[PATCH v5 13/50] multi-process: introduce proxy object

[Jagannathan Raman <jag.raman@oracle.com>]

From: Elena Ufimtseva <elena.ufimtseva@oracle.com>

Defines a PCI Device proxy object as a parent of TYPE_PCI_DEVICE.
PCI Proxy Object is responsible for registering PCI BARs,i
MemoryRegionOps to handle access to the BARs and forwarding those
to the remote device.
PCI Proxy object intercepts config space reads and writes. In case
of pci config write it forwards it to the remote device using
communication channel set by proxy-link object.

Signed-off-by: Elena Ufimtseva <elena.ufimtseva@oracle.com>
Signed-off-by: Jagannathan Raman <jag.raman@oracle.com>
Signed-off-by: John G Johnson <john.g.johnson@oracle.com>
---
 v4 -> v5:
  - Switched to execv

 hw/Makefile.objs              |   2 +
 hw/proxy/Makefile.objs        |   1 +
 hw/proxy/qemu-proxy.c         | 226 ++++++++++++++++++++++++++++++++++++++++++
 include/hw/proxy/qemu-proxy.h |  54 ++++++++++
 include/io/mpqemu-link.h      |   1 +
 remote/remote-main.c          |  28 ++++++
 6 files changed, 312 insertions(+)
 create mode 100644 hw/proxy/Makefile.objs
 create mode 100644 hw/proxy/qemu-proxy.c
 create mode 100644 include/hw/proxy/qemu-proxy.h

diff --git a/hw/Makefile.objs b/hw/Makefile.objs
index af9235b..7b489b1 100644
--- a/hw/Makefile.objs
+++ b/hw/Makefile.objs
@@ -45,6 +45,8 @@ endif
 common-obj-y += $(devices-dirs-y)
 obj-y += $(devices-dirs-y)
 
