[Qemu-devel] [PATCH] spapr_rng: fix race with main loop

[Greg Kurz]

Since commit "60253ed1e6ec rng: add request queue support to rng-random",
the use of a spapr_rng device may hang vCPU threads.

The following path is taken without holding the lock to the main loop mutex:

h_random()
  rng_backend_request_entropy()
    rng_random_request_entropy()
      qemu_set_fd_handler()

The consequence is that entropy_available() may be called before the vCPU
thread could even queue the request: depending on the scheduling, it may
happen that entropy_available() does not call random_recv()->qemu_sem_post().
The vCPU thread will then sleep forever in h_random()->qemu_sem_wait().

This could not happen before 60253ed1e6ec because entropy_available() used
to call random_recv() unconditionally.

This patch ensures the lock is held to avoid the race.

Signed-off-by: Greg Kurz <address@hidden>
---

Thomas,

This is the problem mentioned by Cedric in:

https://lists.nongnu.org/archive/html/qemu-devel/2016-03/msg02526.html

Cheers.

--
Greg

 hw/ppc/spapr_rng.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/hw/ppc/spapr_rng.c b/hw/ppc/spapr_rng.c
index a39d472b66fd..02d6be49f58e 100644
--- a/hw/ppc/spapr_rng.c
+++ b/hw/ppc/spapr_rng.c
@@ -77,13 +77,13 @@ static target_ulong h_random(PowerPCCPU *cpu, 
sPAPRMachineState *spapr,
     hrdata.val.v64 = 0;
     hrdata.received = 0;
 
-    qemu_mutex_unlock_iothread();
     while (hrdata.received < 8) {
         rng_backend_request_entropy(rngstate->backend, 8 - hrdata.received,
                                     random_recv, &hrdata);
+        qemu_mutex_unlock_iothread();
         qemu_sem_wait(&hrdata.sem);
+        qemu_mutex_lock_iothread();
     }
-    qemu_mutex_lock_iothread();
 
     qemu_sem_destroy(&hrdata.sem);
     args[0] = hrdata.val.v64;