GitLab

Many asserts depend on code that is conditionally compiled based on the
DEBUG define. This patch modifies the conditional inclusion of such code
so that it is based on the ENABLE_ASSERTIONS build option.

Change-Id: I6406674788aa7e1ad7c23d86ce94482ad3c382bd
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

The build option `ENABLE_ASSERTIONS` should be used instead. That way
both C and ASM assertions can be enabled or disabled together.

All occurrences of `ASM_ASSERTION` in common code and ARM platforms have
been replaced by `ENABLE_ASSERTIONS`.

ASM_ASSERTION has been removed from the user guide.

Change-Id: I51f1991f11b9b7ff83e787c9a3270c274748ec6f
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Add the new build option `ENABLE_ASSERTIONS` that controls whether or
not assert functions are compiled out. It defaults to 1 for debug builds
and to 0 for release builds.

Additionally, a following patch will be done to allow this build option
to hide auxiliary code used for the checks done in an `assert()`. This
code is is currently under the DEBUG build flag.

Assert messages are now only printed if LOG_LEVEL >= LOG_LEVEL_INFO,
which is the default for debug builds.

This patch also updates the User Guide.

Change-Id: I1401530b56bab25561bb0f274529f1d12c5263bc
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Introduce new build option ENABLE_STACK_PROTECTOR. It enables
compilation of all BL images with one of the GCC -fstack-protector-*
options.

A new platform function plat_get_stack_protector_canary() is introduced.
It returns a value that is used to initialize the canary for stack
corruption detection. Returning a random value will prevent an attacker
from predicting the value and greatly increase the effectiveness of the
protection.

A message is printed at the ERROR level when a stack corruption is
detected.

To be effective, the global data must be stored at an address
lower than the base of the stacks. Failure to do so would allow an
attacker to overwrite the canary as part of an attack which would void
the protection.

FVP implementation of plat_get_stack_protector_canary is weak as
there is no real source of entropy on the FVP. It therefore relies on a
timer's value, which could be predictable.

Change-Id: Icaaee96392733b721fa7c86a81d03660d3c1bc06
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>

The desired behaviour is to call `plat_panic_handler()`, and to use
`no_ret` to do so from ASM.

Change-Id: I88b2feefa6e6c8f9bf057fd51ee0d2e9fb551e4f
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Call console_flush() before execution either terminates or leaves an
exception level.

Fixes: ARM-software/tf-issues#123

Change-Id: I64eeb92effb039f76937ce89f877b68e355588e3
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

This patch adds an additional flag `XLAT_TABLE_NC` which marks the
translation tables as Non-cacheable for MMU accesses.

Change-Id: I7c28ab87f0ce67da237fadc3627beb6792860fd4
Signed-off-by: Summer Qin <summer.qin@arm.com>

ARM erratum 855873 applies to all Cortex-A53 CPUs.
The recommended workaround is to promote "data cache clean"
instructions to "data cache clean and invalidate" instructions.
For core revisions of r0p3 and later this can be done by setting a bit
in the CPUACTLR_EL1 register, so that hardware takes care of the promotion.
As CPUACTLR_EL1 is both IMPLEMENTATION DEFINED and can be trapped to EL3,
we set the bit in firmware.
Also we dump this register upon crashing to provide more debug
information.

Enable the workaround for the Juno boards.

Change-Id: I3840114291958a406574ab6c49b01a9d9847fec8
Signed-off-by: Andre Przywara <andre.przywara@arm.com>

ge, lt, gt and le condition codes in assembly provide a signed test
whereas hs, lo, hi and ls provide the unsigned counterpart. Signed tests
should only be used when strictly necessary, as using them on logically
unsigned values can lead to inverting the test for high enough values.
All offsets, addresses and usually counters are actually unsigned
values, and should be tested as such.

Replace the occurrences of signed condition codes where it was
unnecessary by an unsigned test as the unsigned tests allow the full
range of unsigned values to be used without inverting the result with
some large operands.

