_worker);
> +
> +    pvscsi_cleanup_msi(s);
> +
> +    memory_region_destroy(&s->io_space);
> +}
> +
> +static void
> +pvscsi_reset(DeviceState *dev)
> +{
> +    PCIDevice *d = PCI_DEVICE(dev);
> +    PVSCSIState *s = PVSCSI(d);
> +
> +    trace_pvscsi_state("reset");
> +    pvscsi_reset_adapter(s);
> +}
> +
> +static void
> +pvscsi_pre_save(void *opaque)
> +{
> +    PVSCSIState *s = (PVSCSIState *) opaque;
> +
> +    trace_pvscsi_state("presave");
> +
> +    assert(QTAILQ_EMPTY(&s->pending_queue));
> +    assert(QTAILQ_EMPTY(&s->completion_queue));

If you implement request serialization, you can still assert that the
completion queue is empty.  The pending queue will be reconstructed by
the load_request callbacks.

> +}
> +
> +static int
> +pvscsi_post_load(void *opaque, int version_id)
> +{
> +    trace_pvscsi_state("postload");
> +    return 0;
> +}
> +
> +static const VMStateDescription vmstate_pvscsi = {
> +    .name = TYPE_PVSCSI,
> +    .version_id = 0,
> +    .minimum_version_id = 0,
> +    .minimum_version_id_old = 0,
> +    .pre_save = pvscsi_pre_save,
> +    .post_load = pvscsi_post_load,
> +    .fields      = (VMStateField[]) {
> +        VMSTATE_PCI_DEVICE(parent_obj, PVSCSIState),
> +        VMSTATE_UINT8(msi_used, PVSCSIState),
> +        VMSTATE_UINT64(reg_interrupt_status, PVSCSIState),
> +        VMSTATE_UINT64(reg_interrupt_enabled, PVSCSIState),
> +        VMSTATE_UINT64(reg_command_status, PVSCSIState),
> +        VMSTATE_UINT64(curr_cmd, PVSCSIState),
> +        VMSTATE_UINT32(curr_cmd_data_cntr, PVSCSIState),
> +        VMSTATE_UINT32_ARRAY(curr_cmd_data, PVSCSIState,
> +                             ARRAY_SIZE(((PVSCSIState 
> *)NULL)->curr_cmd_data)),
> +        VMSTATE_UINT8(rings_info_valid, PVSCSIState),
> +        VMSTATE_UINT8(msg_ring_info_valid, PVSCSIState),
> +
> +        VMSTATE_UINT64(rings.rs_pa, PVSCSIState),
> +        VMSTATE_UINT32(rings.txr_len_mask, PVSCSIState),
> +        VMSTATE_UINT32(rings.rxr_len_mask, PVSCSIState),
> +        VMSTATE_UINT64_ARRAY(rings.req_ring_pages_pa, PVSCSIState,
> +                             PVSCSI_SETUP_RINGS_MAX_NUM_PAGES),
> +        VMSTATE_UINT64_ARRAY(rings.cmp_ring_pages_pa, PVSCSIState,
> +                             PVSCSI_SETUP_RINGS_MAX_NUM_PAGES),
> +        VMSTATE_UINT64(rings.consumed_ptr, PVSCSIState),
> +        VMSTATE_UINT64(rings.filled_cmp_ptr, PVSCSIState),
> +
> +        VMSTATE_END_OF_LIST()
> +    }
> +};
> +
> +static void
> +pvscsi_write_config(PCIDevice *pci, uint32_t addr, uint32_t val, int len)
> +{
> +    pci_default_write_config(pci, addr, val, len);
> +    msi_write_config(pci, addr, val, len);
> +}
> +
> +static void pvscsi_class_init(ObjectClass *klass, void *data)
> +{
> +    DeviceClass *dc = DEVICE_CLASS(klass);
> +    PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
> +
> +    k->init = pvscsi_init;
> +    k->exit = pvscsi_uninit;
> +    k->vendor_id = PCI_VENDOR_ID_VMWARE;
> +    k->device_id = PCI_DEVICE_ID_VMWARE_PVSCSI;
> +    k->class_id = PCI_CLASS_STORAGE_SCSI;
> +    k->subsystem_id = 0x1000;
> +    dc->reset = pvscsi_reset;
> +    dc->vmsd = &vmstate_pvscsi;
> +    k->config_write = pvscsi_write_config;
> +}
> +
> +static const TypeInfo pvscsi_info = {
> +    .name          = "pvscsi",
> +    .parent        = TYPE_PCI_DEVICE,
> +    .instance_size = sizeof(PVSCSIState),
