qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Maxim Levitsky <mlevitsk@redhat.com>
To: qemu-devel@nongnu.org
Cc: "Fam Zheng" <fam@euphon.net>,
	"Daniel P. Berrangé" <berrange@redhat.com>,
	"Eduardo Habkost" <ehabkost@redhat.com>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	"Maxim Levitsky" <mlevitsk@redhat.com>,
	"Paolo Bonzini" <pbonzini@redhat.com>
Subject: [PATCH v4 6/9] scsi/scsi-bus: scsi_device_find: don't return unrealized devices
Date: Mon, 31 Aug 2020 18:01:21 +0300	[thread overview]
Message-ID: <20200831150124.206267-7-mlevitsk@redhat.com> (raw)
In-Reply-To: <20200831150124.206267-1-mlevitsk@redhat.com>

The device core first places a device on the bus and then realizes it.
Make scsi_device_find avoid returing such devices to avoid
races in drivers that use an iothread (currently virtio-scsi)

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1812399

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
---
 hw/scsi/scsi-bus.c | 88 ++++++++++++++++++++++++++++------------------
 1 file changed, 53 insertions(+), 35 deletions(-)

diff --git a/hw/scsi/scsi-bus.c b/hw/scsi/scsi-bus.c
index 92d412b65c..7ceae2c92b 100644
--- a/hw/scsi/scsi-bus.c
+++ b/hw/scsi/scsi-bus.c
@@ -51,6 +51,56 @@ static const TypeInfo scsi_bus_info = {
 };
 static int next_scsi_bus;
 
+static SCSIDevice *_scsi_device_find(SCSIBus *bus, int channel, int id, int lun,
+                                     bool include_unrealized)
+{
+    BusChild *kid;
+    SCSIDevice *retval = NULL;
+
+    rcu_read_lock();
+
+    QTAILQ_FOREACH_RCU(kid, &bus->qbus.children, sibling) {
+        DeviceState *qdev = kid->child;
+        SCSIDevice *dev = SCSI_DEVICE(qdev);
+
+        if (dev->channel == channel && dev->id == id) {
+            if (dev->lun == lun) {
+                retval = dev;
+                goto out;
+            }
+
+            /*
+             * If we don't find exact match (channel/bus/lun),
+             * we will return the first device which matches channel/bus
+             */
+
+            if (!retval) {
+                retval = dev;
+            }
+        }
+    }
+out:
+    /*
+     * This function might run on the IO thread and we might race against
+     * main thread hot-plugging the device.
+     * We assume that as soon as .realized is set to true we can let
+     * the user access the device.
+     */
+
+    if (retval && !include_unrealized &&
+        !atomic_load_acquire(&retval->qdev.realized)) {
+        retval = NULL;
+    }
+
+    rcu_read_unlock();
+    return retval;
+}
+
+SCSIDevice *scsi_device_find(SCSIBus *bus, int channel, int id, int lun)
+{
+    return _scsi_device_find(bus, channel, id, lun, false);
+}
+
 static void scsi_device_realize(SCSIDevice *s, Error **errp)
 {
     SCSIDeviceClass *sc = SCSI_DEVICE_GET_CLASS(s);
@@ -186,7 +236,7 @@ static void scsi_qdev_realize(DeviceState *qdev, Error **errp)
             dev->lun = 0;
         }
         do {
-            d = scsi_device_find(bus, dev->channel, ++id, dev->lun);
+            d = _scsi_device_find(bus, dev->channel, ++id, dev->lun, true);
         } while (d && d->lun == dev->lun && id < bus->info->max_target);
         if (d && d->lun == dev->lun) {
             error_setg(errp, "no free target");
@@ -196,7 +246,7 @@ static void scsi_qdev_realize(DeviceState *qdev, Error **errp)
     } else if (dev->lun == -1) {
         int lun = -1;
         do {
-            d = scsi_device_find(bus, dev->channel, dev->id, ++lun);
+            d = _scsi_device_find(bus, dev->channel, dev->id, ++lun, true);
         } while (d && d->lun == lun && lun < bus->info->max_lun);
         if (d && d->lun == lun) {
             error_setg(errp, "no free lun");
@@ -204,7 +254,7 @@ static void scsi_qdev_realize(DeviceState *qdev, Error **errp)
         }
         dev->lun = lun;
     } else {
-        d = scsi_device_find(bus, dev->channel, dev->id, dev->lun);
+        d = _scsi_device_find(bus, dev->channel, dev->id, dev->lun, true);
         assert(d);
         if (d->lun == dev->lun && dev != d) {
             error_setg(errp, "lun already used by '%s'", d->qdev.id);
@@ -1573,38 +1623,6 @@ static char *scsibus_get_fw_dev_path(DeviceState *dev)
                            qdev_fw_name(dev), d->id, d->lun);
 }
 
-SCSIDevice *scsi_device_find(SCSIBus *bus, int channel, int id, int lun)
-{
-    BusChild *kid;
-    SCSIDevice *target_dev = NULL;
-
-    rcu_read_lock();
-
-    QTAILQ_FOREACH_RCU(kid, &bus->qbus.children, sibling) {
-        DeviceState *qdev = kid->child;
-        SCSIDevice *dev = SCSI_DEVICE(qdev);
-
-        if (dev->channel == channel && dev->id == id) {
-            if (dev->lun == lun) {
-                rcu_read_unlock();
-                return dev;
-            }
-
-            /*
-             * If we don't find exact match (channel/bus/lun),
-             * we will return the first device which matches channel/bus
-             */
-
-            if (!target_dev) {
-                target_dev = dev;
-            }
-        }
-    }
-
-    rcu_read_unlock();
-    return target_dev;
-}
-
 /* SCSI request list.  For simplicity, pv points to the whole device */
 
 static int put_scsi_requests(QEMUFile *f, void *pv, size_t size,
-- 
2.26.2



  parent reply	other threads:[~2020-08-31 15:11 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-08-31 15:01 [PATCH v4 0/9] Fix scsi devices plug/unplug races w.r.t virtio-scsi iothread Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 1/9] scsi/scsi_bus: switch search direction in scsi_device_find Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 2/9] rcu: Implement drain_call_rcu Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 3/9] device_core: use drain_call_rcu in in hmp_device_del/qmp_device_add Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 4/9] device-core: use RCU for list of childs of a bus Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 5/9] device-core: use atomic_set on .realized property Maxim Levitsky
2020-08-31 15:01 ` Maxim Levitsky [this message]
2020-09-08 15:00   ` [PATCH v4 6/9] scsi/scsi-bus: scsi_device_find: don't return unrealized devices Stefan Hajnoczi
2020-09-09  8:15     ` Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 7/9] scsi/scsi_bus: Add scsi_device_get Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 8/9] virtio-scsi: use scsi_device_get Maxim Levitsky
2020-08-31 15:01 ` [PATCH v4 9/9] scsi/scsi_bus: fix races in REPORT LUNS Maxim Levitsky
2020-09-01  9:15   ` Maxim Levitsky
2020-09-08 15:27   ` Stefan Hajnoczi
2020-09-09  8:20     ` Maxim Levitsky
2020-09-11 15:12       ` Stefan Hajnoczi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200831150124.206267-7-mlevitsk@redhat.com \
    --to=mlevitsk@redhat.com \
    --cc=berrange@redhat.com \
    --cc=ehabkost@redhat.com \
    --cc=fam@euphon.net \
    --cc=mst@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).