project/bcm63xx/u-boot.git
9 years agomalloc_simple: Add Kconfig option for using only malloc_simple in the SPL
Hans de Goede [Sun, 13 Sep 2015 12:45:15 +0000 (14:45 +0200)]
malloc_simple: Add Kconfig option for using only malloc_simple in the SPL

common/dlmalloc.c is quite big, both in .text and .data usage, therefor
on some boards the SPL is build to use only malloc_simple.c and not the
dlmalloc.c code. This is done in various include/configs/foo.h with the
following construct:

 #ifdef CONFIG_SPL_BUILD
 #define CONFIG_SYS_MALLOC_SIMPLE
 #endif

This commit introduces a SPL_MALLOC_SIMPLE Kconfig bool which allows
selecting this functionality through Kconfig instead.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agospl: spl_relocate_stack_gd: Do not unnecessarily clear bss
Hans de Goede [Sun, 13 Sep 2015 13:36:18 +0000 (15:36 +0200)]
spl: spl_relocate_stack_gd: Do not unnecessarily clear bss

spl_relocate_stack_gd only gets called from arch/arm/lib/crt0.S which
clears the bss directly after calling it, so there is no need to clear
it from spl_relocate_stack_gd.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
9 years agoPrepare v2015.10
Tom Rini [Mon, 19 Oct 2015 22:46:28 +0000 (18:46 -0400)]
Prepare v2015.10

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agopowerpc: Drop old non-generic-board code
Simon Glass [Sat, 17 Oct 2015 18:58:50 +0000 (12:58 -0600)]
powerpc: Drop old non-generic-board code

This code is no-longer used. Drop it.

Signed-off-by: Simon Glass <sjg@chromium.org>
9 years agosbc8641d: enable and test CONFIG_SYS_GENERIC_BOARD
Paul Gortmaker [Sat, 17 Oct 2015 20:40:30 +0000 (16:40 -0400)]
sbc8641d: enable and test CONFIG_SYS_GENERIC_BOARD

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
9 years agosbc8641d: increase monitor size from 256k to 384k
Paul Gortmaker [Sat, 17 Oct 2015 20:40:31 +0000 (16:40 -0400)]
sbc8641d: increase monitor size from 256k to 384k

Between v2015.07-rc1 and v2015.07-rc2 this board started
silent boot failure.  A bisect led to commit 6eed3786c68c8a49d
("net: Move the CMD_NET config to defconfigs").  This commit
looks harmless in itself, but it did implicitly add a feature
to the image which led to this:

 u-boot$git describe 6eed3786c68c8a49d
 v2015.07-rc1-412-g6eed3786c68c
              ^^^

 u-boot$ls -l ../41*/u-boot.bin
 -rwxrwxr-x 1 paul paul 261476 Oct 16 16:47 ../411/u-boot.bin
 -rwxrwxr-x 1 paul paul 266392 Oct 16 16:43 ../412/u-boot.bin
 u-boot$bc
 bc 1.06.95
 Copyright 1991-1994, 1997, 1998, 2000, 2004, 2006 Free Software Foundation, Inc.
 This is free software with ABSOLUTELY NO WARRANTY.
 For details type `warranty'.
 256*1024
 262144

i.e. we finally broke through the 256k monitor size.  Jump it
up to 384k and fix the hard coded value used in the env offset
at the same time.

We were probably flirting with the 256k size issue without
knowing it when testing on different baselines in earlier
commits, but since this is all board specific, a rebase or
reorder to put this commit 1st is of little value.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
9 years agosbc8641d: add basic flash setup instructions to README file
Paul Gortmaker [Sat, 17 Oct 2015 20:40:29 +0000 (16:40 -0400)]
sbc8641d: add basic flash setup instructions to README file

...so that I don't have to go work them out from scratch again
by peering at the manual.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
9 years agosbc8641d: set proper environment sector size.
Paul Gortmaker [Sat, 17 Oct 2015 20:40:28 +0000 (16:40 -0400)]
sbc8641d: set proper environment sector size.

When debugging an env fail due to too small a malloc pool, it
was noted that the env write was 256k.  But the device sector
size is 1/2 that, as can be seen from "fli" output:

Bank # 1: CFI conformant flash (16 x 16)  Size: 16 MB in 131 Sectors
  Intel Extended command set, Manufacturer ID: 0x89, Device ID: 0x1888
  Erase timeout: 4096 ms, write timeout: 1 ms
  Buffer write timeout: 2 ms, buffer size: 64 bytes

  Sector Start Addresses:
  FF000000 E RO   FF020000 E RO   FF040000 E RO   FF060000 E RO   FF080000 E RO
  FF0A0000 E RO   FF0C0000 E RO   FF0E0000 E RO   FF100000 E RO   FF120000 E RO
  [...]
  FFF00000   RO   FFF20000   RO   FFF40000   RO   FFF60000   RO   FFF80000   RO
  FFFA0000   RO   FFFC0000   RO   FFFE0000 E RO   FFFE8000   RO   FFFF0000 E RO
  FFFF8000   RO
=>

The desired env sector is FFF40000->FFF60000, or 0x20000 in length,
just after the 256k u-boot image which starts @ FFF00000.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
9 years agosbc8641d: increase malloc pool size to a sane default
Paul Gortmaker [Sat, 17 Oct 2015 20:40:27 +0000 (16:40 -0400)]
sbc8641d: increase malloc pool size to a sane default

Currently the board fails to save its env, since the env size
is much smaller than the sector size, and the malloc fails for
the pad buffer, giving the user visible symptom of:

Unable to save the rest of sector (253952)

Allow for 1M malloc pool, the same as used on the sbc8548 board.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
9 years agosbc8641d: enable command line editing
Paul Gortmaker [Sat, 17 Oct 2015 20:40:26 +0000 (16:40 -0400)]
sbc8641d: enable command line editing

It is just too painful to use interactively without it.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
9 years agoimage-fit: Fix signature checking
Andrej Rosano [Wed, 14 Oct 2015 15:45:40 +0000 (17:45 +0200)]
image-fit: Fix signature checking

On signature verification failures fit_image_verify() should
exit with error.

Signed-off-by: Andrej Rosano <andrej@inversepath.com>
9 years agoigep00x0: Use BCH8 ECC
Ladislav Michl [Mon, 12 Oct 2015 16:09:14 +0000 (18:09 +0200)]
igep00x0: Use BCH8 ECC

Used NAND chips requires at least 4-bit error correction, so use BCH8
as it is what kernel uses.

