hikey960: Fix race condition between hotplug and idles
From the hotplug testing on Hikey960, in some case cores fail to become
online in the system. When some cores are hotplugged off, if other
cores in the same cluster enter into CPU idle states at the meantime,
the cluster will be powered off. This introduces the state machine
malfunction in the power controller, thus when hotplug on the core
afterwards, it fails to boot up the core because the power controller
thinks the cluster is powered on.
This patch is to avoid race condition between hotplug and idles by
preventing cluster power off when some of cores in the cluster are
hotplugged off, if all cores in the same cluster are hotplugged off,
the cluster can be powered off.
Change-Id: Ib2feeb093357c70409a3536cb4f9da9b504fdcbe Signed-off-by: Wei Yu <yuwei3@hisilicon.com> Signed-off-by: Leo Yan <leo.yan@linaro.org>
Joel Hutton [Tue, 9 Apr 2019 13:45:34 +0000 (14:45 +0100)]
Add note about erratum 814220 for A7
On Cortex-A7 an L2 set/way cache maintenance operation can overtake
an L1 set/way cache maintenance operation. The mitigation for this is
to use a `DSB` instruction before changing cache. The cache cleaning
code happens to already be doing this, so only a comment was added.
Change-Id: Ia1ffb8ca8b6bbbba422ed6f6818671ef9fe02d90 Signed-off-by: Joel Hutton <Joel.Hutton@Arm.com>
Paul Beesley [Mon, 8 Apr 2019 14:24:43 +0000 (15:24 +0100)]
services/spm: Fix service UUID lookup
The spm_sp_get_by_uuid() function is used to look up the secure
partition that provides a given service.
Within this function, memcmp() is used to compare the service
UUIDs but it uses the size of the rdsvc->uuid pointer instead of
the size of its content (missing dereference). This means that only
a partial comparison is performed as UUIDs are 128 bits in length and
rdsvc->uuid is a uint32_t typed pointer.
Instead, use the size of the array pointed to by the svc_uuid parameter,
which will be the full 128 bits, for the comparison.
Change-Id: I258fb0cca3bf19f97b8f2a4c133981647cd050e4 Signed-off-by: Paul Beesley <paul.beesley@arm.com>
Joel Hutton [Mon, 11 Mar 2019 11:37:38 +0000 (11:37 +0000)]
cot-desc: optimise memory further
This changes the auth_img_desc_t struct to have pointers to struct
arrays instead of struct arrays. This saves memory as many of these
were never used, and can be NULL pointers. Note the memory savings are
only when these arrays are not initialised, as it is assumed these
arrays are fixed length. A possible future optimisation could allow for
variable length.
memory diff:
bl1: bl2:
text text
-12 -12
bss bss
-1463 0
data data
-56 -48
rodata rodata
-5688 -2592
total total
-7419 -2652
Change-Id: I8f9bdedf75048b8867f40c56381e3a6dc6402bcc Signed-off-by: Joel Hutton <Joel.Hutton@Arm.com>
Joel Hutton [Wed, 20 Feb 2019 11:56:46 +0000 (11:56 +0000)]
Reduce memory needed for CoT description
When Trusted Board Boot is enabled, we need to specify the Chain of
Trust (CoT) of the BL1 and BL2 images. A CoT consists of an array
of image descriptors. The authentication module assumes that each
image descriptor in this array is indexed by its unique image
identifier. For example, the Trusted Boot Firmware Certificate has to
be at index [TRUSTED_BOOT_FW_CERT_ID].
Unique image identifiers may not necessarily be consecutive. Also,
a given BL image might not use all image descriptors. For example, BL1
does not need any of the descriptors related to BL31. As a result, the
CoT array might contain holes, which unnecessarily takes up space in
the BL binary.
Using pointers to auth_img_desc_t structs (rather than structs
themselves) means these unused elements only use 1 pointer worth of
space, rather than one struct worth of space. This patch also changes
the code which accesses this array to reflect the change to pointers.
