[Qemu-devel] [PATCH 09/15] xen: Initialize event channels and io rings

[stefano.stabellini@eu.citrix.com]

From: Anthony PERARD <anthony.perard@citrix.com>

Open and bind event channels; map ioreq and buffered ioreq rings.

Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
---
 hw/xen_machine_fv.c   |   25 ++++
 target-xen/cpu.h      |    1 +
 target-xen/helper.c   |  362 +++++++++++++++++++++++++++++++++++++++++++++++++
 target-xen/qemu-xen.h |    2 +
 4 files changed, 390 insertions(+), 0 deletions(-)

diff --git a/hw/xen_machine_fv.c b/hw/xen_machine_fv.c
index a6e778a..b1bc88d 100644
--- a/hw/xen_machine_fv.c
+++ b/hw/xen_machine_fv.c
@@ -22,6 +22,9 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  * THE SOFTWARE.
  */
+#include "config.h"
+
+#include <sys/mman.h>
 
 #include "hw.h"
 #include "pc.h"
@@ -71,12 +74,34 @@ static void xen_init_fv(ram_addr_t ram_size,
 
     CPUState *env;
 
+    unsigned long ioreq_pfn;
+    extern void *shared_page;
+    extern void *buffered_io_page;
+
     /* Initialize backend core & drivers */
     if (xen_dm_init() != 0) {
         fprintf(stderr, "%s: xen backend core setup failed\n", __FUNCTION__);
         exit(1);
     }
 
+    xc_get_hvm_param(xen_xc, xen_domid, HVM_PARAM_IOREQ_PFN, &ioreq_pfn);
+    fprintf(stderr, "shared page at pfn %lx\n", ioreq_pfn);
+    shared_page = xc_map_foreign_range(xen_xc, xen_domid, XC_PAGE_SIZE,
+            PROT_READ|PROT_WRITE, ioreq_pfn);
+    if (shared_page == NULL) {
+        fprintf(stderr, "map shared IO page returned error %d handle=%p\n", errno, xen_xc);
+        exit(-1);
+    }
+
+    xc_get_hvm_param(xen_xc, xen_domid, HVM_PARAM_BUFIOREQ_PFN, &ioreq_pfn);
+    fprintf(stderr, "buffered io page at pfn %lx\n", ioreq_pfn);
+    buffered_io_page = xc_map_foreign_range(xen_xc, xen_domid, XC_PAGE_SIZE,
+                                            PROT_READ|PROT_WRITE, ioreq_pfn);
+    if (buffered_io_page == NULL) {
+        fprintf(logfile, "map buffered IO page returned error %d\n", errno);
+        exit(-1);
+    }
+
     /* Initialize a dummy CPU */
     if (cpu_model == NULL) {
 #ifdef TARGET_X86_64
diff --git a/target-xen/cpu.h b/target-xen/cpu.h
index 5a45d1c..573241f 100644
--- a/target-xen/cpu.h
+++ b/target-xen/cpu.h
@@ -72,6 +72,7 @@ typedef struct CPUXenState {
 
 CPUXenState *cpu_xen_init(const char *cpu_model);
 int cpu_xen_exec(CPUXenState *s);
+void cpu_xen_close(CPUXenState *s);
 
 int cpu_get_pic_interrupt(CPUXenState *s);
 void cpu_set_ferr(CPUX86State *s);
diff --git a/target-xen/helper.c b/target-xen/helper.c
index 8cb7771..4571ac0 100644
--- a/target-xen/helper.c
+++ b/target-xen/helper.c
@@ -18,25 +18,77 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 
+#include "config.h"
+
+#include <inttypes.h>
+
+#include <xenctrl.h>
+#include <xen/hvm/ioreq.h>
+
 #include "cpu.h"
 #include "qemu-xen.h"
 #include "xenstore.h"
+#include "hw/xen_backend.h"
+
+long time_offset = 0;
+
+shared_iopage_t *shared_page = NULL;
+
+#define BUFFER_IO_MAX_DELAY  100
+buffered_iopage_t *buffered_io_page = NULL;
+QEMUTimer *buffered_io_timer;
+
+/* the evtchn fd for polling */
+int xce_handle = -1;
+
+/* which vcpu we are serving */
+int send_vcpu = 0;
+
+/* the evtchn port for polling the notification, */
+evtchn_port_t *ioreq_local_port;
 
