On Wed, Jun 29, 2016 at 12:03:16PM +0200, Ard Biesheuvel wrote:

On 29 June 2016 at 11:39, Catalin Marinas [off-list ref] wrote:

quoted

On Tue, Jun 28, 2016 at 06:12:22PM +0200, Ard Biesheuvel wrote:

quoted

On 28 June 2016 at 18:05, Catalin Marinas [off-list ref] wrote:

quoted

(Restarting the thread before I forget the entire discussion)

On Mon, Jun 06, 2016 at 11:18:14PM +0200, Ard Biesheuvel wrote:

quoted

quoted

quoted

quoted

On 31 May 2016 at 17:14, Catalin Marinas [off-list ref] wrote:

quoted

diff --git a/arch/arm64/kernel/efi.c b/arch/arm64/kernel/efi.c
index 78f52488f9ff..0d5753c31c7f 100644
--- a/arch/arm64/kernel/efi.c
+++ b/arch/arm64/kernel/efi.c

@@ -62,10 +62,26 @@ struct screen_info screen_info __section(.data);
 int __init efi_create_mapping(struct mm_struct *mm, efi_memory_desc_t *md)
 {
        pteval_t prot_val = create_mapping_protection(md);
+       phys_addr_t length = md->num_pages << EFI_PAGE_SHIFT;
+       efi_memory_desc_t *next = md;

-       create_pgd_mapping(mm, md->phys_addr, md->virt_addr,
-                          md->num_pages << EFI_PAGE_SHIFT,
-                          __pgprot(prot_val | PTE_NG));
+       /*
+        * Search for the next EFI runtime map and check for any overlap with
+        * the current map when aligned to PAGE_SIZE. In such case, defer
+        * mapping the end of the current range until the next
+        * efi_create_mapping() call.
+        */
+       for_each_efi_memory_desc_continue(next) {
+               if (!(next->attribute & EFI_MEMORY_RUNTIME))
+                       continue;
+               if (next->phys_addr < PAGE_ALIGN(md->phys_addr + length))
+                       length -= (md->phys_addr + length) & ~PAGE_MASK;

[...]

quoted

Another thing I failed to mention is that the new Memory Attributes
table support may map all of the RuntimeServicesCode regions a second
time, but with a higher granularity, using RO for .text and .rodata
and NX for .data and .bss (and the PE/COFF header).

Can this not be done in a single go without multiple passes? That's what
we did for the core arm64 code, the only one left being EFI run-time
mappings.

Well, we probably could, but it is far from trivial.

quoted

quoted

Due to the higher
granularity, regions that were mapped using the contiguous bit the
first time around may be split into smaller regions. Your current code
does not address that case.

If the above doesn't work, the only solution would be to permanently map
these ranges as individual pages, no large blocks.

That is not unreasonable, since regions >2MB are unusual.

We'll have the contiguous bit supported at some point and we won't be
able to use it for EFI run-time mappings. But I don't think that's
essential, minor improvement on a non-critical path.

I'll post some patches to always use PAGE_SIZE granularity for EFI
run-time mappings.

Given that contiguous bit mappings only affect the TLB footprint, I'd
be more concerned about not using block mappings for EfiMemoryMappedIo
regions (since they may cover fairly sizable NOR flashes like the 64
MB one QEMU mach-virt exposes).

Good point.

So I would recommend to only use PAGE_SIZE granularity for
EfiRuntimeServicesCode and EfiRuntimeServicesData regions, since those
are the only ones that can be expected to appear in the Memory
Attributes table, and all other regions will only be mapped a single
time.

Is there a possibility that EfiMemoryMappedIo share the same 64K page
with EfiRuntimeServicesCode? If it does, it won't help much with
avoiding splitting. Unless I keep a combination of these series
(checking the end/start overlap) with a forced page-only mapping for
EfiRuntimeServicesCode/Data.

-- 
Catalin