Re: [PATCH 1/3] x86, mm: Include the entire kernel memory map in trampoline_pgd

[Matt Fleming <matt.fleming@intel.com>]

On Wed, 2012-10-03 at 14:31 +0100, Jan Beulich wrote:
> >>> Matt Fleming <matt@console-pimps.org> 10/03/12 2:59 PM >>>
> >+static int insert_identity_mapping(resource_size_t paddr, unsigned long vaddr,
> >+                    unsigned long size)
> >+{
> >+    unsigned long end = vaddr + size;
> >+    unsigned long next;
> >+    pgd_t *vpgd, *ppgd;
> >+
> >+#ifdef CONFIG_X86_32
> >+    ppgd = initial_page_table + pgd_index(paddr);
> >+
> >+    if (paddr >= PAGE_OFFSET || paddr + size > PAGE_OFFSET)
> >+        return 1;
> >+#else
> >+    ppgd = __va(real_mode_header->trampoline_pgd) + pgd_index(paddr);
> 
> Missing equivalent code (to the 32-bit one above) here - after all, you're trying
> to potentially insert a 52-bit physical address into 48-bit virtual space.

Ah, true.

> >+#endif
> >+
> >+    vpgd = pgd_offset_k(vaddr);
> >+    do {
> >+        next = pgd_addr_end(vaddr, end);
> >+
> >+        if (!pgd_present(*ppgd)) {
> >+            pud_t *ppud = (pud_t *)get_zeroed_page(GFP_KERNEL);
> >+            if (!ppud)
> >+                return 1;
> >+
> >+            set_pgd(ppgd, __pgd(_KERNPG_TABLE | __pa(ppud)));
> >+        }
> >+
> >+        if (ident_pud_range(paddr, vaddr, ppgd, vpgd, next))
> >+            return 1;
> >+    } while (ppgd++, vpgd++, vaddr = next, vaddr != end);
> >+
> >+    return 0;
> >+}
> >+
>  >/*
> >* Remap an arbitrary physical address space into the kernel virtual
> >* address space. Needed when the kernel wants to access high addresses
> >@@ -163,6 +258,10 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr,
> >    ret_addr = (void __iomem *) (vaddr + offset);
> >    mmiotrace_ioremap(unaligned_phys_addr, unaligned_size, ret_addr);
>  >
> >+    if (insert_identity_mapping(phys_addr, vaddr, size))
> >+        printk(KERN_WARNING "ioremap: unable to map 0x%llx in identity pagetable\n",
> >+                    (unsigned long long)phys_addr);
> 
> Isn't that going to trigger quite frequently on 32-bit kernels?

Hmmm... yeah, probably, though it didn't during my testing. If it is
likely to trigger a lot then we might be best only inserting the
identity mmio mapping for 64-bit, and addressing the 32-bit case if we
ever actually need the identity pagetable.