Merge remote-tracking branch 'remotes/bonzini/tags/for-upstream' into staging

- vhost-scsi: add bootindex property
- RCU: fix MemoryRegion lifetime issues in PCI; document the rules;
convert of AddressSpaceDispatch and RAMList
- KVM: add kvm_exit reasons for aarch64

# gpg: Signature made Mon Feb 16 16:32:32 2015 GMT using RSA key ID 78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>"
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini/tags/for-upstream: (21 commits)
  Convert ram_list to RCU
  exec: convert ram_list to QLIST
  cosmetic changes preparing for the following patches
  exec: protect mru_block with RCU
  rcu: add g_free_rcu
  rcu: introduce RCU-enabled QLIST
  exec: RCUify AddressSpaceDispatch
  exec: make iotlb RCU-friendly
  exec: introduce cpu_reload_memory_map
  docs: clarify memory region lifecycle
  pci: split shpc_cleanup and shpc_free
  pcie: remove mmconfig memory leak and wrap mmconfig update with transaction
  memory: keep the owner of the AddressSpace alive until do_address_space_destroy
  rcu: run RCU callbacks under the BQL
  rcu: do not let RCU callbacks pile up indefinitely
  vhost-scsi: set the bootable value of channel/target/lun
  vhost-scsi: add a property for booting
  vhost-scsi: expose the TYPE_FW_PATH_PROVIDER interface
  vhost-scsi: add bootindex property
  qdev: support to get a device firmware path directly
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

1 files changed, 33 insertions, 0 deletions

diff --git a/cpu-exec.c b/cpu-exec.c
index 6738117..2ffeb6e 100644
--- a/cpu-exec.c
+++ b/cpu-exec.c
@@ -24,6 +24,9 @@
 #include "qemu/atomic.h"
 #include "sysemu/qtest.h"
 #include "qemu/timer.h"
+#include "exec/address-spaces.h"
+#include "exec/memory-internal.h"
+#include "qemu/rcu.h"
 
 /* -icount align implementation. */
 
@@ -141,6 +144,33 @@ void cpu_resume_from_signal(CPUState *cpu, void *puc)
     cpu->exception_index = -1;
     siglongjmp(cpu->jmp_env, 1);
 }
+
+void cpu_reload_memory_map(CPUState *cpu)
+{
+    AddressSpaceDispatch *d;
+
+    if (qemu_in_vcpu_thread()) {
+        /* Do not let the guest prolong the critical section as much as it
+         * as it desires.
+         *
+         * Currently, this is prevented by the I/O thread's periodinc kicking
+         * of the VCPU thread (iothread_requesting_mutex, qemu_cpu_kick_thread)
+         * but this will go away once TCG's execution moves out of the global
+         * mutex.
+         *
+         * This pair matches cpu_exec's rcu_read_lock()/rcu_read_unlock(), which
+         * only protects cpu->as->dispatch.  Since we reload it below, we can
+         * split the critical section.
+         */
+        rcu_read_unlock();
+        rcu_read_lock();
+    }
+
+    /* The CPU and TLB are protected by the iothread lock.  */
+    d = atomic_rcu_read(&cpu->as->dispatch);
+    cpu->memory_dispatch = d;
+    tlb_flush(cpu, 1);
+}
 #endif
 
 /* Execute a TB, and fix up the CPU state afterwards if necessary */
@@ -352,6 +382,8 @@ int cpu_exec(CPUArchState *env)
      * an instruction scheduling constraint on modern architectures.  */
     smp_mb();
 
+    rcu_read_lock();
+
     if (unlikely(exit_request)) {
         cpu->exit_request = 1;
     }
@@ -548,6 +580,7 @@ int cpu_exec(CPUArchState *env)
     } /* for(;;) */
 
     cc->cpu_exec_exit(cpu);
+    rcu_read_unlock();
 
     /* fail safe : never use current_cpu outside cpu_exec() */
     current_cpu = NULL;