oweals/u-boot.git
10 years agoARM: zynq: Fix sparse warning in ddrc.c
Michal Simek [Fri, 25 Apr 2014 12:19:00 +0000 (14:19 +0200)]
ARM: zynq: Fix sparse warning in ddrc.c

Warning:
arch/arm/cpu/armv7/zynq/ddrc.c:43:24: warning: Using plain integer as NULL pointer

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoARM: zynq: Fix sparse warnings in slcr.c
Michal Simek [Fri, 25 Apr 2014 11:48:08 +0000 (13:48 +0200)]
ARM: zynq: Fix sparse warnings in slcr.c

Warnings:
arch/arm/cpu/armv7/zynq/slcr.c:21:6: warning: symbol 'zynq_slcr_lock' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:27:6: warning: symbol 'zynq_slcr_unlock' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:34:6: warning: symbol 'zynq_slcr_cpu_reset' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:54:6: warning: symbol 'zynq_slcr_gem_clk_setup' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:81:6: warning: symbol 'zynq_slcr_devcfg_disable' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:94:6: warning: symbol 'zynq_slcr_devcfg_enable' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:107:5: warning: symbol 'zynq_slcr_get_boot_mode' was not declared. Should it be static?
arch/arm/cpu/armv7/zynq/slcr.c:113:5: warning: symbol 'zynq_slcr_get_idcode' was not declared. Should it be static?

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoMerge branch 'fpga' of git://www.denx.de/git/u-boot-microblaze
Tom Rini [Tue, 13 May 2014 11:34:08 +0000 (07:34 -0400)]
Merge branch 'fpga' of git://denx.de/git/u-boot-microblaze

10 years agoMerge branch 'master' of git://git.denx.de/u-boot-video
Tom Rini [Tue, 13 May 2014 11:31:00 +0000 (07:31 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-video

10 years agofpga: zynq: Use helper function zynq_validate_bitstream
Siva Durga Prasad Paladugu [Thu, 13 Mar 2014 06:27:34 +0000 (11:57 +0530)]
fpga: zynq: Use helper function zynq_validate_bitstream

Use helper function zynq_validate_bitstream so that the
code can be reused easily for different cases of dma transfer.

Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: zynq: Use helper functions for zynq dma
Siva Durga Prasad Paladugu [Wed, 12 Mar 2014 11:39:26 +0000 (17:09 +0530)]
fpga: zynq: Use helper functions for zynq dma

Use zynq_dma_xfer_init, zynq_align_dma_buffer,
zynq_dma_transfer helper function performing dma
transfers so that the code can be reused easily for
different cases of dma transfer.

Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: zynq: Remove sparse warnings
Michal Simek [Fri, 25 Apr 2014 11:51:58 +0000 (13:51 +0200)]
fpga: zynq: Remove sparse warnings

Warnings:
drivers/fpga/zynqpl.c:150:32: warning: Using plain integer as NULL pointer
drivers/fpga/zynqpl.c:152:16: warning: Using plain integer as NULL pointer

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: xilinx: Simplify load/dump/info function handling
Michal Simek [Thu, 13 Mar 2014 12:07:57 +0000 (13:07 +0100)]
fpga: xilinx: Simplify load/dump/info function handling

Connect FPGA version with appropriate operations
to remove huge switch-cases for every FPGA family.
Tested on Zynq. Spartan2/Spartan3/Virtex2 just compile test.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: xilinx: Fix the rest of CamelCases
Michal Simek [Thu, 13 Mar 2014 11:58:20 +0000 (12:58 +0100)]
fpga: xilinx: Fix the rest of CamelCases

No functional changes.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: xilinx: Avoid CamelCase for in Xilinx_desc
Michal Simek [Thu, 13 Mar 2014 11:49:21 +0000 (12:49 +0100)]
fpga: xilinx: Avoid CamelCase for in Xilinx_desc

No functional changes.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: virtex2: Avoid CamelCase
Michal Simek [Thu, 13 Mar 2014 10:33:36 +0000 (11:33 +0100)]
fpga: virtex2: Avoid CamelCase

No functional changes.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: spartan3: Avoid CamelCase
Michal Simek [Thu, 13 Mar 2014 10:28:42 +0000 (11:28 +0100)]
fpga: spartan3: Avoid CamelCase

No functional changes.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agofpga: spartan2: Avoid CamelCase
Michal Simek [Thu, 13 Mar 2014 10:23:43 +0000 (11:23 +0100)]
fpga: spartan2: Avoid CamelCase

No functional changes.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agokbuild: Fix trailing whitespaces
Michal Simek [Thu, 24 Apr 2014 13:24:28 +0000 (15:24 +0200)]
kbuild: Fix trailing whitespaces

Trivial fix.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agonet/designware: call phy_connect_dev() to properly setup phylib device
Ian Campbell [Mon, 28 Apr 2014 19:14:05 +0000 (20:14 +0100)]
net/designware: call phy_connect_dev() to properly setup phylib device

This sets up the linkage from the phydev back to the ethernet device. This
symptom of not doing this which I noticed was:
    <NULL> Waiting for PHY auto negotiation to complete....
rather than:
    dwmac.1c50000 Waiting for PHY auto negotiation to complete....

