Merge tag 'pull-xenfv-stable-20231106' of git://git.infradead.org/users/dwmw2/qemu into staging

Bugfixes for emulated Xen support

Selected bugfixes for mainline and stable, especially to the per-vCPU
local APIC vector delivery mode for event channel notifications, which
was broken in a number of ways.

The xen-block driver has been defaulting to the wrong protocol for x86
guest, and this fixes that — which is technically an incompatible change
but I'm fairly sure nobody relies on the broken behaviour (and in
production I *have* seen guests which rely on the correct behaviour,
which now matches the blkback driver in the Linux kernel).

A handful of other simple fixes for issues which came to light as new
features (qv) were being developed.

# -----BEGIN PGP SIGNATURE-----
#
# iQJIBAABCAAyFiEEvgfZ/VSAmrLEsP9fY3Ys2mfi81kFAmVIvv4UHGR3bXcyQGlu
# ZnJhZGVhZC5vcmcACgkQY3Ys2mfi81nFmRAAvK3VNuGDV56TJqFdtEWD+3jzSZU0
# CoL1mxggvwnlFn1SdHvbC5jl+UscknErcNbqlxMTTg9jQiiQqzFuaWujJnL0dEOY
# RJiS2scKln/1gv9NRbLE31FjPwoNz+zJI/iMvdutjT7Ll//v34jY0vd1Y5Wo53ay
# MBschuuxD1sUUTHNj5f9afrgZaetJfgBSNZraiLR5T2HEadJVJuhItdGxW1+KaPI
# zBIcflIeZmJl9b/L1a2bP3KJmRo8QzHB56X3uzwkPhYhYSU2dnCaJTLCkiNfK+Qh
# SgCBMlzsvJbIZqDA9YPOGdKK1ArfTJRmRDwAkqH0YQknQGoIkpN+7eQiiSv6PMS5
# U/93V7r6MfaftIs6YdWSnFozWeBuyKZL9H2nAXqZgL5t6uEMVR8Un/kFnGfslTFY
# 9gQ1o4IM6ECLiXhIP/sPNOprrbFb0HU7QPtEDJOxrJzBM+IfLbldRHn4p9CccqQA
# LHvJF98VhX1d0nA0iZBT3qqfKPbmUhRV9Jrm+WamqNrRXhiGdF8EidsUf8RWX+JD
# xZWJiqhTwShxdLE6TC/JgFz4cQCVHG8QiZstZUbdq59gtz9YO5PGByMgI3ds7iNQ
# lGXAPFm+1wU85W4dZOH7qyim6d9ytFm2Fm110BKM8l9B6UKEuKHpsxXMqdo65JXI
# 7uBKbVpdPKul0DY=
# =dQ7h
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 06 Nov 2023 18:25:02 HKT
# gpg:                using RSA key BE07D9FD54809AB2C4B0FF5F63762CDA67E2F359
# gpg:                issuer "dwmw2@infradead.org"
# gpg: Good signature from "David Woodhouse <dwmw2@infradead.org>" [unknown]
# gpg:                 aka "David Woodhouse <dwmw2@exim.org>" [unknown]
# gpg:                 aka "David Woodhouse <david@woodhou.se>" [unknown]
# gpg:                 aka "David Woodhouse <dwmw2@kernel.org>" [unknown]
# gpg: WARNING: The key's User ID is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: BE07 D9FD 5480 9AB2 C4B0  FF5F 6376 2CDA 67E2 F359

* tag 'pull-xenfv-stable-20231106' of git://git.infradead.org/users/dwmw2/qemu:
  hw/xen: use correct default protocol for xen-block on x86
  hw/xen: take iothread mutex in xen_evtchn_reset_op()
  hw/xen: fix XenStore watch delivery to guest
  hw/xen: don't clear map_track[] in xen_gnttab_reset()
  hw/xen: select kernel mode for per-vCPU event channel upcall vector
  i386/xen: fix per-vCPU upcall vector for Xen emulation
  i386/xen: Don't advertise XENFEAT_supervisor_mode_kernel