 CPUXenState *cpu_xen_init(const char *cpu_model)
 {
     CPUXenState *env = NULL;
     static int inited;
+    int i, rc;
 
     env = qemu_mallocz(sizeof(CPUXenState));
     if (!env)
         return NULL;
     cpu_exec_init(env);
 
+    /* There is no shared_page for PV, we're done now */
+    if (shared_page == NULL)
+        return env;
+
+    ioreq_local_port =
+        (evtchn_port_t *)qemu_mallocz(smp_cpus * sizeof(evtchn_port_t));
+    if (!ioreq_local_port)
+        return NULL;
+
     /* init various static tables */
     if (!inited) {
         inited = 1;
 
         cpu_single_env = env;
+
+        xce_handle = xc_evtchn_open();
+        if (xce_handle == -1) {
+            perror("open");
+            return NULL;
+        }
+
+        /* FIXME: how about if we overflow the page here? */
+        for (i = 0; i < smp_cpus; i++) {
+            rc = xc_evtchn_bind_interdomain(
+                    xce_handle, xen_domid, shared_page->vcpu_ioreq[i].vp_eport);
+            if (rc == -1) {
+                fprintf(stderr, "bind interdomain ioctl error %d\n", errno);
+                return NULL;
+            }
+            ioreq_local_port[i] = rc;
+        }
     }
 
     return env;
@@ -70,7 +122,317 @@ target_phys_addr_t cpu_get_phys_page_debug(CPUState *env, target_ulong addr)
     return addr;
 }
 