+common-obj-$(CONFIG_MPQEMU) += proxy/
+
 remote-pci-obj-$(CONFIG_MPQEMU) += core/
 remote-pci-obj-$(CONFIG_MPQEMU) += block/
 remote-pci-obj-$(CONFIG_MPQEMU) += pci/
diff --git a/hw/proxy/Makefile.objs b/hw/proxy/Makefile.objs
new file mode 100644
index 0000000..eb81624
--- /dev/null
+++ b/hw/proxy/Makefile.objs
@@ -0,0 +1 @@
+common-obj-$(CONFIG_MPQEMU) += qemu-proxy.o
diff --git a/hw/proxy/qemu-proxy.c b/hw/proxy/qemu-proxy.c
new file mode 100644
index 0000000..828bbd7
--- /dev/null
+++ b/hw/proxy/qemu-proxy.c
@@ -0,0 +1,226 @@
+/*
+ * Copyright © 2018, 2020 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "io/mpqemu-link.h"
+#include "exec/memory.h"
+#include "exec/cpu-common.h"
+#include "exec/address-spaces.h"
+#include "hw/pci/pci.h"
+#include "qapi/qmp/qjson.h"
+#include "qapi/qmp/qstring.h"
+#include "hw/proxy/qemu-proxy.h"
+
+static void pci_proxy_dev_realize(PCIDevice *dev, Error **errp);
+
+static int remote_spawn(PCIProxyDev *pdev, const char *opts,
+                        const char *exec_name, Error **errp)
+{
+    char *args[3];
+    pid_t rpid;
+    int fd[2] = {-1, -1};
+    Error *local_error = NULL;
+    int rc = -EINVAL;
+
+    if (pdev->managed) {
+        /* Child is forked by external program (such as libvirt). */
+        error_setg(errp, "Remote processed is managed and launched by external program");
+        return -1;
+    }
+
+    if (!exec_name) {
+        error_setg(errp, "The remote exec name is NULL.");
+        return rc;
+    }
+
+    if (socketpair(AF_UNIX, SOCK_STREAM, 0, fd)) {
+        error_setg(errp, "Unable to create unix socket.");
+        return -1;
+    }
+    /* TODO: Restrict the forked process' permissions and capabilities. */
+    rpid = qemu_fork(&local_error);
+
+    if (rpid == -1) {
+        error_setg(errp, "Unable to spawn emulation program.");
+        close(fd[0]);
+        close(fd[1]);
+        return -1;
+    }
+
+    if (rpid == 0) {
+        close(fd[0]);
+
+        args[0] = g_strdup(exec_name);
+        args[1] = g_strdup_printf("%d", fd[1]);
+        args[2] = NULL;
+        execvp(args[0], (char *const *)args);
+        exit(1);
+    }
+    pdev->remote_pid = rpid;
+
+    close(fd[1]);
+
+    return 0;
+}
+
+static int get_proxy_sock(PCIDevice *dev)
+{
+    PCIProxyDev *pdev;
+
+    pdev = PCI_PROXY_DEV(dev);
+
+    return pdev->socket;
+}
+
+static void set_proxy_sock(PCIDevice *dev, int socket)
+{
+    PCIProxyDev *pdev;
+
+    pdev = PCI_PROXY_DEV(dev);
+
+    pdev->socket = socket;
+    pdev->managed = true;
+}
+
+static int config_op_send(PCIProxyDev *dev, uint32_t addr, uint32_t *val, int l,
+                          unsigned int op)
+{
+    MPQemuMsg msg;
+    struct conf_data_msg conf_data;
+    int wait;
+
+    memset(&msg, 0, sizeof(MPQemuMsg));
+    conf_data.addr = addr;
+    conf_data.val = (op == PCI_CONFIG_WRITE) ? *val : 0;
+    conf_data.l = l;
+
+    msg.data2 = (uint8_t *)&conf_data;
+    if (!msg.data2) {
+        return -ENOMEM;
+    }
+
+    msg.size = sizeof(conf_data);
+    msg.cmd = op;
+    msg.bytestream = 1;
+
+    if (op == PCI_CONFIG_WRITE) {
+        msg.num_fds = 0;
+    } else {
+        /* TODO: Dont create fd each time for send. */
+        wait = GET_REMOTE_WAIT;
+        msg.num_fds = 1;
+        msg.fds[0] = wait;
+    }
+
+    mpqemu_msg_send(&msg, dev->mpqemu_link->com);
+
+    if (op == PCI_CONFIG_READ) {
+        *val = (uint32_t)wait_for_remote(wait);
+        PUT_REMOTE_WAIT(wait);
+    }
+
+    return 0;
+}
+
+static uint32_t pci_proxy_read_config(PCIDevice *d, uint32_t addr, int len)
+{
+    uint32_t val;
+
+    (void)pci_default_read_config(d, addr, len);
+
+    config_op_send(PCI_PROXY_DEV(d), addr, &val, len, PCI_CONFIG_READ);
+
+    return val;
+}
+
+static void pci_proxy_write_config(PCIDevice *d, uint32_t addr, uint32_t val,
+                                   int l)
+{
+    pci_default_write_config(d, addr, val, l);
+
+    config_op_send(PCI_PROXY_DEV(d), addr, &val, l, PCI_CONFIG_WRITE);
+}
+
+static void pci_proxy_dev_class_init(ObjectClass *klass, void *data)
+{
+    PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
+
+    k->realize = pci_proxy_dev_realize;
+    k->config_read = pci_proxy_read_config;
+    k->config_write = pci_proxy_write_config;
+}
+
+static const TypeInfo pci_proxy_dev_type_info = {
+    .name          = TYPE_PCI_PROXY_DEV,
+    .parent        = TYPE_PCI_DEVICE,
+    .instance_size = sizeof(PCIProxyDev),
+    .abstract      = true,
+    .class_size    = sizeof(PCIProxyDevClass),
+    .class_init    = pci_proxy_dev_class_init,
+    .interfaces = (InterfaceInfo[]) {
+        { INTERFACE_CONVENTIONAL_PCI_DEVICE },
+        { },
+    },
+};
+
+static void pci_proxy_dev_register_types(void)
+{
+    type_register_static(&pci_proxy_dev_type_info);
+}
+
+type_init(pci_proxy_dev_register_types)
+
+static void init_proxy(PCIDevice *dev, char *command, char *exec_name,
+                       Error **errp)
+{
+    PCIProxyDev *pdev = PCI_PROXY_DEV(dev);
+    Error *local_error = NULL;
+
+    if (!pdev->managed) {
+        if (command) {
+            remote_spawn(pdev, command, exec_name, &local_error);
+        } else {
+            return;
+        }
+    } else {
+        pdev->remote_pid = atoi(pdev->rid);
+        if (pdev->remote_pid == -1) {
+            error_setg(errp, "Remote PID is -1");
+            return;
+        }
+    }
+
+    pdev->mpqemu_link = mpqemu_link_create();
+
+    if (!pdev->mpqemu_link) {
+        error_setg(errp, "Failed to create proxy link");
+        return;
+    }
+
+    mpqemu_init_channel(pdev->mpqemu_link, &pdev->mpqemu_link->com,
+                        pdev->socket);
+}
+
+static void pci_proxy_dev_realize(PCIDevice *device, Error **errp)
+{
+    PCIProxyDev *dev = PCI_PROXY_DEV(device);
+    PCIProxyDevClass *k = PCI_PROXY_DEV_GET_CLASS(dev);
+    Error *local_err = NULL;
+
+    if (k->realize) {
+        k->realize(dev, &local_err);
+        if (local_err) {
+            error_propagate(errp, local_err);
+        }
+    }
+
+    dev->set_proxy_sock = set_proxy_sock;
+    dev->get_proxy_sock = get_proxy_sock;
+    dev->init_proxy = init_proxy;
+}
diff --git a/include/hw/proxy/qemu-proxy.h b/include/hw/proxy/qemu-proxy.h
new file mode 100644
index 0000000..28b0114
--- /dev/null
+++ b/include/hw/proxy/qemu-proxy.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright © 2018, 2020 Oracle and/or its affiliates.
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef QEMU_PROXY_H
+#define QEMU_PROXY_H
+
+#include "io/mpqemu-link.h"
+
+#define TYPE_PCI_PROXY_DEV "pci-proxy-dev"
+
+#define PCI_PROXY_DEV(obj) \
+            OBJECT_CHECK(PCIProxyDev, (obj), TYPE_PCI_PROXY_DEV)
+
+#define PCI_PROXY_DEV_CLASS(klass) \
+            OBJECT_CLASS_CHECK(PCIProxyDevClass, (klass), TYPE_PCI_PROXY_DEV)
+
+#define PCI_PROXY_DEV_GET_CLASS(obj) \
+            OBJECT_GET_CLASS(PCIProxyDevClass, (obj), TYPE_PCI_PROXY_DEV)
+
+typedef struct PCIProxyDev {
+    PCIDevice parent_dev;
+
+    MPQemuLinkState *mpqemu_link;
+
+    pid_t remote_pid;
+    int socket;
+
+    char *rid;
+
+    bool managed;
+
+    void (*set_proxy_sock) (PCIDevice *dev, int socket);
+    int (*get_proxy_sock) (PCIDevice *dev);
+
+    void (*proxy_ready) (PCIDevice *dev);
+    void (*init_proxy) (PCIDevice *dev, char *command, char *exec_name,
+                        Error **errp);
+
+} PCIProxyDev;
+
+typedef struct PCIProxyDevClass {
+    PCIDeviceClass parent_class;
+
+    void (*realize)(PCIProxyDev *dev, Error **errp);
+
+    char *command;
+} PCIProxyDevClass;
+
+#endif /* QEMU_PROXY_H */
diff --git a/include/io/mpqemu-link.h b/include/io/mpqemu-link.h
index 78c0818..5a2be48 100644
--- a/include/io/mpqemu-link.h
+++ b/include/io/mpqemu-link.h
@@ -14,6 +14,7 @@
 #include "qemu/osdep.h"
 #include "qemu-common.h"
 
