plat-ti: Cleanup platform configuration

Reorganize platform configuration to assist in addition of new
platforms. No functional changes.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Joa

plat-ti: Cleanup platform configuration

Reorganize platform configuration to assist in addition of new
platforms. No functional changes.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Joakim Bech <joakim.bech@linaro.org>

show more ...

core: mm: print memory type name instead of numerical value

Improve the legibility of the memory manager debug traces by
converting the memory types to strings before printing them in
dump_mmap_tabl

core: mm: print memory type name instead of numerical value

Improve the legibility of the memory manager debug traces by
converting the memory types to strings before printing them in
dump_mmap_table(), add_phys_mem() and add_va_space().

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Volodymyr Babchuk <vlad.babchuk@gmail.com>

show more ...

core_mmu_v7: Allow cache memory attributes to match non-SMP Linux

On non-SMP ARM Linux the default cache policy is inner/outer write-back,
no write-allocate not sharable. When compiled with SMP supp

core_mmu_v7: Allow cache memory attributes to match non-SMP Linux

On non-SMP ARM Linux the default cache policy is inner/outer write-back,
no write-allocate not sharable. When compiled with SMP support the
policy is updated to inner/outer write-back with write-allocate sharable.

OP-TEE makes the assumption that SMP will be enabled, allow overriding
this for the non-SMP cases.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core_mmu_v7: Rename index to normal cached memory

The index into cache attribute registers for device memory is called
ATTR_DEVICE_INDEX, but the normal cached memory is referred to as
ATTR_IWBWA_OW

core_mmu_v7: Rename index to normal cached memory

The index into cache attribute registers for device memory is called
ATTR_DEVICE_INDEX, but the normal cached memory is referred to as
ATTR_IWBWA_OWBWA_INDEX, this implies the caching type. This is not
always the type of cache we will use. Rename it to a more generic
ATTR_NORMAL_CACHED_INDEX.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

plat-ti: Move TZDRAM area to better align with other DRAM uses

The area currently reserved for OP-TEE overlaps an area that is used
by another existing device use-case, move OP-TEE to a non-interfer

plat-ti: Move TZDRAM area to better align with other DRAM uses

The area currently reserved for OP-TEE overlaps an area that is used
by another existing device use-case, move OP-TEE to a non-interfering
address.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: assert no null physical address is used in core static mapping

Current implementation of core mapping assumes value 0 denotes an
invalid physical address. Hence this change asserts (in debug m

core: assert no null physical address is used in core static mapping

Current implementation of core mapping assumes value 0 denotes an
invalid physical address. Hence this change asserts (in debug mode)
that no null physical address is used in the core static mapping.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: skip registered physical memory with a null size

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

core: debug trace the number of xlat tables used

These debug traces can be quite handy to monitor the number of
translation tables effectively used at runtime.

Signed-off-by: Etienne Carriere <etie

core: debug trace the number of xlat tables used

These debug traces can be quite handy to monitor the number of
translation tables effectively used at runtime.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: minor cleanup in core_mmu_lpae.c

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

core: more flexible static memory mapping

This change allows remove dependency of the static mapping
description on memory type IDs ordering and the requirements
for such memory types to match uniqu

core: more flexible static memory mapping

This change allows remove dependency of the static mapping
description on memory type IDs ordering and the requirements
for such memory types to match unique memories. This change
will be required by a later change that will define several
memory types for the TEE RAM (read-only/executable, read-only,
read/write).

The setup of the static memory mapping array is somewhat changed:
- 1st step: fill the array from registered "phys_mem" (unchanged)
- 2nd step: assign "region size" to arrays cells (unchanged)
- 3rd step: sort array by region size. This allows to group
small page mappings together to prevent wasting xlat tables.
- 4th step: for non LPAE mapping only: separate secure mapping
from non-secure mapping for the small page mapped areas.
- 5th step: assign virtual addresses for flat-mapped areas.
This sequence saves the flat mapped areas base address which is
used in the next step to assign non-flat mapped virtual memories.
- 6th step: assign virtual addresses for the non-flat mapped areas.
This step considers the cases where non-flat mapped areas are
mapped before or after flat mapped areas as done before this
change.
- a last, yet not mandatory step sorts the static mapping
description array by virtual address ranges.

With this change, the static mapping does not expects memory type
ID value ordering reflect any ordering of the areas virtual address
ranges. Instead, the memory type IDs only reflect the expected
mapping attributes which drive how virtual mapping is built.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: default define stack alignment and core vmem size

Default define CFG_TEE_RAM_VA_SIZE if not defined from platform.

As a side effect of bringing a default value for CFG_TEE_RAM_VA_SIZE
into co

core: default define stack alignment and core vmem size

Default define CFG_TEE_RAM_VA_SIZE if not defined from platform.

