1 files changed, 42 insertions, 0 deletions
diff --git a/results/classifier/118/VMM/2635 b/results/classifier/118/VMM/2635
new file mode 100644
index 000000000..80bbf99c9
--- /dev/null
+++ b/results/classifier/118/VMM/2635
@@ -0,0 +1,42 @@
+VMM: 0.869
+graphic: 0.867
+ppc: 0.812
+device: 0.809
+files: 0.771
+architecture: 0.747
+network: 0.701
+PID: 0.696
+vnc: 0.690
+socket: 0.614
+register: 0.597
+kernel: 0.595
+performance: 0.569
+semantic: 0.568
+risc-v: 0.541
+boot: 0.517
+arm: 0.505
+virtual: 0.499
+mistranslation: 0.473
+debug: 0.467
+TCG: 0.433
+permissions: 0.419
+i386: 0.411
+x86: 0.380
+assembly: 0.372
+hypervisor: 0.321
+peripherals: 0.291
+user-level: 0.271
+KVM: 0.244
+
+A use-after-free bug in pflash_cfi01 snapshot implementation
+Description of problem:
+The flash snapshot restore does not function correctly. Basically when you use “if=pflash,format=raw,unit=0,file=OVMF_VAR.fd", it crashes when trying to restore a snapshot.
+
+The root cause is:
+
+1. In system/runstate.c, function vm_state_notify loops through vm_change_state_head list and calls the callback function for each entry.
+2. One of the callback function pointer points to function postload_update_cb in hw/block/pflash_cfi01.c.
+3. In function postload_update_cb, it calls qemu_del_vm_change_state_handler in which the entry element memory is freed.
+4. Note that, it is still running in the loop, the entry will be visited and get executed, the function pointer may point to a wide memory.
+Additional information:
+