Signed-off-by: Ian Campbell <ijc@hellion.org.uk>
Cc: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
10 years agoboards.cfg: Keep entries sorted
Fabio Estevam [Mon, 12 May 2014 16:01:51 +0000 (13:01 -0300)]
boards.cfg: Keep entries sorted

Run "tools/reformat.py -i -d '-' -s 8 <boards.cfg >boards0.cfg && mv boards0
in order to keep the entries sorted.

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
10 years agofs/fat: correct FAT16/12 file finding in root dir
Wu, Josh [Thu, 8 May 2014 08:14:07 +0000 (16:14 +0800)]
fs/fat: correct FAT16/12 file finding in root dir

When write a file into FAT file system, it will search a match file in
root dir. So the find_directory_entry() will get the first cluster of
root dir content and search the directory item one by one. If the file
is not found, we will call get_fatent_value() to get next cluster of root
dir via lookup the FAT table and continue the search.

The issue is in FAT16/12 system, we cannot get root dir's next clust
from FAT table. The FAT table only be use to find the clust of data
aera in FAT16/12.

In FAT16/12 if the clust is in root dir, the clust number is a negative
number or 0, 1. Since root dir is located in front of the data area.
Data area start clust #2. So the root dir clust number should < 2.

This patch will check above situation before call get_fatenv_value().
If curclust is < 2, include minus number, we just increase one on the
curclust since root dir is in continous cluster.

The patch also add a sanity check for entry in get_fatenv_value().

Signed-off-by: Josh Wu <josh.wu@atmel.com>
10 years agofs: fat_write: fix the incorrect last cluster checking
Wu, Josh [Thu, 8 May 2014 08:14:06 +0000 (16:14 +0800)]
fs: fat_write: fix the incorrect last cluster checking

In fat_write.c, the last clust condition check is incorrect:

  if ((curclust >= 0xffffff8) || (curclust >= 0xfff8)) {
   ... ...
  }

For example, in FAT32 if curclust is 0x11000. It is a valid clust.
But on above condition check, it will be think as a last clust.

So the correct last clust check should be:
  in fat32, curclust >= 0xffffff8
  in fat16, curclust >= 0xfff8
  in fat12, curclust >= 0xff8

This patch correct the last clust check.

Signed-off-by: Josh Wu <josh.wu@atmel.com>
10 years agofs/fat: add fat12 cluster check
Wu, Josh [Thu, 8 May 2014 08:14:05 +0000 (16:14 +0800)]
fs/fat: add fat12 cluster check

Signed-off-by: Josh Wu <josh.wu@atmel.com>
10 years agofs:ext4:write:fix: Reinitialize global variables after updating a file
Łukasz Majewski [Tue, 6 May 2014 07:36:05 +0000 (09:36 +0200)]
fs:ext4:write:fix: Reinitialize global variables after updating a file

This bug shows up when file stored on the ext4 file system is updated.

The ext4fs_delete_file() is responsible for deleting file's (e.g. uImage)
data.
However some global data (especially ext4fs_indir2_block), which is used
during file deletion are left unchanged.

The ext4fs_indir2_block pointer stores reference to old ext4 double
indirect allocated blocks. When it is unchanged, after file deletion,
ext4fs_write_file() uses the same pointer (since it is already initialized
- i.e. not NULL) to return number of blocks to write. This trunks larger
file when previous one was smaller.

Lets consider following scenario:

1. Flash target with ext4 formatted boot.img (which has uImage [*] on itself)
2. Developer wants to upload their custom uImage [**]
- When new uImage [**] is smaller than the [*] - everything works
correctly - we are able to store the whole smaller file with corrupted
ext4fs_indir2_block pointer
- When new uImage [**] is larger than the [*] - theCRC is corrupted,
since truncation on data stored at eMMC was done.
3. When uImage CRC error appears, then reboot and LTHOR/DFU reflashing causes
proper setting of ext4fs_indir2_block() and after that uImage[**]
is successfully stored (correct uImage [*] metadata is stored at an
eMMC on the first flashing).