+#include "sys/eventfd.h"
 #include "qom/object.h"
 #include "qemu/thread.h"
 #include "exec/cpu-common.h"
diff --git a/remote/remote-main.c b/remote/remote-main.c
index 56315cd..7b4cf2f 100644
--- a/remote/remote-main.c
+++ b/remote/remote-main.c
@@ -37,6 +37,32 @@
 static MPQemuLinkState *mpqemu_link;
 PCIDevice *remote_pci_dev;
 
+static void process_config_write(MPQemuMsg *msg)
+{
+    struct conf_data_msg *conf = (struct conf_data_msg *)msg->data2;
+
+    qemu_mutex_lock_iothread();
+    pci_default_write_config(remote_pci_dev, conf->addr, conf->val, conf->l);
+    qemu_mutex_unlock_iothread();
+}
+
+static void process_config_read(MPQemuMsg *msg)
+{
+    struct conf_data_msg *conf = (struct conf_data_msg *)msg->data2;
+    uint32_t val;
+    int wait;
+
+    wait = msg->fds[0];
+
+    qemu_mutex_lock_iothread();
+    val = pci_default_read_config(remote_pci_dev, conf->addr, conf->l);
+    qemu_mutex_unlock_iothread();
+
+    notify_proxy(wait, val);
+
+    PUT_REMOTE_WAIT(wait);
+}
+
 static void process_msg(GIOCondition cond, MPQemuChannel *chan)
 {
     MPQemuMsg *msg = NULL;
@@ -57,8 +83,10 @@ static void process_msg(GIOCondition cond, MPQemuChannel *chan)
     case INIT:
         break;
     case PCI_CONFIG_WRITE:
+        process_config_write(msg);
         break;
     case PCI_CONFIG_READ:
+        process_config_read(msg);
         break;
     default:
         error_setg(&err, "Unknown command");
-- 
1.8.3.1