On 9/18/19 11:02 AM, Richard Henderson wrote: > However this time the changes are extremely minimal, and now > I really *really* don't understand why they don't work, because > as far as I can tell the new locking is *identical* with the > current i/o path.
The difference was failing to set cpu->mem_io_pc, so that TARGET_HAS_PRECISE_SMC within tb_invalidate_phys_page_range__locked could look up the current TB, and potentially restore state and exit to the main loop. Version 3 will have this fixed. r~