Due to above the bug was very difficult to reproduce.
This patch sets default values for all ext4fs_indir* pointers/variables.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
10 years agofs:ext4:cleanup: Remove superfluous code
Łukasz Majewski [Tue, 6 May 2014 07:36:04 +0000 (09:36 +0200)]
fs:ext4:cleanup: Remove superfluous code

Code responsible for handling situation when ext4 has block size of 1024B
can be ordered to take less space.

This patch does that for ext4 common and write files.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
10 years agodisk: part_efi: add support for the Backup GPT
Steve Rae [Mon, 5 May 2014 20:00:08 +0000 (13:00 -0700)]
disk: part_efi: add support for the Backup GPT

Check the Backup GPT table if the Primary GPT table is invalid.
Renamed "Secondary GPT" to "Backup GPT" as per:
  UEFI Specification (Version 2.3.1, Errata A)

Signed-off-by: Steve Rae <srae@broadcom.com>
10 years agocommon/board_f: Fix size variable
York Sun [Sat, 3 May 2014 00:28:05 +0000 (17:28 -0700)]
common/board_f: Fix size variable

DRAM size should use 64-bit variable when the size could be more than 4GB.
Caught and verified on P4080DS with 4GB DDR.

Signed-off-by: York Sun <yorksun@freescale.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agocommon/board_f: Initialized global data for generic board
York Sun [Sat, 3 May 2014 00:28:04 +0000 (17:28 -0700)]
common/board_f: Initialized global data for generic board

Some platforms (tested on mpc85xx, mpc86xx) use global data before calling
function baord_inti_f(). The data should not be cleared later. Any arch
which uses global data in generic board board_init_f() should define
CONFIG_SYS_GENERIC_GLOBAL_DATA.

Signed-off-by: York Sun <yorksun@freescale.com>
CC: Scott Wood <scottwood@freescale.com>
CC: Simon Glass <sjg@chromium.org>
CC: Albert ARIBAUD <albert.u.boot@aribaud.net>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agoTrivial fix to .gitignore for spl/Makefile
Ralph Siemsen [Thu, 1 May 2014 18:18:41 +0000 (14:18 -0400)]
Trivial fix to .gitignore for spl/Makefile

Trivial fix to .gitignore for spl/Makefile

According to the gitignore man page:

"An optional prefix "!" which negates the pattern; any matching file
excluded by a previous pattern will become included again."
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
So the directory exclude "/spl/*" must come before the exception
for spl/Makefile otherwise it has no effect.

Signed-off-by: Ralph Siemsen <ralphs@netwinder.org>
Tested-by: Masahiro Yamada <yamada.m@jp.panasonic.com> [on git v1.7.9.5 / v1.8.3.2]
10 years agocmd_bootm.c: Only say XIP image when load is image_start
Tom Rini [Thu, 1 May 2014 14:01:08 +0000 (10:01 -0400)]
cmd_bootm.c: Only say XIP image when load is image_start

We say we have an XIP (in this case, image loaded at desired execution
address) when the image header has been offset in the load.  It's
possible that in some cases executing the header is non-fatal but that's
not true in many other cases.

Signed-off-by: Tom Rini <trini@ti.com>
10 years agotools: env: Add aes.c placeholder
Marek Vasut [Mon, 28 Apr 2014 09:56:30 +0000 (11:56 +0200)]
tools: env: Add aes.c placeholder

Add missing aes.c placeholder which includes lib/aes.c . Without this
one, tools/env/ will fail to build.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
Cc: Masahiro Yamada <yamada.m@jp.panasonic.com>
Tested-by: Alexey Brodkin <abrodkin@synopsys.com>
Tested-by: Heiko Schocher <hs@denx.de>
10 years agonand_spl: remove unused linker scripts u-boot-nand.lds
Masahiro Yamada [Mon, 28 Apr 2014 08:45:05 +0000 (17:45 +0900)]
nand_spl: remove unused linker scripts u-boot-nand.lds

Commit 345b77ba removed some nand_spl boards but
it missed to delete linker scripts.

These linker scripts are not used now.

And one more fix:
amcc/acadia does not support nand_spl anymore, so remove
  #if !defined(CONFIG_NAND_U_BOOT) ... #endif

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Stefan Roese <sr@denx.de>
Acked-by: Stefan Roese <sr@denx.de>
10 years agoconfig: remove platform CONFIG_SYS_HZ definition part 4
Masahiro Yamada [Mon, 28 Apr 2014 01:18:34 +0000 (10:18 +0900)]
config: remove platform CONFIG_SYS_HZ definition part 4

Some new boards define CONFIG_SYS_HZ again! Remove.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Bo Shen <voice.shen@atmel.com>
10 years agocosmetic: delete misleading comment /* CONFIG_BOARDDIR */
Masahiro Yamada [Mon, 28 Apr 2014 01:17:10 +0000 (10:17 +0900)]
cosmetic: delete misleading comment /* CONFIG_BOARDDIR */

CONFIG_BOARDDIR is not referenced in these linker scripts.
The comment /* CONFIG_BOARDDIR */ is misleading.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
10 years agohush shell: Avoid string write overflow when entering max cmd length
Kristian Otnes [Fri, 25 Apr 2014 13:35:43 +0000 (15:35 +0200)]
hush shell: Avoid string write overflow when entering max cmd length

console_buffer array is defined to be CONFIG_SYS_CBSIZE + 1 long,
whereas the_command array only CONFIG_SYS_CBSIZE long. Subsequent
use of strcpy(the_command, console_buffer) will write final \0
terminating byte outside the_command array when entering a command
of max length.

Signed-off-by: Kristian Otnes <kotnes <at> cisco <dot> com>
10 years agokbuild: allow null board for spl build
Masahiro Yamada [Fri, 25 Apr 2014 12:54:31 +0000 (21:54 +0900)]
kbuild: allow null board for spl build

Commit 33a02da0 supported "<none>" for the board field of boards.cfg.
But it missed to modify spl/Makefile.

This commit provides the flexibility so we can use "<none>" board
in SPL too.

10 years agotps6586x: staticize funtions
Manish Badarkhe [Thu, 24 Apr 2014 03:25:07 +0000 (08:55 +0530)]
tps6586x: staticize funtions

Make funtions static which are locally used in file.

Signed-off-by: Manish Badarkhe <badarkhe.manish@gmail.com>
10 years agodoc: README.generic-board: Fix typo
Fabio Estevam [Tue, 22 Apr 2014 18:45:42 +0000 (15:45 -0300)]
doc: README.generic-board: Fix typo

Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agolib/sha256: fix compile on some hosts
Andreas Bießmann [Sun, 20 Apr 2014 08:34:15 +0000 (10:34 +0200)]
lib/sha256: fix compile on some hosts

Commit 2842c1c24269a05142802d25520e7cb9035e456c introduced lib/sha256 into
mkimage. Since then it will be compiled with HOSTCC which may produce errors
on some systems. Most BSD systems (like OS X for me) do not ship a
linux/string.h which will lead to take the U-Boot provided
include/linux/string.h in the end. This header howver is completely wrong
here. Just take the string.h if compiling with HOSTCC and linux/string.h when
not.

Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
Cc: Heiko Schocher <hs@denx.de>
Acked-by: Heiko Schocher<hs@denx.de>
10 years agokbuild: build with -Werror=date-time if the compiler supports it
Masahiro Yamada [Fri, 18 Apr 2014 10:09:51 +0000 (19:09 +0900)]
kbuild: build with -Werror=date-time if the compiler supports it

Using __DATE__, __TIME__ would make the build non-deterministic.

If the code needs to refer to build date/time, use U_BOOT_DATE and
U_BOOT_TIME in include/generated/timestamp_autogenerated.h instead.

This commit has been imported from Linux Kernel,
which should be applied to U-Boot too:

    commit fe7c36c7bde12190341722af69358e42171162f3
    Author: Josh Triplett <josh@joshtriplett.org>
    Date:   Mon Dec 23 13:56:06 2013 -0800

    Makefile: Build with -Werror=date-time if the compiler supports it

    GCC 4.9 and newer have a new warning -Wdate-time, which warns on any use
    of __DATE__, __TIME__, or __TIMESTAMP__, which would make the build
    non-deterministic.  Now that the kernel does not use any of those
    macros, turn on -Werror=date-time if available, to keep it that way.

    The kernel already (optionally) records this information at build time
    in a single place; other kernel code should not duplicate that.

Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
10 years agofs: ubifs: drop __DATE__ and __TIME__
Masahiro Yamada [Fri, 18 Apr 2014 10:09:50 +0000 (19:09 +0900)]
fs: ubifs: drop __DATE__ and __TIME__

__DATE__ and __TIME__ makes the build non-deterministic.
Drop the debug message using them.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
10 years agoblackfin: replace bfin_gen_rand_mac() with eth_random_addr()
Masahiro Yamada [Fri, 18 Apr 2014 10:09:49 +0000 (19:09 +0900)]
blackfin: replace bfin_gen_rand_mac() with eth_random_addr()

bfin_gen_rand_mac() uses __DATE__ as the seed for random ethernet
address. This makes the build non-deterministic.

In the first place, it should not be implemented as a Bfin-specific
function. Use eth_random_addr() instead.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Sonic Zhang <sonic.zhang@analog.com>
10 years agonet: rename and refactor eth_rand_ethaddr() function
Masahiro Yamada [Fri, 18 Apr 2014 10:09:48 +0000 (19:09 +0900)]
net: rename and refactor eth_rand_ethaddr() function

Some functions in include/net.h are ported from
include/linux/etherdevice.h of Linux Kernel.

For ex.
  is_zero_ether_addr()
  is_multicast_ether_addr()
  is_broadcast_ether_addr()
  is_valid_ether_addr();

So, we should use the same function name as that of Linux Kernel,
eth_rand_addr(), for consistency.

Besides, eth_rand_addr() has been implemented as an inline function.
So it should not be surrounded by #ifdef CONFIG_RANDOM_MACADDR.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
10 years agorand: do not surround function declarations by #ifdef
Masahiro Yamada [Fri, 18 Apr 2014 10:09:47 +0000 (19:09 +0900)]
rand: do not surround function declarations by #ifdef

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
10 years agocmd_time: do not show ticks
Masahiro Yamada [Fri, 18 Apr 2014 08:46:13 +0000 (17:46 +0900)]
cmd_time: do not show ticks

The command "time" shows the execution time of the command given
to the argument, like this:

    time: 45.293 seconds, 45293 ticks

Since we adopted CONFIG_SYS_HZ = 1000 for all boards,
we always have a simple formula: "1 tick = 0.0001 second".

Showing ticks looks almost redundant.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
10 years agonet/phy: enable get_phy_id redefinable
Shengzhou Liu [Fri, 11 Apr 2014 08:14:17 +0000 (16:14 +0800)]
net/phy: enable get_phy_id redefinable

As some PHYs have non-standard PHY ID registers, PHY Id can't
be read correctly by current get_phy_id function, so we enable
get_phy_id redefinable to permit specific PHY driver having own
specific get_phy_id function.

Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com>
10 years agoAR8035/phy: Enable autonegotiation function for ar8035
Xiaobo Xie [Fri, 11 Apr 2014 08:03:11 +0000 (16:03 +0800)]
AR8035/phy: Enable autonegotiation function for ar8035

Function "genphy_parse_link()" used "if (mii_reg & BMSR_ANEGCAPABLE)" before,
but used "if (phydev->supported & SUPPORTED_Autoneg)" now.
So assign "phydev->supported" to "phydev->drv->features" for ar8035
to enable autonegotiation. Then removed the genphy_config_aneg() function.

Signed-off-by: Xie Xiaobo <X.Xie@freescale.com>
10 years agoserial: nsl16550: add hw flow control support
Karicheri, Muralidharan [Wed, 9 Apr 2014 19:38:46 +0000 (15:38 -0400)]
serial: nsl16550: add hw flow control support

keystone serial hw support hw flow control. This patch
enables hw flow control for keystone EVMs as an optional
feature based on CONFIG_SERIAL_HW_FLOW_CONTROL.

Signed-off-by: Murali Karicheri <m-karicheri2@ti.com>
10 years agofs: fat: Fix cache align error message in fatwrite
Nobuhiro Iwamatsu [Tue, 8 Apr 2014 02:12:46 +0000 (11:12 +0900)]
fs: fat: Fix cache align error message in fatwrite

Use of malloc of do_fat_write() causes cache error on ARM v7 platforms.
Perhaps, the same problem will occur at any other CPUs.
This replaces malloc with memalign to fix cache buffer alignment.

Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
Signed-off-by: Yoshiyuki Ito <yoshiyuki.ito.ub@renesas.com>
Tested-by: Hector Palacios <hector.palacios@digi.com>
10 years agobd_info: remove bi_barudrate member from struct bd_info
Masahiro Yamada [Fri, 4 Apr 2014 11:09:58 +0000 (20:09 +0900)]
bd_info: remove bi_barudrate member from struct bd_info

gd->bd->bi_baudrate is a copy of gd->baudrate.

Since baudrate is a common feature for all architectures,
keep gd->baudrate only.

It is true that bi_baudrate was passed to the kernel in that structure
but it was a long time ago.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Tom Rini <trini@ti.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Heiko Schocher <hs@denx.de>
Acked-by: Michal Simek <monstr@monstr.eu> (For microblaze)
10 years agoPrepare v2014.07-rc1 v2014.07-rc1
Tom Rini [Mon, 12 May 2014 12:54:13 +0000 (08:54 -0400)]
Prepare v2014.07-rc1

Signed-off-by: Tom Rini <trini@ti.com>
10 years agoMerge branch 'tom' of git://git.denx.de/u-boot-x86
Tom Rini [Fri, 9 May 2014 22:48:26 +0000 (18:48 -0400)]
Merge branch 'tom' of git://git.denx.de/u-boot-x86

10 years agosandbox: ignore sandbox.dtb
Masahiro Yamada [Mon, 21 Apr 2014 09:50:41 +0000 (18:50 +0900)]
sandbox: ignore sandbox.dtb

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Simon Glass <sjg@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
10 years agosandbox: move source files from board/ to arch/sandbox/
Masahiro Yamada [Mon, 21 Apr 2014 09:39:35 +0000 (18:39 +0900)]
sandbox: move source files from board/ to arch/sandbox/

Prior to commit 33a02da0, all boards must have board/${BOARD}/
or board/${VENDOR}/${BOARD}/ directory.
Now this rule is obsolete.

It looks weird that sandbox defines "vendor" and "board" just for
meeting the old U-Boot directory structure.

Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Simon Glass <sjg@chromium.org>
10 years agopatman: Avoid duplicate sign-offs
Simon Glass [Sun, 20 Apr 2014 16:50:14 +0000 (10:50 -0600)]
patman: Avoid duplicate sign-offs

Keep track of all Signed-off-by tags in a commit and silently suppress any
duplicates.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agopatman: Deal with 'git apply' failures correctly
Simon Glass [Sun, 20 Apr 2014 16:50:13 +0000 (10:50 -0600)]
patman: Deal with 'git apply' failures correctly

This sort of failure is rare, but the code to deal with it is wrong.
Fix it.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agosandbox: Update and expand the README
Simon Glass [Sat, 22 Mar 2014 23:12:59 +0000 (17:12 -0600)]
sandbox: Update and expand the README

Now that sandbox has a good base of features, the README is quite out of
date. Update it, and document the new features.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agosandbox: Provide a build option to avoid using SDL
Simon Glass [Sat, 22 Mar 2014 23:12:58 +0000 (17:12 -0600)]
sandbox: Provide a build option to avoid using SDL

Some machines do not have SDL libraries installed, and it is still useful
to build sandbox without LCD/keyboard support.

Add an option for this, used as follows:

    make sandbox_config all NO_SDL=1

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agoMake 'run' use run_command_list() instead of run_command()
Simon Glass [Sat, 26 Oct 2013 05:01:32 +0000 (23:01 -0600)]
Make 'run' use run_command_list() instead of run_command()

In the case where an environment variable spans multiple lines, we should
use run_command_list() so that all lines are executed. This shold be
backwards compatible with existing behaviour for existing scripts.

Signed-off-by: Simon Glass <sjg@chromium.org>
10 years agosandbox: Support 'env import' and 'env export'
Simon Glass [Sat, 26 Oct 2013 05:01:31 +0000 (23:01 -0600)]
sandbox: Support 'env import' and 'env export'

Adjust the code for these commands so that they work on sandbox.

Signed-off-by: Simon Glass <sjg@chromium.org>
(Adjusted to fix minor merge comflict, when applied)

Change-Id: I987dee6194cd5c83f82604caf894fc85e4eb71a8

10 years agoMerge branch 'master' of git://git.denx.de/u-boot-arm
Tom Rini [Fri, 9 May 2014 15:18:45 +0000 (11:18 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-arm

10 years agoMerge branch 'master' of git://git.denx.de/u-boot-sparc
Tom Rini [Fri, 9 May 2014 15:09:59 +0000 (11:09 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-sparc

10 years agoMerge branch 'u-boot/master'
Albert ARIBAUD [Fri, 9 May 2014 08:47:05 +0000 (10:47 +0200)]
Merge branch 'u-boot/master'

Conflicts:
drivers/net/Makefile

(trivial merge)

10 years agoleon: implement missing get_tbclk()
Daniel Hellstrom [Thu, 8 May 2014 16:52:37 +0000 (18:52 +0200)]
leon: implement missing get_tbclk()

Without this patch SPARC/LEON does not build.

Reported-by: Tom Rini <trini@ti.com>
Signed-off-by: Daniel Hellstrom <daniel@gaisler.com>
10 years agoleon: use CONFIG_SYS_HZ to config timer prescaler
Daniel Hellstrom [Thu, 8 May 2014 17:16:14 +0000 (19:16 +0200)]
leon: use CONFIG_SYS_HZ to config timer prescaler

Before it was hardcoded to 1000 ticks per second.