Image descriptors not needed in BL1 or BL2 respectively are also
ifdef'd out in this patch. For example, verifying the BL31 image is
the responsibility of BL2 so BL1 does not need any of the data
structures describing BL31.
memory diff:
bl1: bl2:
text text
-20 -20
bss bss
-1463 0
data data
-256 -48
rodata rodata
-5240 -1952
total total
-6979 -2020
Change-Id: I163668b174dc2b9bbb183acec817f2126864aaad Signed-off-by: Joel Hutton <Joel.Hutton@Arm.com>
aarch32: Allow compiling with soft-float toolchain
ARMv7 and Cortex-A32(ARMv8/aarch32) uses "arm-linux-gnueabi" toolchain which
has both soft-float and hard-float variants and so there could be scenarios
where soft-float toolchain is used.Even though TF-A documentation recommends
to use hard-float toolchain for aarch32 but there are external projects where
we cannot mandate the selection of toolchain and for those projects at least
the build should not fail.
Current TF-A source fails to build with soft-float toolchain because assembler
does not recognizes "vmsr" instruction which is required to enable floating
point unit.
To avoid this piece of code being compiled with soft-float toolchain add
predefined macro guard " __SOFTFP__" exposed by soft-float toolchain.
Paul Beesley [Mon, 11 Feb 2019 17:58:21 +0000 (17:58 +0000)]
doc: Fix heading levels for changelog
The change-log.rst file does not use the same symbols
as the other documents when organising its headings,
sub-headings and so on. In order to compile these
documents together with Sphinx, forming a top-level
index and a consistent structure, it is necessary to
normalise these.
Change-Id: Ib4620ff03a9e76fec9e36e95549016c7b3fe12bb Signed-off-by: Paul Beesley <paul.beesley@arm.com>
SPM: Adjust size of virtual address space per partition
Rather than using a fixed virtual address space size, read all regions
in the resource description of each partition and restrict the virtual
address space size to the one the partition actually needs.
This also allows SPM to take advantage of the extension ARMv8.4-TTST if
the virtual address space size is small enough.
Change-Id: I8646aa95e659136b58b44b040364cdee631f7e82 Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Right now the virtual address space is fixed to
PLAT_VIRT_ADDR_SPACE_SIZE, so all base translation tables are the same
size and need the same alignment. The current code allocates the exact
space needed by this initial table.
However, a following patch is going to allow each partition to choose
the size of its address space based on the memory regions defined in
their resource description, so it isn't possible to determine this at
build time. As this optimization no longer applies, it has to be
removed.
Change-Id: Ia8d19f4981e1017e4ffe0ba136de73d701044cb0 Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
This gives each Secure Partition complete freedom on its address space.
Previously, the memory used by the exception vectors was reserved and
couldn't be used. Also, it always had to be mapped, forcing SPM to
generate translation tables that included the exception vectors as well
as the Partition memory regions. With this change, partitions can reduce
their address space size easily.
Change-Id: I67fb5e9bdf2870b73347f23bff702fab0a8f8711 Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Remi Pommarel [Sat, 30 Mar 2019 14:58:45 +0000 (15:58 +0100)]
meson/gxl: Add tool to create bl31 bootable images
GXL platforms need to have a specific header at the beginning of bl31
image to be able to boot. This adds a tool to create that and calls it at
build time.
Remi Pommarel [Thu, 28 Mar 2019 22:34:18 +0000 (23:34 +0100)]
meson/gxl: Add support for SHA256 DMA engine
In order to configure and boot SCP, BL31 has to compute and send
the SHA-256 of the firmware data via scpi. Luckily Amlogic GXL SOC
has a DMA facility that could be used to offload SHA-256
computations. This adds basic support of this hardware SHA-256
engine.
Add R-Car V3M support. This is based on the original
V3M support patch for Yocto v2.23.1 by Vladimir Barinov.