Change-Id: I58b7e98d03e3a4476dfb45230311f296d224980a
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>

TLBI instructions for EL3 won't have the desired effect under specific
circumstances in Cortex-A57 r0p0. The workaround is to execute DSB and
TLBI twice each time.

Even though this errata is only needed in r0p0, the current errata
framework is not prepared to apply run-time workarounds. The current one
is always applied if compiled in, regardless of the CPU or its revision.

This errata has been enabled for Juno.

The `DSB` instruction used when initializing the translation tables has
been changed to `DSB ISH` as an optimization and to be consistent with
the barriers used for the workaround.

Change-Id: Ifc1d70b79cb5e0d87e90d88d376a59385667d338
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Added APIs to add and remove regions to the translation tables
dynamically while the MMU is enabled. Only static regions are allowed
to overlap other static ones (for backwards compatibility).

A new private attribute (MT_DYNAMIC / MT_STATIC) has been added to
flag each region as such.

The dynamic mapping functionality can be enabled or disabled when
compiling by setting the build option PLAT_XLAT_TABLES_DYNAMIC to 1
or 0. This can be done per-image.

TLB maintenance code during dynamic table mapping and unmapping has
also been added.

Fixes ARM-software/tf-issues#310

Change-Id: I19e8992005c4292297a382824394490c5387aa3b
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

The printed output has been improved in two ways:

- Whenever multiple invalid descriptors are found, only the first one
  is printed, and a line is added to inform about how many descriptors
  have been omitted.

- At the beginning of each line there is an indication of the table
  level the entry belongs to. Example of the new output:
  `[LV3] VA:0x1000 PA:0x1000 size:0x1000 MEM-RO-S-EXEC`

Change-Id: Ib6f1cd8dbd449452f09258f4108241eb11f8d445
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

The folder lib/xlat_tables_v2 has been created to store a new version
of the translation tables library for further modifications in patches
to follow. At the moment it only contains a basic implementation that
supports static regions.

This library allows different translation tables to be modified by
using different 'contexts'. For now, the implementation defaults to
the translation tables used by the current image, but it is possible
to modify other tables than the ones in use.

Added a new API to print debug information for the current state of
the translation tables, rather than printing the information while
the tables are being created. This allows subsequent debug printing
of the xlat tables after they have been changed, which will be useful
when dynamic regions are implemented in a patch to follow.

The common definitions stored in `xlat_tables.h` header have been moved
to a new file common to both versions, `xlat_tables_defs.h`.

All headers related to the translation tables library have been moved to
a the subfolder `xlat_tables`.

Change-Id: Ia55962c33e0b781831d43a548e505206dffc5ea9
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

This patch fixes a compilation issue with bakery locks when
PSCI library is compiled with USE_COHERENT_MEM = 0 build option.

Change-Id: Ic7f6cf9f2bb37f8a946eafbee9cbc3bf0dc7e900
Signed-off-by: Soby Mathew <soby.mathew@arm.com>

The current PSCI implementation can apply certain optimizations upon the
assumption that all PSCI participants are cache-coherent.

  - Skip performing cache maintenance during power-up.

  - Skip performing cache maintenance during power-down:

    At present, on the power-down path, CPU driver disables caches and
    MMU, and performs cache maintenance in preparation for powering down
    the CPU. This means that PSCI must perform additional cache
    maintenance on the extant stack for correct functioning.

    If all participating CPUs are cache-coherent, CPU driver would
    neither disable MMU nor perform cache maintenance. The CPU being
    powered down, therefore, remain cache-coherent throughout all PSCI
    call paths. This in turn means that PSCI cache maintenance
    operations are not required during power down.

  - Choose spin locks instead of bakery locks:

    The current PSCI implementation must synchronize both cache-coherent
    and non-cache-coherent participants. Mutual exclusion primitives are
    not guaranteed to function on non-coherent memory. For this reason,
    the current PSCI implementation had to resort to bakery locks.

    If all participants are cache-coherent, the implementation can
    enable MMU and data caches early, and substitute bakery locks for
    spin locks. Spin locks make use of architectural mutual exclusion
    primitives, and are lighter and faster.

