> > + /* Under 64-bit hypervisor with a 32-bit domain, the hypervisor > > + * offsets the pt_base by two pages. Hence the reservation that is done > > + * in mmu.c misses two pages. We correct it here if we detect this. */ > > + if (last_phys < __pa(xen_start_info->pt_base)) > > + memblock_reserve(last_phys, __pa(xen_start_info->pt_base) - > > last_phys); > > } > > What are these two pages used for? They are not documented in xen.h, why > should we reserve them? > > After all we still have: > > memblock_reserve(PFN_PHYS(pt_base), (pt_end - pt_base) * PAGE_SIZE); > > that should protect what we are interested in anyway...
You are looking at the x86_64 piece of code. This issue only appears on 32-bit which was not modified by my patches and has: 2003 memblock_reserve(__pa(xen_start_info->pt_base), 2004 xen_start_info->nr_pt_frames * PAGE_SIZE); and as I found out, the pt_base is wrong. The cr3 we load and use is actually two pages back! -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