> +    .class_init    = pvscsi_class_init,
> +};
> +
> +static void
> +pvscsi_register_types(void)
> +{
> +    type_register_static(&pvscsi_info);
> +
> +    trace_pvscsi_register();
> +}
> +
> +type_init(pvscsi_register_types);
> diff --git a/hw/vmw_pvscsi.h b/hw/vmw_pvscsi.h
> new file mode 100644
> index 0000000..17fcf66
> --- /dev/null
> +++ b/hw/vmw_pvscsi.h
> @@ -0,0 +1,434 @@
> +/*
> + * VMware PVSCSI header file
> + *
> + * Copyright (C) 2008-2009, VMware, Inc. All Rights Reserved.
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of the GNU General Public License as published by the
> + * Free Software Foundation; version 2 of the License and no later version.
> + *
> + * This program is distributed in the hope that it will be useful, but
> + * WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or
> + * NON INFRINGEMENT.  See the GNU General Public License for more
> + * details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
> + *
> + * Maintained by: Arvind Kumar <address@hidden>
> + *
> + */
> +
> +#ifndef VMW_PVSCSI_H
> +#define VMW_PVSCSI_H
> +
> +#define VMW_PAGE_SIZE  (4096)
> +#define VMW_PAGE_SHIFT (12)
> +
> +#define MASK(n)        ((1 << (n)) - 1)        /* make an n-bit mask */
> +
> +/*
> + * host adapter status/error codes
> + */
> +enum HostBusAdapterStatus {
> +   BTSTAT_SUCCESS       = 0x00,  /* CCB complete normally with no errors */
> +   BTSTAT_LINKED_COMMAND_COMPLETED           = 0x0a,
> +   BTSTAT_LINKED_COMMAND_COMPLETED_WITH_FLAG = 0x0b,
> +   BTSTAT_DATA_UNDERRUN = 0x0c,
> +   BTSTAT_SELTIMEO      = 0x11,  /* SCSI selection timeout */
> +   BTSTAT_DATARUN       = 0x12,  /* data overrun/underrun */
> +   BTSTAT_BUSFREE       = 0x13,  /* unexpected bus free */
> +   BTSTAT_INVPHASE      = 0x14,  /* invalid bus phase or sequence */
> +                                 /* requested by target           */
> +   BTSTAT_LUNMISMATCH   = 0x17,  /* linked CCB has different LUN  */
> +                                 /* from first CCB                */
> +   BTSTAT_SENSFAILED    = 0x1b,  /* auto request sense failed */
> +   BTSTAT_TAGREJECT     = 0x1c,  /* SCSI II tagged queueing message */
> +                                 /* rejected by target              */
> +   BTSTAT_BADMSG        = 0x1d,  /* unsupported message received by */
> +                                 /* the host adapter                */
> +   BTSTAT_HAHARDWARE    = 0x20,  /* host adapter hardware failed */
> +   BTSTAT_NORESPONSE    = 0x21,  /* target did not respond to SCSI ATN, */
> +                                 /* sent a SCSI RST                     */
> +   BTSTAT_SENTRST       = 0x22,  /* host adapter asserted a SCSI RST */
> +   BTSTAT_RECVRST       = 0x23,  /* other SCSI devices asserted a SCSI RST */
> +   BTSTAT_DISCONNECT    = 0x24,  /* target device reconnected improperly */
> +                                 /* (w/o tag)                            */
> +   BTSTAT_BUSRESET      = 0x25,  /* host adapter issued BUS device reset */
> +   BTSTAT_ABORTQUEUE    = 0x26,  /* abort queue generated */
> +   BTSTAT_HASOFTWARE    = 0x27,  /* host adapter software error */
> +   BTSTAT_HATIMEOUT     = 0x30,  /* host adapter hardware timeout error */
> +   BTSTAT_SCSIPARITY    = 0x34,  /* SCSI parity error detected */
> +};
> +
> +/*
> + * Register offsets.
> + *
> + * These registers are accessible both via i/o space and mm i/o.
> + */
> +
> +enum PVSCSIRegOffset {
> +    PVSCSI_REG_OFFSET_COMMAND        =    0x0,
> +    PVSCSI_REG_OFFSET_COMMAND_DATA   =    0x4,
> +    PVSCSI_REG_OFFSET_COMMAND_STATUS =    0x8,
> +    PVSCSI_REG_OFFSET_LAST_STS_0     =  0x100,
> +    PVSCSI_REG_OFFSET_LAST_STS_1     =  0x104,
> +    PVSCSI_REG_OFFSET_LAST_STS_2     =  0x108,
> +    PVSCSI_REG_OFFSET_LAST_STS_3     =  0x10c,
> +    PVSCSI_REG_OFFSET_INTR_STATUS    = 0x100c,
> +    PVSCSI_REG_OFFSET_INTR_MASK      = 0x2010,
> +    PVSCSI_REG_OFFSET_KICK_NON_RW_IO = 0x3014,
> +    PVSCSI_REG_OFFSET_DEBUG          = 0x3018,
> +    PVSCSI_REG_OFFSET_KICK_RW_IO     = 0x4018,
> +};
> +
> +/*
> + * Virtual h/w commands.
> + */
> +
> +enum PVSCSICommands {
> +    PVSCSI_CMD_FIRST             = 0, /* has to be first */
> +
> +    PVSCSI_CMD_ADAPTER_RESET     = 1,
> +    PVSCSI_CMD_ISSUE_SCSI        = 2,
> +    PVSCSI_CMD_SETUP_RINGS       = 3,
> +    PVSCSI_CMD_RESET_BUS         = 4,
> +    PVSCSI_CMD_RESET_DEVICE      = 5,
> +    PVSCSI_CMD_ABORT_CMD         = 6,
> +    PVSCSI_CMD_CONFIG            = 7,
> +    PVSCSI_CMD_SETUP_MSG_RING    = 8,
> +    PVSCSI_CMD_DEVICE_UNPLUG     = 9,
> +
> +    PVSCSI_CMD_LAST              = 10  /* has to be last */
> +};
> +
> +#define PVSCSI_COMMAND_PROCESSING_SUCCEEDED   (0)
> +#define PVSCSI_COMMAND_PROCESSING_FAILED     (-1)
> +#define PVSCSI_COMMAND_NOT_ENOUGH_DATA       (-2)
> +
> +/*
> + * Command descriptor for PVSCSI_CMD_RESET_DEVICE --
> + */
> +
> +struct PVSCSICmdDescResetDevice {
> +    uint32_t    target;
> +    uint8_t     lun[8];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSICmdDescResetDevice PVSCSICmdDescResetDevice;
> +
> +/*
> + * Command descriptor for PVSCSI_CMD_ABORT_CMD --
> + *
> + * - currently does not support specifying the LUN.
> + * - pad should be 0.
> + */
> +
> +struct PVSCSICmdDescAbortCmd {
> +    uint64_t    context;
> +    uint32_t    target;
> +    uint32_t    pad;
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSICmdDescAbortCmd PVSCSICmdDescAbortCmd;
> +
> +/*
> + * Command descriptor for PVSCSI_CMD_SETUP_RINGS --
> + *
> + * Notes:
> + * - reqRingNumPages and cmpRingNumPages need to be power of two.