+// get the ioreq packets from share mem
+static ioreq_t *__cpu_get_ioreq(int vcpu)
+{
+    ioreq_t *req = &shared_page->vcpu_ioreq[vcpu];
+
+    if (req->state != STATE_IOREQ_READY) {
+        fprintf(stderr, "I/O request not ready: "
+                "%x, ptr: %x, port: %"PRIx64", "
+                "data: %"PRIx64", count: %u, size: %u\n",
+                req->state, req->data_is_ptr, req->addr,
+                req->data, req->count, req->size);
+        return NULL;
+    }
+
+    xen_rmb(); /* see IOREQ_READY /then/ read contents of ioreq */
+
+    req->state = STATE_IOREQ_INPROCESS;
+    return req;
+}
+
+// use poll to get the port notification
+// ioreq_vec--out,the
+// retval--the number of ioreq packet
+static ioreq_t *cpu_get_ioreq(void)
+{
+    int i;
+    evtchn_port_t port;
+
+    port = xc_evtchn_pending(xce_handle);
+    if (port != -1) {
+        for ( i = 0; i < smp_cpus; i++ )
+            if ( ioreq_local_port[i] == port )
+                break;
+
+        if ( i == smp_cpus ) {
+            fprintf(stderr, "Fatal error while trying to get io event!\n");
+            exit(1);
+        }
+
+        // unmask the wanted port again
+        xc_evtchn_unmask(xce_handle, port);
+
+        // get the io packet from shared memory
+        send_vcpu = i;
+        return __cpu_get_ioreq(i);
+    }
+
+    // read error or read nothing
+    return NULL;
+}
+
+static unsigned long do_inp(CPUState *env, unsigned long addr,
+        unsigned long size)
+{
+    switch(size) {
+        case 1:
+            return cpu_inb(addr);
+        case 2:
+            return cpu_inw(addr);
+        case 4:
+            return cpu_inl(addr);
+        default:
+            fprintf(stderr, "inp: bad size: %lx %lx\n", addr, size);
+            exit(-1);
+    }
+}
+
+static void do_outp(CPUState *env, unsigned long addr,
+        unsigned long size, unsigned long val)
+{
+    switch(size) {
+        case 1:
+            return cpu_outb(addr, val);
+        case 2:
+            return cpu_outw(addr, val);
+        case 4:
+            return cpu_outl(addr, val);
+        default:
+            fprintf(stderr, "outp: bad size: %lx %lx\n", addr, size);
+            exit(-1);
+    }
+}
+
+static inline void read_physical(uint64_t addr, unsigned long size, void *val)
+{
+    return cpu_physical_memory_rw((target_phys_addr_t)addr, val, size, 0);
+}
+
+static inline void write_physical(uint64_t addr, unsigned long size, void *val)
+{
+    return cpu_physical_memory_rw((target_phys_addr_t)addr, val, size, 1);
+}
+
+static void cpu_ioreq_pio(CPUState *env, ioreq_t *req)
+{
+    int i, sign;
+
+    sign = req->df ? -1 : 1;
+
+    if (req->dir == IOREQ_READ) {
+        if (!req->data_is_ptr) {
+            req->data = do_inp(env, req->addr, req->size);
+        } else {
+            unsigned long tmp;
+
+            for (i = 0; i < req->count; i++) {
+                tmp = do_inp(env, req->addr, req->size);
+                write_physical((target_phys_addr_t) req->data
+                        + (sign * i * req->size),
+                        req->size, &tmp);
+            }
+        }
+    } else if (req->dir == IOREQ_WRITE) {
+        if (!req->data_is_ptr) {
+            do_outp(env, req->addr, req->size, req->data);
+        } else {
+            for (i = 0; i < req->count; i++) {
+                unsigned long tmp = 0;
+
+                read_physical((target_phys_addr_t) req->data
+                        + (sign * i * req->size),
+                        req->size, &tmp);
+                do_outp(env, req->addr, req->size, tmp);
+            }
+        }
+    }
+}
+
+static void cpu_ioreq_move(CPUState *env, ioreq_t *req)
+{
+    int i, sign;
+
+    sign = req->df ? -1 : 1;
+
+    if (!req->data_is_ptr) {
+        if (req->dir == IOREQ_READ) {
+            for (i = 0; i < req->count; i++) {
+                read_physical(req->addr
+                        + (sign * i * req->size),
+                        req->size, &req->data);
+            }
+        } else if (req->dir == IOREQ_WRITE) {
+            for (i = 0; i < req->count; i++) {
+                write_physical(req->addr
+                        + (sign * i * req->size),
+                        req->size, &req->data);
+            }
+        }
+    } else {
+        target_ulong tmp;
+
+        if (req->dir == IOREQ_READ) {
+            for (i = 0; i < req->count; i++) {
+                read_physical(req->addr
+                        + (sign * i * req->size),
+                        req->size, &tmp);
+                write_physical((target_phys_addr_t )req->data
+                        + (sign * i * req->size),
+                        req->size, &tmp);
+            }
+        } else if (req->dir == IOREQ_WRITE) {
+            for (i = 0; i < req->count; i++) {
+                read_physical((target_phys_addr_t) req->data
+                        + (sign * i * req->size),
+                        req->size, &tmp);
+                write_physical(req->addr