The optimizations are applied when HW_ASSISTED_COHERENCY build option is
enabled, as it's expected that all PSCI participants are cache-coherent
in those systems.

Change-Id: Iac51c3ed318ea7e2120f6b6a46fd2db2eae46ede
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>

The PSCI implementation performs cache maintenance operations on its
data structures to ensure their visibility to both cache-coherent and
non-cache-coherent participants. These cache maintenance operations
can be skipped if all PSCI participants are cache-coherent. When
HW_ASSISTED_COHERENCY build option is enabled, we assume PSCI
participants are cache-coherent.

For usage abstraction, this patch introduces wrappers for PSCI cache
maintenance and barrier operations used for state coordination: they are
effectively NOPs when HW_ASSISTED_COHERENCY is enabled, but are
applied otherwise.

Also refactor local state usage and associated cache operations to make
it clearer.

Change-Id: I77f17a90cba41085b7188c1345fe5731c99fad87
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>

This patch removes unnecessary `isb` from the enable DCO sequence as
there is no need to synchronize this operation.

Change-Id: I0191e684bbc7fdba635c3afbc4e4ecd793b6f06f
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>

This patch moves the code to disable DCO operations out from common
CPU files. This allows the platform code to call thsi API as and
when required. There are certain CPU power down states which require
the DCO to be kept ON and platforms can decide selectively now.

Change-Id: Icb946fe2545a7d8c5903c420d1ee169c4921a2d1
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>

The errata is enabled by default on r0p4, which is confusing given that
we state we do not enable errata by default.

This patch clarifies this sentence by saying it is enabled in hardware
by default.

Change-Id: I70a062d93e1da2416d5f6d5776a77a659da737aa
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>

This patch adds support for all variants of the Denver CPUs. The
variants export their cpu_ops to allow all Denver platforms to run
the Trusted Firmware stack.

Change-Id: I1488813ddfd506ffe363d8a32cda1b575e437035
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>

The ARMv8v.1 architecture extension has introduced support for far
atomics, which includes compare-and-swap. Compare and Swap instruction
is only available for AArch64.

Introduce build options to choose the architecture versions to target
ARM Trusted Firmware:

  - ARM_ARCH_MAJOR: selects the major version of target ARM
    Architecture. Default value is 8.

  - ARM_ARCH_MINOR: selects the minor version of target ARM
    Architecture. Default value is 0.

When:

  (ARM_ARCH_MAJOR > 8) || ((ARM_ARCH_MAJOR == 8) && (ARM_ARCH_MINOR >= 1)),

for AArch64, Compare and Swap instruction is used to implement spin
locks. Otherwise, the implementation falls back to using
load-/store-exclusive instructions.

Update user guide, and introduce a section in Firmware Design guide to
summarize support for features introduced in ARMv8 Architecture
Extensions.

Change-Id: I73096a0039502f7aef9ec6ab3ae36680da033f16
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>

Perform stat accounting for retention/standby states also when
requested at multiple power levels.

Change-Id: I2c495ea7cdff8619bde323fb641cd84408eb5762
Signed-off-by: dp-arm <dimitris.papastamos@arm.com>

This patch introduces the following three platform interfaces:

* void plat_psci_stat_accounting_start(const psci_power_state_t *state_info)

  This is an optional hook that platforms can implement in order
  to perform accounting before entering a low power state.  This
  typically involves capturing a timestamp.

* void plat_psci_stat_accounting_stop(const psci_power_state_t *state_info)

  This is an optional hook that platforms can implement in order
  to perform accounting after exiting from a low power state.  This
  typically involves capturing a timestamp.

* u_register_t plat_psci_stat_get_residency(unsigned int lvl,
	const psci_power_state_t *state_info,
	unsigned int last_cpu_index)

  This is an optional hook that platforms can implement in order
  to calculate the PSCI stat residency.

If any of these interfaces are overridden by the platform, it is
recommended that all of them are.

By default `ENABLE_PSCI_STAT` is disabled.  If `ENABLE_PSCI_STAT`
is set but `ENABLE_PMF` is not set then an alternative PSCI stat
collection backend must be provided.  If both are set, then default
weak definitions of these functions are provided, using PMF to
calculate the residency.

NOTE: Previously, platforms did not have to explicitly set
`ENABLE_PMF` since this was automatically done by the top-level
Makefile.

Change-Id: I17b47804dea68c77bc284df15ee1ccd66bc4b79b
Signed-off-by: dp-arm <dimitris.papastamos@arm.com>

Replace all use of memset by zeromem when zeroing moderately-sized
structure by applying the following transformation:
memset(x, 0, sizeof(x)) => zeromem(x, sizeof(x))

As the Trusted Firmware is compiled with -ffreestanding, it forbids the
compiler from using __builtin_memset and forces it to generate calls to
the slow memset implementation. Zeromem is a near drop in replacement
for this use case, with a more efficient implementation on both AArch32
and AArch64.

Change-Id: Ia7f3a90e888b96d056881be09f0b4d65b41aa79e
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>

Introduce zeromem_dczva function on AArch64 that can handle unaligned
addresses and make use of DC ZVA instruction to zero a whole block at a
time. This zeroing takes place directly in the cache to speed it up
without doing external memory access.

Remove the zeromem16 function on AArch64 and replace it with an alias to
zeromem. This zeromem16 function is now deprecated.

Remove the 16-bytes alignment constraint on __BSS_START__ in
firmware-design.md as it is now not mandatory anymore (it used to comply
with zeromem16 requirements).

Change the 16-bytes alignment constraints in SP min's linker script to a
8-bytes alignment constraint as the AArch32 zeromem implementation is now
more efficient on 8-bytes aligned addresses.

Introduce zero_normalmem and zeromem helpers in platform agnostic header
that are implemented this way:
* AArch32:
	* zero_normalmem: zero using usual data access
	* zeromem: alias for zero_normalmem
* AArch64:
	* zero_normalmem: zero normal memory  using DC ZVA instruction
	                  (needs MMU enabled)
	* zeromem: zero using usual data access

Usage guidelines: in most cases, zero_normalmem should be preferred.

There are 2 scenarios where zeromem (or memset) must be used instead:
* Code that must run with MMU disabled (which means all memory is
  considered device memory for data accesses).
* Code that fills device memory with null bytes.

Optionally, the following rule can be applied if performance is
important:
* Code zeroing small areas (few bytes) that are not secrets should use
  memset to take advantage of compiler optimizations.

  Note: Code zeroing security-related critical information should use
  zero_normalmem/zeromem instead of memset to avoid removal by
  compilers' optimizations in some cases or misbehaving versions of GCC.

Fixes ARM-software/tf-issues#408

Change-Id: Iafd9663fc1070413c3e1904e54091cf60effaa82
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>

The errata reporting policy is as follows:

  - If an errata workaround is enabled:

    - If it applies (i.e. the CPU is affected by the errata), an INFO
      message is printed, confirming that the errata workaround has been
      applied.

    - If it does not apply, a VERBOSE message is printed, confirming
      that the errata workaround has been skipped.

  - If an errata workaround is not enabled, but would have applied had
    it been, a WARN message is printed, alerting that errata workaround
    is missing.

The CPU errata messages are printed by both BL1 (primary CPU only) and
runtime firmware on debug builds, once for each CPU/errata combination.

Relevant output from Juno r1 console when ARM Trusted Firmware is built
with PLAT=juno LOG_LEVEL=50 DEBUG=1:

  VERBOSE: BL1: cortex_a57: errata workaround for 806969 was not applied
  VERBOSE: BL1: cortex_a57: errata workaround for 813420 was not applied
  INFO:    BL1: cortex_a57: errata workaround for disable_ldnp_overread was applied
  WARNING: BL1: cortex_a57: errata workaround for 826974 was missing!
  WARNING: BL1: cortex_a57: errata workaround for 826977 was missing!
  WARNING: BL1: cortex_a57: errata workaround for 828024 was missing!
  WARNING: BL1: cortex_a57: errata workaround for 829520 was missing!
  WARNING: BL1: cortex_a57: errata workaround for 833471 was missing!
  ...
  VERBOSE: BL31: cortex_a57: errata workaround for 806969 was not applied
  VERBOSE: BL31: cortex_a57: errata workaround for 813420 was not applied
  INFO:    BL31: cortex_a57: errata workaround for disable_ldnp_overread was applied
  WARNING: BL31: cortex_a57: errata workaround for 826974 was missing!
  WARNING: BL31: cortex_a57: errata workaround for 826977 was missing!
  WARNING: BL31: cortex_a57: errata workaround for 828024 was missing!
  WARNING: BL31: cortex_a57: errata workaround for 829520 was missing!
  WARNING: BL31: cortex_a57: errata workaround for 833471 was missing!
  ...
  VERBOSE: BL31: cortex_a53: errata workaround for 826319 was not applied
  INFO:    BL31: cortex_a53: errata workaround for disable_non_temporal_hint was applied

Also update documentation.

Change-Id: Iccf059d3348adb876ca121cdf5207bdbbacf2aba
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>

Some side-channel attacks involve an attacker inferring something from
the time taken for a memory compare operation to complete, for example
when comparing hashes during image authentication. To mitigate this,
timingsafe_bcmp() must be used for such operations instead of the
standard memcmp().

This function executes in constant time and so doesn't leak any timing
information to the caller.

Change-Id: I470a723dc3626a0ee6d5e3f7fd48d0a57b8aa5fd
Signed-off-by: dp-arm <dimitris.papastamos@arm.com>
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

This code has been imported and slightly adapted from FreeBSD:
https://github.com/freebsd/freebsd/blob/6253393ad8df55730481bf2aafd76bdd6182e2f5/lib/libc/string/strnlen.c



Change-Id: Ie5ef5f92e6e904adb88f8628077fdf1d27470eb3
Signed-off-by: Sandrine Bailleux <sandrine.bailleux@arm.com>

One nasty part of ATF is some of boolean macros are always defined
as 1 or 0, and the rest of them are only defined under certain
conditions.

For the former group, "#if FOO" or "#if !FOO" must be used because
"#ifdef FOO" is always true.  (Options passed by $(call add_define,)
are the cases.)

For the latter, "#ifdef FOO" or "#ifndef FOO" should be used because
checking the value of an undefined macro is strange.

Here, IMAGE_BL* is handled by make_helpers/build_macro.mk like
follows:

  $(eval IMAGE := IMAGE_BL$(call uppercase,$(3)))

  $(OBJ): $(2)
          @echo "  CC      $$<"
          $$(Q)$$(CC) $$(TF_CFLAGS) $$(CFLAGS) -D$(IMAGE) -c $$< -o $$@

This means, IMAGE_BL* is defined when building the corresponding
image, but *undefined* for the other images.

So, IMAGE_BL* belongs to the latter group where we should use #ifdef
or #ifndef.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>

NOTE - this is patch does not address all occurrences of system
includes not being in alphabetical order, just this one case.

Change-Id: I3cd23702d69b1f60a4a9dd7fd4ae27418f15b7a3

Delete old version of libfdt at lib/libfdt. Move new libfdt API
headers to include/lib/libfdt and all other files to lib/libfdt.

Change-Id: I32b7888f1f20d62205310e363accbef169ad7b1b
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

* Add libfdt.mk helper makefile
* Remove unused libfdt files
* Minor changes to fdt.h and libfdt.h to make them C99 compliant

Adapted from 754d78b1

.

Change-Id: I0847f1c2e6e11f0c899b0b7ecc522c0ad7de210c
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Import libfdt code from https://git.kernel.org/cgit/utils/dtc/dtc.git


tag "v1.4.2" commit ec02b34c05be04f249ffaaca4b666f5246877dea.

This version includes commit d0b3ab0a0f46ac929b4713da46f7fdcd893dd3bd,
which fixes a buffer overflow in fdt_offset_ptr().