Signed-off-by: Vladimir Barinov <vladimir.barinov@cogentembedded.com> Signed-off-by: Valentine Barshak <valentine.barshak@cogentembedded.com> Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
---
Marek: Update on top of mainline ATF/master
Marek Vasut [Sat, 30 Mar 2019 03:01:41 +0000 (04:01 +0100)]
rcar_gen3: plat: Print DRAM bank size in MiB if below 1 GiB
Print the DRAM bank size in MiB instead of GiB in case the bank size
is smaller than 1 GiB. This prevents printing zeroes on systems with
small DRAM sizes.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Paul Beesley [Fri, 29 Mar 2019 10:14:56 +0000 (10:14 +0000)]
doc: Clarify draft status of SPCI and SPRT specs
These SPM-related specifications are mentioned in the readme and
the change log. Update references to these specs to make it clear
that they are in draft form and are expected to change.
Change-Id: Ia2791c48c371a828246d96f102a402747cd69f96 Signed-off-by: Paul Beesley <paul.beesley@arm.com>
The latest version of GCC are required to use the new features of TF-A.
Suggest to use the latest version available on developer.arm.com instead
of the version specified on the Linaro Release notes.
At the time of writing, GCC 8.2-2019.01 is the latest version available.
Change-Id: Idd5c00749e39ca9dc8b7c5623b5d64356c9ce6e5 Signed-off-by: Louis Mayencourt <louis.mayencourt@arm.com>
Paul Beesley [Mon, 25 Mar 2019 16:42:19 +0000 (16:42 +0000)]
doc: Add contents.rst to link to pages
This is the temporary contents page that links
to all other documents (except platform ports).
This page is needed during the
trustedfirmware.org migration, before we have a
Sphinx rendering pipeline set up, because cgit
doesn't offer a good way to view rendered docs
while browsing the tree. We need to have a links
page that can be opened from the cgit 'about'
view.
Change-Id: I3ad87a9fa8a14dc8e371aac7ee473575fed316bf Signed-off-by: Paul Beesley <paul.beesley@arm.com>
Tien Hock, Loh [Fri, 22 Mar 2019 04:54:31 +0000 (12:54 +0800)]
driver: synosys: Fix SD MMC not initializing correctly
dw_params.mmc_dev_type should be assigned before mmc_init, otherwise SDMMC
initialization will fail as the initialization treats the device as EMMC
instead of SD.
Signed-off-by: Tien Hock, Loh <tien.hock.loh@intel.com>
Fixed the below bugs:
1) Bug related to build flag V=1: if the flag was V=0, building with
ROMLIB would fail.
2) Due to a syntax bug in genwrappers.sh, index file entries marked as
"patch" or "reserved" were ignored.
3) Added a prepending hash to constants that genwrappers is generating.
4) Due to broken dependencies, currently the inclusion functionality is
intentionally not utilised. This is why the contents of romlib/jmptbl.i
have been copied to platform specific jmptbl.i files. As a result of the
broken dependencies, when changing the index files, e.g. patching
functions, a clean build is always required. This is a known issue that
will be fixed in the future.
Change-Id: I9d92aa9724e86d8f90fcd3e9f66a27aa3cab7aaa Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
xlat_tables_v2: Revert recent changes to remove recursion
This commit reverts the following commits:
- c54c7fc35842 ("xlat_tables_v2: print xlat tables without recursion")
- db8cac2d986a ("xlat_tables_v2: unmap region without recursion.")
- 0ffe269215bd ("xlat_tables_v2: map region without recursion.")
This was part of PR#1843.
A problem has been detected in one of our test run configurations
involving dynamic mapping of regions and it is blocking the next
release. Until the problem can be solved, it is safer to revert
the changes.
Change-Id: I3d5456e4dbebf291c8b74939c6fb02a912e0903b Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
Alexei Fedorov [Wed, 13 Mar 2019 11:05:07 +0000 (11:05 +0000)]
Declare PAuth for Secure world as experimental
Declare ENABLE_PAUTH and CTX_INCLUDE_PAUTH_REGS
build options as experimental.
Pointer Authentication is enabled for Non-secure world
irrespective of the value of these build flags if the
CPU supports it.
The patch also fixes the description of fiptool 'help' command.
projects / arm-tf.git / log