project/bcm63xx/atf.git
5 years agoMerge "PSCI: Lookup list of parent nodes to lock only once" into integration
Soby Mathew [Mon, 10 Jun 2019 09:39:23 +0000 (09:39 +0000)]
Merge "PSCI: Lookup list of parent nodes to lock only once" into integration

5 years agoMerge "Update Allwinner SoC names in documentation" into integration
Soby Mathew [Mon, 10 Jun 2019 09:14:13 +0000 (09:14 +0000)]
Merge "Update Allwinner SoC names in documentation" into integration

5 years agoUpdate Allwinner SoC names in documentation
Samuel Holland [Sat, 8 Jun 2019 21:18:02 +0000 (16:18 -0500)]
Update Allwinner SoC names in documentation

Provide the friendly marketing names, not just the platform name.

Signed-off-by: Samuel Holland <samuel@sholland.org>
Change-Id: Id4427abb73d0c1be4ac1709b2a8e87beffc20dd5

5 years agoMerge "Neoverse N1: Introduce workaround for Neoverse N1 erratum 1315703" into integr...
John Tsichritzis [Fri, 7 Jun 2019 15:20:45 +0000 (15:20 +0000)]
Merge "Neoverse N1: Introduce workaround for Neoverse N1 erratum 1315703" into integration

5 years agoPSCI: Lookup list of parent nodes to lock only once
Andrew F. Davis [Tue, 4 Jun 2019 14:46:54 +0000 (10:46 -0400)]
PSCI: Lookup list of parent nodes to lock only once

When acquiring or releasing the power domain locks for a given CPU the
parent nodes are looked up by walking the up the PD tree list on both the
acquire and release path, only one set of lookups is needed. Fetch the
parent nodes first and pass this list into both the acquire and release
functions to avoid the double lookup.

This also allows us to not have to do this lookup after coherency has
been exited during the core power down sequence. The shared struct
psci_cpu_pd_nodes is not placed in coherent memory like is done
for psci_non_cpu_pd_nodes and doing so would negatively affect
performance. With this patch we remove the need to have it in coherent
memory by moving the access out of psci_release_pwr_domain_locks().

Signed-off-by: Andrew F. Davis <afd@ti.com>
Change-Id: I7b9cfa9d31148dea0f5e21091c8b45ef7fe4c4ab

5 years agoMerge "FVP: Remove GIC initialisation from secondary core cold boot" into integration
John Tsichritzis [Thu, 6 Jun 2019 15:19:50 +0000 (15:19 +0000)]
Merge "FVP: Remove GIC initialisation from secondary core cold boot" into integration

5 years agoNeoverse N1: Introduce workaround for Neoverse N1 erratum 1315703
Andre Przywara [Mon, 20 May 2019 13:57:06 +0000 (14:57 +0100)]
Neoverse N1: Introduce workaround for Neoverse N1 erratum 1315703

Neoverse N1 erratum 1315703 is a Cat A (rare) erratum [1], present in
older revisions of the Neoverse N1 processor core.
The workaround is to set a bit in the implementation defined CPUACTLR2_EL1
system register, which will disable the load-bypass-store feature.

[1] http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.pjdocpjdoc-466751330-1032/index.html

Change-Id: I5c708dbe0efa4daa0bcb6bd9622c5efe19c03af9
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
5 years agoMerge "Introduce BTI support in ROMLIB" into integration
John Tsichritzis [Thu, 6 Jun 2019 10:26:10 +0000 (10:26 +0000)]
Merge "Introduce BTI support in ROMLIB" into integration

5 years agomediatek: mt8183: add mcsi driver
kenny liang [Thu, 2 May 2019 14:01:39 +0000 (22:01 +0800)]
mediatek: mt8183: add mcsi driver

add mcsi driver to support cache coherence.

Change-Id: I94f5922783e5dbc6b7e92aa06464bc1f0177f00a
Signed-off-by: kenny liang <kenny.liang@mediatek.com>
5 years agomediatek: mt8183: add GIC driver
kenny liang [Thu, 2 May 2019 14:26:22 +0000 (22:26 +0800)]
mediatek: mt8183: add GIC driver

Add Mediatek GIC driver to support interrupt functions.

Signed-off-by: kenny liang <kenny.liang@mediatek.com>
Change-Id: I967a18f2e45b7bbc88c506dd4f1f40a745227ad9

5 years agoFVP: Remove GIC initialisation from secondary core cold boot
John Tsichritzis [Mon, 3 Jun 2019 15:20:46 +0000 (16:20 +0100)]
FVP: Remove GIC initialisation from secondary core cold boot

During the secondary cores' cold boot path, the cores initialise the GIC
CPU interface. However this is a redundant action since 1) the cores are
powered down immediately after that, 2) the GIC CPU interface is
initialised from scratch when the secondary cores are powered up again
later.

Moreover, this part of code was introducing a bug. In a GICv3 system,
the GIC's CPU interface system registers must not be written without the
core being marked as "awake" in the redistributor. However, this
sequence was performing such accesses and this would cause those cores
to hang. The hang was caused by the DSB instruction that would never
complete because of the GIC not recognising those writes.

For the two aforementioned reasons, the entire part of the GIC CPU
interface initialisation is removed.