Change-Id: I05a30511ea68417ee7ff26477da3f99e0bd4e06b
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Various CPU drivers in ARM Trusted Firmware register functions to handle
power-down operations. At present, separate functions are registered to
power down individual cores and clusters.

This scheme operates on the basis of core and cluster, and doesn't cater
for extending the hierarchy for power-down operations. For example,
future CPUs might support multiple threads which might need powering
down individually.

This patch therefore reworks the CPU operations framework to allow for
registering power down handlers on specific level basis. Henceforth:

  - Generic code invokes CPU power down operations by the level
    required.

  - CPU drivers explicitly mention CPU_NO_RESET_FUNC when the CPU has no
    reset function.

  - CPU drivers register power down handlers as a list: a mandatory
    handler for level 0, and optional handlers for higher levels.

All existing CPU drivers are adapted to the new CPU operations framework
without needing any functional changes within.

Also update firmware design guide.

Change-Id: I1826842d37a9e60a9e85fdcee7b4b8f6bc1ad043
Signed-off-by: Jeenu Viswambharan <jeenu.viswambharan@arm.com>

Unsigned conditions should be used instead of signed ones when comparing
addresses or sizes in assembly.
Signed-off-by: Douglas Raillard <douglas.raillard@arm.com>
Change-Id: Id3bd9ccaf58c37037761af35ac600907c4bb0580

Testing showed that the time spent in a cluster power down
operation is dominated by cache flushes.  Add two more timestamps
in runtime instrumentation to keep track of the time spent
flushing the L1/L2 caches.

Change-Id: I4c5a04e7663543225a85d3c6b271d7b706deffc4
Signed-off-by: dp-arm <dimitris.papastamos@arm.com>

In AArch64, depending on the granularity of the translation tables,
level 0 and/or level 1 of the translation tables may not support block
descriptors, only table descriptors.

This patch introduces a check to make sure that, even if theoretically
it could be possible to create a block descriptor to map a big memory
region, a new subtable will be created to describe its mapping.

Change-Id: Ieb9c302206bfa33fbaf0cdc6a5a82516d32ae2a7
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Added the definitions `PLAT_PHY_ADDR_SPACE_SIZE` and
`PLAT_VIRT_ADDR_SPACE_SIZE` which specify respectively the physical
and virtual address space size a platform can use.

`ADDR_SPACE_SIZE` is now deprecated. To maintain compatibility, if any
of the previous defines aren't present, the value of `ADDR_SPACE_SIZE`
will be used instead.

For AArch64, register ID_AA64MMFR0_EL1 is checked to calculate the
max PA supported by the hardware and to verify that the previously
mentioned definition is valid. For AArch32, a 40 bit physical
address space is considered.

Added asserts to check for overflows.

Porting guide updated.

Change-Id: Ie8ce1da5967993f0c94dbd4eb9841fc03d5ef8d6
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

Each translation table level entry can only map a given virtual
address onto physical addresses of the same granularity. For example,
with the current configuration, a level 2 entry maps blocks of 2 MB,
so the physical address must be aligned to 2 MB. If the address is not
aligned, the MMU will just ignore the lower bits.

This patch adds an assertion to make sure that physical addresses are
always aligned to the correct boundary.

Change-Id: I0ab43df71829d45cdbe323301b3053e08ca99c2c
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>

There is no guarantee on the signedness of char.  It can be either
signed or unsigned.  On ARM it is unsigned and hence this memcmp()
implementation works as intended.

On other machines, char can be signed (x86 for example).  In that case
(and assuming a 2's complement implementation), interpreting a
bit-pattern of 0xFF as signed char can yield -1.  If *s1 is 0 and *s2
is 255 then the difference *s1 - *s2 should be negative.  The C
integer promotion rules guarantee that the unsigned chars will be
converted to int before the operation takes place.  The current
implementation will return a positive value (0 - (-1)) instead, which
is wrong.

Fix it by changing the signedness to unsigned to avoid surprises for
anyone using this code on non-ARM systems.

Change-Id: Ie222fcaa7c0c4272d7a521a6f2f51995fd5130cc
Signed-off-by: dp-arm <dimitris.papastamos@arm.com>