project/bcm63xx/atf.git
5 years agoxilinx: zynqmp: Move zynqmp_ipi.h to include directory
Jolly Shah [Tue, 8 Jan 2019 19:21:29 +0000 (11:21 -0800)]
xilinx: zynqmp: Move zynqmp_ipi.h to include directory

Move zynqmp_ipi.h to platform specific include directory.
Rename it to plat_ipi.h instead of platform name. So, it can
be used to common source files which needs platform specific
data.

Signed-off-by: Tejas Patel <tejas.patel@xilinx.com>
Reviewed-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agoplat: xilinx: zynqmp: Move zynqmp_def.h to include directory
Jolly Shah [Tue, 8 Jan 2019 19:16:16 +0000 (11:16 -0800)]
plat: xilinx: zynqmp: Move zynqmp_def.h to include directory

Move zynqmp_def.h to platform specific include directory.
Also, update source file to include header file from updated
path of zynqmp_def.h

Signed-off-by: Tejas Patel <tejas.patel@xilinx.com>
Reviewed-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agoplat: xilinx: zynqmp: Move zynqmp_private.h to include directory
Jolly Shah [Tue, 8 Jan 2019 19:10:47 +0000 (11:10 -0800)]
plat: xilinx: zynqmp: Move zynqmp_private.h to include directory

Move zynqmp_private.h to platform specific include directory.
Also, rename it to plat_private.h instead of having platform
name. So, it can be used to common source files which needs
platform specific data.

Signed-off-by: Tejas Patel <tejas.patel@xilinx.com>
Reviewed-by: Siva Durga Prasad Paladugu <siva.durga.paladugu@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agoMerge pull request #1743 from JoelHutton/jh/fixcssdriver
Antonio Niño Díaz [Wed, 9 Jan 2019 11:44:44 +0000 (11:44 +0000)]
Merge pull request #1743 from JoelHutton/jh/fixcssdriver

juno:Fix CSS_USE_SCMI_SDS_DRIVER=0 configuration

5 years agojuno:Fix CSS_USE_SCMI_SDS_DRIVER=0 configuration
Joel Hutton [Tue, 8 Jan 2019 15:18:28 +0000 (15:18 +0000)]
juno:Fix CSS_USE_SCMI_SDS_DRIVER=0 configuration

A previous commit 89f2e589856f ("plat/arm: remove weak implemention of
'plat_arm_psci_override_pm_ops' function") introduced a problem with the
CSS_USE_SCMI_SDS_DRIVER configuration. In juno_pm.c the
css_scmi_override_pm_ops function was used regardless of whether the
flag was set. This patch ifdefs the function to restore the original
behaviour.

Change-Id: I508025ba70cf3a9250cc6270c1df209179c37ae7
Signed-off-by: Joel Hutton <Joel.Hutton@Arm.com>
5 years agoMerge pull request #1734 from marex/arm/master/update-rcar-2.0.0
Antonio Niño Díaz [Tue, 8 Jan 2019 15:46:29 +0000 (15:46 +0000)]
Merge pull request #1734 from marex/arm/master/update-rcar-2.0.0

Arm/master/update rcar 2.0.0

5 years agorcar_gen3: plat: update renesas version to 2.0.0
Marek Vasut [Mon, 31 Dec 2018 13:51:14 +0000 (14:51 +0100)]
rcar_gen3: plat: update renesas version to 2.0.0

Update the reported Renesas version to 2.0.0, since the DDR/QoS/PFC
table versions match the ones from that release.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: pwrc: Switch to common delay implementation
Marek Vasut [Fri, 28 Dec 2018 19:15:33 +0000 (20:15 +0100)]
rcar_gen3: drivers: pwrc: Switch to common delay implementation

Replace the ad-hoc implementation of delay in PWRC driver
with common R-Car delay code.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: delay: Rewrite from assembler to C
Marek Vasut [Wed, 26 Dec 2018 14:57:08 +0000 (15:57 +0100)]
rcar_gen3: drivers: delay: Rewrite from assembler to C

Rewrite the delay code from assembler to C.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Add generic timer init
Marek Vasut [Fri, 28 Dec 2018 19:12:13 +0000 (20:12 +0100)]
rcar_gen3: plat: Add generic timer init

Add code to determine the platform timer frequency and configure
the generic timer accordingly early in BL2.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: scif: Use TEND flag for transmission end detection
Marek Vasut [Thu, 27 Dec 2018 19:31:22 +0000 (20:31 +0100)]
rcar_gen3: drivers: scif: Use TEND flag for transmission end detection

Use the SCIF SCFSR:TEND bit to check that all data were transmitted by
the SCIF and that there are no more valid data to transmit in the FIFO.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: qos: Synchronize tables
Marek Vasut [Wed, 12 Dec 2018 15:35:00 +0000 (16:35 +0100)]
rcar_gen3: drivers: qos: Synchronize tables

Synchronize the QoS tables with Renesas ATF release 2.0.0 .

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: pfc: Synchronize tables
Marek Vasut [Wed, 12 Dec 2018 16:40:10 +0000 (17:40 +0100)]
rcar_gen3: drivers: pfc: Synchronize tables

Synchronize the pin control tables with Renesas ATF release 2.0.0 .

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: ddr-b: Synchronize tables
Marek Vasut [Wed, 12 Dec 2018 17:06:39 +0000 (18:06 +0100)]
rcar_gen3: drivers: ddr-b: Synchronize tables

Synchronize the R-Car DDR-B driver, used on R-Car H3/M3W/M3N,
with Renesas ATF release 2.0.0 .

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: ddr-a: Checkpatch cleanup
Marek Vasut [Sun, 16 Dec 2018 18:28:59 +0000 (19:28 +0100)]
rcar_gen3: drivers: ddr-a: Checkpatch cleanup

Run Linux kernel checkpatch on the DDR-A init code to clean it up:
$ checkpatch.pl --fix --fix-inplace -f drivers/staging/renesas/rcar/ddr/ddr_a/*

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: ddr-a: Synchronize tables
Marek Vasut [Wed, 12 Dec 2018 17:06:39 +0000 (18:06 +0100)]
rcar_gen3: drivers: ddr-a: Synchronize tables

Synchronize the R-Car DDR-A driver, used on R-Car E3,
with Renesas ATF release 2.0.0 .

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: cpld: Move rcar_cpld_reset_cpu() into header
Marek Vasut [Fri, 28 Dec 2018 10:26:03 +0000 (11:26 +0100)]
rcar_gen3: drivers: cpld: Move rcar_cpld_reset_cpu() into header

Move the rcar_cpld_reset_cpu() function into header file and zap the externs.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: swdt: Access SCR in EL3
Marek Vasut [Thu, 27 Dec 2018 19:28:45 +0000 (20:28 +0100)]
rcar_gen3: drivers: swdt: Access SCR in EL3

The code runs in EL3, use EL3 accessors to manipulate the interrupt bit.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: drivers: auth-mod: Access SCTLR in EL3
Marek Vasut [Thu, 27 Dec 2018 19:26:01 +0000 (20:26 +0100)]
rcar_gen3: drivers: auth-mod: Access SCTLR in EL3

The code runs in EL3, use EL3 accessors to manipulate the cache bits.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Dump EL3 interrupt error registers
Marek Vasut [Fri, 28 Dec 2018 19:23:36 +0000 (20:23 +0100)]
rcar_gen3: plat: Dump EL3 interrupt error registers

Since the interrupts are handled in EL3, dump the EL3 error registers
in case an error happens.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Enable programmable CPU reset address
Marek Vasut [Mon, 31 Dec 2018 14:57:11 +0000 (15:57 +0100)]
rcar_gen3: plat: Enable programmable CPU reset address

The reset address is programmable on the R-Car Gen3, enable it.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Disable IPMMU PV0 cache on E3
Marek Vasut [Mon, 31 Dec 2018 15:48:04 +0000 (16:48 +0100)]
rcar_gen3: plat: Disable IPMMU PV0 cache on E3

Disable the IPMMU PV0 cache on E3 rev. 1.x .

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Add E3 rev. 1.1 support
Marek Vasut [Fri, 28 Dec 2018 19:11:38 +0000 (20:11 +0100)]
rcar_gen3: plat: Add E3 rev. 1.1 support

Add support for R-Car E3 silicon rev. 1.1

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Add missing platform auto-detection name
Marek Vasut [Fri, 28 Dec 2018 19:11:26 +0000 (20:11 +0100)]
rcar_gen3: plat: Add missing platform auto-detection name

Add missing TARGET_NAME for the case where RCAR_LSI is set to AUTO,
which is platform auto-detection.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Clean up rcar_pwrc_code_copy_to_system_ram()
Marek Vasut [Mon, 31 Dec 2018 16:12:45 +0000 (17:12 +0100)]
rcar_gen3: plat: Clean up rcar_pwrc_code_copy_to_system_ram()

Call the function only from architecture setup and at the end of
suspend cycle instead of calling it all over the place.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Fix BL2 size check
Marek Vasut [Wed, 26 Dec 2018 14:57:08 +0000 (15:57 +0100)]
rcar_gen3: plat: Fix BL2 size check

Rename BL2_LIMIT to BL2_IMAGE_LIMIT and BL2_SYSRAM_LIMIT to BL2_LIMIT to
correctly set BL2_LIMIT value. Set correct DEVICE_SRAM_BASE to match the
hardware. Use BL2_END in rcar_configure_mmu_el3() to mark the cacheable
BL2 area.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Staticize memory maps
Marek Vasut [Sun, 30 Dec 2018 16:19:03 +0000 (17:19 +0100)]
rcar_gen3: plat: Staticize memory maps

Staticize the platform memory map tables as they are only used within
the platform_common.c file.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Function cleanup
Marek Vasut [Wed, 26 Dec 2018 14:57:08 +0000 (15:57 +0100)]
rcar_gen3: plat: Function cleanup

Replace foo_t with struct foo.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Fix cache line size
Marek Vasut [Sun, 30 Dec 2018 16:21:39 +0000 (17:21 +0100)]
rcar_gen3: plat: Fix cache line size

The CPU has cache line size of 64 Bytes, fix the cache line size.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Disable SVE
Marek Vasut [Fri, 28 Dec 2018 19:38:09 +0000 (20:38 +0100)]
rcar_gen3: plat: Disable SVE

Apply 3872fc2d1fc5 ("Do not enable SVE on pre-v8.2 platforms") to
R-Car Gen3 too.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Add missing dependency to rcar_srecord
Marek Vasut [Mon, 31 Dec 2018 18:47:37 +0000 (19:47 +0100)]
rcar_gen3: plat: Add missing dependency to rcar_srecord

Add missing dependency on the bl2.elf and bl31.elf into the rcar_srecord
target, which uses those ELF files to generate the SRECs.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Rename H3 label
Marek Vasut [Tue, 8 Jan 2019 13:05:39 +0000 (14:05 +0100)]
rcar_gen3: plat: Rename H3 label

Rename the H3 label to avoid confusing clang, which generates an error
if the label is just H3. Rename it to RCARH3.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agorcar_gen3: plat: Drop ddr_regdef_len()
Marek Vasut [Tue, 8 Jan 2019 13:04:08 +0000 (14:04 +0100)]
rcar_gen3: plat: Drop ddr_regdef_len()

This function is unused and triggers clang error, drop it.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
5 years agoMerge pull request #1740 from soby-mathew/sm/restrict_pie_to_fvp
Antonio Niño Díaz [Tue, 8 Jan 2019 11:28:44 +0000 (11:28 +0000)]
Merge pull request #1740 from soby-mathew/sm/restrict_pie_to_fvp

plat/arm: Restrict PIE support to FVP

5 years agoMerge pull request #1742 from sandrine-bailleux-arm/sb/fix-arm-mk
Antonio Niño Díaz [Tue, 8 Jan 2019 11:28:24 +0000 (11:28 +0000)]
Merge pull request #1742 from sandrine-bailleux-arm/sb/fix-arm-mk

Fix a variable expansion in Arm platform makefiles

5 years agoMerge pull request #1741 from deepan02/deepan02-arm-uart-definations
Antonio Niño Díaz [Tue, 8 Jan 2019 11:28:12 +0000 (11:28 +0000)]
Merge pull request #1741 from deepan02/deepan02-arm-uart-definations

plat/arm/n1sdp: define the uart constants for N1SDP

5 years agoMerge pull request #1739 from Yann-lms/includes
Antonio Niño Díaz [Tue, 8 Jan 2019 09:10:32 +0000 (09:10 +0000)]
Merge pull request #1739 from Yann-lms/includes

stm32mp1: do not include platform header files directly in drivers

5 years agoMerge pull request #1736 from antonio-nino-diaz-arm/an/maintainers
Antonio Niño Díaz [Tue, 8 Jan 2019 09:10:22 +0000 (09:10 +0000)]
Merge pull request #1736 from antonio-nino-diaz-arm/an/maintainers

maintainers: Fix path of Marvell documentation

5 years agoMerge pull request #1732 from jollysxilinx/integration
Antonio Niño Díaz [Tue, 8 Jan 2019 09:09:23 +0000 (09:09 +0000)]
Merge pull request #1732 from jollysxilinx/integration

plat: xilinx: Clock and PLL EEMI API Support

5 years agoplat/arm/n1sdp: define the uart constants for N1SDP
Deepak Pandey [Tue, 18 Dec 2018 11:40:24 +0000 (17:10 +0530)]
plat/arm/n1sdp: define the uart constants for N1SDP

This patch removes the dependency of the N1SDP on soc
css defines in order to let the N1SDP platform port
define the uart related constants.

Change-Id: If13796f278586a01512ee99615502b30e478189e
Signed-off-by: Deepak Pandey <Deepak.Pandey@arm.com>
5 years agoFix a variable expansion in Arm platform makefiles
Sandrine Bailleux [Mon, 7 Jan 2019 14:35:37 +0000 (15:35 +0100)]
Fix a variable expansion in Arm platform makefiles

The top level makefile defines the PLAT variable, not PLATFORM.
This mistake was causing an empty variable expansion and showing an
incomplete error message.

Change-Id: I5da1275c73c61a7c1823643a76300f255841719d
Signed-off-by: Sandrine Bailleux <sandrine.bailleux@arm.com>
5 years agoplat/arm: Restrict PIE support to FVP
Soby Mathew [Mon, 7 Jan 2019 14:07:58 +0000 (14:07 +0000)]
plat/arm: Restrict PIE support to FVP

The patch SHA 55cf015c enabled PIE support when RESET_TO_BL31=1 for
all ARM platforms. But it seems n1sdp platform doesn't work with PIE
support yet. Hence restrict the ENABLE_PIE=1 to fvp platform.

Change-Id: If44e42528e4b0b57c69084503f346576fe0748bd
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
5 years agoMerge pull request #1737 from antonio-nino-diaz-arm/an/asm-assert
Antonio Niño Díaz [Mon, 7 Jan 2019 11:06:19 +0000 (11:06 +0000)]
Merge pull request #1737 from antonio-nino-diaz-arm/an/asm-assert

Remove ASM_ASSERTION check in Makefile

5 years agoMerge pull request #1735 from antonio-nino-diaz-arm/an/load-image-v2
Antonio Niño Díaz [Mon, 7 Jan 2019 11:06:09 +0000 (11:06 +0000)]
Merge pull request #1735 from antonio-nino-diaz-arm/an/load-image-v2

plat/arm: Remove comment that mentions LOAD_IMAGE_V2

5 years agostm32mp1: do not include platform header files directly in drivers
Yann Gautier [Mon, 7 Jan 2019 10:17:24 +0000 (11:17 +0100)]
stm32mp1: do not include platform header files directly in drivers

Instead, only platform_def.h is included.
The required files to be included are added in stm32mp1_def.h.

Signed-off-by: Yann Gautier <yann.gautier@st.com>
5 years agozynqmp: pm: Invalidate unused APLL_TO_LPD clock
Jolly Shah [Wed, 2 Jan 2019 21:53:33 +0000 (13:53 -0800)]
zynqmp: pm: Invalidate unused APLL_TO_LPD clock

This clock does not drive any clock in LPD so there is no need for
Linux to try to initialize it.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Invalidate several clocks that Linux doesn't need to control
Jolly Shah [Fri, 4 Jan 2019 20:03:44 +0000 (12:03 -0800)]
zynqmp: pm: Invalidate several clocks that Linux doesn't need to control

Linux has no reason to use these system and debug clocks and therefore
shouldn't access them. These clocks are marked as invalid in order to
prevent Linux from registering and querying them.

Note that despite clocks being marked as invalid a security issue
still remains in place as there is nothing that prevents the
non-secure world from gating these clocks and that way causing
damage to the system.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Add ACPU_FULL and ACPU_HALF clocks in the invalid list
Jolly Shah [Wed, 2 Jan 2019 21:48:19 +0000 (13:48 -0800)]
zynqmp: pm: Add ACPU_FULL and ACPU_HALF clocks in the invalid list

These clocks are marked as invalid in order to prevent Linux from
registering them.

Note that despite clocks being marked as invalid a security issue
still remains in place as there is nothing that prevents the
non-secure world from gating these clocks and that way halt
the whole APU subsystem.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Fix model of ACPU clocks
Jolly Shah [Wed, 2 Jan 2019 21:45:53 +0000 (13:45 -0800)]
zynqmp: pm: Fix model of ACPU clocks

In the existing model for ACPU clock the mux, divider, and gate were
represented as one clock and ACPU_HALF was modelled as child of
ACPU clock. This is not correct. ACPU clock model contains only
mux and the divider, and it has 2 children: ACPU_FULL and ACPU_HALF
clocks which have only gates. The models of ACPU and ACPU_HALF clocks
are fixed and ACPU_FULL clock is added.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock get parent EEMI API
Jolly Shah [Wed, 2 Jan 2019 21:44:25 +0000 (13:44 -0800)]
zynqmp: pm: Reimplement clock get parent EEMI API

Clock get parent EEMI API is reimplemented to use system-level clock
and pll EEMI APIs rather than direct MMIO read/write accesses to clock
and pll control registers.
Since linux still uses clock set parent API to get pre_src, post_src, div2
and bypasss, in the implementation of pm_clock_get_parent() we need to
workaround this by distinguishing two cases:
1) if the given clock ID corresponds to a PLL-related clock ID (*_PRE_SRC,
   *_POST_SRC, *_INT_MUX or *_PLL clock IDs); or
2) given clock ID is truly an on-chip clock.
For case 1) we'll map the call onto PLL-specific EEMI API with the
respective parameter ID. For case 2) the call is passed to the PMU.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock set parent EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:57:40 +0000 (11:57 -0800)]
zynqmp: pm: Reimplement clock set parent EEMI API

Clock set parent EEMI API is reimplemented to use system-level clock
and pll EEMI APIs rather than direct MMIO read/write accesses to clock
and pll control registers.
Since linux still uses clock set parent API to set pre_src, post_src, div2
and bypass, in the implementation of pm_clock_set_parent() we need to
workaround this by distinguishing two cases:
1) if the given clock ID corresponds to a PLL-related clock ID (*_PRE_SRC,
*_POST_SRC, *_INT_MUX or *PLL clock IDs); or 2) given clock ID is truly
an on-chip clock.
For case 1) we'll map the call onto PLL set parameter EEMI API with the
respective parameter ID. Since clock set parent interface to EL1/2 receives
parent index (mux select value), the value is just passed to PMU.
Functions that appear to be unused after this change is made are removed.

Setting the parent of *PLL clocks, that actually model bypass, is not
possible. This is already ensured by the existing clock model having the
CLK_SET_RATE_NO_REPARENT flag. The API also doesn't allow changing the
bypass parent. Bypass is controlled only by the PMU firmware.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Cleanup for clock set/get rate EEMI API
Jolly Shah [Wed, 2 Jan 2019 21:39:30 +0000 (13:39 -0800)]
zynqmp: pm: Cleanup for clock set/get rate EEMI API

Clock set/get rate are not implemented and will likely never be.
Remove empty function stubs.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock get divider EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:53:37 +0000 (11:53 -0800)]
zynqmp: pm: Reimplement clock get divider EEMI API

Clock get divider EEMI API is reimplemented to use system-level clock
get divider EEMI API rather than direct MMIO read/write accesses to clock
control registers.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock set divider EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:49:46 +0000 (11:49 -0800)]
zynqmp: pm: Reimplement clock set divider EEMI API

Clock set divider EEMI API is reimplemented to use system-level clock
set divider EEMI API rather than direct MMIO read/write accesses to clock
control registers.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock get state (status) EEMI API
Jolly Shah [Wed, 2 Jan 2019 20:55:41 +0000 (12:55 -0800)]
zynqmp: pm: Reimplement clock get state (status) EEMI API

Clock get state EEMI API is reimplemented to use system-level clock
and pll EEMI APIs rather than direct MMIO read/write accesses to clock
and pll control registers.
Since linux is_enabled method for PLLs still uses clock get state API
get the PLL state, in the implementation of pm_clock_getstate() we need
to workaround this by distinguishing two cases: 1) if the given clock ID
corresponds to a PLL output clock ID; or 2) given clock ID is truly an
on-chip clock whose state of the gate should be returned.
For case 1) we'll call pm_api_clock_pll_getstate() implemented in
pm_api_clock.h/c. This function will query the PLL state from PMU using
the system-level PLL get mode EEMI API.
For case 2) we'll call the PMU to query the clock gate state using
system-level clock get status EEMI API.
Functions that appear to be unused after this change is made are removed.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock disable EEMI API
Jolly Shah [Wed, 2 Jan 2019 20:54:40 +0000 (12:54 -0800)]
zynqmp: pm: Reimplement clock disable EEMI API

Clock disable EEMI API is reimplemented to use system-level clock
and pll EEMI APIs rather than direct MMIO read/write accesses to clock
and pll control registers.
Since linux still uses clock disable API to reset the PLL in the
implementation of pm_clock_disable() we need to workaround this by
distinguishing two cases: 1) if the given clock ID corresponds to a PLL
output clock ID; or 2) given clock ID is truly an on-chip clock that can
be gated.
For case 1) we'll call pm_api_clock_pll_disable() implemented in
pm_api_clock.h/c. This function will reset the PLL using the system-level
PLL set mode EEMI API with the reset mode argument.
For case 2) we'll call the PMU to configure the clock gate. This is done
using system-level clock disable EEMI API.
Functions that appear to be unused after this change is made are removed.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Reimplement clock enable EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:45:59 +0000 (11:45 -0800)]
zynqmp: pm: Reimplement clock enable EEMI API

Clock enable EEMI API is reimplemented to use system-level clock
and pll EEMI APIs rather than direct MMIO read/write accesses to clock
and pll control registers.
Since linux still uses clock enable API to trigger locking of the PLLs
in the pm_clock_enable() implementation we need to workaround this by
distinguishing two cases: 1) if the given clock ID corresponds to a PLL
output clock ID; or 2) given clock ID is truly an on-chip clock that can
be gated.
For case 1) we'll call pm_api_clock_pll_enable() implemented in
pm_api_clock.h/c. This function checks what is the buffered PLL mode and
calls the system-level PLL set mode EEMI API with the buffered mode value
specified as argument. Long term, if linux driver get fixed to use PLL
EEMI API to control PLLs, this case could be removed from ATF.
For case 2) we'll call the PMU to configure the clock gate. This is done
using system-level clock enable EEMI API.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Return the buffered PLL mode through IOCTL PLL get mode API
Jolly Shah [Wed, 2 Jan 2019 20:49:21 +0000 (12:49 -0800)]
zynqmp: pm: Return the buffered PLL mode through IOCTL PLL get mode API

When linux calls pm_ioctl_get_pll_frac_mode() it doesn't expect the actual
mode to be read from hardware, but the value that it is intending to
program. Therefore, we return the buffered value to linux.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Buffer the PLL mode that is set using IOCTL API
Jolly Shah [Wed, 2 Jan 2019 20:46:46 +0000 (12:46 -0800)]
zynqmp: pm: Buffer the PLL mode that is set using IOCTL API

When linux calls pm_ioctl_set_pll_frac_mode() it doesn't expect the
fractional mode to be changed in hardware. Furthermore, even before this
patch setting the mode which is done by writing into register takes
no effect until the PLL reset is deasserted, i.e. until linux "enables"
the PLL. To adjust the code to system-level PLL EEMI API and avoid
unnecessary IPIs that would otherwise be issued, we buffer the mode
value set via IOCTL until the PLL mode really needs to be set.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Set PLL fractional data using PLL set parameter EEMI API
Jolly Shah [Wed, 2 Jan 2019 20:42:56 +0000 (12:42 -0800)]
zynqmp: pm: Set PLL fractional data using PLL set parameter EEMI API

Fractional data should be set using PLL set parameter EEMI API. This
stands for system-level communication (APU to PMU). Since linux
already uses a specific IOCTL function to do this and we need to
keep it that way, the pll clock ID given by linux has to be mapped
to the pll node ID that is communicated at the system-level (argument
of PLL set parameter API).
With this modification the function pm_api_clk_set_pll_frac_data is
removed from pm_api_clock.c/h because it became unused.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Get PLL fractional data using PLL get parameter EEMI API
Jolly Shah [Wed, 2 Jan 2019 20:40:17 +0000 (12:40 -0800)]
zynqmp: pm: Get PLL fractional data using PLL get parameter EEMI API

Fractional data should be get using PLL get parameter EEMI API. This
stands for system-level communication (APU to PMU). Since linux
already uses a specific IOCTL function to do this and we need to
keep it that way, the pll clock ID given by linux has to be mapped
to the pll node ID that is communicated at the system-level (argument
of PLL get parameter API).
With this modification the function pm_api_clk_get_pll_frac_data is
removed from pm_api_clock.c/h because it became unused.
The clock enum is defined as 'enum clock_id'.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Implement PLL get mode EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:35:48 +0000 (11:35 -0800)]
zynqmp: pm: Implement PLL get mode EEMI API

This API will be used to get the currently configured PLL mode:
reset (bypassed and unlocked), integer or fractional (locked).

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Implement PLL set mode EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:32:31 +0000 (11:32 -0800)]
zynqmp: pm: Implement PLL set mode EEMI API

This API will be used to set the PLL mode: reset (unlocked), integer
or fractional (locked). If reset mode is set the PM controller will
bypass the target PLL prior to asserting the reset. If integer or
fractional mode is set the PM controller will program and trigger
locking of the PLL. If success status is returned the PLL is locked
and its bypass is deasserted. If fractional mode is set the fractional
divider (data parameter) has to have a non-zero value prior to issuing
pll set fractional mode. The caller need to ensure that the data
parameter is properly set using pll get/set parameter EEMI API.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Implement PLL get parameter EEMI API
Jolly Shah [Fri, 4 Jan 2019 19:28:38 +0000 (11:28 -0800)]
zynqmp: pm: Implement PLL get parameter EEMI API

This API will be used to get a parameter for the PLL. Parameter values
represent the values as defined in the Zynq MPSoC register reference
manual ug1087.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agozynqmp: pm: Implement PLL set parameter EEMI API
Jolly Shah [Wed, 2 Jan 2019 20:27:00 +0000 (12:27 -0800)]
zynqmp: pm: Implement PLL set parameter EEMI API

This API will be used to set a parameter for the PLL. The parameter
value that is set will have effect once the PLL mode is set to integer
or fractional mode. Parameter values represent the values as defined
in the Zynq MPSoC register reference manual ug1087.

Signed-off-by: Mirela Simonovic <mirela.simonovic@aggios.com>
Acked-by: Will Wong <WILLW@xilinx.com>
Signed-off-by: Jolly Shah <jollys@xilinx.com>
5 years agoRemove ASM_ASSERTION check in Makefile
Antonio Nino Diaz [Fri, 4 Jan 2019 14:05:26 +0000 (14:05 +0000)]
Remove ASM_ASSERTION check in Makefile

ASM_ASSERTION was deprecated long ago, this check is no longer needed.

Change-Id: I2a5770f76ea1317461c0059dad8dba9dc0e5af32
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge pull request #1726 from antonio-nino-diaz-arm/an/includes
Antonio Niño Díaz [Fri, 4 Jan 2019 11:42:16 +0000 (11:42 +0000)]
Merge pull request #1726 from antonio-nino-diaz-arm/an/includes

Sanitise includes across codebase

5 years agomaintainers: Fix path of Marvell documentation
Antonio Nino Diaz [Fri, 4 Jan 2019 10:58:30 +0000 (10:58 +0000)]
maintainers: Fix path of Marvell documentation

Change-Id: Idd5a2b700b936b22e94194dfadee68b968d03380
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agorpi3: Remove unneeded include paths from Makefile
Antonio Nino Diaz [Mon, 17 Dec 2018 17:49:35 +0000 (17:49 +0000)]
rpi3: Remove unneeded include paths from Makefile

Change-Id: I20d164f7573ebc24935e92a924472e29c8a0e3a0
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoSanitise includes across codebase
Antonio Nino Diaz [Fri, 14 Dec 2018 00:18:21 +0000 (00:18 +0000)]
Sanitise includes across codebase

Enforce full include path for includes. Deprecate old paths.

The following folders inside include/lib have been left unchanged:

- include/lib/cpus/${ARCH}
- include/lib/el3_runtime/${ARCH}

The reason for this change is that having a global namespace for
includes isn't a good idea. It defeats one of the advantages of having
folders and it introduces problems that are sometimes subtle (because
you may not know the header you are actually including if there are two
of them).

For example, this patch had to be created because two headers were
called the same way: e0ea0928d5b7 ("Fix gpio includes of mt8173 platform
to avoid collision."). More recently, this patch has had similar
problems: 46f9b2c3a282 ("drivers: add tzc380 support").

This problem was introduced in commit 4ecca33988b9 ("Move include and
source files to logical locations"). At that time, there weren't too
many headers so it wasn't a real issue. However, time has shown that
this creates problems.

Platforms that want to preserve the way they include headers may add the
removed paths to PLAT_INCLUDES, but this is discouraged.

Change-Id: I39dc53ed98f9e297a5966e723d1936d6ccf2fc8f
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoReorganize architecture-dependent header files
Antonio Nino Diaz [Mon, 17 Dec 2018 17:20:57 +0000 (17:20 +0000)]
Reorganize architecture-dependent header files

The architecture dependant header files in include/lib/${ARCH} and
include/common/${ARCH} have been moved to /include/arch/${ARCH}.

Change-Id: I96f30fdb80b191a51448ddf11b1d4a0624c03394
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoplat/arm: Always define TSP memory region
Antonio Nino Diaz [Fri, 14 Dec 2018 01:27:19 +0000 (01:27 +0000)]
plat/arm: Always define TSP memory region

Even though this is not used unless SPD=tspd, only defining it when
SPD_tspd is defined doesn't have any advantage and it makes it harder to
read the code.

Change-Id: I3d93135e05f39be071d16f8a47394a9a3ff54bc8
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge pull request #1731 from miyatsu/doc-fix-20181225
Antonio Niño Díaz [Fri, 4 Jan 2019 09:14:22 +0000 (09:14 +0000)]
Merge pull request #1731 from miyatsu/doc-fix-20181225

docs: marvell: Fix typo in file build.txt

5 years agoMerge pull request #1728 from satheesbalya-arm/sb1/sb1_2497_v84_dit
Antonio Niño Díaz [Fri, 4 Jan 2019 09:13:32 +0000 (09:13 +0000)]
Merge pull request #1728 from satheesbalya-arm/sb1/sb1_2497_v84_dit

Enable DIT if supported

5 years agoEnable DIT if supported
Sathees Balya [Thu, 6 Dec 2018 13:33:24 +0000 (13:33 +0000)]
Enable DIT if supported

This patch enables the Data Independent Timing
functionality (DIT) in EL3 if supported
by the platform.

Change-Id: Ia527d6aa2ee88a9a9fe1c941220404b9ff5567e5
Signed-off-by: Sathees Balya <sathees.balya@arm.com>
5 years agodocs: marvell: Fix typo in file build.txt
Ding Tao [Tue, 25 Dec 2018 06:42:08 +0000 (14:42 +0800)]
docs: marvell: Fix typo in file build.txt

Replace "SECURE=0" with "MARVELL_SECURE_BOOT=0".

Signed-off-by: Ding Tao <miyatsu@qq.com>
5 years agoplat/arm: Remove comment that mentions LOAD_IMAGE_V2
Antonio Nino Diaz [Wed, 19 Dec 2018 17:07:32 +0000 (17:07 +0000)]
plat/arm: Remove comment that mentions LOAD_IMAGE_V2

This option was removed in v2.0.

Change-Id: I0c929fddb57d60c5eff9e45dd5ee8c636b28e6c3
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge pull request #1730 from antonio-nino-diaz-arm/an/spdx
Antonio Niño Díaz [Wed, 19 Dec 2018 14:08:23 +0000 (15:08 +0100)]
Merge pull request #1730 from antonio-nino-diaz-arm/an/spdx

Use SPDX identifier in checkpatch config file

5 years agoUse SPDX identifier in checkpatch config file
Antonio Nino Diaz [Wed, 19 Dec 2018 13:13:33 +0000 (13:13 +0000)]
Use SPDX identifier in checkpatch config file

Change-Id: I4113604ac69ba07bd90d8268be65cae8aa43138e
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge pull request #1725 from Yann-lms/clang_aarch32
Antonio Niño Díaz [Wed, 19 Dec 2018 12:43:17 +0000 (13:43 +0100)]
Merge pull request #1725 from Yann-lms/clang_aarch32

clang: 32 bit compilation should include march32-directive

5 years agoMerge pull request #1727 from antonio-nino-diaz-arm/an/gitignore
Antonio Niño Díaz [Wed, 19 Dec 2018 09:52:27 +0000 (10:52 +0100)]
Merge pull request #1727 from antonio-nino-diaz-arm/an/gitignore

gitignore: Ignore sptool and doimage binaries

5 years agoMerge pull request #1729 from vwadekar/tegra-native-gicv2
Antonio Niño Díaz [Wed, 19 Dec 2018 09:51:28 +0000 (10:51 +0100)]
Merge pull request #1729 from vwadekar/tegra-native-gicv2

Tegra native gicv2

5 years agoTegra: support for native GICv2 drivers
Varun Wadekar [Fri, 19 Oct 2018 18:42:28 +0000 (11:42 -0700)]
Tegra: support for native GICv2 drivers

This patch converts Tegra platforms to support native
GICv2 drivers. This involves removes Tegra's GIC driver
port platforms to use interrupt_props

Change-Id: I83d8a690ff276dd97928dc60824a4fd36999bb30
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
5 years agoTegra: support to compile native GICv2 drivers
Varun Wadekar [Fri, 19 Oct 2018 18:44:31 +0000 (11:44 -0700)]
Tegra: support to compile native GICv2 drivers

This patch adds a macro to allow platforms to compile native
GICv2 drivers along with Tegra handlers.

Change-Id: I8281796c09dae5704cff2daab831395d65e261b7
Signed-off-by: Varun Wadekar <vwadekar@nvidia.com>
5 years agogitignore: Ignore sptool and doimage binaries
Antonio Nino Diaz [Fri, 14 Dec 2018 10:47:52 +0000 (10:47 +0000)]
gitignore: Ignore sptool and doimage binaries

Change-Id: I119b85179f68148e128cc194f11d2cccb1e33d6d
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge pull request #1721 from marex/arm/master/memsize-passing-v2
Antonio Niño Díaz [Tue, 18 Dec 2018 14:24:52 +0000 (15:24 +0100)]
Merge pull request #1721 from marex/arm/master/memsize-passing-v2

plat: rcar: Move FDT from x3 to x1

5 years agoMerge pull request #1717 from satheesbalya-arm/sb1/sb1_2629_romlib_ifc
Antonio Niño Díaz [Tue, 18 Dec 2018 14:22:49 +0000 (15:22 +0100)]
Merge pull request #1717 from satheesbalya-arm/sb1/sb1_2629_romlib_ifc

romlib: Add platform specific jump table list

5 years agoromlib: Add platform specific jump table list
Sathees Balya [Wed, 31 Oct 2018 14:05:08 +0000 (14:05 +0000)]
romlib: Add platform specific jump table list

This patch allows platforms to define their
own jump table list for library at ROM. The
file has the list of functions to be used
from library at ROM. It can also include
other list files.

Change-Id: I721c35d7dad3dcadbb3a7f3277bfd5d3e1f6e00a
Signed-off-by: Sathees Balya <sathees.balya@arm.com>
5 years agoMerge pull request #1722 from antonio-nino-diaz-arm/an/arch
Antonio Niño Díaz [Tue, 18 Dec 2018 12:50:38 +0000 (13:50 +0100)]
Merge pull request #1722 from antonio-nino-diaz-arm/an/arch

Synchronize architectural headers with TF-A-Tests

5 years agoImport exception helpers from TF-A-Tests
Antonio Nino Diaz [Fri, 23 Nov 2018 15:04:01 +0000 (15:04 +0000)]
Import exception helpers from TF-A-Tests

This is done in order to keep the files in both repositories in sync.

Change-Id: Ie1a9f321cbcfe8d7d14f206883fa718872271218
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoDeprecate some unused AArch64 assembly functions
Antonio Nino Diaz [Fri, 23 Nov 2018 13:54:41 +0000 (13:54 +0000)]
Deprecate some unused AArch64 assembly functions

Change-Id: I8a0be21783a0c12325e6ab22e9e53ab5466ed9e0
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMove helpers from arch.h to arch_helpers.h
Antonio Nino Diaz [Fri, 23 Nov 2018 13:54:00 +0000 (13:54 +0000)]
Move helpers from arch.h to arch_helpers.h

Some of the affected macros can only be used from C code. In general, we
use arch_helpers.h for any C helpers to access registers. For
consistency, the other macros have been moved as well.

Also, import some AArch32 helpers from TF-A-Tests.

Change-Id: Ie8fe1ddeadba5336c12971ddc39a7883121386b1
Signed-off-by: Antonio Nino Diaz <antonio.ninodiaz@arm.com>
5 years agoMerge pull request #1718 from vwadekar/cc-variable-findstring
Antonio Niño Díaz [Tue, 18 Dec 2018 10:55:45 +0000 (11:55 +0100)]
Merge pull request #1718 from vwadekar/cc-variable-findstring

build: find "armclang" string in the 'CC' variable

5 years agoMerge pull request #1719 from Yann-lms/useless_flags
Antonio Niño Díaz [Tue, 18 Dec 2018 10:55:11 +0000 (11:55 +0100)]
Merge pull request #1719 from Yann-lms/useless_flags

stm32mp1: remove useless compilation flags

5 years agoclang: 32 bit compilation should include march32-directive
Yann Gautier [Tue, 18 Dec 2018 09:46:34 +0000 (10:46 +0100)]
clang: 32 bit compilation should include march32-directive

This is done for other compilers, and without this some code does not
compile, like inline assembly code.

Fixes ARM-software/tf-issues#657.

Signed-off-by: Yann Gautier <yann.gautier@st.com>
5 years agoMerge pull request #1720 from jeenu-arm/gic-priority
Antonio Niño Díaz [Tue, 18 Dec 2018 09:05:22 +0000 (10:05 +0100)]
Merge pull request #1720 from jeenu-arm/gic-priority

GIC: Remove lowest priority constants

5 years agoMerge pull request #1723 from soby-mathew/sm/reset_bl31_fix
Antonio Niño Díaz [Tue, 18 Dec 2018 09:04:57 +0000 (10:04 +0100)]
Merge pull request #1723 from soby-mathew/sm/reset_bl31_fix

FVP: Fixes for RESET_TO_BL31

5 years agodocs: User-guide corrections for RESET_TO_BL31
Soby Mathew [Wed, 12 Dec 2018 14:54:23 +0000 (14:54 +0000)]
docs: User-guide corrections for RESET_TO_BL31

This patch updates the user guide instructions for
RESET_TO_SP_MIN and RESET_TO_BL31 cases. The load
address for BL31 had to be updated because of increase
in code size. Also, information about PIE support when
RESET_TO_BL31=1 for FVP is added.

In the case of RESET_TO_SP_MIN, the RVBAR address
was wrong in the instruction. This is also corrected
in the patch.

Change-Id: I65fe6d28c5cf79bee0a11fbde320d49fcc1aacf5
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
5 years agoBL31: correct GOT section omission
Soby Mathew [Wed, 12 Dec 2018 14:33:11 +0000 (14:33 +0000)]
BL31: correct GOT section omission

When the patch SHA 931f7c6 introduced PIE support for BL31,
adding the GOT section when the SEPARATE_CODE_AND_RODATA=0
to the linker script was erroneously omitted. This patch corrects
the same.

Also the patch reduces the alignment requirement for GOT and RELA
sections from 16 bytes to 8. Comments are added explain the
intent for alignment.

Change-Id: I8035cbf75f346f99bd56b13f32e0b3b70dd2fe6c
Signed-off-by: Soby Mathew <soby.mathew@arm.com>
5 years agoFVP: Change BL31_BASE when RESET_TO_BL31=1
Soby Mathew [Wed, 12 Dec 2018 14:13:52 +0000 (14:13 +0000)]
FVP: Change BL31_BASE when RESET_TO_BL31=1

This patch defines BL31_BASE to 0x0 when RESET_TO_BL31=1 as the
executable is built with PIE support and can be loaded anywhere
in SRAM for execution.

Change-Id: I4007f4626322f1200a6304c9c565987d3357986c
Signed-off-by: Soby Mathew <soby.mathew@arm.com>