stm32mp1: release constraint on core count

Minor cleaning and allow CFG_TEE_CORE_NB_CORE to be set to 1.

Signed-off-by: Etienne Carriere <etienne.carriere@st.com>
Acked-by: Jens Wiklander <jens.wik

stm32mp1: release constraint on core count

Minor cleaning and allow CFG_TEE_CORE_NB_CORE to be set to 1.

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

show more ...

core: serial8250_uart_dev_init(): use calloc()

Allocate pl011_data with calloc() instead of malloc() get initialized
memory.

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Fixes: d2769

core: serial8250_uart_dev_init(): use calloc()

Allocate pl011_data with calloc() instead of malloc() get initialized
memory.

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Fixes: d276907c8c4f ("core: drivers: serial8250_uart: Add DT support")
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: pl011_dev_alloc(): use calloc()

Allocate pl011_data with calloc() instead of malloc() get initialized
memory.

Without this could pd->base.va contain garbage when pl011_init() is
called.

Revi

core: pl011_dev_alloc(): use calloc()

Allocate pl011_data with calloc() instead of malloc() get initialized
memory.

Without this could pd->base.va contain garbage when pl011_init() is
called.

Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Fixes: ddf45954360c ("pl011: dt: Add DT support")
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

documentation: Add some text to describe DTB overlays to optee_design

This patch adds some documentary text to optee_design.md describing how
OPTEE DTB overlay works.

Signed-off-by: Bryan O'Donoghu

documentation: Add some text to describe DTB overlays to optee_design

This patch adds some documentary text to optee_design.md describing how
OPTEE DTB overlay works.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>

show more ...

core: generic_boot: Add init_dt_overlay()

This patch adds init_dt_overlay() into to the DT boot flow.

init_dt_overlay() operates in one of two ways.

1. By appending to a DT overlay passed from a p

core: generic_boot: Add init_dt_overlay()

This patch adds init_dt_overlay() into to the DT boot flow.

init_dt_overlay() operates in one of two ways.

1. By appending to a DT overlay passed from a prior boot stage such as ATF.
In this case OPTEE DT nodes will be appended to the existing DT overlay.

2. By creating an OPTEE specific DT overlay at CFG_DT_ADDR.

A subsequent boot phase must then pick up the DT overlay in-memory and
merge that overlay into a main DTB structure.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>

show more ...

core: generic_boot: Add device-tree fragments

This patch adds a call into add_dt_overlay_fragment() at
add_dt_path_subnode() affecting at the time of writing

- /firmware/optee
- /psci
- /memory/res

core: generic_boot: Add device-tree fragments

This patch adds a call into add_dt_overlay_fragment() at
add_dt_path_subnode() affecting at the time of writing

- /firmware/optee
- /psci
- /memory/reserved

if CFG_EXTERNAL_DTB_OVERLAY is defined then the set of OPTEE DTB entries is
treated as a set of DT overlay fragments to be populated into the specified
address at CFG_DTB_ADDR.

This allows ATF or u-boot to pass a blank DTB instead of a populated DTB
into OPTEE and for OPTEE then to return back a DTB with overlay fragments.

A subsequent boot stage can then merge the OPTEE provided overlay into the
main DTB.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: generic_boot: Add add_dt_overlay_fragment()

This patch adds a dt routine add_dt_overlay_fragment(). This purpose of
which is to encapsulate the dynamic FDT node entries OPTEE provides inside
o

core: generic_boot: Add add_dt_overlay_fragment()

This patch adds a dt routine add_dt_overlay_fragment(). This purpose of
which is to encapsulate the dynamic FDT node entries OPTEE provides inside
of a

fragment@0 {
target-path = "/";
__overlay__ {
/* OPTEE nodes go here */
};
};

A subsequent set of patches will wrapper up existing dynamic OPTEE nodes

- /firmware/optee
- /psci
- /reserved/memory
optee@0xaddress

Once done it will be possible for a DTB in memory to be populated
with OPTEE fragments and for a subsequent bootloader to merge the OPTEE
populated overlay into a DTB loaded by a later stage.

if CFG_EXTERNAL_DTB_OVERLAY is not defined then this code has no effect.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: generic_boot: Utilize add_dt_path_subnode where appropriate

This patch replaces some repetitive code of the form

offs = fdt_path_offset(fdt, "path");
if (offs < 0)
return -1;
offs = fdt_add_

core: generic_boot: Utilize add_dt_path_subnode where appropriate

This patch replaces some repetitive code of the form

offs = fdt_path_offset(fdt, "path");
if (offs < 0)
return -1;
offs = fdt_add_subnode(fdt, offs, "subnode");

with

offs = add_dt_path_subnode(fdt, "path", "subnode");