As a side effect of bringing a default value for CFG_TEE_RAM_VA_SIZE
into core_mmu.h and thus including 'platform_config.h', the macro
STACK_ALIGNMENT defined in user_ta.c must not conflict with the macro
defined by the platform. Hence this change also default defines
STACK_ALIGNMENT if not defined from platform.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: debug trace for non-LPAE mapping

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

core: fix non-LPAE mapping against unmapped areas

Core defines some virtual memory than should not be default
mapped. Yet core_mmu_v7.c loads a non null descriptor in MMU
tables for such memories: a

core: fix non-LPAE mapping against unmapped areas

Core defines some virtual memory than should not be default
mapped. Yet core_mmu_v7.c loads a non null descriptor in MMU
tables for such memories: attributes are null (which makes
the page effectively not mapped) but a meaningless non null
physical page address is loaded. This change

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: move place where TEE_RAM mapping region size is forced

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

core: move mmap trace into a specific routine

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

core: fix memory mapping overlapping sequence

Before this patch, the overlapping memory ranges where detected
only if the lower physical range was registered before the
higher overlapping physical r

core: fix memory mapping overlapping sequence

Before this patch, the overlapping memory ranges where detected
only if the lower physical range was registered before the
higher overlapping physical range.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: fix plat-stm iomem mapping

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

core: remove useless rounding for registered io memory

This change removes useless rounding on registered io memory for
the platforms maintained by Linaro.

Also remove registering of GIC iomem on p

core: remove useless rounding for registered io memory

This change removes useless rounding on registered io memory for
the platforms maintained by Linaro.

Also remove registering of GIC iomem on plat-mediatek as the platform
does not use the GIC resources.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: deprecate DEVICEx_TYPE/_PA_BASE/_SIZE

Macros DEVICEx_TYPE, DEVICEx_PA_BASE and DEVICEx__SIZE used to
help platform to register their address range mapping requirements.
These are now deprecate

core: deprecate DEVICEx_TYPE/_PA_BASE/_SIZE

Macros DEVICEx_TYPE, DEVICEx_PA_BASE and DEVICEx__SIZE used to
help platform to register their address range mapping requirements.
These are now deprecated since platform should use the more flexible
register_phys_mem() macro.

This change removes all occurrences of DEVICEx_TYPE/_PA_BASE/_SIZE
and use the register_phys_mem() instead.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: fix register_XXX_mem() against physical address

Use __COUNTER__ instead of the registered physical address to
generate the label of the structure defined by the macros
__register_phys_mem() an

core: fix register_XXX_mem() against physical address

Use __COUNTER__ instead of the registered physical address to
generate the label of the structure defined by the macros
__register_phys_mem() and __register_sdp_mem().

Before this change, when argument "addr" is used, one cannot use these
macros providing an address that is the result of a local operation.

I.e This implementation was not possible:
__register_phys_mem(<any-id>, ROUNDUP(<addr>, <value>), <size>);
and one needed to use a temporary macro for the address computation:
#define MY_BASE_ADDRESS ROUNDUP(<addr>, <value>)
__register_phys_mem(<any-id>, MY_BASE_ADDRESS, <size>);

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

tee_mm: add locking while working with pool->entry

This fixes race condition in linked list handling.

Signed-off-by: Volodymyr Babchuk <vlad.babchuk@gmail.com>
Reviewed-by: Jens Wiklander <jens.wik

tee_mm: add locking while working with pool->entry

This fixes race condition in linked list handling.

Signed-off-by: Volodymyr Babchuk <vlad.babchuk@gmail.com>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>

show more ...

spinlock: add cpu_spin_lock_xsave()/xrestore() functions

These are functions like spin_lock_irqsave() functions in linux kernel.
They should replace xxx_lock() functions in different modules.

Signe

spinlock: add cpu_spin_lock_xsave()/xrestore() functions

These are functions like spin_lock_irqsave() functions in linux kernel.
They should replace xxx_lock() functions in different modules.

Signed-off-by: Volodymyr Babchuk <vlad.babchuk@gmail.com>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>

show more ...

core: increase size of the exception stack in AArch32 mode

'stack_abt' is too small when pager is enabled and needs to be increased.
'stack_abt' is used for aborts (prefetch/data/undef) and native/f

core: increase size of the exception stack in AArch32 mode

'stack_abt' is too small when pager is enabled and needs to be increased.
'stack_abt' is used for aborts (prefetch/data/undef) and native/foreign
interruptions exceptions.

Tests on ARMv7 targets show that between 1000 and 1500 bytes of these
stacks are consumed at runtime. Tests considered configurations with
trace/debug/LPAE/VFP enabled and disabled. Tests were based on running
OP-TEE with the xtest non-regression suite.

Signed-off-by: Etienne Carriere <etienne.carriere@st.com>
Tested-by: Etienne Carriere <etienne.carriere@st.com> (qemu_virt/b2260)
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core_mmu: round VAs to region size

Entries in memory map are sorted by address, not by region size.
So, if entries have region size 2M, 4k, 2M, ... then all
maps after second one will be misaligned.

core_mmu: round VAs to region size

Entries in memory map are sorted by address, not by region size.
So, if entries have region size 2M, 4k, 2M, ... then all
maps after second one will be misaligned. Thus, we need to
explicitly round virtual addresses.

Signed-off-by: Volodymyr Babchuk <vlad.babchuk@gmail.com>
Reviewed-by: Jens Wiklander <jens.wiklander@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>
Tested-by: Etienne Carriere <etienne.carriere@linaro.org> (qemu_virt)

show more ...

core: fix compile error with CFG_WITH_USER_TA=n

Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>
Fixes: 040bc0f04138 ("core: add test case for hash-tree")
Signed-off-by: Jens Wiklander <j

core: fix compile error with CFG_WITH_USER_TA=n

Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>
Fixes: 040bc0f04138 ("core: add test case for hash-tree")
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...