Signed-off-by: Ladislav Michl <ladis@linux-mips.org>
Acked-by: Javier Martinez Canillas <javier@osg.samsung.com>
9 years agovexpress64: Juno: Add initialisation code for Juno R1 PCIe host bridge.
Liviu Dudau [Mon, 19 Oct 2015 10:08:32 +0000 (11:08 +0100)]
vexpress64: Juno: Add initialisation code for Juno R1 PCIe host bridge.

Juno R1 has an XpressRICH3 PCIe host bridge that needs to be initialised
in order for the Linux kernel to be able to enumerate the bus. Add
support code here that enables the host bridge, trains the links and
sets up the Address Translation Tables.

Signed-off-by: Liviu Dudau <Liviu.Dudau@foss.arm.com>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
[trini: Always declare vexpress64_pcie_init and continue handling logic
inside the function]
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agovexpress64: Juno: Declare all 8GB of RAM and make them visible to the kernel.
Liviu Dudau [Mon, 19 Oct 2015 10:08:31 +0000 (11:08 +0100)]
vexpress64: Juno: Declare all 8GB of RAM and make them visible to the kernel.

Juno comes with 8GB RAM, but U-Boot only passes 2GB to the kernel.
Declare a secondary memory bank and set the sizes correctly.

Signed-off-by: Liviu Dudau <Liviu.Dudau@foss.arm.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Ryan Harkin <ryan.harkin@linaro.org>
Tested-by: Ryan Harkin <ryan.harkin@linaro.org>
9 years agodfu: dfu_sf: Take the start address into account
Fabio Estevam [Wed, 23 Sep 2015 03:50:39 +0000 (00:50 -0300)]
dfu: dfu_sf: Take the start address into account

The dfu_alt_info_spl variable allows passing a starting point
for the binary to be flashed in the SPI NOR.

For example, if we have 'dfu_alt_info_spl=spl raw 0x400', this means
that we want to flash the binary starting at address 0x400.

In order to do so we need to erase the entire sector and write to
the the subsequent SPI NOR sectors taking such start address
into account for the address calculations.

Tested by succesfully writing SPL binary into 0x400 offset and
the u-boot.img at offset 64 kiB of a SPL NOR.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
[trini: Use lldiv for the math]
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agodfu: dfu_sf: Use the erase sector size for erase operations
Fabio Estevam [Tue, 22 Sep 2015 03:55:00 +0000 (00:55 -0300)]
dfu: dfu_sf: Use the erase sector size for erase operations

SPI NOR flashes need to erase the entire sector size and we cannot pass
any arbitrary length for the erase operation.

To illustrate the problem:

Copying data from PC to DFU device
Download    [=========================] 100%       478208 bytes
Download done.
state(7) = dfuMANIFEST, status(0) = No error condition is present
state(10) = dfuERROR, status(14) = Something went wrong, but the
device does not know what it was
Done!

In this case, the binary has 478208 bytes and the M25P32 SPI NOR
has an erase sector of 64kB.

478208  = 7 entire sectors of 64kiB + 19456 bytes.

Erasing the first seven 64 kB sectors works fine, but when trying
to erase the remainding 19456 causes problem and the board hangs.

Fix the issue by always erasing with the erase sector size.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
9 years agodoc/README.scrapyard: Add more entries
Tom Rini [Mon, 19 Oct 2015 17:32:09 +0000 (13:32 -0400)]
doc/README.scrapyard: Add more entries

- Add deletions from August 30 2015.
- A few from Sept 12, one from Oct 2nd.

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agoRevert "arm: Remove inetspace_v2_cmc board"
Tom Rini [Mon, 19 Oct 2015 16:24:52 +0000 (12:24 -0400)]
Revert "arm: Remove inetspace_v2_cmc board"

Upon further review when populating README.scrapyard, inetspace_v2_cmc
is a variant on netspace_v2 and not just an orphan config.

This reverts commit 653600a715db49859c06ba5dfb858c15c4108d54.

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agoMerge branch 'master' of git://git.denx.de/u-boot-arm
Tom Rini [Mon, 19 Oct 2015 15:30:38 +0000 (11:30 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-arm

9 years agoRevert "arm: Remove d2net_v2 defconfig file"
Tom Rini [Mon, 19 Oct 2015 15:20:54 +0000 (11:20 -0400)]
Revert "arm: Remove d2net_v2 defconfig file"

Upon further review when populating README.scrapyard, d2net_v2 is a
variant on net2big_v2 and not just an orphan config.  To help in the
future also add this to board/LaCie/net2big_v2/MAINTAINERS which needed
a little consolidation anyhow.

This reverts commit 1363740e7948a8e4bee8d5adcdf0f63f7782879d.

Cc: Simon Guinot <simon.guinot@sequanux.org>
Cc: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agodoc/README.scrapyard: Populate recent removals
Tom Rini [Mon, 19 Oct 2015 15:05:47 +0000 (11:05 -0400)]
doc/README.scrapyard: Populate recent removals

Add in the commit IDs / dates for boards removed on Sept 2nd.

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agoARM: rpi: add another revision of Raspberry Pi A+
Lubomir Rintel [Wed, 14 Oct 2015 15:17:54 +0000 (17:17 +0200)]
ARM: rpi: add another revision of Raspberry Pi A+

Seen this one in the wild. Is labelled "Raspberry Pi Model A+ V1.1,
(C) Raspberry Pi 2014". A standard A+ board, much like the one with
version 0x12, didn't notice any differencies.

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
9 years agoARM: dockstar: move start of environment area
Eric Cooper [Mon, 12 Oct 2015 23:18:52 +0000 (19:18 -0400)]
ARM: dockstar: move start of environment area

The default dockstar configuration for U-Boot currently causes it to
overrun the environment area, so that a "saveenv" command bricks the
device.  This patch moves the environment to a higher address to avoid
that.

Signed-off-by: Eric Cooper <ecc@cmu.edu>
9 years agoARM: k2e/l: Apply WA for selecting PA clock source
Lokesh Vutla [Thu, 8 Oct 2015 06:01:47 +0000 (11:31 +0530)]
ARM: k2e/l: Apply WA for selecting PA clock source

On keystone2 Lamarr and Edison platforms, the PA clocksource
mux in PLL REG1, can be changed only after enabling its clock
domain.
So selecting the output of PASS PLL as input to PA only after
enabling the clockdomain.
This is as per the debug done by "Vitaly Andrianov <vitalya@ti.com>"
and based on the previous work done by "Hao Zhang <hzhang@ti.com>"

Fixes: d634a0775bcf ("ARM: keystone2: Cleanup PLL init code")
Reported-by: Vitaly Andrianov <vitalya@ti.com>
Tested-by: Vitaly Andrianov <vitalya@ti.com>
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
9 years agoarch/powerpc/config.mk: Pass -fno-ira-hoist-pressure when possible
Tom Rini [Sat, 17 Oct 2015 12:04:11 +0000 (08:04 -0400)]
arch/powerpc/config.mk: Pass -fno-ira-hoist-pressure when possible

There are various toolchain issues that cause us to produce invalid
binaries with certain gcc 4.8.x and 4.9.x versions when we don't pass
this flag in.