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

6 files changed, 31 insertions, 13 deletions

diff --git a/hw/block/xen-block.c b/hw/block/xen-block.c
index a07cd7e..bfa5396 100644
--- a/hw/block/xen-block.c
+++ b/hw/block/xen-block.c
@@ -115,9 +115,13 @@ static void xen_block_connect(XenDevice *xendev, Error **errp)
         return;
     }
 
-    if (xen_device_frontend_scanf(xendev, "protocol", "%ms",
-                                  &str) != 1) {
-        protocol = BLKIF_PROTOCOL_NATIVE;
+    if (xen_device_frontend_scanf(xendev, "protocol", "%ms", &str) != 1) {
+        /* x86 defaults to the 32-bit protocol even for 64-bit guests. */
+        if (object_dynamic_cast(OBJECT(qdev_get_machine()), "x86-machine")) {
+            protocol = BLKIF_PROTOCOL_X86_32;
+        } else {
+            protocol = BLKIF_PROTOCOL_NATIVE;
+        }
     } else {
         if (strcmp(str, XEN_IO_PROTO_ABI_X86_32) == 0) {
             protocol = BLKIF_PROTOCOL_X86_32;
diff --git a/hw/i386/kvm/xen_evtchn.c b/hw/i386/kvm/xen_evtchn.c
index a731738..b2b4be9 100644
--- a/hw/i386/kvm/xen_evtchn.c
+++ b/hw/i386/kvm/xen_evtchn.c
@@ -490,6 +490,12 @@ int xen_evtchn_set_callback_param(uint64_t param)
         break;
     }
 
+    /* If the guest has set a per-vCPU callback vector, prefer that. */
+    if (gsi && kvm_xen_has_vcpu_callback_vector()) {
+        in_kernel = kvm_xen_has_cap(EVTCHN_SEND);
+        gsi = 0;
+    }
+
     if (!ret) {
         /* If vector delivery was turned *off* then tell the kernel */
         if ((s->callback_param >> CALLBACK_VIA_TYPE_SHIFT) ==
@@ -1129,6 +1135,7 @@ int xen_evtchn_reset_op(struct evtchn_reset *reset)
         return -ESRCH;
     }
 
+    QEMU_IOTHREAD_LOCK_GUARD();
     return xen_evtchn_soft_reset();
 }
 
diff --git a/hw/i386/kvm/xen_gnttab.c b/hw/i386/kvm/xen_gnttab.c
index 21c30e3..839ec92 100644
--- a/hw/i386/kvm/xen_gnttab.c
+++ b/hw/i386/kvm/xen_gnttab.c
@@ -541,7 +541,5 @@ int xen_gnttab_reset(void)
     s->entries.v1[GNTTAB_RESERVED_XENSTORE].flags = GTF_permit_access;
     s->entries.v1[GNTTAB_RESERVED_XENSTORE].frame = XEN_SPECIAL_PFN(XENSTORE);
 
-    memset(s->map_track, 0, s->max_frames * ENTRIES_PER_FRAME_V1);
-
     return 0;
 }
diff --git a/hw/i386/kvm/xen_xenstore.c b/hw/i386/kvm/xen_xenstore.c
index 660d0b7..8e716a7 100644
--- a/hw/i386/kvm/xen_xenstore.c
+++ b/hw/i386/kvm/xen_xenstore.c
@@ -1357,10 +1357,12 @@ static void fire_watch_cb(void *opaque, const char *path, const char *token)
     } else {
         deliver_watch(s, path, token);
         /*
-         * If the message was queued because there was already ring activity,
-         * no need to wake the guest. But if not, we need to send the evtchn.
+         * Attempt to queue the message into the actual ring, and send
+         * the event channel notification if any bytes are copied.
          */
-        xen_be_evtchn_notify(s->eh, s->be_port);
+        if (s->rsp_pending && put_rsp(s) > 0) {
+            xen_be_evtchn_notify(s->eh, s->be_port);
+        }
     }
 }
 