> + * - reqRingNumPages and cmpRingNumPages need to be different from 0,
> + * - reqRingNumPages and cmpRingNumPages need to be inferior to
> + *   PVSCSI_SETUP_RINGS_MAX_NUM_PAGES.
> + */
> +
> +#define PVSCSI_SETUP_RINGS_MAX_NUM_PAGES        32
> +struct PVSCSICmdDescSetupRings {
> +    uint32_t    reqRingNumPages;
> +    uint32_t    cmpRingNumPages;
> +    uint64_t    ringsStatePPN;
> +    uint64_t    reqRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES];
> +    uint64_t    cmpRingPPNs[PVSCSI_SETUP_RINGS_MAX_NUM_PAGES];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSICmdDescSetupRings PVSCSICmdDescSetupRings;
> +
> +/*
> + * Command descriptor for PVSCSI_CMD_SETUP_MSG_RING --
> + *
> + * Notes:
> + * - this command was not supported in the initial revision of the h/w
> + *   interface. Before using it, you need to check that it is supported by
> + *   writing PVSCSI_CMD_SETUP_MSG_RING to the 'command' register, then
> + *   immediately after read the 'command status' register:
> + *       * a value of -1 means that the cmd is NOT supported,
> + *       * a value != -1 means that the cmd IS supported.
> + *   If it's supported the 'command status' register should return:
> + *      sizeof(PVSCSICmdDescSetupMsgRing) / sizeof(uint32_t).
> + * - this command should be issued _after_ the usual SETUP_RINGS so that the
> + *   RingsState page is already setup. If not, the command is a nop.
> + * - numPages needs to be a power of two,
> + * - numPages needs to be different from 0,
> + * - pad should be zero.
> + */
> +
> +#define PVSCSI_SETUP_MSG_RING_MAX_NUM_PAGES  16
> +
> +struct PVSCSICmdDescSetupMsgRing {
> +    uint32_t    numPages;
> +    uint32_t    pad;
> +    uint64_t    ringPPNs[PVSCSI_SETUP_MSG_RING_MAX_NUM_PAGES];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSICmdDescSetupMsgRing PVSCSICmdDescSetupMsgRing;
> +
> +enum PVSCSIMsgType {
> +    PVSCSI_MSG_DEV_ADDED          = 0,
> +    PVSCSI_MSG_DEV_REMOVED        = 1,
> +    PVSCSI_MSG_LAST               = 2,
> +};
> +
> +/*
> + * Msg descriptor.
> + *
> + * sizeof(struct PVSCSIRingMsgDesc) == 128.
> + *
> + * - type is of type enum PVSCSIMsgType.
> + * - the content of args depend on the type of event being delivered.
> + */
> +
> +struct PVSCSIRingMsgDesc {
> +    uint32_t    type;
> +    uint32_t    args[31];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSIRingMsgDesc PVSCSIRingMsgDesc;
> +
> +struct PVSCSIMsgDescDevStatusChanged {
> +    uint32_t    type;  /* PVSCSI_MSG_DEV _ADDED / _REMOVED */
> +    uint32_t    bus;
> +    uint32_t    target;
> +    uint8_t     lun[8];
> +    uint32_t    pad[27];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSIMsgDescDevStatusChanged PVSCSIMsgDescDevStatusChanged;
> +
> +/*
> + * Rings state.
> + *
> + * - the fields:
> + *    . msgProdIdx,
> + *    . msgConsIdx,
> + *    . msgNumEntriesLog2,
> + *   .. are only used once the SETUP_MSG_RING cmd has been issued.
> + * - 'pad' helps to ensure that the msg related fields are on their own
> + *   cache-line.