Tested-by: Joakim Tjernlund <joakim.tjernlund@transmode.se>
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agoMerge git://git.denx.de/u-boot-socfpga
Tom Rini [Sat, 17 Oct 2015 00:21:04 +0000 (20:21 -0400)]
Merge git://git.denx.de/u-boot-socfpga

9 years agoarm: dts: socfpga: add "u-boot,dm-pre-reloc" to socfpga_cyclone5_socdk dts
Dinh Nguyen [Mon, 12 Oct 2015 16:59:04 +0000 (11:59 -0500)]
arm: dts: socfpga: add "u-boot,dm-pre-reloc" to socfpga_cyclone5_socdk dts

We need "u-boot,dm-pre-reloc" in the socfpga_cyclone5_socdk.dts file in
order for the SPL to use SD/MMC.

Signed-off-by: Dinh Nguyen <dinguyen@opensource.altera.com>
9 years agoarm: socfpga: enable data/inst prefetch and shared override in the L2
Dinh Nguyen [Thu, 15 Oct 2015 15:13:36 +0000 (10:13 -0500)]
arm: socfpga: enable data/inst prefetch and shared override in the L2

Update the L2 AUX CTRL settings for the SoCFPGA.

Enabling D and I prefetch bits helps improve SDRAM performance on the
platform.

Also, we need to enable bit 22 of the L2. By not having bit 22 set in the
PL310 Auxiliary Control register (shared attribute override enable) has the
side effect of transforming Normal Shared Non-cacheable reads into Cacheable
no-allocate reads.

Coherent DMA buffers in Linux always have a Cacheable alias via the
kernel linear mapping and the processor can speculatively load cache
lines into the PL310 controller. With bit 22 cleared, Non-cacheable
reads would unexpectedly hit such cache lines leading to buffer
corruption.

Signed-off-by: Dinh Nguyen <dinguyen@opensource.altera.com>
9 years agovf610twr: Fix typo in DRAM init
Anthony Felice [Fri, 9 Oct 2015 20:38:39 +0000 (16:38 -0400)]
vf610twr: Fix typo in DRAM init

This commit fixes a typo in vf610twr DRAM init that was causing a hang in
U-Boot for the Vybrid Tower. This typo was introduced in commit 3f353cecc
(vf610: refactor DDRMC code).

Signed-off-by: Anthony Felice <tony.felice@timesys.com>
Reviewed-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agoMerge branch 'master' of git://git.denx.de/u-boot-samsung
Tom Rini [Fri, 16 Oct 2015 11:19:47 +0000 (07:19 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-samsung

9 years agoarm: mmu: Add missing volatile for reading SCTLR register
Alison Wang [Wed, 9 Sep 2015 02:22:02 +0000 (10:22 +0800)]
arm: mmu: Add missing volatile for reading SCTLR register

Add 'volatile' qualifier to the asm statement in get_cr()
so that the statement is not optimized out by the compiler.

(http://comments.gmane.org/gmane.linux.linaro.toolchain/5163)

Without the 'volatile', get_cr() returns a wrong value which
prevents enabling the MMU  and later causes a PCIE VA access
failure.

Signed-off-by: Alison Wang <alison.wang@freescale.com>
9 years agoMerge branch 'master' of git://git.denx.de/u-boot-arm
Tom Rini [Thu, 15 Oct 2015 21:45:39 +0000 (17:45 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-arm

9 years agopci: pcie_imx: Fix hang on mx6qp
Fabio Estevam [Tue, 13 Oct 2015 14:01:27 +0000 (11:01 -0300)]
pci: pcie_imx: Fix hang on mx6qp

PCI driver currently hangs on mx6qp.

Toggle the reset bit with the appropriate timings to fix the issue.

Based on the FSL kernel driver implementation.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Stefano Babic <sbabic@denx.de>
9 years agoarmv8/gic: Fix GIC v2 initialization
Thierry Reding [Thu, 20 Aug 2015 09:52:15 +0000 (11:52 +0200)]
armv8/gic: Fix GIC v2 initialization

Initialize all GICD_IGROUPRn registers and set up GICC_CTLR to enable
interrupts to the primary CPU. This fixes issues seen after booting a
Linux kernel from U-Boot.

Suggested-by: Marc Zyngier <marc.zyngier@arm.com>
Suggested-by: Mark Rutland <mark.rutland@arm.com>
Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoarmv8/mmu: Set bits marked RES1 in TCR
Thierry Reding [Thu, 20 Aug 2015 09:52:14 +0000 (11:52 +0200)]
armv8/mmu: Set bits marked RES1 in TCR

For EL3 and EL2, the documentation says that bits 31 and 23 are reserved
but should be written as 1.

For EL1, only bit 23 is not reserved, so only write bit 31 as 1.

Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoMerge branch 'master' of git://www.denx.de/git/u-boot-imx
Tom Rini [Thu, 15 Oct 2015 12:43:38 +0000 (08:43 -0400)]
Merge branch 'master' of git://denx.de/git/u-boot-imx

9 years agoARM: uniphier: fix address mapping in README.uniphier
Masahiro Yamada [Wed, 30 Sep 2015 12:15:58 +0000 (21:15 +0900)]
ARM: uniphier: fix address mapping in README.uniphier

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
9 years agoarmv8/mmu: Clean up TCR programming
Thierry Reding [Thu, 20 Aug 2015 09:52:13 +0000 (11:52 +0200)]
armv8/mmu: Clean up TCR programming

Use the inner shareable attribute for memory, which makes more sense
considering that this code is called when caches are being enabled.

Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
9 years agoarm: vf610twr: improve memory layout
Stefan Agner [Wed, 14 Oct 2015 17:58:43 +0000 (10:58 -0700)]
arm: vf610twr: improve memory layout

Currently, the device tree relocation is disabled, likely to
keep some DDR3 RAM at the end for Cortex-M4 firmwares. This
can be archived using bootm_size, which limits the image
processing range of the boot commands.

Move the device tree standard load address to a higher address
which aligns better with what we are doing on other boards.

Signed-off-by: Stefan Agner <stefan@agner.ch>
Acked-by: Otavio Salvador <otavio@ossystems.com.br>
9 years agocolibri_vf: Fix bstlen field
Fabio Estevam [Wed, 14 Oct 2015 02:54:32 +0000 (23:54 -0300)]
colibri_vf: Fix bstlen field

Commit 3f353cecc ("vf610: refactor DDRMC code") changed the original
bstlen field from 3 to 0.

Restore the original value for proper behaviour.

Based on the patch from Anthony Felice <tony.felice@timesys.com>
for the vf610twr board.

Reported-by: Stefan Agner <stefan@agner.ch>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agomtd: nand: vf610_nfc: resync with upstream Linux version
Stefan Agner [Wed, 14 Oct 2015 05:11:42 +0000 (22:11 -0700)]
mtd: nand: vf610_nfc: resync with upstream Linux version

This resyncs the driver changes with the Linux version of the
driver. The driver received some feedback in the LKML and got
recently acceppted, the latest version can be found here:
https://lkml.org/lkml/2015/9/2/678

Notable changes are:
- On ECC error, reread OOB and count bit flips in OOB too.
  If flipped bits are below threshold, also return an empty
  OOB buffer.
- Return the amount of bit flips in vf610_nfc_read_page.
- Use endianness aware vf610_nfc_read to read ECC status.
- Do not enable IDLE IRQ (since we do not operate with an
  interrupt service routine).
- Use type safe struct for buffer variants (vf610_nfc_alt_buf).
- Renamed variables in struct vf610_nfc (column and page_sz)
  to reflect better what they really representing.

The U-Boot version currently does not support RAW NAND write
when using the HW ECC engine.

Signed-off-by: Bhuvanchandra DV <bhuvanchandra.dv@toradex.com>
Signed-off-by: Stefan Agner <stefan@agner.ch>
Tested-by: Albert ARIBAUD (3ADEV) <albert.aribaud@3adev.fr>
Tested-by: Stefan Agner <stefan@agner.ch>
Acked-by: Scott Wood <scottwood@freescale.com>
9 years agoMerge remote-tracking branch 'u-boot/master'
Albert ARIBAUD [Wed, 14 Oct 2015 08:46:36 +0000 (10:46 +0200)]
Merge remote-tracking branch 'u-boot/master'

9 years agoMerge branch 'master' of git://www.denx.de/git/u-boot-imx
Tom Rini [Tue, 13 Oct 2015 12:37:38 +0000 (08:37 -0400)]
Merge branch 'master' of git://denx.de/git/u-boot-imx

9 years agoexynos: more debug and cleanup in do_sdhci_init()
Tobias Jakobi [Mon, 5 Oct 2015 11:47:53 +0000 (13:47 +0200)]
exynos: more debug and cleanup in do_sdhci_init()

Add more debug printfs in do_sdhci_init() for calls
that can potentially fail.

Acked-by: Przemyslaw Marczak <p.marczak@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoexynos: be more verbose in process_nodes()
Tobias Jakobi [Mon, 5 Oct 2015 11:47:52 +0000 (13:47 +0200)]
exynos: be more verbose in process_nodes()

In case sdhci_get_config() or do_sdhci_init() fail, show
the error code that was returned.

Acked-by: Przemyslaw Marczak <p.marczak@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoexynos: Fix passing of errors in exynos_mmc_init()
Tobias Jakobi [Mon, 5 Oct 2015 11:47:51 +0000 (13:47 +0200)]
exynos: Fix passing of errors in exynos_mmc_init()

exynos_mmc_init() always returns zero, so for the caller
it looks like it never fails.

Correct this by returning the error code of process_nodes().
For process_nodes() do something similar and return early
when do_sdhci_init() fails.

v2: Only fail in process_nodes() if we fail on all
    available nodes.

Acked-by: Przemyslaw Marczak <p.marczak@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoexynos: Properly zero initialize host in s5p_sdhci_init()
Tobias Jakobi [Mon, 5 Oct 2015 11:47:50 +0000 (13:47 +0200)]
exynos: Properly zero initialize host in s5p_sdhci_init()

This makes sure that setting the host_caps in s5p_sdhci_core_init()
doesn't operate on potentially uninitialized memory.

Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoodroid: Add boot script (boot.scr) support
Guillaume GARDET [Fri, 9 Oct 2015 12:26:23 +0000 (14:26 +0200)]
odroid: Add boot script (boot.scr) support

Add boot script (boot.scr) support. If no boot script are
found, it boots as usual.

Signed-off-by: Guillaume GARDET <guillaume.gardet@free.fr>
Tested-by: Przemyslaw Marczak <p.marczak@samsung.com>
Acked-by: Przemyslaw Marczak <p.marczak@samsung.com>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agoodroid: replace 'fatload' with 'load' to be able to use EXT* partitions
Guillaume GARDET [Fri, 9 Oct 2015 12:26:22 +0000 (14:26 +0200)]
odroid: replace 'fatload' with 'load' to be able to use EXT* partitions

Replace 'fatload' command by 'load', to be able to use EXT*
partitions while keeping FAT partition compatibility.

Signed-off-by: Guillaume GARDET <guillaume.gardet@free.fr>
Tested-by: Przemyslaw Marczak <p.marczak@samsung.com>
Acked-by: Przemyslaw Marczak <p.marczak@samsung.com>
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
9 years agols102xa: Fix reset hang
Fabio Estevam [Sat, 3 Oct 2015 17:21:00 +0000 (14:21 -0300)]
ls102xa: Fix reset hang

Since commit 623d96e89aca6("imx: wdog: correct wcr register settings")
issuing a 'reset' command causes the system to hang.

Unlike i.MX and Vybrid, the watchdog controller on LS102x is big-endian.

This means that the watchdog on LS1021 has been working by accident as
it does not use the big-endian accessors in drivers/watchdog/imx_watchdog.c.
Commit 623d96e89aca6("imx: wdog: correct wcr register settings") only
revelead the endianness problem on LS102x.

In order to fix the reset hang, introduce a reset_cpu() implementation that
is specific for ls102x, which accesses the watchdog WCR register in big-endian
format. All that is required to reset LS102x is to clear the SRS bit.

This approach is a temporary workaround to avoid a regression for LS102x
in the 2015.10 release. The proper fix is to make the watchdog driver
endian-aware, so that it can work for i.MX, Vybrid and LS102x.

Reported-by: Sinan Akman <sinan@writeme.com>
Tested-by: Sinan Akman <sinan@writeme.com>
Reviewed-by: Wolfgang Denk <wd@denx.de>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agoimx_watchdog: Add a header file for watchdog registers
Fabio Estevam [Sat, 3 Oct 2015 17:20:59 +0000 (14:20 -0300)]
imx_watchdog: Add a header file for watchdog registers

Create fsl_wdog.h to store the watchdog registers and bit fields.

This can be useful when accesses to the watchdog block are made from other
parts, such as arch/arm/ cpu code.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agoPrepare v2015.10-rc5
Tom Rini [Mon, 12 Oct 2015 15:14:27 +0000 (11:14 -0400)]
Prepare v2015.10-rc5

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agoenv_eeprom.c: Correct using saved environment
Ludger Dreier [Mon, 12 Oct 2015 11:34:24 +0000 (13:34 +0200)]
env_eeprom.c: Correct using saved environment

The changes in ed6a5d4 unintentionally broke support for reading the
environment saved to eeprom back.  To correct this the crc-check and
decision on which environment to use is now moved to env_relocate_spec.
This is done for both the "redundant env" and the "single env" case.

Signed-off-by: Ludger Dreier <ludger.dreier@keymile.com>
9 years agopcm052: fix MTD partitioning
Albert ARIBAUD (3ADEV) [Sun, 11 Oct 2015 18:06:39 +0000 (20:06 +0200)]
pcm052: fix MTD partitioning

MTD partitioning in current pcm052 configuration is inconsistent.
Fix it across MTDPARTS_DEFAULT, CONFIG_EXTRA_ENV_SETTINGS, and
CONFIG_ENV_OFFSET[_REDUND].

Signed-off-by: Albert ARIBAUD (3ADEV) <albert.aribaud@3adev.fr>
9 years agotest/fs/fs-test.sh: Update expected results and TC10 logic
Tom Rini [Sun, 11 Oct 2015 20:48:36 +0000 (16:48 -0400)]
test/fs/fs-test.sh: Update expected results and TC10 logic

With the changes in 7a3e70c we now get read(2) behavior so trying to
read 2MB with 1MB left in the file results in 1MB read and a warning.
We update the test logic here to make sure we read back 1MB as expected.
This change however changes the overall summary as while EXT4 continues
to not have offset support the test now fails when expected to pass
rather than fails when expected to fail (and we report that as pass).

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agolpc32xx: fix calculation of HCLK PLL output clock
Vladimir Zapolskiy [Sun, 4 Oct 2015 22:18:45 +0000 (23:18 +0100)]
lpc32xx: fix calculation of HCLK PLL output clock

Execution branches on feedback mode are swapped, this has no effect
if default direct mode is on (then p_div is equal to 1 and Fout equals
to Fcco), that's why the problem remained unnoticed for a long time.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
9 years agolpc32xx: remove surplus clock cycle in PL175 WAIT_OEN config
Vladimir Zapolskiy [Sun, 4 Oct 2015 22:18:24 +0000 (23:18 +0100)]
lpc32xx: remove surplus clock cycle in PL175 WAIT_OEN config

According to ARM PrimeCell PL175 documentation WAIT_OEN config value
is defined without any additional clocks added to the value set by a
client, the change fixes the wrong interface to WAIT_OEN config.

The change also touches a single user of LPC32xx EMC and corrects
configured "output enable delay" value on its side according to the
changed interface.

No functional change intended.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
9 years agonand: omap_gpmc: Change correctable bit-flips messages to debug()
Ezequiel García [Sun, 4 Oct 2015 21:34:42 +0000 (18:34 -0300)]
nand: omap_gpmc: Change correctable bit-flips messages to debug()

Messages on corrected bit-flips are not really useful,
as bit-flips are perfectly normal. Let's avoid cluttering
the console and make them debug.

Signed-off-by: Ezequiel Garcia <ezequiel@vanguardiasur.com.ar>
9 years agoFix variation in timestamps caused by timezone differences.
Vagrant Cascadian [Fri, 2 Oct 2015 16:11:51 +0000 (09:11 -0700)]
Fix variation in timestamps caused by timezone differences.

When building with SOURCE_DATE_EPOCH set, avoid use of mktime in
default_image.c, which converts the timestamp into localtime. This
causes variation based on timezone when building u-boot.img and
u-boot-sunxi-with-spl.bin targets.

Signed-off-by: Vagrant Cascadian <vagrant@debian.org>
Tested-by: Paul Kocialkowski <contact@paulk.fr>
Acked-by: Paul Kocialkowski <contact@paulk.fr>
9 years agocommon/image.c: Make boot_get_ramdisk() perform a check for Android images
Tom Rini [Thu, 27 Aug 2015 19:42:41 +0000 (15:42 -0400)]
common/image.c: Make boot_get_ramdisk() perform a check for Android images

In 2dd4632 the check for where a ramdisk is found on an Android image
was got moved into the "normal" loop here, causing people to have to
pass the kernel address in the ramdisk address location in order to have
Android boot still.  This changed previous behavior so perform a check
early in the function to see if we have an Android image and if so use
that as where to look for the ramdisk (which is what the rest of the
code here expects).  We allow for this to still be overridden with an
explicit ramdisk address to be passed as normal.

Cc: Rob Herring <robh@kernel.org>
Reported-by: Paul Kocialkowski <contact@paulk.fr>
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agotools/mkimage.c: Clarify help text for -D slightly
Tom Rini [Wed, 26 Aug 2015 19:21:23 +0000 (15:21 -0400)]
tools/mkimage.c: Clarify help text for -D slightly

Try and make it clear that -D will replace all arguments passed to dtc
and is not appending them.

Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agoarndale: Apply Cortex-A15 errata #773022 and #774769
Ian Campbell [Tue, 29 Sep 2015 09:27:09 +0000 (10:27 +0100)]
arndale: Apply Cortex-A15 errata #773022 and #774769

We run 4 Arndale boards in our automated test framework, they have
been running quite happily for quite some time using a Debian Wheezy
userspace.

However when upgrading to a Debian Jessie we started seeing frequent
segmentation faults from gcc when building the kernel, to the extent
that it is unable to successfully build the kernel twice in a row, and
often fails on the first attempt.

Searching around I found https://bugs.launchpad.net/arndale/+bug/1081417
which pointed towards http://www.spinics.net/lists/kvm-arm/msg03723.html
and CPU Errata 773022 and 774769.

This errata needs to be applied to all processors in an SMP system,
meaning that the usual strategy of applying them in
arch/arm/cpu/armv7/start.S is not appropriate (since that applies to
the boot processor only). Instead we apply these errata in the secure
monitor which is code that is traversed by all processors as they are
brought up.

The net affect on Arndale is that ACTLR changes from 0x40 to
0x2000042. I ran 17 kernel compile iterations overnight with no
segfaults.

Runtime testing was done on our v2014.10 based branch and forward
ported (with only minimal and trivial contextual conflicts) to current
master, where it has been build tested only.

I suppose in theory these errata apply to any Exynos5250 based boards,
but Arndale is the only one I have access to and I have therefore
chosen to be conservative and only apply it there.

Also, reorder CONFIG_ARM_ERRATA_794072 in README to make the list
numerically sorted.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
9 years agoimage: fix support for Android boot images with no ramdisk
Rob Herring [Mon, 5 Oct 2015 19:37:07 +0000 (14:37 -0500)]
image: fix support for Android boot images with no ramdisk

If an Android boot image does not contain a ramdisk, make sure rd_len
and rd_data are returned to indicate no ramdisk rather than just relying
on returning an error.

Signed-off-by: Rob Herring <robh@kernel.org>
9 years agoAdd support for LZ4 decompression algorithm
Julius Werner [Wed, 7 Oct 2015 03:03:53 +0000 (20:03 -0700)]
Add support for LZ4 decompression algorithm

This patch adds support for LZ4-compressed FIT image contents. This
algorithm has a slightly worse compression ration than LZO while being
nearly twice as fast to decompress. When loading images from a fast
storage medium this usually results in a boot time win.

Sandbox-tested only since I don't have a U-Boot development system set
up right now. The code was imported unchanged from coreboot where it's
proven to work, though. I'm mostly interested in getting this recognized
by mkImage for use in a downstream project.

Signed-off-by: Julius Werner <jwerner@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agoppc4xx: Remove lcd4_lwmon5 support
Stefan Roese [Fri, 2 Oct 2015 06:20:37 +0000 (08:20 +0200)]
ppc4xx: Remove lcd4_lwmon5 support

This platform has not gone into production. So lets remove it.

Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
9 years agoppc4xx: Convert lwmon5 board to generic board
Stefan Roese [Fri, 2 Oct 2015 06:20:36 +0000 (08:20 +0200)]
ppc4xx: Convert lwmon5 board to generic board

Add CONFIG_SYS_GENERIC_BOARD to lwmon5.h and CONFIG_DISPLAY_BOARDINFO
to Kconfig file.

Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
9 years agoRevert "powerpc: ppc4xx: remove lwmon5 support"
Stefan Roese [Fri, 2 Oct 2015 06:20:35 +0000 (08:20 +0200)]
Revert "powerpc: ppc4xx: remove lwmon5 support"

This reverts commit 8fe11b8901a31d11990488c82bc23612589d57be.

I'll add support to lwmon5 in the next patch and will remove
support for the broken lcd4_lwmon5 as well.

Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Masahiro Yamada <yamada.masahiro@socionext.com>
9 years agofs/fat/fat_write: Fix management of empty files
Benoît Thébaudeau [Mon, 28 Sep 2015 13:45:32 +0000 (15:45 +0200)]
fs/fat/fat_write: Fix management of empty files

Overwriting an empty file not created by U-Boot did not work, and it
could even corrupt the FAT. Moreover, creating empty files or emptying
existing files allocated a cluster, which is not standard.

Fix this by always keeping empty files clusterless as specified by
Microsoft (the start cluster must be set to 0 in the directory entry in
that case), and by supporting overwriting such files.

Signed-off-by: Benoît Thébaudeau <benoit@wsystem.com>
9 years agofs/fat/fat_write: Factor out duplicate code
Benoît Thébaudeau [Mon, 28 Sep 2015 13:45:31 +0000 (15:45 +0200)]
fs/fat/fat_write: Factor out duplicate code

Signed-off-by: Benoît Thébaudeau <benoit@wsystem.com>
9 years agofs/fat/fat_write: Fix curclust/newclust mix-up
Benoît Thébaudeau [Mon, 28 Sep 2015 13:45:30 +0000 (15:45 +0200)]
fs/fat/fat_write: Fix curclust/newclust mix-up

curclust was used instead of newclust in the debug() calls and in one
CHECK_CLUST() call, which could skip a failure case.

Signed-off-by: Benoît Thébaudeau <benoit@wsystem.com>
9 years agofs/fat/fat_write: Merge calls to set_cluster()
Benoît Thébaudeau [Mon, 28 Sep 2015 13:45:29 +0000 (15:45 +0200)]
fs/fat/fat_write: Merge calls to set_cluster()

set_contents() had uselessly split calls to set_cluster(). Merge these
calls, which removes some cases of set_cluster() being called with a
size of zero.

Signed-off-by: Benoît Thébaudeau <benoit@wsystem.com>
9 years agofs/fat/fat_write: Fix buffer alignments
Benoît Thébaudeau [Mon, 28 Sep 2015 13:45:28 +0000 (15:45 +0200)]
fs/fat/fat_write: Fix buffer alignments

set_cluster() was using a temporary buffer without enforcing its
alignment for DMA and cache. Moreover, it did not check the alignment of
the passed buffer, which can come directly from applicative code or from
the user.

This could cause random data corruption, which has been observed on
i.MX25 writing to an SD card.

Fix this by only passing ARCH_DMA_MINALIGN-aligned buffers to
disk_write(), which requires the introduction of a buffer bouncing
mechanism for the misaligned buffers passed to set_cluster().

By the way, improve the handling of the corresponding return values from
disk_write():
 - print them with debug() in case of error,
 - consider that there is an error is disk_write() returns a smaller
   block count than the requested one, not only if its return value is
   negative.

After this change, set_cluster() and get_cluster() are almost
symmetrical.

Signed-off-by: Benoît Thébaudeau <benoit@wsystem.com>
9 years agoti: omap3: config: remove 1 from boolean define
Igor Grinberg [Thu, 8 Oct 2015 18:12:25 +0000 (21:12 +0300)]
ti: omap3: config: remove 1 from boolean define

CONFIG_TWL4030_POWER is a boolean define variable. It is either defined
or not defined and should not have a value assigned to it.
Remove the value.

Signed-off-by: Igor Grinberg <grinberg@compulab.co.il>
9 years agovexpress64: juno: use /dev/sda2
Ryan Harkin [Fri, 9 Oct 2015 16:18:08 +0000 (17:18 +0100)]
vexpress64: juno: use /dev/sda2

This patch changes the default "root=" parameter to "/dev/sda2".

Many linux based distros use /dev/sda1 for their boot partition; this is
often not a rootfs that can be used by the "root=" parameter.

Linaro images use /dev/sda1 as a boot partition, although this of a
different nature to a distro image.  Linaro uses /dev/sda2 for the rootfs
partition.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agovexpress64: juno: add alternate kernel and device tree filenames
Ryan Harkin [Fri, 9 Oct 2015 16:18:07 +0000 (17:18 +0100)]
vexpress64: juno: add alternate kernel and device tree filenames

The latest Juno firmware stores the files in NOR flash as "norkern" for
kernel binary, "board.dtb" for the device tree binary.

The "old" firmware used the name "Image" for the kernel binary and
"juno" for the device tree binary.

Rather than just change the default U-Boot configuration to use the new
names, breaking users with the old firmware, attempt to load the default
filename first.  If that fails, attempt to load the alternate filename.

I've echo'd that we are loading the alternate file to counter the
output from "afs load" shown if the first load attempt fails.  For
example, I see output like this on my Juno board when it's configured
the with the "old" firmware:

    image "norkern" not found in flash
    Loading Image instead of norkern
    loaded region 0 from 08500000 to 8000000000AB6318 bytes
    image "board.dtb" not found in flash
    Loading juno instead of board.dtb
    loaded region 0 from 0A000000 to 8300000000003188 bytes

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agovexpress64: juno: add optional initrd
Ryan Harkin [Fri, 9 Oct 2015 16:18:06 +0000 (17:18 +0100)]
vexpress64: juno: add optional initrd

Some OS images require an initrd on Juno.

If the file ramdisk.img exists in NOR flash, then we load it and pass
the address to the kernel.  Otherwise, we pass the "-" parameter as
before.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agocommon/armflash: load_image returns success or failure
Ryan Harkin [Fri, 9 Oct 2015 16:18:05 +0000 (17:18 +0100)]
common/armflash: load_image returns success or failure

Change the load_image so that it returns success or failure of the
command (using CMD_RET_SUCCESS or CMD_RET_FAILURE).

This way, hush scripts can optionally load different files depending
upon the system configuration.

A simple example:

if afs load ${kernel_name} ${kernel_addr}; then echo loaded; else echo \
not loaded; fi

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agocommon/armflash: add command to check if image exists
Ryan Harkin [Fri, 9 Oct 2015 16:18:04 +0000 (17:18 +0100)]
common/armflash: add command to check if image exists

Add a command to the ARM flash support to check if an image exists or
not.

If the image is found, it will return CMD_RET_SUCCESS, else
CMD_RET_FAILURE.  This allows hush scripts to conditionally load images.

A simple example:

if afs exists ${kernel_name}; then echo found; else echo \
not found; fi

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agovexpress64: juno: add androidboot.hardware=juno
Ryan Harkin [Fri, 9 Oct 2015 16:18:03 +0000 (17:18 +0100)]
vexpress64: juno: add androidboot.hardware=juno

Linaro's Juno Android builds requires the androidboot.hardware parameter
be set to a know board name.

Non-Android kernels ignore this extra parameter because they don't
contain code to parse it.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agovexpress64: fvp dram: add DRAM configuration
Ryan Harkin [Fri, 9 Oct 2015 16:18:02 +0000 (17:18 +0100)]
vexpress64: fvp dram: add DRAM configuration

Create an additional FVP configuration to boot images pre-loaded into
DRAM.

Sometimes it's preferential to boot the model by loading the files
directly into DRAM via model parameters, rather than using
SemiHosting.

An example of model parmaters that are used to pre-load the files
into DRAM:
    --data cluster0.cpu0=Image@0x80080000 \
    --data cluster0.cpu0=fvp-base-gicv2-psci.dtb@0x83000000 \
    --data cluster0.cpu0=uInitrd@0x84000000

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
[trini: Update board/armltd/vexpress64/Kconfig logic]
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agovexpress64: increase max gunzip size
Ryan Harkin [Fri, 9 Oct 2015 16:18:01 +0000 (17:18 +0100)]
vexpress64: increase max gunzip size

vexpress64 kernels are usually over 8 MBytes in length, so setting the
max uImage length to 64 Mbytes should give us plenty of scope for
expansion.

I mostly chose this length to match other board configs that use
"(64 << 20)".

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agovexpress64: Kconfig: tidy up
Ryan Harkin [Fri, 9 Oct 2015 16:18:00 +0000 (17:18 +0100)]
vexpress64: Kconfig: tidy up

The FVP and Juno settings were identical, but duplicated, so I removed
the duplication with this patch.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
[trini: Adjust logic to keep if/endif in the file]
Signed-off-by: Tom Rini <trini@konsulko.com>
9 years agovexpress64: fix checkpatch warnings
Ryan Harkin [Fri, 9 Oct 2015 16:17:59 +0000 (17:17 +0100)]
vexpress64: fix checkpatch warnings

This patch fixes a couple of checkpatch warnings on the vexpress64 config.

Signed-off-by: Ryan Harkin <ryan.harkin@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
9 years agoconfigs: ls1021atwr: Enable DSPI for LS1021ATWR
Yao Yuan [Tue, 15 Sep 2015 10:28:20 +0000 (18:28 +0800)]
configs: ls1021atwr: Enable DSPI for LS1021ATWR

DSPI2 can be verified when boot from QSPI now.

Signed-off-by: Yuan Yao <yao.yuan@freescale.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agomtd: sf: Add support AT26DF081A chip
Yao Yuan [Tue, 15 Sep 2015 10:28:19 +0000 (18:28 +0800)]
mtd: sf: Add support AT26DF081A chip

AT26DF081A is the spi flash type of TWR-MEM(SCH-26248) card.
We can access the flash through DSPI2 on LS1021ATWR board.

Signed-off-by: Yuan Yao <yao.yuan@freescale.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agodm: dts: ls1021a-twr: Enable DSPI2 on LS1021ATWR
Yuan Yao [Wed, 30 Sep 2015 07:35:15 +0000 (13:05 +0530)]
dm: dts: ls1021a-twr: Enable DSPI2 on LS1021ATWR

Erratum A-008022 has been fixed on LS1021A Rev2.0.
So we can use DSPI2 now, this patch enable DSPI2
in dts for LS1021ATWR.

Signed-off-by: Yuan Yao <yao.yuan@freescale.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agospi: tegra20: Add support for mode selection
Mirza Krak [Tue, 8 Sep 2015 08:30:49 +0000 (10:30 +0200)]
spi: tegra20: Add support for mode selection

Respect the mode passed in set_mode ops.

Signed-off-by: Mirza Krak <mirza.krak@hostmobility.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
9 years agospi: zynq_spi: Fix to configure CPOL, CPHA mask
Jagan Teki [Mon, 7 Sep 2015 20:08:50 +0000 (01:38 +0530)]
spi: zynq_spi: Fix to configure CPOL, CPHA mask

priv->mode is initialized when .set_speed triggers
with mode value, so checking mode for configuring
CPOL, CPHA using priv->mode is invalid hence use
mode from .set_speed argument, and at the end
priv->mode will initialized with mode.

This patch also replaces formatting string to use
speed instead of mode in .set_speed ops.

Signed-off-by: Jagan Teki <jteki@openedev.com>
9 years agospi: xilinx_spi: Fix to configure CPOL, CPHA mask
Jagan Teki [Mon, 7 Sep 2015 19:56:29 +0000 (01:26 +0530)]
spi: xilinx_spi: Fix to configure CPOL, CPHA mask

priv->mode is initialized when .set_speed triggers
with mode value, so checking mode for configuring
CPOL, CPHA using priv->mode is invalid hence use
mode from .set_speed argument, and at the end
priv->mode will initialized with mode.

This patch also replaces formatting string to use
speed instead of mode in .set_speed ops.

Signed-off-by: Jagan Teki <jteki@openedev.com>
9 years agosunxi: Fix USB regulators in Linksprite_pcDuino_defconfig
Siarhei Siamashka [Thu, 8 Oct 2015 20:22:46 +0000 (23:22 +0300)]
sunxi: Fix USB regulators in Linksprite_pcDuino_defconfig

The pcDuino1 board unconditionally provides 5V to USB host
receptacles. The pcDuino2 board has a voltage regulator,
controlled by the PD2 pin which is pulled-up by default
(so that the USB power is also enabled by default).

Not specifying pins for enabling USB power in the defconfig
means that the PH3 and PH6 pins are driven high by default.
The PH6 pin is available on the Arduino-compatible expansion
header and touching it is not nice (this may be even dangerous,
depending on what kind of role is assigned to this particular
pin by various Arduino shields).

This patch explicitly configures the USB VBUS pins to "",
which means that no pins should be touched. The patch has
been tested on a pcDuino2 board and USB still works.

Signed-off-by: Siarhei Siamashka <siarhei.siamashka@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
9 years agosunxi: Fix pcDuino reliability by downclocking DRAM to 360MHz
Siarhei Siamashka [Thu, 8 Oct 2015 20:22:45 +0000 (23:22 +0300)]
sunxi: Fix pcDuino reliability by downclocking DRAM to 360MHz

Linksprite_pcDuino_defconfig is a generic config for pcDuino1 and
pcDuino2 boards. The pcDuino2 board exists at least in two variants
(with DDR3 chips from HYNIX or NANYA). At least one pcDuino2 board
with HYNIX DDR3 fails the lima-memtester reliability test unless
the DRAM clock speed is reduced to 360MHz.

A detailed analysis report, generated by the a10-tpr3-scan tool with
the explanations why the DRAM is failing at 408MHz, is available at:
    http://linux-sunxi.org/index.php?title=User:Ssvb/pcDuino2_with_HYNIX_DDR3_reliability_test&oldid=15152
    http://web.archive.org/web/20151008190210/http://linux-sunxi.org/User:Ssvb/pcDuino2_with_HYNIX_DDR3_reliability_test

Signed-off-by: Siarhei Siamashka <siarhei.siamashka@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
9 years agoMerge git://git.denx.de/u-boot-x86
Tom Rini [Fri, 9 Oct 2015 13:55:33 +0000 (09:55 -0400)]
Merge git://git.denx.de/u-boot-x86

9 years agopci: Fix expansion ROM programming for multi-function devices
Bin Meng [Wed, 7 Oct 2015 09:13:18 +0000 (02:13 -0700)]
pci: Fix expansion ROM programming for multi-function devices

PCI_HEADER_TYPE register (offset 0x0e) bit 7 is an indicator
for multi-function devices. We should mask it off before using
it as the header type.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
9 years agoMerge git://git.denx.de/u-boot-arc
Tom Rini [Thu, 8 Oct 2015 07:04:36 +0000 (03:04 -0400)]
Merge git://git.denx.de/u-boot-arc

9 years agoMerge git://git.denx.de/u-boot-dm
Tom Rini [Thu, 8 Oct 2015 07:03:41 +0000 (03:03 -0400)]
Merge git://git.denx.de/u-boot-dm

9 years agoboard: axs10x - cap max SDIO clock value to bus/2
Alexey Brodkin [Sun, 4 Oct 2015 13:10:26 +0000 (16:10 +0300)]
board: axs10x - cap max SDIO clock value to bus/2

It turned out with some boards (FPGA firmwares?) and cards combos
current clock settings doesn't work as expected leading to strange
card freezes or corrupted data being read from the card.

Especially this was seen with Transcend 2Gb cards shipped as a part of
ARC SDP:
----------------->8---------------
AXS# mmcinfo
Device: Synopsys Mobile storage
Manufacturer ID: 74
OEM: 4a60
Name: SDC
Tran Speed: 50000000
Rd Block Len: 512
SD version 3.0
High Capacity: No
Capacity: 1.8 GiB
Bus Width: 4-bit
Erase Group Size: 512 Bytes
AXS# fatload mmc 0
** Unrecognized filesystem type **
----------------->8---------------

With this change that problem is fixed.
Note "Tran Speed" above doesn't match clock value set in DW MMC.
It is max value for card's speed class.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
9 years agoimximage: fix commands other than write_data
Troy Kisky [Tue, 15 Sep 2015 01:06:31 +0000 (18:06 -0700)]
imximage: fix commands other than write_data

When CHECK_BITS_SET was added, they forgot to add
a new command table, and instead overwrote the
previous table.

Signed-off-by: Troy Kisky <troy.kisky@boundarydevices.com>
Tested-by: Fabio Estevam <fabio.estevam@freescale.com>
9 years agoimximage: header.length of 4 is valid
Troy Kisky [Mon, 21 Sep 2015 21:02:48 +0000 (14:02 -0700)]
imximage: header.length of 4 is valid

Signed-off-by: Troy Kisky <troy.kisky@boundarydevices.com>
9 years agosandbox: Correct operaion of 'reset' command
Simon Glass [Sat, 3 Oct 2015 17:21:16 +0000 (11:21 -0600)]
sandbox: Correct operaion of 'reset' command

Currently 'reset' only works with the test device tree. When run without a
device tree, or with the normal device tree, the following error is
displayed:

   Reset not supported on this platform

Fix the driver and the standard device tree to avoid this.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reported-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
9 years agodm: test: Show the amount of leaked memory on error
Simon Glass [Sat, 12 Sep 2015 14:45:20 +0000 (08:45 -0600)]
dm: test: Show the amount of leaked memory on error

Adjust the memory leak tests to show the amount of memory leaked. This can
be a useful signal as to what is wrong.

Signed-off-by: Simon Glass <sjg@chromium.org>