I have an iMac, late 2009, with a Core i7 & ATI Radeon HD 4850 512 MB graphics.
In the past few days, I've had two cases where my system hung after the display had gone to sleep (but before the computer had slept). In both cases, I had a Windows 7 VM open, with 3D acceleration enabled (I don’t recall if that was the default). The system was still responsive remotely after the hang, so I logged in with ssh and poked around a bit before eventually rebooting.
In both instances, the vmware-vmx process responsible for the Windows 7 VM appeared to be looping, consuming slightly over 100% of CPU. It wasn't obvious from a profiling sample exactly what was going on, but in both hangs I saw this thread active, which pretty much never shows up when the system is running properly: [I don't know why sample recorded only 12 samples — it was consistent across all threads, 12 on this sample, 15 from an earlier one.]
12 Thread_190191
+ 12 thread_start (in libsystem_c.dylib) + 13 [0x7fff8a9611e1]
+ 12 _pthread_start (in libsystem_c.dylib) + 327 [0x7fff8a9747a2]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x1b91be [0x1001b91be]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x18a06b [0x10018a06b]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x175d19 [0x100175d19]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x17dc6e [0x10017dc6e]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x180e76 [0x100180e76]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x1828c5 [0x1001828c5]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x1e6cee [0x1001e6cee]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x1eca6a [0x1001eca6a]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x1ed938 [0x1001ed938]
+ 12 ??? (in vmware-vmx) load address 0x100000000 + 0x1ed644 [0x1001ed644]
+ 12 glTexSubImage2D (in libGL.dylib) + 77 [0x7fff8aae51c5]
+ 12 glTexSubImage2D_Exec (in GLEngine) + 1137 [0x1089b9074]
+ 12 ??? (in ATIRadeonX2000GLDriver) load address 0x109800000 + 0x1f8aaf [0x1099f8aaf]
+ 12 IOConnectCallMethod (in IOKit) + 346 [0x7fff8868763e]
+ 12 io_connect_method (in IOKit) + 380 [0x7fff886d3fcf]
+ 12 mach_msg (in libsystem_kernel.dylib) + 70 [0x7fff90fafc42]
+ 12 mach_msg_trap (in libsystem_kernel.dylib) + 10 [0x7fff90fb0686]
I don’t know if this is possibly a normal state when the screen saver is running, but the 100% CPU is suspicious.
I’m suspecting that the update to the ATI drivers in 10.8.4 is probably what is causing this hang, but I also suspect that VMware engineers have more connections into Apple’s graphics engineering than I do. :-)
I’ll try disabling 3D acceleration to see if that helps (by reducing the load on the ATI if nothing else), but there does seem to be an issue here. (I never saw this with 10.8.3.) So far, I haven't seen this when the screen saver hadn't kicked in, either. [Hmmm — I had “random” screen saver, and the screen was blank, so I don’t actually know what was running. I’ll set it to “Message”, again on the hopes of not pushing the driver too hard.]
If there’s more information from my system that would be helpful, feel free to ask.
Anton