diff --git a/include/sysemu/kvm_xen.h b/include/sysemu/kvm_xen.h
index 595abfb..961c702 100644
--- a/include/sysemu/kvm_xen.h
+++ b/include/sysemu/kvm_xen.h
@@ -22,6 +22,7 @@
 int kvm_xen_soft_reset(void);
 uint32_t kvm_xen_get_caps(void);
 void *kvm_xen_get_vcpu_info_hva(uint32_t vcpu_id);
+bool kvm_xen_has_vcpu_callback_vector(void);
 void kvm_xen_inject_vcpu_callback_vector(uint32_t vcpu_id, int type);
 void kvm_xen_set_callback_asserted(void);
 int kvm_xen_set_vcpu_virq(uint32_t vcpu_id, uint16_t virq, uint16_t port);
diff --git a/target/i386/kvm/xen-emu.c b/target/i386/kvm/xen-emu.c
index 76348f9..75b2c55 100644
--- a/target/i386/kvm/xen-emu.c
+++ b/target/i386/kvm/xen-emu.c
@@ -267,7 +267,6 @@ static bool kvm_xen_hcall_xen_version(struct kvm_xen_exit *exit, X86CPU *cpu,
             fi.submap |= 1 << XENFEAT_writable_page_tables |
                          1 << XENFEAT_writable_descriptor_tables |
                          1 << XENFEAT_auto_translated_physmap |
-                         1 << XENFEAT_supervisor_mode_kernel |
                          1 << XENFEAT_hvm_callback_vector |
                          1 << XENFEAT_hvm_safe_pvclock |
                          1 << XENFEAT_hvm_pirqs;
@@ -307,7 +306,7 @@ static int kvm_xen_set_vcpu_callback_vector(CPUState *cs)
 
     trace_kvm_xen_set_vcpu_callback(cs->cpu_index, vector);
 
-    return kvm_vcpu_ioctl(cs, KVM_XEN_HVM_SET_ATTR, &xva);
+    return kvm_vcpu_ioctl(cs, KVM_XEN_VCPU_SET_ATTR, &xva);
 }
 
 static void do_set_vcpu_callback_vector(CPUState *cs, run_on_cpu_data data)
@@ -425,6 +424,13 @@ void kvm_xen_set_callback_asserted(void)
     }
 }
 
+bool kvm_xen_has_vcpu_callback_vector(void)
+{
+    CPUState *cs = qemu_get_cpu(0);
+
+    return cs && !!X86_CPU(cs)->env.xen_vcpu_callback_vector;
+}
+
 void kvm_xen_inject_vcpu_callback_vector(uint32_t vcpu_id, int type)
 {
     CPUState *cs = qemu_get_cpu(vcpu_id);
@@ -441,7 +447,8 @@ void kvm_xen_inject_vcpu_callback_vector(uint32_t vcpu_id, int type)
          * deliver it as an MSI.
          */
         MSIMessage msg = {
-            .address = APIC_DEFAULT_ADDRESS | X86_CPU(cs)->apic_id,
+            .address = APIC_DEFAULT_ADDRESS |
+                       (X86_CPU(cs)->apic_id << MSI_ADDR_DEST_ID_SHIFT),
             .data = vector | (1UL << MSI_DATA_LEVEL_SHIFT),
         };
         kvm_irqchip_send_msi(kvm_state, msg);
@@ -850,8 +857,7 @@ static bool kvm_xen_hcall_hvm_op(struct kvm_xen_exit *exit, X86CPU *cpu,
     int ret = -ENOSYS;
     switch (cmd) {
     case HVMOP_set_evtchn_upcall_vector:
-        ret = kvm_xen_hcall_evtchn_upcall_vector(exit, cpu,
-                                                 exit->u.hcall.params[0]);
+        ret = kvm_xen_hcall_evtchn_upcall_vector(exit, cpu, arg);
         break;
 
     case HVMOP_pagetable_dying: