oweals/u-boot.git
9 years agoExynos: Fix L2 cache timings on Exynos5420 and Exynos5800
Doug Anderson [Fri, 20 Feb 2015 07:57:20 +0000 (13:27 +0530)]
Exynos: Fix L2 cache timings on Exynos5420 and Exynos5800

It was found that the L2 cache timings that we had before could cause
freezes and hangs.  We should make things more robust with better
timings.  Currently the production ChromeOS kernel applies these
timings, but it's nice to fixup firmware too (and upstream probably
won't take our kernel hacks).

This also provides a big cleanup of the L2 cache init code avoiding
some duplication.  The way things used to work:
* low_power_start() was installed by the SPL (both at boot and resume
  time) and left resident in iRAM for the kernel to use when bringing
  up additional CPUs.  It used configure_l2_ctlr() and
  configure_l2_actlr() when it detected it was on an A15.  This was
  needed (despite the L2 cache registers being shared among all A15s)
  because we might have been the first man in after the whole A15
  cluster was shutdown.
* secondary_cores_configure() was called on at boot time and at resume
  time.  Strangely this called configure_l2_ctlr() but not
  configure_l2_actlr() which was almost certainly wrong.  Given that
  we'll call both (see next bullet) later in the boot process it
  didn't matter for normal boot, but I guess this is how L2 cache
  settings got set on 5420/5800 (but not 5250?) at resume time.
* exynos5_set_l2cache_params() was called as part of cache enablement.
  This should happen at boot time (normally in the SPL except for USB
  boot where it happens in main U-Boot).

Note that the old code wasn't setting ECC/parity in the cache
enablement code but we happened to get it anyway because we'd call
secondary_cores_configure() at boot time.  For resume time we'd get it
anyway when the 2nd A15 core came up.

Let's make this a whole lot simpler.  Now we always set these
parameters in the same place for all boots and use the same code for
setting up secondary CPUs.

Intended net effects of this change (other than cleanup):
* Timings go from before:
    data: 0 cycle setup, 3 cycles (0x2) latency
    tag:  0 cycle setup, 3 cycles (0x2) latency
  after:
    data: 1 cycle setup, 4 cycles (0x3) latency
    tag:  1 cycle setup, 4 cycles (0x3) latency
* L2ACTLR is properly initted on 5420/5800 in all cases.

One note is that we're still relying on luck to keep low_power_start()
working.  The compiler is being nice and not storing anything on the
stack.

Another note is that on its own this patch won't help to fix cache
settings in an RW U-Boot update where we still have the RO SPL.  The
plan for that is:
* Have RW U-Boot re-init the cache right before calling the kernel
  (after it has turned the L2 cache off).  This is why the functions
  are in a header file instead of lowlevel_init.c.

* Have the kernel save the L2 cache settings of the boot CPU and apply
  them to all other CPUs.  We get a little lucky here because the old
  code was using "|=" to modify the registers and all of the bits that
  it's setting are also present in the new settings (!).  That means
  that when the 2nd CPU in the A15 cluster comes up it doesn't
  actually mess up the settings of the 1st CPU in the A15 cluster.  An
  alternative option is to have the kernel write its own
  low_power_start() code.

Signed-off-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: Make A7s boot with thumb-mode U-Boot on warm reset
Akshay Saraswat [Fri, 20 Feb 2015 07:57:19 +0000 (13:27 +0530)]
Exynos542x: Make A7s boot with thumb-mode U-Boot on warm reset

On warm reset, all cores jump to the low_power_start function because iRAM
data is retained and because while executing iROM code all cores find
the jump flag 0x02020028 set. In low_power_start, cores check the reset
status and if true they clear the jump flag and jump back to 0x0.

The A7 cores do jump to 0x0 but consider following instructions as a Thumb
instructions which in turn makes them loop inside the iROM code instead of
jumping to power_down_core.

This issue is fixed by replacing the "mov pc" instruction with a "bx"
instruction which switches state along with the jump to make the execution
unit consider the branch target as an ARM instruction.

Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: Fix secondary core booting for thumb
Akshay Saraswat [Fri, 20 Feb 2015 07:57:18 +0000 (13:27 +0530)]
Exynos542x: Fix secondary core booting for thumb

When compiled SPL for Thumb secondary cores failed to boot
at the kernel boot up. Only one core came up out of 4.
This was happening because the code relocated to the
address 0x02073000 by the primary core was an ARM asm
code which was executed by the secondary cores as if it
was a thumb code.
This patch fixes the issue of secondary cores considering
relocated code as Thumb instructions and not ARM instructions
by jumping to the relocated with the help of "bx" ARM instruction.
"bx" instruction changes the 5th bit of CPSR which allows
execution unit to consider the following instructions as ARM
instructions.

Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: add L2 control register configuration
Akshay Saraswat [Fri, 20 Feb 2015 07:57:17 +0000 (13:27 +0530)]
Exynos542x: add L2 control register configuration

This patch does 3 things:
1. Enables ECC by setting 21st bit of L2CTLR.
2. Restore data and tag RAM latencies to 3 cycles because iROM sets
   0x3000400 L2CTLR value during switching.
3. Disable clean/evict push to external by setting 3rd bit of L2ACTLR.
   We need to restore this here due to switching.

Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: cache: Disable clean/evict push to external
Akshay Saraswat [Fri, 20 Feb 2015 07:57:16 +0000 (13:27 +0530)]
Exynos542x: cache: Disable clean/evict push to external

L2 Auxiliary Control Register provides configuration
and control options for the L2 memory system. Bit 3
of L2ACTLR stands for clean/evict push to external.
Setting bit 3 disables clean/evict which is what
this patch intends to do.

Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: Add workaround for exynos iROM errata
Akshay Saraswat [Fri, 20 Feb 2015 07:57:15 +0000 (13:27 +0530)]
Exynos542x: Add workaround for exynos iROM errata

iROM logic provides undesired jump address for CPU2.
This patch adds a programmable susbstitute for a part of
iROM logic which wakes up cores and provides jump addresses.
This patch creates a logic to make all secondary cores jump
to a particular address which evades the possibility of CPU2
jumping to wrong address and create undesired results.

Logic of the workaround:

Step-1: iROM code checks value at address 0x2020028.
Step-2: If value is 0xc9cfcfcf, it jumps to the address (0x202000+CPUid*4),
else, it continues executing normally.
Step-3: Primary core puts secondary cores in WFE and store 0xc9cfcfcf in
0x2020028 and jump address (pointer to function low_power_start)
in (0x202000+CPUid*4).
Step-4: When secondary cores recieve event signal they jump to this address
and continue execution.

Signed-off-by: Kimoon Kim <kimoon.kim@samsung.com>
Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: Add workaround for ARM errata 799270
Akshay Saraswat [Fri, 20 Feb 2015 07:57:14 +0000 (13:27 +0530)]
Exynos542x: Add workaround for ARM errata 799270

This patch adds workaround for the ARM errata 799270 which says
"If the L2 cache logic clock is stopped because of L2 inactivity,
setting or clearing the ACTLR.SMP bit might not be effective. The bit is
modified in the ACTLR, meaning a read of the register returns the
updated value. However the logic that uses that bit retains the previous
value."

Signed-off-by: Kimoon Kim <kimoon.kim@samsung.com>
Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: Add workaround for ARM errata 798870
Akshay Saraswat [Fri, 20 Feb 2015 07:57:13 +0000 (13:27 +0530)]
Exynos542x: Add workaround for ARM errata 798870

This patch adds workaround for ARM errata 798870 which says
"If back-to-back speculative cache line fills (fill A and fill B) are
issued from the L1 data cache of a CPU to the L2 cache, the second
request (fill B) is then cancelled, and the second request would have
detected a hazard against a recent write or eviction (write B) to the
same cache line as fill B then the L2 logic might deadlock."

Signed-off-by: Kimoon Kim <kimoon.kim@samsung.com>
Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: CPU: Power down all secondary cores
Akshay Saraswat [Fri, 20 Feb 2015 07:57:12 +0000 (13:27 +0530)]
Exynos542x: CPU: Power down all secondary cores

This patch adds code to shutdown secondary cores.
When U-boot comes up, all secondary cores appear powered on,
which is undesirable and causes side effects while
initializing these cores in kernel.

Secondary core power down happens in following steps:

Step-1: After Exynos power-on, primary core starts executing first.
Step-2: In iROM code every core has to check 2 flags i.e.
addresses 0x02020028 & 0x02020004.
Step-3: Initially 0x02020028 is 0 for all cores and 0x02020004 has a
jump address for primary core and 0 for all secondary cores.
Step-4: Therefore, primary core follows normal iROM execution and jumps
to BL1 eventually, whereas all secondary cores enter WFE.
Step-5: When primary core comes into function secondary_cores_configure,
it puts pointer to function power_down_core into 0x02020004
and provides DSB and SEV for all cores so that they may come out
of WFE and jump to power_down_core function.
Step-6: And ultimately because of power_down_core all
secondary cores shut-down.

Signed-off-by: Kimoon Kim <kimoon.kim@samsung.com>
Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoExynos542x: Config: Add various configs
Akshay Saraswat [Fri, 20 Feb 2015 07:57:11 +0000 (13:27 +0530)]
Exynos542x: Config: Add various configs

This patch adds "iRAM, CPU state and low power" configs
which are the addresses acting as flag registers.

iROM code checks CONFIG_LOWPOWER_FLAG address. If it is equal
to CONFIG_LOWPOWER_EN then it jumps to the address (0x02020000+CPUID*4).
This is a part of iROM logic. Rest other flags are being used at
various places during kernel switching and reset.
They are nowhere documented because they are part programming.
These configs are serving as flags for us because they are
representing the addresses in iRAM which we are using for
storing and extracting CPU Status and GIC status.

Signed-off-by: Akshay Saraswat <akshay.s@samsung.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoMerge branch 'master' of git://git.denx.de/u-boot-fsl-qoriq
Tom Rini [Wed, 25 Feb 2015 23:14:18 +0000 (18:14 -0500)]
Merge branch 'master' of git://git.denx.de/u-boot-fsl-qoriq

9 years agocrypto/fsl - Add progressive hashing support using hardware acceleration.
gaurav rana [Fri, 20 Feb 2015 07:21:46 +0000 (12:51 +0530)]
crypto/fsl - Add progressive hashing support using hardware acceleration.

Currently only normal hashing is supported using hardware acceleration.
Added support for progressive hashing using hardware.

Signed-off-by: Ruchika Gupta <ruchika.gupta@freescale.com>
Signed-off-by: Gaurav Rana <gaurav.rana@freescale.com>
CC: Simon Glass <sjg@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agocrypto/fsl: Make function names consistent for blob encapsulation/decapsulation.
gaurav rana [Wed, 25 Feb 2015 04:07:09 +0000 (09:37 +0530)]
crypto/fsl: Make function names consistent for blob encapsulation/decapsulation.

This patch does the following:

1. The function names for encapsulation and decapsulation
were inconsitent in freescale's implementation and cmd_blob file.
This patch corrects the issues.
2. The function protopye is also modified to change the length parameter
from u8 to u32 to allow encapsulation and decapsulation of larger images.
3. Modified the description of km paramter in the command usage for better
readability.

Signed-off-by: Gaurav Rana <gaurav.rana@freescale.com>
Reviewed-by: Ruchika Gupta <ruchika.gupta@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agozmx25: Convert to generic board
Fabio Estevam [Mon, 23 Feb 2015 11:51:38 +0000 (08:51 -0300)]
zmx25: Convert to generic board

Boards need to select CONFIG_SYS_GENERIC_BOARD in order to prevent removal
from the project.

Cc: Matthias Weisser <weisserm@arcor.de>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agoimx31_phycore: Convert to generic board
Fabio Estevam [Mon, 23 Feb 2015 11:51:37 +0000 (08:51 -0300)]
imx31_phycore: Convert to generic board

Boards need to select CONFIG_SYS_GENERIC_BOARD in order to prevent removal
from the project.

Acked-by: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agomx31ads: Convert to generic board
Fabio Estevam [Mon, 23 Feb 2015 11:51:36 +0000 (08:51 -0300)]
mx31ads: Convert to generic board

Boards need to select CONFIG_SYS_GENERIC_BOARD in order to prevent removal
from the project.

Acked-by: Anatolij Gustschin <agust@denx.de>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agoARM: davinci: remove hawkboard support
Masahiro Yamada [Tue, 24 Feb 2015 02:45:10 +0000 (11:45 +0900)]
ARM: davinci: remove hawkboard support

This is still a non-generic board.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Sughosh Ganu <urwithsughosh@gmail.com>
Cc: Syed Mohammed Khasim <sm.khasim@gmail.com>
Acked-by: Marek Vasut <marex@denx.de>
9 years agoARM: remove tnetv107x board support
Masahiro Yamada [Tue, 24 Feb 2015 02:45:09 +0000 (11:45 +0900)]
ARM: remove tnetv107x board support

This is still a non-generic board.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Chan-Taek Park <c-park@ti.com>
Acked-by: Marek Vasut <marex@denx.de>
9 years agoARM: remove a320evb board support
Masahiro Yamada [Tue, 24 Feb 2015 02:45:08 +0000 (11:45 +0900)]
ARM: remove a320evb board support

This is still a non-generic board.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Po-Yu Chuang <ratbert@faraday-tech.com>
Acked-by: Marek Vasut <marex@denx.de>
9 years agoARM: remove cm4008 and cm41xx board support
Masahiro Yamada [Tue, 24 Feb 2015 02:45:07 +0000 (11:45 +0900)]
ARM: remove cm4008 and cm41xx board support

These are still non-generic boards.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Greg Ungerer <greg.ungerer@opengear.com>
Acked-by: Marek Vasut <marex@denx.de>
9 years agoARM: remove dkb board support
Masahiro Yamada [Tue, 24 Feb 2015 02:45:06 +0000 (11:45 +0900)]
ARM: remove dkb board support

This is still a non-generic board.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Lei Wen <leiwen@marvell.com>
Acked-by: Marek Vasut <marex@denx.de>
9 years agoARM: remove jadecpu board support
Masahiro Yamada [Tue, 24 Feb 2015 02:45:05 +0000 (11:45 +0900)]
ARM: remove jadecpu board support

This is still a non-generic board.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Matthias Weisser <weisserm@arcor.de>
Acked-by: Marek Vasut <marex@denx.de>
9 years agokconfig: remove unneeded dependency on !SPL_BUILD
Masahiro Yamada [Tue, 24 Feb 2015 13:26:21 +0000 (22:26 +0900)]
kconfig: remove unneeded dependency on !SPL_BUILD

Now CONFIG_SPL_BUILD is not defined in Kconfig, so
"!depends on SPL_BUILD" and "if !SPL_BUILD" are redundant.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
9 years agokconfig: switch to single .config configuration
Masahiro Yamada [Tue, 24 Feb 2015 13:26:20 +0000 (22:26 +0900)]
kconfig: switch to single .config configuration

When Kconfig for U-boot was examined, one of the biggest issues was
how to support multiple images (Normal, SPL, TPL).  There were
actually two options, "single .config" and "multiple .config".
After some discussions and thought experiments, I chose the latter,
i.e. to create ".config", "spl/.config", "tpl/.config" for Normal,
SPL, TPL, respectively.

It is true that the "multiple .config" strategy provided us the
maximum flexibility and helped to avoid duplicating CONFIGs among
Normal, SPL, TPL, but I have noticed some fatal problems:

[1] It is impossible to share CONFIG options across the images.
  If you change the configuration of Main image, you often have to
  adjust some SPL configurations correspondingly.  Currently, we
  cannot handle the dependencies between them.  It means one of the
  biggest advantages of Kconfig is lost.

[2] It is too painful to change both ".config" and "spl/.config".
  Sunxi guys started to work around this problem by creating a new
  configuration target.  Commit cbdd9a9737cc (sunxi: kconfig: Add
  %_felconfig rule to enable FEL build of sunxi platforms.) added
  "make *_felconfig" to enable CONFIG_SPL_FEL on both images.
  Changing the configuration of multiple images in one command is a
  generic demand.  The current implementation cannot propose any
  good solution about this.

[3] Kconfig files are getting ugly and difficult to understand.
  Commit b724bd7d6349 (dm: Kconfig: Move CONFIG_SYS_MALLOC_F_LEN to
  Kconfig) has sprinkled "if !SPL_BUILD" over the Kconfig files.

[4] The build system got more complicated than it should be.
  To adjust Linux-originated Kconfig to U-Boot, the helper script
  "scripts/multiconfig.sh" was introduced.  Writing a complicated
  text processor is a shell script sometimes caused problems.

Now I believe the "single .config" will serve us better.  With it,
all the problems above would go away.  Instead, we will have to add
some CONFIG_SPL_* (and CONFIG_TPL_*) options such as CONFIG_SPL_DM,
but we will not have much.  Anyway, this is what we do now in
scripts/Makefile.spl.

I admit my mistake with my apology and this commit switches to the
single .config configuration.

It is not so difficult to do that:

 - Remove unnecessary processings from scripts/multiconfig.sh
  This file will remain for a while to support the current defconfig
  format.  It will be removed after more cleanups are done.

 - Adjust some makefiles and Kconfigs

 - Add some entries to include/config_uncmd_spl.h and the new file
   scripts/Makefile.uncmd_spl.  Some CONFIG options that are not
   supported on SPL must be disabled because one .config is shared
   between SPL and U-Boot proper going forward.  I know this is not
   a beautiful solution and I think we can do better, but let's see
   how much we will have to describe them.

 - update doc/README.kconfig

More cleaning up patches will follow this.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
9 years agokconfig: Adjust ordering so that defaults work as expected
Simon Glass [Tue, 24 Feb 2015 13:26:19 +0000 (22:26 +0900)]
kconfig: Adjust ordering so that defaults work as expected

At present defaults in arch-specific Kconfig files are ignored if the
top-level item comes ahead of it in include order. This means that it is
not possible to have a U-Boot default that architectures and boards can
override. This does not seem very useful.

Move the include earlier to support this.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reported-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Reviewed-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
9 years agoARM: UniPhier: set CONFIG_SYS_MALLOC_F to the global default value
Masahiro Yamada [Tue, 24 Feb 2015 13:26:18 +0000 (22:26 +0900)]
ARM: UniPhier: set CONFIG_SYS_MALLOC_F to the global default value

It is true that malloc is necessary for Driver Model before
relocation, but there is no good reason to reserve the malloc
space more than enough.  The default value 0x400 works well.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
9 years agoOMAP3: igep0032: Fix regression due commit f3b4bc45.
Enric Balletbo i Serra [Tue, 24 Feb 2015 18:27:15 +0000 (19:27 +0100)]
OMAP3: igep0032: Fix regression due commit f3b4bc45.

Commit referenced in subject breaks IGEP0032 build with the following
error:

  drivers/misc/status_led.c:30:7: error: 'RED_LED_GPIO' undeclared here (not in a function)
  scripts/Makefile.build:275: recipe for target 'drivers/misc/status_led.o' failed
  make[2]: *** [drivers/misc/status_led.o] Error 1
  scripts/Makefile.build:420: recipe for target 'drivers/misc' failed
  make[1]: *** [drivers/misc] Error 2
  Makefile:1093: recipe for target 'drivers' failed
  make: *** [drivers] Error 2

Fix this by skipping the status led on IGEP0032 machine as is not available
and throw an error for future machines if the status led is not configured
to avoid build breakage.

Reported-by: Albert ARIBAUD <albert.u.boot@aribaud.net>
Signed-off-by: Enric Balletbo i Serra <enric.balletbo@collabora.com>
9 years agommc: fsl_esdhc: Add support for DDR mode
Volodymyr Riazantsev [Tue, 20 Jan 2015 15:16:44 +0000 (10:16 -0500)]
mmc: fsl_esdhc: Add support for DDR mode

Add support of the DDR mode for eSDHC driver.
Enable it for i.MX6 SoC family only.

Signed-off-by: Volodymyr Riazantsev <volodymyr.riazantsev@globallogic.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm: ls1021x: Add support for initializing CAAM's stream id
Alison Wang [Fri, 16 Jan 2015 09:21:34 +0000 (17:21 +0800)]
arm: ls1021x: Add support for initializing CAAM's stream id

There 4 JRs, 4 RTICs and 8 DECOs, and set them the same stream id
for using the same SMMU3 on LS1021A.

Signed-off-by: Xiubo Li <Li.Xiubo@freescale.com>
Signed-off-by: Alison Wang <alison.wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm: ls102xa: workaround for cache coherency problem
chenhui zhao [Fri, 23 Jan 2015 07:53:53 +0000 (15:53 +0800)]
arm: ls102xa: workaround for cache coherency problem

The RCPM FSM may not be reset after power-on, for example,
in the cases of cold boot and wakeup from deep sleep.
It causes cache coherency problem and may block deep sleep.
Therefore, reset them if they are not be reset.

Signed-off-by: Chenhui Zhao <chenhui.zhao@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agodriver/pci: add Layerscape PCIe driver
Minghuan Lian [Wed, 21 Jan 2015 09:29:20 +0000 (17:29 +0800)]
driver/pci: add Layerscape PCIe driver

The patch adds Freescale Layerscape PCIe driver and provides
up to 4 controllers support.

Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/ls1021a: add PCIe settings
Minghuan Lian [Wed, 21 Jan 2015 09:29:19 +0000 (17:29 +0800)]
arm/ls1021a: add PCIe settings

The patch enables and adds PCIe settings for boards LS1021AQDS
and LS1021ATWR.

Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/ls102xa: use a array to define pexmscportsr
Minghuan Lian [Wed, 21 Jan 2015 09:29:18 +0000 (17:29 +0800)]
arm/ls102xa: use a array to define pexmscportsr

Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm/ls102xa: create TLB to map PCIe region
Minghuan Lian [Wed, 21 Jan 2015 09:29:17 +0000 (17:29 +0800)]
arm/ls102xa: create TLB to map PCIe region

LS1021A's PCIe1 region begins 0x40_00000000; PCIe2 begins
0x48_00000000. In order to access PCIe device, we must create
TLB to map the 40bit physical address to 32bit virtual address.
This patch will enable MMU after DDR is available and creates MMU
table in DRAM to map all 4G space; then, re-use the reserved space
to map PCIe region. The following the mapping layout.

VA mapping:
    -------  <---- 0GB
   |       |
   |       |
   |-------| <---- 0x24000000
   |///////|  ===> 192MB VA map for PCIe1 with offset 0x40_0000_0000
   |-------| <---- 0x300000000
   |       |
   |-------| <---- 0x34000000
   |///////|  ===> 192MB VA map for PCIe2 with offset 0x48_0000_0000
   |-------| <---- 0x40000000
   |       |
   |-------| <---- 0x80000000 DDR0 space start
   |\\\\\\\|
   |\\\\\\\|  ===> 2GB VA map for 2GB DDR0 Memory space
   |\\\\\\\|
   -------  <---- 4GB DDR0 space end

Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarm: ls102xa: Define default values for some CCSR macros
Alison Wang [Fri, 16 Jan 2015 09:23:04 +0000 (17:23 +0800)]
arm: ls102xa: Define default values for some CCSR macros

This patch is to define default values for some CCSR macros
to make header files cleaner.

Signed-off-by: Alison Wang <alison.wang@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agodrivers/mc: Migrated MC Flibs to 0.5.2
J. German Rivera [Tue, 6 Jan 2015 21:19:02 +0000 (13:19 -0800)]
drivers/mc: Migrated MC Flibs to 0.5.2

Upgrade Manage Complex (MC) flib API to 0.5.2. Rename directory
fsl_mc to fsl-mc. Change the fsl-mc node in Linux device tree
from "fsl,dprcr" to "fsl-mc". Print MC version info when
appropriate.

Signed-off-by: J. German Rivera <German.Rivera@freescale.com>
Signed-off-by: Lijun Pan <Lijun.Pan@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085a_emu: Enable sync of refresh
York Sun [Tue, 6 Jan 2015 21:19:01 +0000 (13:19 -0800)]
armv8/ls2085a_emu: Enable sync of refresh

Enable sync of DDR refresh for LS2085a platform. GPP DDR controllers
stay in sync. DP-DDR has only one controller so it does no harm.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Enable erratum workround for A008514
York Sun [Tue, 6 Jan 2015 21:19:00 +0000 (13:19 -0800)]
armv8/fsl-lsch3: Enable erratum workround for A008514

Erratum A008514 appleis to ls2085a.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Enable workaround for A008336
York Sun [Tue, 6 Jan 2015 21:18:59 +0000 (13:18 -0800)]
armv8/fsl-lsch3: Enable workaround for A008336

Erratum A008336 applied to LS2085A.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agols2085/configs: Ensure right banners are printed for EMU and SIMU
Bhupesh Sharma [Tue, 6 Jan 2015 21:18:58 +0000 (13:18 -0800)]
ls2085/configs: Ensure right banners are printed for EMU and SIMU

This patch enusres that right banners are printed for LS2085A
emulator and simulator platforms.

Signed-off-by: Bhupesh Sharma <bhupesh.sharma@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoARMv8/ls2085a: Move kernel image load address
Stuart Yoder [Tue, 6 Jan 2015 21:18:57 +0000 (13:18 -0800)]
ARMv8/ls2085a: Move kernel image load address

Move the load address of the kernel image to get it away from the
region of the uncompressed kernel.

Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoARMv8/ls2085a: Switch to passing earlycon to kernel
Arnab Basu [Tue, 6 Jan 2015 21:18:56 +0000 (13:18 -0800)]
ARMv8/ls2085a: Switch to passing earlycon to kernel

Since Linux v3.16-rc1 earlyprintk has been removed for arm64.
Switch to using earlycon.

Signed-off-by: Arnab Basu <arnab.basu@freescale.com>
Signed-off-by: York Sun <yorksun@freescale.com>
9 years agodriver/ddr/fsl: Add sync of refresh
York Sun [Tue, 6 Jan 2015 21:18:55 +0000 (13:18 -0800)]
driver/ddr/fsl: Add sync of refresh

Add sync of refresh for multiple DDR controllers. DDRC initialization
needs to complete first. Code is re-ordered to keep refresh close.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoARMv8/LS2085A: Adjust system clock and DDR clock
York Sun [Tue, 6 Jan 2015 21:18:54 +0000 (13:18 -0800)]
ARMv8/LS2085A: Adjust system clock and DDR clock

Set system clock to 100MHz and DDR clock to 133MHz.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoARMv8/LS2085A: HugeTLB support is required by default in LS NADK
Kuldip Giroh [Tue, 6 Jan 2015 21:18:53 +0000 (13:18 -0800)]
ARMv8/LS2085A: HugeTLB support is required by default in LS NADK

LS NADK memory manager by default works on HugeTLB. Hence bootargs
must include parameters default_hugepagesz (default hugepagesize,
hugepagesz (hugepage size) and hugepages (number of hugepages to be
reserved in kernel for the given size).

Signed-off-by: Kuldip Giroh <kuldip.giroh@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agodriver/ddr/fsl: Fix a typo in timing_cfg_8 calculation
York Sun [Tue, 6 Jan 2015 21:18:52 +0000 (13:18 -0800)]
driver/ddr/fsl: Fix a typo in timing_cfg_8 calculation

wwt_bg should match rrt_bg. It was a typo in driver.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoARMv8/LS2085A: Enable auto precharge for DP-DDR
York Sun [Tue, 6 Jan 2015 21:18:51 +0000 (13:18 -0800)]
ARMv8/LS2085A: Enable auto precharge for DP-DDR

DP-DDR benefits from auto precharge because of its specific
application.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agodriver/ddr/fsl: Add support for multiple DDR clocks
York Sun [Tue, 6 Jan 2015 21:18:50 +0000 (13:18 -0800)]
driver/ddr/fsl: Add support for multiple DDR clocks

Controller number is passed for function calls to support individual
DDR clock, depending on SoC implementation. It is backward compatible
with exising platforms. Multiple clocks have been verifyed on LS2085A
emulator.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Add support for second DDR clock
York Sun [Tue, 6 Jan 2015 21:18:49 +0000 (13:18 -0800)]
armv8/fsl-lsch3: Add support for second DDR clock

FSL-LSCH3 platforms can have multiple DDR clocks. LS2085A has one clock for
general DDR controlers, and another clock for DP-DDR. DDR driver needs to
change to support multiple clocks.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agodriver/ddr/fsl: Add workround for erratumn A008514
York Sun [Tue, 6 Jan 2015 21:18:48 +0000 (13:18 -0800)]
driver/ddr/fsl: Add workround for erratumn A008514

Erratum A008514 workround requires writing register eddrtqcr1 with
value 0x63b20002.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agodriver/ddr/fsl: Add workaround for A008336
York Sun [Tue, 6 Jan 2015 21:18:47 +0000 (13:18 -0800)]
driver/ddr/fsl: Add workaround for A008336

Erratum A008336 requires setting EDDRTQCR1[2] in DDRC DCSR space
for 64-bit DDR controllers.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agodriver/ddr/fsl: Adjust CAS to preamble override for emulator
York Sun [Tue, 6 Jan 2015 21:18:45 +0000 (13:18 -0800)]
driver/ddr/fsl: Adjust CAS to preamble override for emulator

On ZeBu emulator, CAS to preamble overrides need to be set to
satisfy the timing. This only impact platforms with CONFIG_EMU.

These should be set before MEM_EN is set.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Add fdt-fixup for clock frequency of the DUART nodes
Bhupesh Sharma [Tue, 6 Jan 2015 21:18:44 +0000 (13:18 -0800)]
armv8/fsl-lsch3: Add fdt-fixup for clock frequency of the DUART nodes

This patch adds the fdt-fixup logic for the clock frequency of the
NS16550A related device tree nodes.

Signed-off-by: Bhupesh Sharma <bhupesh.sharma@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/ls2085a: Enable cluster timebase for all clusters
York Sun [Tue, 6 Jan 2015 21:18:43 +0000 (13:18 -0800)]
armv8/ls2085a: Enable cluster timebase for all clusters

LS2085A and its variants can have up to four clusters. It is safe
to enable timebase for all even some may be disabled.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Convert flushing L3 to assembly to avoid using stack
York Sun [Tue, 6 Jan 2015 21:18:42 +0000 (13:18 -0800)]
armv8/fsl-lsch3: Convert flushing L3 to assembly to avoid using stack

Flushing L3 cache in CCN-504 requries d-cache to be disabled. Using
assembly function to guarantee stack is not used before flushing is
completed. Timeout is needed for simualtor on which CCN-504 is not
implemented. Return value can be checked for timeout situation.

Change bootm.c to disable dcache instead of simply flushing, required
by flushing L3.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoARMv8/fsl-lsch3: Patch cpu node properties in DT for online cores
Arnab Basu [Tue, 6 Jan 2015 21:18:41 +0000 (13:18 -0800)]
ARMv8/fsl-lsch3: Patch cpu node properties in DT for online cores

U-Boot should only add "enable-method" and "cpu-release-address"
properties to the "cpu" node of the online cores.

Signed-off-by: Arnab Basu <arnab.basu@freescale.com>
Signed-off-by: York Sun <yorksun@freescale.com>
9 years agoarmv8/fsl-lsch3: Change normal memory shareability
York Sun [Tue, 6 Jan 2015 21:11:22 +0000 (13:11 -0800)]
armv8/fsl-lsch3: Change normal memory shareability

According to hardware implementation, a single outer shareable global
coherence group is defined. Inner shareable has not bee enabled.

Signed-off-by: York Sun <yorksun@freescale.com>
9 years agofsl-ch3/lowlevel: TZPC and TZASC programming to configure non-secure accesses
Bhupesh Sharma [Tue, 6 Jan 2015 21:11:21 +0000 (13:11 -0800)]
fsl-ch3/lowlevel: TZPC and TZASC programming to configure non-secure accesses

This patch ensures that the TZPC (BP147) and TZASC-400 programming
happens for LS2085A SoC only when the desired config flags are
enabled and ensures that the TZPC programming is done to allow Non-secure
(NS) + secure (S) transactions only for DCGF registers.

The TZASC component is not present on LS2085A-Rev1, so the TZASC-400
config flag is turned OFF for now.

Signed-off-by: Bhupesh Sharma <bhupesh.sharma@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
9 years agoMerge branch 'master' of git://git.denx.de/u-boot-mmc
Tom Rini [Mon, 23 Feb 2015 21:18:06 +0000 (16:18 -0500)]
Merge branch 'master' of git://git.denx.de/u-boot-mmc

9 years agommc: sdhci: fix bus width switching on Samsung SoCs
Matt Reimer [Thu, 19 Feb 2015 18:22:53 +0000 (11:22 -0700)]
mmc: sdhci: fix bus width switching on Samsung SoCs

Fix bus width switching from 8-bit mode down to 4-bit or 1-bit modes on
Samsung SoCs using SDHCI_QUIRK_USE_WIDE8.  These SoCs report controller
version 2.0 yet they support 8-bit bus widths.  If 8-bit mode was
previously enabled and then an operation like "mmc dev" caused a switch
back down to 4-bit or 1-bit mode, WIDE8 was left set, causing failures.

This problem was manifested by "mmc dev" timing out.

Signed-off-by: Matt Reimer <mreimer@sdgsystems.com>
9 years agommc: print SD/eMMC type for inited mmc devices
Przemyslaw Marczak [Fri, 20 Feb 2015 11:29:27 +0000 (12:29 +0100)]
mmc: print SD/eMMC type for inited mmc devices

Depending on the boot priority, the eMMC/SD cards,
can be initialized with the same numbers for each boot.

To be sure which mmc device is SD and which is eMMC,
this info is printed by 'mmc list' command, when
the init is done.

Signed-off-by: Przemyslaw Marczak <p.marczak@samsung.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
9 years agommc: exynos dwmmc: check boot mode before init dwmmc
Przemyslaw Marczak [Fri, 20 Feb 2015 11:29:26 +0000 (12:29 +0100)]
mmc: exynos dwmmc: check boot mode before init dwmmc

Before this commit, the mmc devices were always registered
in the same order. So dwmmc channel 0 was registered as mmc 0,
channel 1 as mmc 1, etc.
In case of possibility to boot from more then one device,
the CONFIG_SYS_MMC_ENV_DEV should always point to right mmc device.

This can be achieved by init boot device as first, so it will be
always registered as mmc 0. Thanks to this, the 'saveenv' command
will work fine for all mmc boot devices.

Exynos based boards usually uses mmc host channels configuration:
- 0, or 0+1 for 8 bit  - as a default boot device (usually eMMC)
- 2 for 4bit - as an optional boot device (usually SD card slot)

And usually the boot order is defined by OM pin configuration,
which can be changed in a few ways, eg.
- Odroid U3     - eMMC card insertion -> first boot from eMMC
- Odroid X2/XU3 - boot priority jumper

By this commit, Exynos dwmmc driver will check the OM pin configuration,
and then try to init the boot device and register it as mmc 0.

Signed-off-by: Przemyslaw Marczak <p.marczak@samsung.com>
Cc: Minkyu Kang <mk7.kang@samsung.com>
Cc: Jaehoon Chung <jh80.chung@samsung.com>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Akshay Saraswat <akshay.s@samsung.com>
9 years agosunxi: mmc: Always declare High Capacity capability
Hans de Goede [Thu, 19 Feb 2015 20:03:21 +0000 (21:03 +0100)]
sunxi: mmc: Always declare High Capacity capability

High Capacity (e)MMC cards work fine on sun4i / sun5i, and not having this
capability set causes u-boot to not recognize the eMMC on an Utoo P66 A13
tablet, so always set it thereby fixing this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agommc: exynos_dw_mmc: use the exynos specific data structure
Jaehoon Chung [Wed, 4 Feb 2015 06:48:40 +0000 (15:48 +0900)]
mmc: exynos_dw_mmc: use the exynos specific data structure

Clksel value is exynos specific value.
It removed "clksel_val" into dwmci_host and created the
"dwmci_exynos_priv_data" structure for exynos specific data.

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
9 years agommc: exynos_dw_mmc: set to clksel_val into board-init function
Jaehoon Chung [Wed, 4 Feb 2015 06:48:39 +0000 (15:48 +0900)]
mmc: exynos_dw_mmc: set to clksel_val into board-init function

"clksel_val" is assigned to property of mmc or defined value.
But it doesn't write at initial sequence.
There is a reason that get the wrong source-clock value.
This patch fixed it.

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
9 years agommc: dw_mmc: fixed the wrong bit control
Jaehoon Chung [Wed, 14 Jan 2015 08:37:53 +0000 (17:37 +0900)]
mmc: dw_mmc: fixed the wrong bit control

If mode is not DDR-mode, then it needs to clear it.

Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
9 years agommc: Implement SD/MMC versioning properly
Pantelis Antoniou [Fri, 23 Jan 2015 10:12:01 +0000 (12:12 +0200)]
mmc: Implement SD/MMC versioning properly

The SD/MMC version scheme was buggy when dealing with standard
major.minor.change cases. Fix it by using something similar to
the linux's kernel versioning method.

Signed-off-by: Pantelis Antoniou <pantelis.antoniou@konsulko.com>
Tested-by: Jaehoon Chung <jh80.chung@samsung.com>
Reported-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Stephen Warren <swarren@nvidia.com>
9 years agoMerge git://git.denx.de/u-boot-sunxi
Tom Rini [Sun, 22 Feb 2015 03:01:09 +0000 (22:01 -0500)]
Merge git://git.denx.de/u-boot-sunxi

9 years agosunxi: Machine id hack to prevent loading buggy sunxi-3.4 kernels
Siarhei Siamashka [Sat, 21 Feb 2015 05:34:09 +0000 (07:34 +0200)]
sunxi: Machine id hack to prevent loading buggy sunxi-3.4 kernels

Right now U-Boot supports the CONFIG_OLD_SUNXI_KERNEL_COMPAT option,
which makes it go out of its way in limiting the selection of PLL clock
frequencies and PMIC voltages in order not to upset outdated buggy
sunxi-3.4 kernel releases. And if the CONFIG_OLD_SUNXI_KERNEL_COMPAT
option is not set, then booting such old kernels exhibits various
failures at runtime. This is very user unfriendly, and there were
already several incidents when people wasted their time being hit
by these runtime failures and trying to debug them.

The right solution is not to add hacks and workarounds to the mainline
U-Boot, but to fix these bugs in the sunxi-3.4 kernel. And in fact,
the updated sunxi-3.4 kernels already exist. Still we need to follow
the 'Principle of Least Surprise' and U-Boot needs to ensure that
the old buggy kernels are not getting happily booted when the
CONFIG_OLD_SUNXI_KERNEL_COMPAT option is not set. And this patch
addresses this particular issue.

This patch makes U-Boot store the 'compatibility revision' number in
the top 4 bits of the machine id and pass it to the kernel. The old
buggy kernels will fail to load with a very much googlable error
message on the serial console (the "r1 = 0x100010bb" part of it):

  "Error: unrecognized/unsupported machine ID (r1 = 0x100010bb)"

This error message can be documented in the linux-sunxi wiki with
proper explanations about how to resolve this situation and where
to get the necessary bugfixes for the sunxi-3.4 kernel.

The fixed sunxi-3.4 kernels implement a revision compatibility check
and clear the top 4 bits of the machine id if everything is alright.
By accepting the machine id with the bits 31:28 set to 1, the sunxi-3.4
kernel effectively certifies that it has the PLL5 clock speed and
AXP209 DCDC3 voltage fixes applied.

It is still possible to set the CONFIG_OLD_SUNXI_KERNEL_COMPAT option
in U-Boot if the user desires to use an outdated unpatched sunxi-3.4
kernel.

Signed-off-by: Siarhei Siamashka <siarhei.siamashka@gmail.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
9 years agosunxi: Set the /chosen/stdout-path fdt property for sunxi boards
Hans de Goede [Fri, 20 Feb 2015 15:55:12 +0000 (16:55 +0100)]
sunxi: Set the /chosen/stdout-path fdt property for sunxi boards

While discussing with some people how to get the Linux kernel to do the
right thing wrt sending output to both the serial console and the
hdmi out / lcd screen when booting on ARM devices, Grant Likely pointed out
that there already is a solution for this.

All we need to do is set the /chosen/stdout-path fdt property, and if no
console= arguments were specified on the kernel commandline the kernel
will honor this and add this device as a console (next to the primary
video output on hdmi).

And u-boot already has support for setting this, all we need to do is
define OF_STDOUT_PATH and then everything will just work ootb, without
people needing to meddle with adding console= arguments in extlinux.conf .

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Reviewed-by: Tom Rini <trini@ti.com>
9 years agosunxi: Fix sun5i mbus speed when booting old kernels
Hans de Goede [Thu, 19 Feb 2015 13:46:44 +0000 (14:46 +0100)]
sunxi: Fix sun5i mbus speed when booting old kernels

Older linux-sunxi-3.4 kernels override our PLL6 setting with 300 MHz,
halving the mbus frequency, so set it to 300 MHz ourselves and base the
mbus divider on that.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agosunxi: musb: Check Vbus-det before enabling otg port power
Hans de Goede [Mon, 16 Feb 2015 21:13:43 +0000 (22:13 +0100)]
sunxi: musb: Check Vbus-det before enabling otg port power

Sending out 5V when there is a charger connected to the otg port is not a
good idea, so check for this and error out.

Note this commit currently breaks otg support on the q8h tablets, as we need
to do some magic with the pmic there to get vbus info, this is deliberate
(better safe then sorry), fixing this is on my TODO list.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agosunxi: Add support for the UTOO P66 tablet
Hans de Goede [Mon, 16 Feb 2015 18:47:43 +0000 (19:47 +0100)]
sunxi: Add support for the UTOO P66 tablet

The UTOO P66 is a 6" A13 tablet / lcd ereader. It features a 6" 480x800 ips
lcd screen, 512MB RAM & 4GB emmc.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agosunxi: mmc: Always declare High Capacity capability
Hans de Goede [Thu, 19 Feb 2015 19:34:32 +0000 (20:34 +0100)]
sunxi: mmc: Always declare High Capacity capability

High Capacity (e)MMC cards work fine on sun4i / sun5i, and not having this
capability set causes u-boot to not recognize the eMMC on an Utoo P66 A13
tablet, so always set it thereby fixing this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agorpi: add support for Raspberry Pi 2 model B
Stephen Warren [Mon, 16 Feb 2015 19:16:15 +0000 (12:16 -0700)]
rpi: add support for Raspberry Pi 2 model B

USB doesn't seem to work yet; the controller detects the on-board Hub/
Ethernet device but can't read the descriptors from it. I haven't
investigated yet.

Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
9 years agobcm2836 SoC support (used in Raspberry Pi 2 model B)
Stephen Warren [Mon, 16 Feb 2015 19:16:14 +0000 (12:16 -0700)]
bcm2836 SoC support (used in Raspberry Pi 2 model B)

The bcm2835 and bcm2836 are essentially identical, except:
- The CPU is an ARM1176 v.s. a quad-core Cortex-A7.
- The physical address of many IO controllers has moved.

Rather than introducing a whole new bcm2836 value for $(SOC) or $(ARCH),
update the existing bcm2835 code to handle the minor differences, and
plumb it into the ARMv7 CPU architecture.

Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
9 years agobcm2835/rpi: add SPDX license tags for some files
Stephen Warren [Mon, 16 Feb 2015 19:16:13 +0000 (12:16 -0700)]
bcm2835/rpi: add SPDX license tags for some files

Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
9 years agoARM: prepare for including <mach/*.h>
Masahiro Yamada [Fri, 20 Feb 2015 08:04:19 +0000 (17:04 +0900)]
ARM: prepare for including <mach/*.h>

This commit adds $(srctree)/arch/arm/$(machdirs)/include/mach to
the headers search path.

It allows us to replace "#include <asm/arch/foo.h>" with
"#include <mach/foo.h>".  As "#include <asm/arch/foo.h>" is still
supported, we can modify each file one by one.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
9 years agoARM: keystone: move SoC headers to mach-keystone/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:18 +0000 (17:04 +0900)]
ARM: keystone: move SoC headers to mach-keystone/include/mach

Move arch/arm/include/asm/arch-keystone/*
  -> arch/arm/mach-keystone/include/mach/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Tom Rini <trini@ti.com>
9 years agoARM: orion5x: move SoC headers to mach-orion5x/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:17 +0000 (17:04 +0900)]
ARM: orion5x: move SoC headers to mach-orion5x/include/mach

Move arch/arm/include/asm/arch-orion5x/*
  -> arch/arm/mach-orion5x/include/mach/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Albert ARIBAUD <albert.u.boot@aribaud.net>
9 years agoARM: nomadik: move SoC headers to mach-nomadik/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:16 +0000 (17:04 +0900)]
ARM: nomadik: move SoC headers to mach-nomadik/include/mach

Move arch/arm/include/asm/arch-nomadik/*
  -> arch/arm/mach-nomadik/include/mach/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Nomadik Linux Team <STN_WMM_nomadik_linux@list.st.com>
Cc: Alessandro Rubini <rubini@unipv.it>
9 years agoARM: kirkwood: move SoC headers to mach-kirkwood/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:15 +0000 (17:04 +0900)]
ARM: kirkwood: move SoC headers to mach-kirkwood/include/mach

Move arch/arm/include/asm/arch-kirkwood/*
  -> arch/arm/mach-kirkwood/include/mach/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Stefan Roese <sr@denx.de>
Cc: Prafulla Wadaskar <prafulla@marvell.com>
Cc: Luka Perkov <luka.perkov@sartura.hr>
9 years agoARM: davinci: move SoC headers to mach-davinci/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:14 +0000 (17:04 +0900)]
ARM: davinci: move SoC headers to mach-davinci/include/mach

Move arch/arm/include/asm/arch-davinci/*
  -> arch/arm/mach-davinci/include/mach/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Tom Rini <trini@ti.com>
9 years agoARM: at91: move SoC headers to mach-at91/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:13 +0000 (17:04 +0900)]
ARM: at91: move SoC headers to mach-at91/include/mach

Move arch/arm/include/asm/arch-at91/*
  -> arch/arm/mach-at91/include/mach/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Andreas Bießmann <andreas.devel@googlemail.com>
9 years agokbuild: prepare for moving headers into mach-*/include/mach
Masahiro Yamada [Fri, 20 Feb 2015 08:04:12 +0000 (17:04 +0900)]
kbuild: prepare for moving headers into mach-*/include/mach

In U-Boot, SoC-specific headers are placed in
arch/$(ARCH)/include/asm/arch-$(SOC) and a symbolic link to that
directory is created at the early stage of the build process.

Creating and removing a symbolic link during the build is not
preferred.  In fact, Linux Kernel did away with include/asm-$(ARCH)
directories a long time time ago.

As for ARM, now it is possible to collect SoC sources into
arch/arm/mach-$(SOC).  It is also reasonable to move SoC headers
into arch/arm/mach-$(SOC)/include/mach.

This commit prepares for that.
If the directory arch/$(ARCH)/mach-$(SOC)/include/mach exists,
a symbolic to that directory is created.  Otherwise, a symbolic link
to arch/$(ARCH)/include/asm/arch-$(SOC) or arch-$(CPU) is created.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
9 years agoARM: keystone: move SoC sources to mach-keystone
Masahiro Yamada [Fri, 20 Feb 2015 08:04:11 +0000 (17:04 +0900)]
ARM: keystone: move SoC sources to mach-keystone

Move
arch/arm/cpu/armv7/keystone/* -> arch/arm/mach-keystone/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Tom Rini <trini@ti.com>
9 years agoARM: versatile: move SoC sources to mach-versatile
Masahiro Yamada [Fri, 20 Feb 2015 08:04:10 +0000 (17:04 +0900)]
ARM: versatile: move SoC sources to mach-versatile

Move
arch/arm/cpu/arm926ejs/versatile/* -> arch/arm/mach-versatile/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
9 years agoARM: orion5x: move SoC sources to mach-orion5x
Masahiro Yamada [Fri, 20 Feb 2015 08:04:09 +0000 (17:04 +0900)]
ARM: orion5x: move SoC sources to mach-orion5x

Move
arch/arm/cpu/arm926ejs/orion5x/* -> arch/arm/mach-orion5x/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Albert ARIBAUD <albert.u.boot@aribaud.net>
9 years agoARM: highbank: move SoC sources to mach-highbank
Masahiro Yamada [Fri, 20 Feb 2015 08:04:08 +0000 (17:04 +0900)]
ARM: highbank: move SoC sources to mach-highbank

Move
arch/arm/cpu/armv7/highbank/* -> arch/arm/mach-highbank/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Rob Herring <robh@kernel.org>
9 years agoARM: nomadik: move SoC sources to mach-nomadik
Masahiro Yamada [Fri, 20 Feb 2015 08:04:07 +0000 (17:04 +0900)]
ARM: nomadik: move SoC sources to mach-nomadik

Move
arch/arm/cpu/arm926ejs/nomadik/* -> arch/arm/mach-nomadik/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Nomadik Linux Team <STN_WMM_nomadik_linux@list.st.com>
Cc: Alessandro Rubini <rubini@unipv.it>
9 years agoARM: kirkwood: move SOC sources to mach-kirkwood
Masahiro Yamada [Fri, 20 Feb 2015 08:04:06 +0000 (17:04 +0900)]
ARM: kirkwood: move SOC sources to mach-kirkwood

Move
arch/arm/cpu/arm926ejs/kirkwood/* -> arch/arm/mach-kirkwood/*

Note:
 Perhaps, can we merge arch/arm/mach-kirkwood and
 arch/arm/mvebu-common into arch/arm/mach-mvebu, like Linux?

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Stefan Roese <sr@denx.de>
Cc: Prafulla Wadaskar <prafulla@marvell.com>
Cc: Luka Perkov <luka.perkov@sartura.hr>
9 years agoARM: davinci: move SoC sources to mach-davinci
Masahiro Yamada [Fri, 20 Feb 2015 08:04:05 +0000 (17:04 +0900)]
ARM: davinci: move SoC sources to mach-davinci

Move
arch/arm/cpu/arm926ejs/davinci/* -> arch/arm/mach-davinci/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Tom Rini <trini@ti.com>
9 years agoARM: tegra: collect SoC sources into mach-tegra
Masahiro Yamada [Fri, 20 Feb 2015 08:04:04 +0000 (17:04 +0900)]
ARM: tegra: collect SoC sources into mach-tegra