+                        + (sign * i * req->size),
+                        req->size, &tmp);
+            }
+        }
+    }
+}
+
+static void cpu_ioreq_timeoffset(CPUState *env, ioreq_t *req)
+{
+    char b[64];
+
+    time_offset += (unsigned long)req->data;
+
+    fprintf(stderr, "Time offset set %ld, added offset %"PRId64"\n",
+            time_offset, req->data);
+    sprintf(b, "%ld", time_offset);
+    xenstore_vm_write(xen_domid, "rtc/timeoffset", b);
+}
+
+static void __handle_ioreq(CPUState *env, ioreq_t *req)
+{
+    if (!req->data_is_ptr && (req->dir == IOREQ_WRITE) &&
+            (req->size < sizeof(target_ulong)))
+        req->data &= ((target_ulong)1 << (8 * req->size)) - 1;
+
+    switch (req->type) {
+        case IOREQ_TYPE_PIO:
+            cpu_ioreq_pio(env, req);
+            break;
+        case IOREQ_TYPE_COPY:
+            cpu_ioreq_move(env, req);
+            break;
+        case IOREQ_TYPE_TIMEOFFSET:
+            cpu_ioreq_timeoffset(env, req);
+            break;
+        case IOREQ_TYPE_INVALIDATE:
+            qemu_invalidate_map_cache();
+            break;
+        default:
+            hw_error("Invalid ioreq type 0x%x\n", req->type);
+    }
+}
+
+static void __handle_buffered_iopage(CPUState *env)
+{
+    buf_ioreq_t *buf_req = NULL;
+    ioreq_t req;
+    int qw;
+
+    if (!buffered_io_page)
+        return;
+
+    while (buffered_io_page->read_pointer !=
+            buffered_io_page->write_pointer) {
+        buf_req = &buffered_io_page->buf_ioreq[
+            buffered_io_page->read_pointer % IOREQ_BUFFER_SLOT_NUM];
+        req.size = 1UL << buf_req->size;
+        req.count = 1;
+        req.addr = buf_req->addr;
+        req.data = buf_req->data;
+        req.state = STATE_IOREQ_READY;
+        req.dir = buf_req->dir;
+        req.df = 1;
+        req.type = buf_req->type;
+        req.data_is_ptr = 0;
+        qw = (req.size == 8);
+        if (qw) {
+            buf_req = &buffered_io_page->buf_ioreq[
+                (buffered_io_page->read_pointer+1) % IOREQ_BUFFER_SLOT_NUM];
+            req.data |= ((uint64_t)buf_req->data) << 32;
+        }
+
+        __handle_ioreq(env, &req);
+
+        xen_mb();
+        buffered_io_page->read_pointer += qw ? 2 : 1;
+    }
+}
+
+static void handle_buffered_io(void *opaque)
+{
+    CPUState *env = opaque;
+
+    __handle_buffered_iopage(env);
+    qemu_mod_timer(buffered_io_timer, BUFFER_IO_MAX_DELAY +
+                   qemu_get_clock(rt_clock));
+}
+
+static void cpu_handle_ioreq(void *opaque)
+{
+    CPUState *env = opaque;
+    ioreq_t *req = cpu_get_ioreq();
+
+    __handle_buffered_iopage(env);
+    if (req) {
+        __handle_ioreq(env, req);
+
+        if (req->state != STATE_IOREQ_INPROCESS) {
+            fprintf(stderr, "Badness in I/O request ... not in service?!: "
+                    "%x, ptr: %x, port: %"PRIx64", "
+                    "data: %"PRIx64", count: %u, size: %u\n",
+                    req->state, req->data_is_ptr, req->addr,
+                    req->data, req->count, req->size);
+            destroy_hvm_domain();
+            return;
+        }
+
+        xen_wmb(); /* Update ioreq contents /then/ update state. */
+
+        req->state = STATE_IORESP_READY;
+        xc_evtchn_notify(xce_handle, ioreq_local_port[send_vcpu]);
+    }
+}
+
 void xen_main_loop_prepare(void)
 {
+    CPUState *env = cpu_single_env;
+
+    int evtchn_fd = xce_handle == -1 ? -1 : xc_evtchn_fd(xce_handle);
+
+    buffered_io_timer = qemu_new_timer(rt_clock, handle_buffered_io,
+                                       cpu_single_env);
+    qemu_mod_timer(buffered_io_timer, qemu_get_clock(rt_clock));
+
+    if (evtchn_fd != -1)
+        qemu_set_fd_handler(evtchn_fd, cpu_handle_ioreq, NULL, env);
+
     xenstore_record_dm_state("running");
 }
+
+void destroy_hvm_domain(void)
+{
+    xc_interface *xcHandle;
+    int sts;
+
+    xcHandle = xc_interface_open(NULL, NULL, 0);
+    if (xcHandle < 0)
+        fprintf(stderr, "Cannot acquire xenctrl handle\n");
+    else {
+        sts = xc_domain_shutdown(xcHandle, xen_domid, SHUTDOWN_poweroff);
+        if (sts != 0)
+            fprintf(stderr, "? xc_domain_shutdown failed to issue poweroff, "
+                    "sts %d, errno %d\n", sts, errno);
+        else
+            fprintf(stderr, "Issued domain %d poweroff\n", xen_domid);
+        xc_interface_close(xcHandle);
+    }
+}
diff --git a/target-xen/qemu-xen.h b/target-xen/qemu-xen.h
index 091ae07..79a4638 100644
--- a/target-xen/qemu-xen.h
+++ b/target-xen/qemu-xen.h
@@ -22,12 +22,14 @@ void     qemu_invalidate_map_cache(void);
 
 /* target-xen/exec-dm.c */
 
+void destroy_hvm_domain(void);
 int cpu_register_io_memory_fixed(int io_index,
                            CPUReadMemoryFunc * const *mem_read,
                            CPUWriteMemoryFunc * const *mem_write,
                            void *opaque);
 
 /* target-xen/helper.c */
+extern int xce_handle;
 void xen_main_loop_prepare(void);
 
 #endif /*QEMU_XEN_H*/
-- 
1.7.0.4