Signed-off-by: Daniel Hellstrom <daniel@gaisler.com>
10 years agoserial: zynq: Fix typo in suffix function name
Michal Simek [Fri, 25 Apr 2014 11:46:28 +0000 (13:46 +0200)]
serial: zynq: Fix typo in suffix function name

's/zynq_serial_initalize/zynq_serial_initialize/g'
serial_initialize is used by all serial drivers.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoserial: zynq: Remove sparse warnings
Michal Simek [Fri, 25 Apr 2014 11:45:08 +0000 (13:45 +0200)]
serial: zynq: Remove sparse warnings

Warnings:
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_init' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_setbrg' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_getc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_tstc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_putc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_puts' was not declared. Should it be static?
drivers/serial/serial_zynq.c:182:22: warning: symbol 'uart_zynq_serial0_device' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_init' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_setbrg' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_getc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_tstc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_putc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_puts' was not declared. Should it be static?
drivers/serial/serial_zynq.c:185:22: warning: symbol 'uart_zynq_serial1_device' was not declared. Should it be static?

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoi2c: zynq: Fixed compilation errors when using DEBUG
Jesper B. Christensen [Fri, 25 Apr 2014 13:46:17 +0000 (15:46 +0200)]
i2c: zynq: Fixed compilation errors when using DEBUG

Signed-off-by: Jesper B. Christensen <jesper.christensen@cobham.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoMerge branch 'serial' of git://www.denx.de/git/u-boot-microblaze
Tom Rini [Tue, 6 May 2014 18:55:45 +0000 (14:55 -0400)]
Merge branch 'serial' of git://denx.de/git/u-boot-microblaze

10 years agoMerge branch 'master' of git://www.denx.de/git/u-boot-microblaze
Tom Rini [Tue, 6 May 2014 18:55:27 +0000 (14:55 -0400)]
Merge branch 'master' of git://denx.de/git/u-boot-microblaze

10 years agoMerge branch 'net' of git://www.denx.de/git/u-boot-microblaze
Tom Rini [Tue, 6 May 2014 18:50:22 +0000 (14:50 -0400)]
Merge branch 'net' of git://denx.de/git/u-boot-microblaze

10 years agoMerge branch 'master' of git://git.denx.de/u-boot-usb
Tom Rini [Tue, 6 May 2014 18:45:51 +0000 (14:45 -0400)]
Merge branch 'master' of git://git.denx.de/u-boot-usb

10 years agoserial: zynq: Fix typo in suffix function name
Michal Simek [Fri, 25 Apr 2014 11:46:28 +0000 (13:46 +0200)]
serial: zynq: Fix typo in suffix function name

's/zynq_serial_initalize/zynq_serial_initialize/g'
serial_initialize is used by all serial drivers.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoserial: zynq: Remove sparse warnings
Michal Simek [Fri, 25 Apr 2014 11:45:08 +0000 (13:45 +0200)]
serial: zynq: Remove sparse warnings

Warnings:
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_init' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_setbrg' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_getc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_tstc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_putc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:181:1: warning: symbol 'uart_zynq0_puts' was not declared. Should it be static?
drivers/serial/serial_zynq.c:182:22: warning: symbol 'uart_zynq_serial0_device' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_init' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_setbrg' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_getc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_tstc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_putc' was not declared. Should it be static?
drivers/serial/serial_zynq.c:184:1: warning: symbol 'uart_zynq1_puts' was not declared. Should it be static?
drivers/serial/serial_zynq.c:185:22: warning: symbol 'uart_zynq_serial1_device' was not declared. Should it be static?

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agonet: zynq: Fix sparse warnings in gem
Michal Simek [Fri, 25 Apr 2014 12:17:38 +0000 (14:17 +0200)]
net: zynq: Fix sparse warnings in gem

Add missing header.

Warnings:
drivers/net/zynq_gem.c:491:5: warning: symbol 'zynq_gem_initialize' was not declared. Should it be static?
drivers/net/zynq_gem.c:542:5: warning: symbol 'zynq_gem_of_init' was not declared. Should it be static?

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agonet: zynq: Use predefined macros instead of hardcoded value
Michal Simek [Tue, 25 Feb 2014 09:25:38 +0000 (10:25 +0100)]
net: zynq: Use predefined macros instead of hardcoded value

MII is used by this driver.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agomicroblaze: Wire up OF support for emaclite
Stephan Linz [Wed, 4 Jul 2012 20:25:31 +0000 (22:25 +0200)]
microblaze: Wire up OF support for emaclite

 - expand the condition with CONFIG_OF_CONTROL

Signed-off-by: Stephan Linz <linz@li-pro.net>
Acked-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
10 years agoat91: video: atmel_hlcdfb.c: fix bad timing configuration
Wu, Josh [Mon, 10 Mar 2014 08:40:41 +0000 (16:40 +0800)]
at91: video: atmel_hlcdfb.c: fix bad timing configuration