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: generic_boot: Add helper add_dt_path_subnode()

A common pattern in the DT code is

offs = fdt_path_offset(fdt, "/");
if (offs < 0)
return -1;
offs = fdt_add_subnode(fdt, offs, "newnode");
if

core: generic_boot: Add helper add_dt_path_subnode()

A common pattern in the DT code is

offs = fdt_path_offset(fdt, "/");
if (offs < 0)
return -1;
offs = fdt_add_subnode(fdt, offs, "newnode");
if (offs < 0)
return -1;

as was pointed out by Jerome in a related PR to-do with adding overlays
this is a candidate for functional decomposition.

This patch adds the necessary helper function.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Suggested-by: Jerome Forissier <jerome.forissier@linaro.org>

show more ...

TA: export CFG_TEE_TA_LOG_LEVEL to dev-kit

Exports CFG_TEE_TA_LOG_LEVEL to TA dev-kit. It can still be overridden
when compiling the TA, but it makes sense to default to the value used
when compilin

TA: export CFG_TEE_TA_LOG_LEVEL to dev-kit

Exports CFG_TEE_TA_LOG_LEVEL to TA dev-kit. It can still be overridden
when compiling the TA, but it makes sense to default to the value used
when compiling the dev-kit.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: base TA store on scatter array

The TA store interface is now implemented using scatter array.

The priorities of the different TA store are also changed to use 2, 4
and 9 instead or 5, 9, 10 a

core: base TA store on scatter array

The TA store interface is now implemented using scatter array.

The priorities of the different TA store are also changed to use 2, 4
and 9 instead or 5, 9, 10 as they are now sorted alphabetically. The new
allocation makes it easier to stick something between "Secure Storage
TA" and the "REE" stores. It's doubtful that anyone would ever want to
make something higher priority than the "early TA" store, but just in
case a few numbers are reserved.

Acked-by: Etienne Carriere <etienne.carriere@linaro.org>
Tested-by: Jerome Forissier <jerome.forissier@linaro.org> (HiKey960)
Reviewed-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: base memory registration on scatter array

The register_*() macros are now implemented using scatter array.

Acked-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jerome Forissi

core: base memory registration on scatter array

The register_*() macros are now implemented using scatter array.

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

show more ...

core: base pseudo_ta_register() on scatter array

The pseudo_ta_register() implementation is now based on scatter array.

Acked-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jerome

core: base pseudo_ta_register() on scatter array

The pseudo_ta_register() implementation is now based on scatter array.

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

show more ...

core: initcall.h use scattered array

Initcalls uses generic scattered array instead of special
implementation.

Acked-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Jerome Forissier

core: initcall.h use scattered array

Initcalls uses generic scattered array instead of special
implementation.

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

show more ...

core: add scattered array

Adds a scattered array which allows defining arrays scattered over
several source files. The implementation is based on some support by the
linker.

This is a generic solut

core: add scattered array

Adds a scattered array which allows defining arrays scattered over
several source files. The implementation is based on some support by the
linker.

This is a generic solution to initcalls and other similar scattered
initializations.

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

show more ...

core: keep.h: avoid duplicate symbols

Prior to this patch if the KEEP_*() macros are used with two static
variables with the same name but in different source files it will
result in a duplicated gl

core: keep.h: avoid duplicate symbols

Prior to this patch if the KEEP_*() macros are used with two static
variables with the same name but in different source files it will
result in a duplicated global symbol. This happens because the internal
trickery uses global variables based on the static symbol it's supposed
to keep. With this patch the global variables uses the file unique
define __FILE_ID__ to avoid the conflict.

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

show more ...

mk/compile.mk: define __FILE_ID__ for all files

Defines __FILE_ID__ with a unique name of the file being compiled.
Useful to create globally unique variables.

Acked-by: Etienne Carriere <etienne.ca

mk/compile.mk: define __FILE_ID__ for all files

Defines __FILE_ID__ with a unique name of the file being compiled.
Useful to create globally unique variables.

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

show more ...

core: asan: support GCC version >= 7.0

With GCC 7.0 there was an ABI change for the address sanitizer. This
patch extends struct asan_global with the new odr_indicator field to
work with this new AB

core: asan: support GCC version >= 7.0

With GCC 7.0 there was an ABI change for the address sanitizer. This
patch extends struct asan_global with the new odr_indicator field to
work with this new ABI.

Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

documentation: description of core/include/

Add missing description of the file-tree content of directory
core/include.