> + */
> +
> +struct PVSCSIRingsState {
> +    uint32_t    reqProdIdx;
> +    uint32_t    reqConsIdx;
> +    uint32_t    reqNumEntriesLog2;
> +
> +    uint32_t    cmpProdIdx;
> +    uint32_t    cmpConsIdx;
> +    uint32_t    cmpNumEntriesLog2;
> +
> +    uint8_t     pad[104];
> +
> +    uint32_t    msgProdIdx;
> +    uint32_t    msgConsIdx;
> +    uint32_t    msgNumEntriesLog2;
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSIRingsState PVSCSIRingsState;
> +
> +/*
> + * Request descriptor.
> + *
> + * sizeof(RingReqDesc) = 128
> + *
> + * - context: is a unique identifier of a command. It could normally be any
> + *   64bit value, however we currently store it in the serialNumber variable
> + *   of struct SCSI_Command, so we have the following restrictions due to the
> + *   way this field is handled in the vmkernel storage stack:
> + *    * this value can't be 0,
> + *    * the upper 32bit need to be 0 since serialNumber is as a uint32_t.
> + *   Currently tracked as PR 292060.
> + * - dataLen: contains the total number of bytes that need to be transferred.
> + * - dataAddr:
> + *   * if PVSCSI_FLAG_CMD_WITH_SG_LIST is set: dataAddr is the PA of the 
> first
> + *     s/g table segment, each s/g segment is entirely contained on a single
> + *     page of physical memory,
> + *   * if PVSCSI_FLAG_CMD_WITH_SG_LIST is NOT set, then dataAddr is the PA of
> + *     the buffer used for the DMA transfer,
> + * - flags:
> + *   * PVSCSI_FLAG_CMD_WITH_SG_LIST: see dataAddr above,
> + *   * PVSCSI_FLAG_CMD_DIR_NONE: no DMA involved,
> + *   * PVSCSI_FLAG_CMD_DIR_TOHOST: transfer from device to main memory,
> + *   * PVSCSI_FLAG_CMD_DIR_TODEVICE: transfer from main memory to device,
> + *   * PVSCSI_FLAG_CMD_OUT_OF_BAND_CDB: reserved to handle CDBs larger than
> + *     16bytes. To be specified.
> + * - vcpuHint: vcpuId of the processor that will be most likely waiting for 
> the
> + *   completion of the i/o. For guest OSes that use lowest priority message
> + *   delivery mode (such as windows), we use this "hint" to deliver the
> + *   completion action to the proper vcpu. For now, we can use the vcpuId of
> + *   the processor that initiated the i/o as a likely candidate for the vcpu
> + *   that will be waiting for the completion..
> + * - bus should be 0: we currently only support bus 0 for now.
> + * - unused should be zero'd.
> + */
> +
> +#define PVSCSI_FLAG_CMD_WITH_SG_LIST        (1 << 0)
> +#define PVSCSI_FLAG_CMD_OUT_OF_BAND_CDB     (1 << 1)
> +#define PVSCSI_FLAG_CMD_DIR_NONE            (1 << 2)
> +#define PVSCSI_FLAG_CMD_DIR_TOHOST          (1 << 3)
> +#define PVSCSI_FLAG_CMD_DIR_TODEVICE        (1 << 4)
> +
> +#define PVSCSI_KNOWN_FLAGS \
> +  (PVSCSI_FLAG_CMD_WITH_SG_LIST     | \
> +   PVSCSI_FLAG_CMD_OUT_OF_BAND_CDB  | \
> +   PVSCSI_FLAG_CMD_DIR_NONE         | \
> +   PVSCSI_FLAG_CMD_DIR_TOHOST       | \
> +   PVSCSI_FLAG_CMD_DIR_TODEVICE)
> +
> +struct PVSCSIRingReqDesc {
> +    uint64_t    context;
> +    uint64_t    dataAddr;
> +    uint64_t    dataLen;
> +    uint64_t    senseAddr;
> +    uint32_t    senseLen;
> +    uint32_t    flags;
> +    uint8_t     cdb[16];
> +    uint8_t     cdbLen;
> +    uint8_t     lun[8];
> +    uint8_t     tag;
> +    uint8_t     bus;
> +    uint8_t     target;
> +    uint8_t     vcpuHint;
> +    uint8_t     unused[59];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSIRingReqDesc PVSCSIRingReqDesc;
> +
> +/*
> + * Scatter-gather list management.
