Jeremy Fitzhardinge wrote: > Do you disable interrupts when lazy_mmu is active? Otherwise, it seems > to me that drivers will have problems using kmap_atomic() within an > interrupt routine... >
Yup, that's a pretty serious bug. Looks like the lazy mmu code needs to save and restore interrupts, or alternatively, have kmap_atomic turn off lazy_mmu if it is enabled. Zach _______________________________________________ Virtualization mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/virtualization
