dma.h provides the prototypes for the different dma_alloc_* functions,
so we should include it to make sure the prototypes are consistent.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Current code for dma_sync_single_for_device(), when called with dir
set to DMA_FROM_DEVICE, will first invalidate given region of memory
as a first step and then clean+invalidate it as a second. While the
second step should be harmless it seems to be an unnecessary no-op
that could probably be avoided.

Analogous code in Linux kernel (4.18) in arch/arm64/mm/cache.S:

ENTRY(__dma_map_area)
	cmp	w2, #DMA_FROM_DEVICE
	b.eq	__dma_inv_area
	b	__dma_clean_area
ENDPIPROC(__dma_map_area)

is written to only perform either invalidate or clean, depending on
the direction, so change dma_sync_single_for_device() to behave in the
same vein and perfom _either_ invlidate or flush of the given region.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Flags passed to arch_remap_range() are architecture independent, so it
can't be passed as is to map_region(). Add code to do the proper
conversion to avoid subtle bugs that this confusion brings.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Make use of create_table() instead of calling xmemalign() and memset()
explicitly.

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Until now it was not possible to enable the MMU in PBL because
create_section needs memory allocations which are not available. With
this patch we move the early MMU support to a separate file and all
necessary aux functions to mmu_64.h. create_sections is reimplmented
for the early case to only create 1st level pagetables.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Change the arguements to type unsigned long which is suitable for both
arm32 and arm64. While at it move the prototype to arch/arm/include/.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

cpu/mmu.h has nothing in common for the 32bit and 64bit variant. Make it
two separate files.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Make zero page faulting which allows us to catch NULL pointer derefs.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

For proper DMA support dma_alloc_coherent and DMA sync operations are
needed. Implement them.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Do it as U-Boot: Disable MMU first, then flush caches and finally
invalidate tlbs. I wish I could reference some document instead of
U-Boot code, but I haven't found anything.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

If the MMU is enabled then we should be able to disable it, no
matter if we initialized it in barebox or not. This change is not
really needed but helps when we are starting second stage from U-Boot
with the 'go' command which leaves the MMU enabled.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

When the MMU is disabled there is no need to disable the icache. Leave
it enabled.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

A BITS_PER_VA value of 33 is a little small. Increase it to 39 which is
the maximum size we can do with 3 level page tables. The TCR value
depends on the current exception level, so we have to calculate the
value during runtime. To do this use a function derived from U-Boots
get_tcr function.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

When we reach level 3 page tables we set the PTE_TYPE_PAGE bit in attr.
This is wrong since in the outer loop we can fall back to a lower level
in which case the PTE_TYPE_PAGE may not be set.
Fix this by not modifying attr and instead compose the *pte value when
needed.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

When we create a higher level page table we have to initialize it
with the settings from the previous lower level page table so that
we do not modify unrelated mappings. split_block() is taken from
U-Boot code and does this job.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

'PMD' (Page Middle Directory) is a Linuxism that is not really
helpful in the barebox MMU code. Use the U-Boot definitions
which only use PTE_* and seem to be more consistent for our
usecase.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Using board code to enable the MMU is not nice. Do it in generic
code. Since mmu_enable() is now done in mmu_64.c we no longer have
to export it.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

The memory attributes register contains the memory attribute settings
for the corresponding to the possible AttrIndx values in the page
table entries. Passing UNCACHED_MEM makes no sense here, pass the
desired attributes instead.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

By default map the whole address space as unshared device memory
as this is what barebox drivers normally expect.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

During initialization of the page tables we want to create a flat
uncached mapping for the whole address space. create_sections() takes
the size in bytes, not in MiB as assumed by the callees. Fix this.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

create_sections() takes size in bytes, not in megabytes, so
drop the _m prefix from the size argument.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Current code allocates 16KiB for page tables. Whenever a new
table is needed an index is increased, but the bounds of this
index are never checked. Allocate the page tables dynamically
to solve this.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

find_pte is currently unused, nevertheless it's useful for debugging
purposes. Add a __maybe_unused to silence the compiler warning.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Instead of #ifdefing the correct pgtable header file to include,
include it where it's needed. Also, move the memory type attributes
into there consumers, namely the mmu.c files.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

This commit adds basic mmu support, ie:
        - DMA cache handling is not supported
        - Remapping memory region also

The current mmu setting is:
        - 4KB granularity
        - 3 level lookup (skipping L0)
        - 33 bits per VA

This is based on coreboot and u-boot mmu configuration.

Signed-off-by: Raphael Poggi <poggi.raph@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>