Change-Id: I6c33a1edda69dd5b6add16a27390a70731b5532a
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "Apply compile-time check for AArch64-only cores" into integration
John Tsichritzis [Wed, 5 Jun 2019 10:26:22 +0000 (10:26 +0000)]
Merge "Apply compile-time check for AArch64-only cores" into integration

5 years agoMerge "Prevent pending G1S interrupt become G0 interrupt" into integration
John Tsichritzis [Wed, 5 Jun 2019 10:08:55 +0000 (10:08 +0000)]
Merge "Prevent pending G1S interrupt become G0 interrupt" into integration

5 years agoPrevent pending G1S interrupt become G0 interrupt
James kung [Fri, 31 May 2019 07:40:05 +0000 (15:40 +0800)]
Prevent pending G1S interrupt become G0 interrupt

According to Arm GIC spec(IHI0069E, section 4.6.1),
when GICD_CTLR.DS == 0, Secure Group 1 interrupts
are treated as Group 0 by a CPU interface if:
- The PE does not implement EL3.
- ICC_SRE_EL1(S).SRE == 0

When a cpu enter suspend or deep idle, it might be
powered off. When the cpu resume, according to
the GIC spec(IHI0069E, section 9.2.15, 9.2.16 and
9.2.22) the ICC_SRE_EL1.SRE reset value is 0 (if
write is allowed) and G0/G1S/G1NS interrupt of the
GIC cpu interface are all disabled.

If a G1S SPI interrupt occurred and the target cpu
of the SPI is assigned to a specific cpu which is
in suspend and is powered off, when the cpu resume
and start to initial the GIC cpu interface, the
initial sequence might affect the interrupt group
type of the pending interrupt on the cpu interface.

Current initial sequence on the cpu interface is:
1. Enable G0 interrupt
2. Enable G1S interrupt
3. Enable ICC_SRE_EL1(S).SRE

It is possible to treat the pending G1S interrupt
as G0 interrupt on the cpu interface if the G1S
SPI interrupt occurred between step2 and step3.

To prevent the above situation happend, the initial
sequence should be changed as follows:
1. Enable ICC_SRE_EL1(S).SRE
2. Enable G0 interrupt
3. Enable G1S interrupt

Change-Id: Ie34f6e0b32eb9a1677ff72571fd4bfdb5cae25b0
Signed-off-by: James Kung <kong1191@gmail.com>
5 years agoApply compile-time check for AArch64-only cores
John Tsichritzis [Mon, 3 Jun 2019 12:54:30 +0000 (13:54 +0100)]
Apply compile-time check for AArch64-only cores

Some cores support only AArch64 mode. In those cores, only a limited
subset of the AArch32 system registers are implemented. Hence, if TF-A
is supposed to run on AArch64-only cores, it must be compiled with
CTX_INCLUDE_AARCH32_REGS=0.

Currently, the default settings for compiling TF-A are with the AArch32
system registers included. So, if we compile TF-A the default way and
attempt to run it on an AArch64-only core, we only get a runtime panic.

Now a compile-time check has been added to ensure that this flag has the
appropriate value when AArch64-only cores are included in the build.

Change-Id: I298ec550037fafc9347baafb056926d149197d4c
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "doc: Enable automatic labels for page titles" into integration
John Tsichritzis [Mon, 3 Jun 2019 16:24:24 +0000 (16:24 +0000)]
Merge "doc: Enable automatic labels for page titles" into integration

5 years agoMerge "Add information about the mailing list in the docs" into integration
John Tsichritzis [Mon, 3 Jun 2019 16:21:50 +0000 (16:21 +0000)]
Merge "Add information about the mailing list in the docs" into integration

5 years agoAdd information about the mailing list in the docs
John Tsichritzis [Mon, 3 Jun 2019 09:51:22 +0000 (10:51 +0100)]
Add information about the mailing list in the docs

Change-Id: I41ce5323c33a81db13c5cc40de1ac4e221a10cd8
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge changes from topic "jts/docs" into integration
Paul Beesley [Fri, 31 May 2019 13:37:55 +0000 (13:37 +0000)]
Merge changes from topic "jts/docs" into integration

* changes:
  Removing IRC related info from the documentation
  Further fixes to documentation links

5 years agoRemoving IRC related info from the documentation
John Tsichritzis [Fri, 31 May 2019 13:26:48 +0000 (14:26 +0100)]
Removing IRC related info from the documentation

Change-Id: I5cf8c70a304bf5869cbeb12fa8d39171cff48ebd
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "rockchip: drop rockchip-specific imported linker symbols for bl31" into integr...
Paul Beesley [Thu, 30 May 2019 14:41:18 +0000 (14:41 +0000)]
Merge "rockchip: drop rockchip-specific imported linker symbols for bl31" into integration

5 years agodoc: Enable automatic labels for page titles
Paul Beesley [Fri, 17 May 2019 10:45:36 +0000 (11:45 +0100)]
doc: Enable automatic labels for page titles

Automatic labelling of document titles is a prerequisite for
converting the format of cross-document links. Sphinx will
generate (via the enabled extension) a hidden link target for
each document title and this can be referred to later, from
another page, to link to the target.

The plugin options being used require Sphinx >= 2.0.0 so a
requirements.txt file has been added. This file is used with
the pip package manager for Python so that the correct
dependencies are installed.

Change-Id: Ic2049db5804aa4a6447608ba4299de958ce0a87d
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agoFurther fixes to documentation links
John Tsichritzis [Tue, 28 May 2019 12:13:39 +0000 (13:13 +0100)]
Further fixes to documentation links

Change-Id: Ib021c721652d96f6c06ea18741f19a72bba1d00f
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "Cortex-A55: workarounds for errata 1221012" into integration
Paul Beesley [Wed, 29 May 2019 11:29:12 +0000 (11:29 +0000)]
Merge "Cortex-A55: workarounds for errata 1221012" into integration

5 years agorockchip: drop rockchip-specific imported linker symbols for bl31
Heiko Stuebner [Wed, 29 May 2019 10:03:38 +0000 (12:03 +0200)]
rockchip: drop rockchip-specific imported linker symbols for bl31

In the rockchip bl31 setup the __RO_START__ and __RO_END__ symbols are
currently imported into special BL31_RO_* constants while the general
code also imports them as BL_CODE_BASE and BL_CODE_END.