> + *
> + * As described above, when PVSCSI_FLAG_CMD_WITH_SG_LIST is set in the
> + * RingReqDesc.flags, then RingReqDesc.dataAddr is the PA of the first s/g
> + * table segment.
> + *
> + * - each segment of the s/g table contain a succession of struct
> + *   PVSCSISGElement.
> + * - each segment is entirely contained on a single physical page of memory.
> + * - a "chain" s/g element has the flag PVSCSI_SGE_FLAG_CHAIN_ELEMENT set in
> + *   PVSCSISGElement.flags and in this case:
> + *     * addr is the PA of the next s/g segment,
> + *     * length is undefined, assumed to be 0.
> + */
> +
> +struct PVSCSISGElement {
> +    uint64_t    addr;
> +    uint32_t    length;
> +    uint32_t    flags;
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSISGElement PVSCSISGElement;
> +
> +/*
> + * Completion descriptor.
> + *
> + * sizeof(RingCmpDesc) = 32
> + *
> + * - context: identifier of the command. The same thing that was specified
> + *   under "context" as part of struct RingReqDesc at initiation time,
> + * - dataLen: number of bytes transferred for the actual i/o operation,
> + * - senseLen: number of bytes written into the sense buffer,
> + * - hostStatus: adapter status,
> + * - scsiStatus: device status,
> + * - pad should be zero.
> + */
> +
> +struct PVSCSIRingCmpDesc {
> +    uint64_t    context;
> +    uint64_t    dataLen;
> +    uint32_t    senseLen;
> +    uint16_t    hostStatus;
> +    uint16_t    scsiStatus;
> +    uint32_t    pad[2];
> +} QEMU_PACKED;
> +
> +typedef struct PVSCSIRingCmpDesc PVSCSIRingCmpDesc;
> +
> +/*
> + * Interrupt status / IRQ bits.
> + */
> +
> +#define PVSCSI_INTR_CMPL_0                 (1 << 0)
> +#define PVSCSI_INTR_CMPL_1                 (1 << 1)
> +#define PVSCSI_INTR_CMPL_MASK              MASK(2)
> +
> +#define PVSCSI_INTR_MSG_0                  (1 << 2)
> +#define PVSCSI_INTR_MSG_1                  (1 << 3)
> +#define PVSCSI_INTR_MSG_MASK               (MASK(2) << 2)
> +
> +#define PVSCSI_INTR_ALL_SUPPORTED          MASK(4)
> +
> +/*
> + * Number of MSI-X vectors supported.
> + */
> +#define PVSCSI_MAX_INTRS        24
> +
> +/*
> + * Enumeration of supported MSI-X vectors
> + */
> +#define PVSCSI_VECTOR_COMPLETION   0
> +
> +/*
> + * Misc constants for the rings.