This commit moves files as follows:

 arch/arm/cpu/arm720t/tegra20/*      -> arch/arm/mach-tegra/tegra20/*
 arch/arm/cpu/arm720t/tegra30/*      -> arch/arm/mach-tegra/tegra30/*
 arch/arm/cpu/arm720t/tegra114/*     -> arch/arm/mach-tegra/tegra114/*
 arch/arm/cpu/arm720t/tegra124*      -> arch/arm/mach-tegra/tegra124/*
 arch/arm/cpu/arm720t/tegra-common/* -> arch/arm/mach-tegra/*
 arch/arm/cpu/armv7/tegra20/*        -> arch/arm/mach-tegra/tegra20/*
 arch/arm/cpu/armv7/tegra30/*        -> arch/arm/mach-tegra/tegra30/*
 arch/arm/cpu/armv7/tegra114/*       -> arch/arm/mach-tegra/tegra114/*
 arch/arm/cpu/armv7/tegra124/*       -> arch/arm/mach-tegra/tegra124/*
 arch/arm/cpu/armv7/tegra-common/*   -> arch/arm/mach-tegra/*
 arch/arm/cpu/tegra20-common/*       -> arch/arm/mach-tegra/tegra20/*
 arch/arm/cpu/tegra30-common/*       -> arch/arm/mach-tegra/tegra30/*
 arch/arm/cpu/tegra114-common/*      -> arch/arm/mach-tegra/tegra114/*
 arch/arm/cpu/tegra124-common/*      -> arch/arm/mach-tegra/tegra124/*
 arch/arm/cpu/tegra-common/*         -> arch/arm/mach-tegra/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Tested-by: Simon Glass <sjg@chromium.org> [ on nyan-big ]
Cc: Stephen Warren <swarren@nvidia.com>
Cc: Tom Warren <twarren@nvidia.com>
9 years agoARM: at91: collect SoC sources into mach-at91
Masahiro Yamada [Fri, 20 Feb 2015 08:04:03 +0000 (17:04 +0900)]
ARM: at91: collect SoC sources into mach-at91

This commit moves source files as follows:

  arch/arm/cpu/arm920t/at91/*   -> arch/arm/mach-at91/arm920t/*
  arch/arm/cpu/arm926ejs/at91/* -> arch/arm/mach-at91/arm926ejs/*
  arch/arm/cpu/armv7/at91/*     -> arch/arm/mach-at91/armv7/*
  arch/arm/cpu/at91-common/*    -> arch/arm/mach-at91/*

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Andreas Bießmann <andreas.devel@googlemail.co>
9 years agoARM: prepare for moving SoC sources into mach-*
Masahiro Yamada [Fri, 20 Feb 2015 08:04:02 +0000 (17:04 +0900)]
ARM: prepare for moving SoC sources into mach-*

In U-boot, the directory structure, arch/$(ARCH)/cpu/$(CPU)/$(SOC)/
has been adopted except that $(CPU) is missing from some
architectures and $(SOC) is missing from some CPUs.

This structure did not fit very well in some cases.

[1] AT91

AT91 SoC family have been developed across some ARM processor
generations.  Generally speaking, some IPs are often re-used in the
same SoC family (same SoC vendor) even when the main processor is
updated.  As a result, a SoC-common directory is needed in the upper
level.  Currently, AT91 source files are placed as follows:

  arch/arm/cpu/arm920t/at91/*
  arch/arm/cpu/arm926ejs/at91/*
  arch/arm/cpu/armv7/at91/*
  arch/arm/cpu/at91-common/*

Once directories are split, the motivation for refactorings across
CPU directories is lost.  Some files in arm920t/at91/ and
arm926ejs/at91/ are so similar that they could be merged.

[2] Tegra

Tegra is a little bit special case where different CPUs are used for
SPL and the main U-boot.  To obey the arch/$(ARCH)/cpu/$(CPU)/$(SOC)
structure, the source files must be placed across the CPUs,
again SoC-common directory is necessary in the upper level.

Moreover, there are several families in Tegra: Tegra20, Tegra30,
Tegra114, Tegra124.  Here again, the tegra-common directory is needed
to contain commonly-used files.

Tegra directories have been sprinkled in the directory structure.

  arch/arm/cpu/arm720t/tegra20
  arch/arm/cpu/arm720t/tegra30
  arch/arm/cpu/arm720t/tegra114
  arch/arm/cpu/arm720t/tegra124
  arch/arm/cpu/arm720t/tegra-common
  arch/arm/cpu/armv7/tegra20
  arch/arm/cpu/armv7/tegra30
  arch/arm/cpu/armv7/tegra114
  arch/arm/cpu/armv7/tegra124
  arch/arm/cpu/armv7/tegra-common
  arch/arm/cpu/tegra20-common
  arch/arm/cpu/tegra30-common
  arch/arm/cpu/tegra114-common
  arch/arm/cpu/tegra124-common
  arch/arm/cpu/tegra-common

As you see, splitting SoC code by the CPU is not going well,
especially for ARM.
Why don't we collect SoC-specific files into a single place?

A good example we can follow is Linux's arch/arm/mach-* structure.

This item was discussed in the following thread:
http://thread.gmane.org/gmane.comp.boot-loaders.u-boot/188548/

Looks like I got some positive responses and we are almost ready to
start this movement.

This commit prepares arch/arm/Makefile for describing machdirs in it.

After this commit, we can move SoC directory to arch/arm/mach-$(SOC)
in simple steps although some cases such as AT91 and Tegra need more
fixes.

What we generally have to do is:

[1] Move files arch/arm/cpu/$(CPU)/$(SOC)/* to arch/arm/mach-$(SOC)/*
[2] Add machine entry into arch/arm/Makefile
[3] Remove "obj-y += $(SOC)" from arch/arm/cpu/$(CPU)/Makefile
[4] Fix the Kconfig file path in arch/arm/Kconfig
[5] Modify MAINTAINERS if necessary

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
9 years agoARM: at91: move board select menu and common settings
Masahiro Yamada [Fri, 20 Feb 2015 08:04:01 +0000 (17:04 +0900)]
ARM: at91: move board select menu and common settings

The board select menu in arch/arm/Kconfig is still big.
To slim down it, this commit moves AT91 boards to
arch/arm/mach-at91/Kconfig.
Also, consolidate "config SYS_SOC" in each board Kconfig.

The Kconfig files under board/ directory were modified with the
following command:

    find board -name Kconfig | xargs sed -i -e '
    /config SYS_SOC/ {
        N
        /default "at91"/ {
            N
            d
        }
    }
    '

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Andreas Bießmann <andreas.devel@googlemail.co>
9 years agoMerge git://git.denx.de/u-boot-dm
Tom Rini [Sat, 21 Feb 2015 13:22:23 +0000 (08:22 -0500)]
Merge git://git.denx.de/u-boot-dm

9 years agosunxi: video: Add support for tl059wv5c0 lcd panels
Hans de Goede [Mon, 16 Feb 2015 16:49:47 +0000 (17:49 +0100)]
sunxi: video: Add support for tl059wv5c0 lcd panels

Add support for the 6" 480x800 tl059wv5c0 panel used on e.g. Utoo P66 and
Aigo M60/M608/M606 tablets.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agosunxi: video: Add support for LCD panels which need to be configured via i2c
Hans de Goede [Mon, 16 Feb 2015 16:23:25 +0000 (17:23 +0100)]
sunxi: video: Add support for LCD panels which need to be configured via i2c

This commits adds support for configuring a a bitbang i2c controller, which
is used on some boards to configure the LCD panel (via i2c).

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
9 years agosunxi: video: Add support for LCD reset pin
Hans de Goede [Mon, 16 Feb 2015 16:26:41 +0000 (17:26 +0100)]
sunxi: video: Add support for LCD reset pin

On some boards there is a gpio to reset the LCD panel, add support for this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>