So we can just use the general symbols and can drop the duplication.

Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Change-Id: Ibf1b48ad80bed897247a1690a32711030479262d

5 years agoMerge "Beautify "make help"" into integration
Paul Beesley [Wed, 29 May 2019 09:35:04 +0000 (09:35 +0000)]
Merge "Beautify "make help"" into integration

5 years agoMerge "Makefile: Add default warning flags" into integration
Paul Beesley [Wed, 29 May 2019 09:31:41 +0000 (09:31 +0000)]
Merge "Makefile: Add default warning flags" into integration

5 years agoCortex-A55: workarounds for errata 1221012
Ambroise Vincent [Tue, 28 May 2019 08:52:48 +0000 (09:52 +0100)]
Cortex-A55: workarounds for errata 1221012

The workaround is added to the Cortex-A55 cpu specific file. The
workaround is disabled by default and have to be explicitly enabled by
the platform integrator.

Change-Id: I3e6fd10df6444122a8ee7d08058946ff1cc912f8
Signed-off-by: Ambroise Vincent <ambroise.vincent@arm.com>
5 years agoMerge "plat: imx8m: Add the aipstz init to config peripheral access" into integration
Soby Mathew [Tue, 28 May 2019 13:18:56 +0000 (13:18 +0000)]
Merge "plat: imx8m: Add the aipstz init to config peripheral access" into integration

5 years agoMerge changes from topic "for-upstream" into integration
Soby Mathew [Tue, 28 May 2019 13:18:41 +0000 (13:18 +0000)]
Merge changes from topic "for-upstream" into integration

* changes:
  ti: k3: common: Set L2 latency on A72 cores
  ti: k3: common: Add support for J721E

5 years agoMerge "Fix documentation links" into integration
John Tsichritzis [Tue, 28 May 2019 11:53:58 +0000 (11:53 +0000)]
Merge "Fix documentation links" into integration

5 years agoFix documentation links
John Tsichritzis [Tue, 28 May 2019 11:45:06 +0000 (12:45 +0100)]
Fix documentation links

Change-Id: Ic09e74f22b43fba51ee17cd02b5e1dc5d8e0bb63
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "Add support for Branch Target Identification" into integration
Paul Beesley [Fri, 24 May 2019 16:47:25 +0000 (16:47 +0000)]
Merge "Add support for Branch Target Identification" into integration

5 years agoMerge changes from topic "jts/docs" into integration
Paul Beesley [Fri, 24 May 2019 16:46:59 +0000 (16:46 +0000)]
Merge changes from topic "jts/docs" into integration

* changes:
  Docs fixes
  Update security documentation

5 years agoAdd support for Branch Target Identification
Alexei Fedorov [Fri, 24 May 2019 11:17:09 +0000 (12:17 +0100)]
Add support for Branch Target Identification

This patch adds the functionality needed for platforms to provide
Branch Target Identification (BTI) extension, introduced to AArch64
in Armv8.5-A by adding BTI instruction used to mark valid targets
for indirect branches. The patch sets new GP bit [50] to the stage 1
Translation Table Block and Page entries to denote guarded EL3 code
pages which will cause processor to trap instructions in protected
pages trying to perform an indirect branch to any instruction other
than BTI.
BTI feature is selected by BRANCH_PROTECTION option which supersedes
the previous ENABLE_PAUTH used for Armv8.3-A Pointer Authentication
and is disabled by default. Enabling BTI requires compiler support
and was tested with GCC versions 9.0.0, 9.0.1 and 10.0.0.
The assembly macros and helpers are modified to accommodate the BTI
instruction.
This is an experimental feature.
Note. The previous ENABLE_PAUTH build option to enable PAuth in EL3
is now made as an internal flag and BRANCH_PROTECTION flag should be
used instead to enable Pointer Authentication.
Note. USE_LIBROM=1 option is currently not supported.

Change-Id: Ifaf4438609b16647dc79468b70cd1f47a623362e
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
5 years agoMakefile: Add default warning flags
Ambroise Vincent [Fri, 24 May 2019 11:47:43 +0000 (12:47 +0100)]
Makefile: Add default warning flags

The flags are taken from the different warning levels of the build
system when they do not generate any error with the current upstreamed
platforms.

Change-Id: Ia70cff83bedefb6d2f0dd266394ef77fe47e7f65
Signed-off-by: Ambroise Vincent <ambroise.vincent@arm.com>
5 years agoDocs fixes
John Tsichritzis [Fri, 24 May 2019 11:27:49 +0000 (12:27 +0100)]
Docs fixes

1) Fix links in "about" page
2) Put back the "contents" page with adjusted links

Change-Id: Id09140b91df5cf0a275149801d05d8cfeeda1c6e
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoUpdate security documentation
John Tsichritzis [Tue, 21 May 2019 09:37:55 +0000 (10:37 +0100)]
Update security documentation

1) Replace references to "Arm Trusted Firmware" with "TF-A"
2) Update issue tracker link

Change-Id: I12d827d49f6cc34e46936d7f7ccf44e32b26a0bd
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoIntroduce BTI support in ROMLIB
John Tsichritzis [Tue, 21 May 2019 14:47:37 +0000 (15:47 +0100)]
Introduce BTI support in ROMLIB

When TF-A is compiled with BTI enabled, the branches in the ROMLIB
jumptable must be preceded by a "bti j" instruction.

Moreover, when the additional "bti" instruction is inserted, the
jumptable entries have a distance of 8 bytes between them instead of 4.
Hence, the wrappers are also modified accordinly.

If TF-A is compiled without BTI enabled, the ROMLIB jumptable and
wrappers are generated as before.

Change-Id: Iaa59897668f8e59888d39046233300c2241d8de7
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoBeautify "make help"
John Tsichritzis [Tue, 21 May 2019 14:57:31 +0000 (15:57 +0100)]
Beautify "make help"

Changes to make the help text a bit more readable:
1) The "usage" part is now a one-liner
2) The supported platforms list is printed separately

Change-Id: I93e48a6cf1d28f0ef9f3db16ce17725e4dff33c9
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "plat/meson/gxl: BL31: remove BL2 dependency" into integration
Sandrine Bailleux [Thu, 23 May 2019 13:19:09 +0000 (13:19 +0000)]
Merge "plat/meson/gxl: BL31: remove BL2 dependency" into integration

5 years agoMerge changes from topic "pb/sphinx-doc" into integration
Sandrine Bailleux [Thu, 23 May 2019 13:14:00 +0000 (13:14 +0000)]
Merge changes from topic "pb/sphinx-doc" into integration

* changes:
  doc: Use proper note and warning annotations
  doc: Refactor contributor acknowledgements
  doc: Reorganise images and update links
  doc: Set correct syntax highlighting style
  doc: Add minimal glossary
  doc: Remove per-page contents lists
  doc: Make checkpatch ignore rst files
  doc: Format security advisory titles and headings
  doc: Reformat platform port documents
  doc: Normalise section numbering and headings
  doc: Reword document titles

5 years agoMerge "drivers: scmi: scmi_sq: Modify wrong payload length" into integration
Sandrine Bailleux [Thu, 23 May 2019 09:09:52 +0000 (09:09 +0000)]
Merge "drivers: scmi: scmi_sq: Modify wrong payload length" into integration

5 years agodrivers: scmi: scmi_sq: Modify wrong payload length
Masahisa Kojima [Thu, 23 May 2019 05:41:35 +0000 (14:41 +0900)]
drivers: scmi: scmi_sq: Modify wrong payload length

Payload length of the get dram mapping information message is 0.
The mbx_mem->len parameter should be 4, it only contains
message header.

Fixes: b67d202 ("plat/synquacer: enable SCMI support")
Signed-off-by: Masahisa Kojima <masahisa.kojima@linaro.org>
Change-Id: If1cd4c855da2dc5dc4b6da3bea152b8441971de7

5 years agoti: k3: common: Set L2 latency on A72 cores
Andrew F. Davis [Fri, 10 May 2019 15:20:50 +0000 (11:20 -0400)]
ti: k3: common: Set L2 latency on A72 cores

The Cortex-A72 based cores on K3 platforms can be clocked fast
enough that an extra latency cycle is needed to ensure correct
L2 access. Set the latency here for all A72 cores.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Change-Id: Id534316dec1c1f326908efbfd964f219cda7386a

5 years agoti: k3: common: Add support for J721E
Nishanth Menon [Fri, 22 Jun 2018 11:36:29 +0000 (06:36 -0500)]
ti: k3: common: Add support for J721E

Enable Cortex-A72 support for J721E.

Change-Id: I5bea5fb6ec45d1a9f8f2192d42da2cc03ae0f7ec
Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Andrew F. Davis <afd@ti.com>
5 years agodoc: Use proper note and warning annotations
Paul Beesley [Wed, 13 Mar 2019 16:20:44 +0000 (16:20 +0000)]
doc: Use proper note and warning annotations

The documentation contains plenty of notes and warnings. Enable
special rendering of these blocks by converting the note prefix
into a .. note:: annotation.

Change-Id: I34e26ca6bf313d335672ab6c2645741900338822
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Refactor contributor acknowledgements
Paul Beesley [Wed, 13 Mar 2019 16:02:25 +0000 (16:02 +0000)]
doc: Refactor contributor acknowledgements

- Make the list of contributors into an actual list
- Use note syntax for the note
- Remove the Individuals heading since there are none

This file could be considered for removal as it is a legacy
document, as its note explains.

Change-Id: Idf984bc192af7a0ec367a6642ab99ccccf5df1a8
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Reorganise images and update links
Paul Beesley [Wed, 13 Mar 2019 15:49:27 +0000 (15:49 +0000)]
doc: Reorganise images and update links

Change-Id: I679d1499376a524bef1cfc33df995b0a719b5ac8
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Set correct syntax highlighting style
Paul Beesley [Wed, 13 Mar 2019 15:11:04 +0000 (15:11 +0000)]
doc: Set correct syntax highlighting style

Several code blocks do not specify a language for syntax
highlighting. This results in Sphinx using a default highlighter
which is Python.

This patch adds the correct language to each code block that doesn't
already specify it.

Change-Id: Icce1949aabfdc11a334a42d49edf55fa673cddc3
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Add minimal glossary
Paul Beesley [Wed, 13 Mar 2019 13:58:02 +0000 (13:58 +0000)]
doc: Add minimal glossary

One of the current issues with the documentation is that terms and
abbreviations are frequently redefined. For example, we might have
a sentence like "... the SCP (System Control Processor) will ...".

These definitions might be repeated several times across pages, or
even within the same document. Equally, some of these abbreviations
are missed and are never expanded.

Sphinx provides a :term: keyword that takes some text and,
if that text is defined in a glossary document, links to its glossary
entry. Using this functionality will prevent repeated definitions
and will make the docs more maintainable by using a single
definition source.

The glossary added in this patch was created from a quick scrub of
the source code - there may be missing entries. The SDEI abbreviation
was used as an example.

Note that a global_substitutions file was created. This file contains
the RST 'replace' statements that convert plain text terms into linked
terms (by adding the ':term:' keyword to them). An example is:

.. |TF-A| replace:: :term:`TF-A`

The 'rst_prolog' variable in conf.py is used to inject this list of
replacements into each page. Terms must be surrounded with the pipe
character to be turned into links - this means that we can still
prevent certain terms from being linked if we don't want them to be.

Change-Id: I87010ed9cfa4a60011a9b4a431b98cb4bb7baa28
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Remove per-page contents lists
Paul Beesley [Thu, 7 Mar 2019 17:03:22 +0000 (17:03 +0000)]
doc: Remove per-page contents lists

These are no longer needed as there will always be a table of contents
rendered to the left of every page.

Some of these lists can be quite long and, when opening a page, the
reader sees nothing but a huge list of contents! After this patch,
the document contents are front-and-centre and the contents are
nicely rendered in the sidebar without duplication.

Change-Id: I444754d548ec91d00f2b04e861de8dde8856aa62
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Make checkpatch ignore rst files
Paul Beesley [Thu, 7 Mar 2019 16:42:31 +0000 (16:42 +0000)]
doc: Make checkpatch ignore rst files

Previously checkpatch was invoked with options to make it ignore
Markdown (md) files as this was the dominant format for TF-A
documents. Now that rst is being used everywhere this option needs
updating.

Change-Id: I59b5a0bcc45d2386df4f880b8d333baef0bbee77
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Format security advisory titles and headings
Paul Beesley [Thu, 7 Mar 2019 16:22:44 +0000 (16:22 +0000)]
doc: Format security advisory titles and headings

Required so that the advisory documents are all valid RST files (with a
header) and that they all integrate into the document tree.

Change-Id: I68ca2b0b9e648e24b460deb772c471a38518da26
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Reformat platform port documents
Paul Beesley [Wed, 22 May 2019 10:22:44 +0000 (11:22 +0100)]
doc: Reformat platform port documents

The platform port documents are not very standardised right now and
they don't integrate properly into the document tree so:

1) Make sure each port has a proper name and title (incl. owner)
2) Correct use of headings, subheadings, etc in each port
3) Resolve any naming conflicts between documents

Change-Id: I4c2da6f57172b7f2af3512e766ae9ce3b840b50f
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Normalise section numbering and headings
Paul Beesley [Thu, 7 Mar 2019 15:53:44 +0000 (15:53 +0000)]
doc: Normalise section numbering and headings

Required work to make all documents sit at the correct levels within
the document tree and any derived content like the table of contents
and the categories in the sidebar.

Change-Id: I4885fbe30864a87c8822ee67482b71fb46a8fbc6
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Reword document titles
Paul Beesley [Thu, 7 Mar 2019 15:47:15 +0000 (15:47 +0000)]
doc: Reword document titles

This patch attempts to standardise the document titles as well as
adding titles to documents that were missing one. The aim is to
remove needless references to "TF-A" or "Trusted Firmware" in the
title of every document and to make sure that the title matches
with the document content.

Change-Id: I9b93ccf43b5d57e8dc793a5311b8ed7c4dd245cc
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agoMerge "rcar_gen3: drivers: qos: update QoS setting" into integration
Sandrine Bailleux [Wed, 22 May 2019 08:01:02 +0000 (08:01 +0000)]
Merge "rcar_gen3: drivers: qos: update QoS setting" into integration

5 years agoMerge changes Icf1ea76c,I9ca3f278 into integration
Sandrine Bailleux [Wed, 22 May 2019 07:06:39 +0000 (07:06 +0000)]
Merge changes Icf1ea76c,I9ca3f278 into integration

* changes:
  imx8: Replace PLAT_IMX8* with automatic PLAT_imx8*
  plat: imx8mq: Implement IMX_SIP_GET_SOC_INFO

5 years agoMerge "Rework smc_unknown return code path in smc_handler" into integration
Sandrine Bailleux [Wed, 22 May 2019 06:22:03 +0000 (06:22 +0000)]
Merge "Rework smc_unknown return code path in smc_handler" into integration

5 years agorcar_gen3: drivers: qos: update QoS setting
Yoshifumi Hosoya [Fri, 12 Apr 2019 08:08:15 +0000 (17:08 +0900)]
rcar_gen3: drivers: qos: update QoS setting

Update M3 Ver.3.0 QoS setting rev.0.03.

Change-Id: I892521d456c606ac3d30f2b2ac6b4e16faa5fc48
Signed-off-by: Yoshifumi Hosoya <yoshifumi.hosoya.wj@renesas.com>
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agoMerge "doc: Move content out of readme and create new index page" into integration
Sandrine Bailleux [Tue, 21 May 2019 14:57:07 +0000 (14:57 +0000)]
Merge "doc: Move content out of readme and create new index page" into integration

5 years agoMerge "doc: Move documents into subdirectories" into integration
Sandrine Bailleux [Tue, 21 May 2019 14:56:08 +0000 (14:56 +0000)]
Merge "doc: Move documents into subdirectories" into integration

5 years agodoc: Move content out of readme and create new index page
Paul Beesley [Thu, 7 Mar 2019 15:25:14 +0000 (15:25 +0000)]
doc: Move content out of readme and create new index page

Previously the readme.rst file served as the entrypoint for the
documentation. With a Sphinx build the top-level document is set
to be index.rst as it contains the primary document index.

This patch moves some content from readme.rst into index.rst and
splits the license information out into license.rst.

Change-Id: I5c50250b81136fe36aa9ceedaae302b44ec11e47
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agodoc: Move documents into subdirectories
Paul Beesley [Mon, 11 Feb 2019 17:54:45 +0000 (17:54 +0000)]
doc: Move documents into subdirectories

This change creates the following directories under docs/
in order to provide a grouping for the content:

- components
- design
- getting_started
- perf
- process

In each of these directories an index.rst file is created
and this serves as an index / landing page for each of the
groups when the pages are compiled. Proper layout of the
top-level table of contents relies on this directory/index
structure.

Without this patch it is possible to build the documents
correctly with Sphinx but the output looks messy because
there is no overall hierarchy.

Change-Id: I3c9f4443ec98571a56a6edf775f2c8d74d7f429f
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agoMerge "doc: Add minimal Sphinx support" into integration
Sandrine Bailleux [Tue, 21 May 2019 13:42:17 +0000 (13:42 +0000)]
Merge "doc: Add minimal Sphinx support" into integration

5 years agoplat: imx8m: Add the aipstz init to config peripheral access
Jacky Bai [Tue, 21 May 2019 12:24:52 +0000 (20:24 +0800)]
plat: imx8m: Add the aipstz init to config peripheral access

AIPSTZ provide access control for all the peripherals connected
to it. In this patch all the perperals are configured accessible
to all the master. it can be customized based the actual use
case.

Signed-off-by: Jacky Bai <ping.bai@nxp.com>
Change-Id: I5ef5baa1da6906f13a60923d27ede336c61e319a

5 years agoMerge "romlib: Improve compilation flags definition" into integration
Sandrine Bailleux [Tue, 21 May 2019 12:33:18 +0000 (12:33 +0000)]
Merge "romlib: Improve compilation flags definition" into integration

5 years agodoc: Add minimal Sphinx support
Paul Beesley [Wed, 23 Jan 2019 15:39:39 +0000 (15:39 +0000)]
doc: Add minimal Sphinx support

Add the essentials for supporting a Sphinx documentation build:

- A makefile under docs/ to invoke Sphinx with the desired output
  format
- A Sphinx master configuration file (conf.py)
- A single, top-level index page (index.rst)
- The TF.org logo that is integrated in the the sidebar of the
  rendered output

Change-Id: I85e67e939658638337ca7972936a354878083a25
Signed-off-by: Paul Beesley <paul.beesley@arm.com>
5 years agoMerge "Fix docs references to header files" into integration
Sandrine Bailleux [Tue, 21 May 2019 08:55:31 +0000 (08:55 +0000)]
Merge "Fix docs references to header files" into integration

5 years agoFix docs references to header files
John Tsichritzis [Mon, 13 May 2019 10:20:05 +0000 (11:20 +0100)]
Fix docs references to header files

Change-Id: I5c06e777d93ac653a853997c2b7c1c9d09b1e49c
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoRework smc_unknown return code path in smc_handler
Madhukar Pappireddy [Wed, 8 May 2019 20:41:41 +0000 (15:41 -0500)]
Rework smc_unknown return code path in smc_handler

The intention of this patch is to leverage the existing el3_exit() return
routine for smc_unknown return path rather than a custom set of instructions.
In order to leverage el3_exit(), the necessary counteraction (i.e., saving the
system registers apart from GP registers) must be performed. Hence a series of
instructions which save system registers( like SPSR_EL3, SCR_EL3 etc) to stack
are moved to the top of group of instructions which essentially decode the OEN
from the smc function identifier and obtain the specific service handler in
rt_svc_descs_array. This ensures that the control flow for both known and
unknown smc calls will be similar.

Change-Id: I67f94cfcba176bf8aee1a446fb58a4e383905a87
Signed-off-by: Madhukar Pappireddy <madhukar.pappireddy@arm.com>
5 years agoMerge "Update docs for FVP v11.6" into integration
Sandrine Bailleux [Tue, 21 May 2019 08:17:16 +0000 (08:17 +0000)]
Merge "Update docs for FVP v11.6" into integration

5 years agoromlib: Improve compilation flags definition
Louis Mayencourt [Mon, 29 Apr 2019 15:35:30 +0000 (16:35 +0100)]
romlib: Improve compilation flags definition

* Optimization flags were only provided for debug build.
* Set optimisation level to -O1
* Remove CFLAGS which is never used for romlib
* Remove the ignored -g flag from LDFLAGS

Change-Id: Id4b69026d8a322ed4cb0acf06c350f13d31571ad
Signed-off-by: Louis Mayencourt <louis.mayencourt@arm.com>
5 years agoUpdate docs for FVP v11.6
John Tsichritzis [Mon, 20 May 2019 12:09:34 +0000 (13:09 +0100)]
Update docs for FVP v11.6

Change-Id: I33c1bf49aa10867e1a2ca4c167112b99bf756dda
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoimx8: Replace PLAT_IMX8* with automatic PLAT_imx8*
Leonard Crestez [Mon, 20 May 2019 08:28:50 +0000 (11:28 +0300)]
imx8: Replace PLAT_IMX8* with automatic PLAT_imx8*

Platform defines are already provided by the build system so let's not
duplicate them.

Change-Id: Icf1ea76c3c3213e27b447c95e2b22b961fa7693e
Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
5 years agoplat: imx8mq: Implement IMX_SIP_GET_SOC_INFO
Leonard Crestez [Fri, 10 May 2019 10:07:41 +0000 (13:07 +0300)]
plat: imx8mq: Implement IMX_SIP_GET_SOC_INFO

The manual documents that 0x3036006c should contains the soc revision
for imx8mq but this always reports A0. Work around this by parsing the
ROM header and checking if OCOTP register 0x40 is stuck at 0xff0055aa.

Determining this inside TF-A makes life easier for OS, see for example
this linux discussion: https://lkml.org/lkml/2019/5/3/465

The soc revision can also be useful inside TF-A itself, for example for
the non-upstream DDR DVFS "busfreq" feature is affected by 8mq erratas.

The clock for OCOTP block can be disabled by OS so only initialize soc
revision once at boot time.

Change-Id: I9ca3f27840229ce8a28b53870e44da29f63c73aa
Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
5 years agoplat/meson/gxl: BL31: remove BL2 dependency
Kevin Hilman [Thu, 16 May 2019 21:38:02 +0000 (14:38 -0700)]
plat/meson/gxl: BL31: remove BL2 dependency

Remove an assert() that assumes a specific value being passed from
BL2.  This value is dependent on BL2 version, so makes this assert()
not portable.

Suggested-by: Remi Pommarel <repk@triplefau.lt>
Signed-off-by: Kevin Hilman <khilman@baylibre.com>
Change-Id: Ife3d934b2fa37fc1c66963dd4eb1afe2ca17d740

5 years agoMerge changes from topic "sami/550_fix_n1sdp_issues_v1" into integration
Soby Mathew [Thu, 16 May 2019 08:33:56 +0000 (08:33 +0000)]
Merge changes from topic "sami/550_fix_n1sdp_issues_v1" into integration

* changes:
  N1SDP: Initialise CNTFRQ in Non Secure CNTBaseN
  N1SDP: Fix DRAM2 start address
  Add option for defining platform DRAM2 base
  Disable speculative loads only if SSBS is supported

5 years agoMerge "plat: imx8mq: Remove duplicated linker symbols" into integration
Soby Mathew [Wed, 15 May 2019 16:00:40 +0000 (16:00 +0000)]
Merge "plat: imx8mq: Remove duplicated linker symbols" into integration

5 years agoMerge "drivers: ufs: Extend the delay after reset to wait for some slower chips"...
Soby Mathew [Wed, 15 May 2019 15:58:17 +0000 (15:58 +0000)]
Merge "drivers: ufs: Extend the delay after reset to wait for some slower chips" into integration

5 years agoMerge "Remove .arch directives from spinlock.S" into integration
Soby Mathew [Wed, 15 May 2019 15:54:32 +0000 (15:54 +0000)]
Merge "Remove .arch directives from spinlock.S" into integration

5 years agoMerge "SMMUv3: Abort DMA transactions" into integration
Soby Mathew [Wed, 15 May 2019 15:54:16 +0000 (15:54 +0000)]
Merge "SMMUv3: Abort DMA transactions" into integration

5 years agoN1SDP: Initialise CNTFRQ in Non Secure CNTBaseN
Sami Mujawar [Fri, 10 May 2019 07:52:07 +0000 (08:52 +0100)]
N1SDP: Initialise CNTFRQ in Non Secure CNTBaseN

N1SDP exhibits the behavior similar to Juno wherein CNTBaseN.CNTFRQ
can be written but does not reflect the value of the CNTFRQ register
in CNTCTLBase frame. This doesn't follow ARM ARM in that the value
updated in CNTCTLBase.CNTFRQ is not reflected in CNTBaseN.CNTFRQ.

Hence enable the workaround (applied to Juno) for N1SDP that updates
the CNTFRQ register in the Non Secure CNTBaseN frame.

Change-Id: Id89ee1bca0f25c9d62f8f794f2c4f4e618cdf092
Signed-off-by: Sami Mujawar <sami.mujawar@arm.com>
5 years agoN1SDP: Fix DRAM2 start address
Sami Mujawar [Thu, 9 May 2019 12:43:30 +0000 (13:43 +0100)]
N1SDP: Fix DRAM2 start address

The default DRAM2 start address for Arm platforms
is 0x880000000. However, for N1SDP platform this is
0x8080000000.

Fix the DRAM2 start address by initialising
PLAT_ARM_DRAM2_BASE.

Without this fix there is a mismatch of the System
memory region view as seen by the BL31 runtime
firmware (PSCI) versus the view of the OS (which
is based on the description provided by UEFI. In
this case UEFI is correctly describing the DRAM2
start address).

This implicates in secondary cores failing to start
on some Operating Systems if the OS decides to place
the secondary start address in the mismatched region.

Change-Id: I57220e753219353dda429868b4c5e1a69944cc64
Signed-off-by: Sami Mujawar <sami.mujawar@arm.com>
5 years agoAdd option for defining platform DRAM2 base
Sami Mujawar [Thu, 9 May 2019 12:35:02 +0000 (13:35 +0100)]
Add option for defining platform DRAM2 base

The default DRAM2 base address for Arm platforms
is 0x880000000. However, on some platforms the
firmware may want to move the start address to
a different value.

To support this introduce PLAT_ARM_DRAM2_BASE that
defaults to 0x880000000; but can be overridden by
a platform (e.g. in platform_def.h).

Change-Id: I0d81195e06070bc98f376444b48ada2db1666e28
Signed-off-by: Sami Mujawar <sami.mujawar@arm.com>
5 years agoDisable speculative loads only if SSBS is supported
Sami Mujawar [Fri, 10 May 2019 13:28:37 +0000 (14:28 +0100)]
Disable speculative loads only if SSBS is supported

Examine the ID_AA64PFR1_EL1 bits 7:4 to see if speculative
loads (SSBS) is implemented, before disabling speculative
loads.

Change-Id: I7607c45ed2889260d22a94f6fd9af804520acf67
Signed-off-by: Sami Mujawar <sami.mujawar@arm.com>
5 years agodrivers: ufs: Extend the delay after reset to wait for some slower chips
John Stultz [Mon, 13 May 2019 23:56:19 +0000 (16:56 -0700)]
drivers: ufs: Extend the delay after reset to wait for some slower chips

We've seen issues with some THG based UFS chips, where
after reset the LUNs don't always enumerate properly.

After some debugging, we found that extending the mdelay
here seems to resolve the issue by giving the chips enough
time to complete reset.

Change-Id: I848f810b2438ed6ad3d33db614c61d2cef9ac400
Signed-off-by: John Stultz <john.stultz@linaro.org>
5 years agoplat: imx8mq: Remove duplicated linker symbols
Jacky Bai [Tue, 9 Apr 2019 02:55:24 +0000 (10:55 +0800)]
plat: imx8mq: Remove duplicated linker symbols

Remove duplicated linker symbols, resue the symbols
defined in bl_common.h

Change-Id: I10de450eccc78c09b61a8ae7126bf4f4029fa682
Signed-off-by: Jacky Bai <ping.bai@nxp.com>
5 years agoRemove .arch directives from spinlock.S
Alexei Fedorov [Fri, 10 May 2019 15:55:16 +0000 (16:55 +0100)]
Remove .arch directives from spinlock.S

This patch removes .arch "arm8.1-a" and "armv8-a"
directives which overwrite ASFLAGS_aarch64 option based
on ARM_ARCH_MINOR passed to Makefile and cause
translation errors like
"selected processor does not support `bti jc'"
for armv8.5-a targets when BTI support is enabled.

Change-Id: Idca5b66ed1e5d86e2188b0c0f16c3819990957c4
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
5 years agoSMMUv3: Abort DMA transactions
Alexei Fedorov [Thu, 9 May 2019 11:14:40 +0000 (12:14 +0100)]
SMMUv3: Abort DMA transactions

For security DMA should be blocked at the SMMU by default
unless explicitly enabled for a device. SMMU is disabled
after reset with all streams bypassing the SMMU, and
abortion of all incoming transactions implements a default
deny policy on reset.
This patch also moves "bl1_platform_setup()" function from
arm_bl1_setup.c to FVP platforms' fvp_bl1_setup.c and
fvp_ve_bl1_setup.c files.

Change-Id: Ie0ffedc10219b1b884eb8af625bd4b6753749b1a
Signed-off-by: Alexei Fedorov <Alexei.Fedorov@arm.com>
5 years agoMerge "Initialize platform for MediaTek mt8183" into integration
Soby Mathew [Fri, 10 May 2019 10:42:18 +0000 (10:42 +0000)]
Merge "Initialize platform for MediaTek mt8183" into integration

5 years agoMerge "maintainers: Step down as sub-maintainer" into integration
Antonio Niño Díaz [Fri, 10 May 2019 08:57:32 +0000 (08:57 +0000)]
Merge "maintainers: Step down as sub-maintainer" into integration

5 years agoMerge "plat: imx8m: Implement IMX_SIP_BUILDINFO" into integration
Antonio Niño Díaz [Fri, 10 May 2019 08:51:29 +0000 (08:51 +0000)]
Merge "plat: imx8m: Implement IMX_SIP_BUILDINFO" into integration

5 years agoInitialize platform for MediaTek mt8183
kenny liang [Wed, 10 Apr 2019 13:09:26 +0000 (21:09 +0800)]
Initialize platform for MediaTek mt8183

- Add basic platform setup
- Add generic CPU helper functions
- Add delay timer platform implementation
- Use TI 16550 uart driver

Change-Id: I1c29569c68fe9fca5e10e88a22a29690bab7141f
Signed-off-by: kenny liang <kenny.liang@mediatek.com>
5 years agoplat: imx8m: Implement IMX_SIP_BUILDINFO
Leonard Crestez [Wed, 8 May 2019 19:29:21 +0000 (22:29 +0300)]
plat: imx8m: Implement IMX_SIP_BUILDINFO

The IMX_SIP_BUILDINFO call was implemented for imx8qm and imx8qx but
it's also applicable to imx8m.

This fixes U-Boot not printing commit hash on 8m with upstream TF-A.

Change-Id: Idcfd9729eaaccf329c24e241da325f1f6cd3c880
Signed-off-by: Leonard Crestez <leonard.crestez@nxp.com>
5 years agomaintainers: Step down as sub-maintainer
Antonio Nino Diaz [Thu, 9 May 2019 13:26:22 +0000 (14:26 +0100)]
maintainers: Step down as sub-maintainer

I'm giving full maintainership of the Raspberry Pi 3 platform port to
Paul. I'm also leaving the GXBB maintainership to Andre, who is also
happy to pass it on to someone else who is more interested in it.

Change-Id: Ieb2212f5fc11ebde9fc0c857e9e305d691d4ee3f
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge "Add Makefile check for PAuth and AArch64" into integration
Antonio Niño Díaz [Thu, 9 May 2019 12:49:37 +0000 (12:49 +0000)]
Merge "Add Makefile check for PAuth and AArch64" into integration

5 years agoAdd Makefile check for PAuth and AArch64
John Tsichritzis [Tue, 7 May 2019 13:09:09 +0000 (14:09 +0100)]
Add Makefile check for PAuth and AArch64

Pointer authentication is supported only in AArch64. A relevant check is
added for that in the Makefile.

Change-Id: I021ba65a9bd5764fd33292bee42617015e04a870
Signed-off-by: John Tsichritzis <john.tsichritzis@arm.com>
5 years agoMerge "docs: Update contribution guidelines for binary components" into integration
Soby Mathew [Wed, 8 May 2019 13:41:56 +0000 (13:41 +0000)]
Merge "docs: Update contribution guidelines for binary components" into integration