> + */
> +
> +#define PVSCSI_MAX_NUM_PAGES_REQ_RING   PVSCSI_SETUP_RINGS_MAX_NUM_PAGES
> +#define PVSCSI_MAX_NUM_PAGES_CMP_RING   PVSCSI_SETUP_RINGS_MAX_NUM_PAGES
> +#define PVSCSI_MAX_NUM_PAGES_MSG_RING   PVSCSI_SETUP_MSG_RING_MAX_NUM_PAGES
> +
> +#define PVSCSI_MAX_NUM_REQ_ENTRIES_PER_PAGE \
> +                (VMW_PAGE_SIZE / sizeof(struct PVSCSIRingReqDesc))
> +
> +#define PVSCSI_MAX_NUM_CMP_ENTRIES_PER_PAGE \
> +                (VMW_PAGE_SIZE / sizeof(PVSCSIRingCmpDesc))
> +
> +#define PVSCSI_MAX_NUM_MSG_ENTRIES_PER_PAGE \
> +                (VMW_PAGE_SIZE / sizeof(PVSCSIRingMsgDesc))
> +
> +#define PVSCSI_MAX_REQ_QUEUE_DEPTH \
> +    (PVSCSI_MAX_NUM_PAGES_REQ_RING * PVSCSI_MAX_NUM_REQ_ENTRIES_PER_PAGE)
> +
> +#define PVSCSI_MEM_SPACE_COMMAND_NUM_PAGES     1
> +#define PVSCSI_MEM_SPACE_INTR_STATUS_NUM_PAGES 1
> +#define PVSCSI_MEM_SPACE_MISC_NUM_PAGES        2
> +#define PVSCSI_MEM_SPACE_KICK_IO_NUM_PAGES     2
> +#define PVSCSI_MEM_SPACE_MSIX_NUM_PAGES        2
> +
> +enum PVSCSIMemSpace {
> +    PVSCSI_MEM_SPACE_COMMAND_PAGE       = 0,
> +    PVSCSI_MEM_SPACE_INTR_STATUS_PAGE   = 1,
> +    PVSCSI_MEM_SPACE_MISC_PAGE          = 2,
> +    PVSCSI_MEM_SPACE_KICK_IO_PAGE       = 4,
> +    PVSCSI_MEM_SPACE_MSIX_TABLE_PAGE    = 6,
> +    PVSCSI_MEM_SPACE_MSIX_PBA_PAGE      = 7,
> +};
> +
> +#define PVSCSI_MEM_SPACE_NUM_PAGES \
> +    (PVSCSI_MEM_SPACE_COMMAND_NUM_PAGES +       \
> +     PVSCSI_MEM_SPACE_INTR_STATUS_NUM_PAGES +   \
> +     PVSCSI_MEM_SPACE_MISC_NUM_PAGES +          \
> +     PVSCSI_MEM_SPACE_KICK_IO_NUM_PAGES +       \
> +     PVSCSI_MEM_SPACE_MSIX_NUM_PAGES)
> +
> +#define PVSCSI_MEM_SPACE_SIZE    (PVSCSI_MEM_SPACE_NUM_PAGES * VMW_PAGE_SIZE)
> +
> +#endif /* VMW_PVSCSI_H */
> diff --git a/trace-events b/trace-events
> index 412f7e4..66037a1 100644
> --- a/trace-events
> +++ b/trace-events
> @@ -761,6 +761,42 @@ pc87312_info_ide(uint32_t base) "base 0x%x"
>  pc87312_info_parallel(uint32_t base, uint32_t irq) "base 0x%x, irq %u"
>  pc87312_info_serial(int n, uint32_t base, uint32_t irq) "id=%d, base 0x%x, 
> irq %u"
>  
> +# hw/pvscsi.c
> +pvscsi_rings_mgr_init_data(uint32_t txr_len_log2, uint32_t rxr_len_log2) 
> "TX/RX rings logarithms set to %d/%d"
> +pvscsi_rings_mgr_init_msg(uint32_t len_log2) "MSG ring logarithm set to %d"
> +pvscsi_rings_mgr_flush_cmp_ring(uint64_t filled_cmp_ptr) "new production 
> counter of completion ring is 0x%"PRIx64""
> +pvscsi_rings_mgr_flush_msg_ring(uint64_t filled_cmp_ptr) "new production 
> counter of message ring is 0x%"PRIx64""
> +pvscsi_update_irq_level(bool raise, uint64_t mask, uint64_t status) 
> "interrupt level set to %d (MASK: 0x%"PRIx64", STATUS: 0x%"PRIx64")"
> +pvscsi_update_irq_msi(void) "sending MSI notification"