Sub-directory core/include/dt-bindings is mentioned even if
currently empty a

documentation: description of core/include/

Add missing description of the file-tree content of directory
core/include.

Sub-directory core/include/dt-bindings is mentioned even if
currently empty as it is where the device tree bindings header
files should be located.

Remove deprecated core/arch/user_mode description.

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

show more ...

core: embedded secure device tree

This change introduces configurations CFG_EMDED_DTB and
CFG_EMBED_DTB_SOURCE_FILE. When CFG_EMDED_DTB=y a device
tree blob (DTB) is embedded in a read-only section

core: embedded secure device tree

This change introduces configurations CFG_EMDED_DTB and
CFG_EMBED_DTB_SOURCE_FILE. When CFG_EMDED_DTB=y a device
tree blob (DTB) is embedded in a read-only section of the
core based on an in-tree device tree source (DTS) file.

CFG_EMBED_DTS_SOURCE_FILE defines the relative path of
the target device in core/arch/$(ARCH)/dts.

Non empty CFG_EMBED_DTS_SOURCE_FILE content implies CFG_EMBED_DTB=y.
CFG_EMBED_DTB=y mandates CFG_EMBED_DTS_SOURCE_FILE definition.
CFG_EMDED_DTB=y mandates CFG_DT=y.

Since the embedded DTB is read-only, core do not attempt to
modify it adding information such as OP-TEE resources nodes and
properties. Core still get generic information such as system
memory address range and debug console configuration
from the embedded DTB.

Documentation includes a DT section in the OP-TEE design
description.

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

show more ...

scripts: bin_to_c.py

bin_to_c.py generates an ANSI-C source file from a binary image file.
The C source defines an unsigned byte array that stores the binary
data with a 64bit alignment constraint.

scripts: bin_to_c.py

bin_to_c.py generates an ANSI-C source file from a binary image file.
The C source defines an unsigned byte array that stores the binary
data with a 64bit alignment constraint.

The label of the array is provided as input argument to the script.

Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Acked-by: Jerome.Forissier <jerome.forissier@linaro.org>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

core: helper for device tree compilation

Add helper function gen-dtb-file in the build sequence. It generates
a DTB file from input DTS file. The DTS file is preprocessed with
CPP before being compi

core: helper for device tree compilation

Add helper function gen-dtb-file in the build sequence. It generates
a DTB file from input DTS file. The DTS file is preprocessed with
CPP before being compiled using the device tree compiler (DTC).

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

show more ...

core: fix bug in vm_map() causing pager panic()

Prior to this patch vm_map() didn't allocate new page tables (PGTs) as
needed, it only checked that it later would be possible to allocate the
needed

core: fix bug in vm_map() causing pager panic()

Prior to this patch vm_map() didn't allocate new page tables (PGTs) as
needed, it only checked that it later would be possible to allocate the
needed PGTs. This is enough if the user_ta_ctx (UTC) isn't active. With
dynamically linked libraries the UTC will be active when vm_map() is
called to make room for the new segments. If the already allocated PGTs
happen to cover even the new memory range it will still work, this is
normally the case with CFG_WITH_LPAE=y since each page table covers 2
MiB. With CFG_WITH_LPAE=n the page tables only covers 1 MiB and that's
not enough when loading the os_test TA in xtest case 1006.

This patch fixes the problem by instead of just checking that it later
will be possible to allocate needed PGTs, it also allocates the PGTs if
the UTC is active.

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

show more ...

core: remove MOBJ_INVALID_COOKIE

Removes MOBJ_INVALID_COOKIE which resulted in an unexpected ABI change
against the normal world driver. Instead 0 is continued to be used as an
invalid/absent cookie

core: remove MOBJ_INVALID_COOKIE

Removes MOBJ_INVALID_COOKIE which resulted in an unexpected ABI change
against the normal world driver. Instead 0 is continued to be used as an
invalid/absent cookie value.

Tested-by: Sumit Garg <sumit.garg@linaro.org>
Acked-by: Jerome Forissier <jerome.forissier@linaro.org>
Reported-by: Sumit Garg <sumit.garg@linaro.org>
Fixes: cd278f78382b ("core: simplify shm cookie handling")
Signed-off-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...

lib.mk: centralize profiling flag (-pg)

Code cleanup, no functional change. This commit avoids the duplication of
the -pg flag in the library makefiles.

Signed-off-by: Jerome Forissier <jerome.fori

lib.mk: centralize profiling flag (-pg)

Code cleanup, no functional change. This commit avoids the duplication of
the -pg flag in the library makefiles.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Acked-by: Jens Wiklander <jens.wiklander@linaro.org>

show more ...