> +pvscsi_cmp_ring_put(unsigned long addr) "got completion descriptor 0x%lx"
> +pvscsi_msg_ring_put(unsigned long addr) "got message descriptor 0x%lx"
> +pvscsi_complete_request(uint64_t context, uint64_t len, uint8_t sense_key) 
> "completion: ctx: 0x%"PRIx64", len: 0x%"PRIx64", sense key: %u"
> +pvscsi_get_sg_list(int nsg, size_t size) "get SG list: depth: %u, size: %lu"
> +pvscsi_get_next_sg_elem(uint32_t flags) "unknown flags in SG element (val: 
> 0x%x)"
> +pvscsi_command_complete_not_found(uint32_t tag) "can't find request for tag 
> 0x%x"
> +pvscsi_command_complete_data_run(void) "not all data required for command 
> transferred"
> +pvscsi_command_complete_sense_len(int len) "sense information length is %d 
> bytes"
> +pvscsi_convert_sglist(uint64_t context, unsigned long addr, uint32_t resid) 
> "element: ctx: 0x%"PRIx64" addr: 0x%lx, len: %ul"
> +pvscsi_process_req_descr(uint8_t cmd, uint64_t ctx) "SCSI cmd 0x%x, ctx: 
> 0x%"PRIx64""
> +pvscsi_process_req_descr_unknown_device(void) "command directed to unknown 
> device rejected"
> +pvscsi_process_req_descr_invalid_dir(void) "command with invalid transfer 
> direction rejected"
> +pvscsi_process_io(unsigned long addr) "got descriptor 0x%lx"
> +pvscsi_on_cmd_noimpl(const char* cmd) "unimplemented command %s ignored"
> +pvscsi_on_cmd_reset_dev(uint32_t tgt, int lun, void* dev) 
> "PVSCSI_CMD_RESET_DEVICE[target %u lun %d (dev 0x%p)]"
> +pvscsi_on_cmd_arrived(const char* cmd) "command %s arrived"
> +pvscsi_on_cmd_abort(uint64_t ctx, uint32_t tgt) "command 
> PVSCSI_CMD_ABORT_CMD for ctx 0x%"PRIx64", target %u"
> +pvscsi_on_cmd_unknown(uint64_t cmd_id) "unknown command %"PRIx64""
> +pvscsi_on_cmd_unknown_data(uint32_t data) "data for unknown command 0x:%x"
> +pvscsi_io_write(const char* cmd, uint64_t val) "%s write: %"PRIx64""
> +pvscsi_io_write_unknown(unsigned long addr, unsigned sz, uint64_t val) 
> "unknown write address: 0x%lx size: %u bytes value: 0x%"PRIx64""
> +pvscsi_io_read(const char* cmd, uint64_t status) "%s read: 0x%"PRIx64""
> +pvscsi_io_read_unknown(unsigned long addr, unsigned sz) "unknown read 
> address: 0x%lx size: %u bytes"
> +pvscsi_init_msi_fail(int res) "failed to initialize MSI, error %d"
> +pvscsi_state(const char* state) "starting %s ..."
> +pvscsi_register(void) "PVSCSI QEMU device emulation registered"
> +pvscsi_tx_rings_ppn(const char* label, uint64_t ppn) "%s page: %"PRIx64""
> +pvscsi_tx_rings_num_pages(const char* label, uint32_t num) "Number of %s 
> pages: %u"
> +
>  # xen-all.c
>  xen_ram_alloc(unsigned long ram_addr, unsigned long size) "requested: %#lx, 
> size %#lx"
>  xen_client_set_memory(uint64_t start_addr, unsigned long size, bool 
> log_dirty) "%#"PRIx64" size %#lx, log_dirty %i"
> 




reply via email to

[Prev in Thread] Current Thread [Next in Thread]