The right correspondance between LCD margins and LCD timings is:
    * upper margin -> vertical back porch
    * lower margin -> vertical front porch
    * left  margin -> horizontal back porch
    * right margin -> horizontal front porch

Signed-off-by: Josh Wu <josh.wu@atmel.com>
10 years agoconfig: enable CMD_BMP when API+LCD is enabled
Stephen Warren [Fri, 18 Apr 2014 17:15:55 +0000 (11:15 -0600)]
config: enable CMD_BMP when API+LCD is enabled

When both CONFIG_API and CONFIG_LCD are enabled, the API code calls
lcd_display_bitmap(). That isn't compiled unless either CONFIG_CMD_BMP
or CONFIG_SPLASH_SCREEN is enabled. In order to prevent build problems,
have config_fallbacks.h enable CONFIG_CMD_BMP when both API and LCD are
enabled.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
10 years agologos: Update of the syteco company logo
Thomas Diener [Wed, 23 Apr 2014 05:52:45 +0000 (07:52 +0200)]
logos: Update of the syteco company logo

Signed-off-by: Thomas Diener <dietho@gmx.de>
Signed-off-by: Anatolij Gustschin <agust@denx.de>
10 years agocommon: fixed linker-list example
Mateusz Zalega [Tue, 29 Apr 2014 18:14:22 +0000 (20:14 +0200)]
common: fixed linker-list example

Last argument shouldn't be there.

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@ti.com>
10 years agoUSB: gadget: save driver name before registering it
Stephen Warren [Thu, 1 May 2014 21:45:16 +0000 (15:45 -0600)]
USB: gadget: save driver name before registering it

g_dnl_register() currently first attempts to register a composite
driver by name, and then saves the driver name once it's registered.
Internally to the registration code, g_dnl_do_config() is called and
attempts to compare the composite device's name with the list of known
device names. This fails since the composite device's name has not yet
been stored. This means that the first time "ums 0 0" is run, it fails,
but subsequent attempts succeed.

Re-order the name-saving and registration code to solve this.

Fixes: e5b834e07f51 ("USB: gadget: added a saner gadget downloader registration API")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agoUSB: gadget: added a saner gadget downloader registration API
Mateusz Zalega [Mon, 28 Apr 2014 19:13:28 +0000 (21:13 +0200)]
USB: gadget: added a saner gadget downloader registration API

Preprocessor definitions and hardcoded implementation selection in
g_dnl core were replaced by a linker list made of (usb_function_name,
bind_callback) pairs.

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Acked-by: Marek Vasut <marex@denx.de>
10 years agoam335x: dfu: disable DFU in am335x_evm SPL build
Mateusz Zalega [Mon, 28 Apr 2014 19:13:27 +0000 (21:13 +0200)]
am335x: dfu: disable DFU in am335x_evm SPL build

Future patches will make DFU too large to fit in this board's SPL build.

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Tom Rini <trini@ti.com>
Reviewed-by: Lukasz Majewski <l.majewski@samsung.com>
10 years agoums: always initialize mmc before ums_disk_init()
Mateusz Zalega [Mon, 28 Apr 2014 19:13:26 +0000 (21:13 +0200)]
ums: always initialize mmc before ums_disk_init()

In cases when MMC hadn't been initialized before, ie. by the user or other
subsystem, it was still uninitialized while UMS media capacity check,
leading to broken ums command.

UMS has to initialize resources it uses.

Tested on Samsung Goni.

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Tested-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Minkyu Kang <mk7.kang@samsung.com>
10 years agodfu: mmc: change offset base handling
Mateusz Zalega [Mon, 28 Apr 2014 19:13:25 +0000 (21:13 +0200)]
dfu: mmc: change offset base handling

Previously offsets handled by dfu_fill_entity_mmc(), defined in boards'
CONFIG_DFU_ALT were treated as hexadecimal regardless of their prefix,
which sometimes led to confusion. This patch forces usage of explicit
numerical base prefixes.

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Cc: Tom Rini <trini@ti.com>
Cc: Minkyu Kang <mk7.kang@samsung.com>
10 years agodfu: mmc: raw data write fix
Mateusz Zalega [Mon, 28 Apr 2014 19:13:24 +0000 (21:13 +0200)]
dfu: mmc: raw data write fix

When user attempted to perform a raw write using DFU (vide
dfu_fill_entity_mmc) with MMC interface not initialized before,
get_mmc_blk_size() reported invalid (zero) block size - it wasn't
possible to write ie. a new u-boot image.

This commit fixes that by initializing MMC device before use in
dfu_fill_entity_mmc().

While fixing initialization sequence, I had to change about half of
dfu_fill_entity_mmc's body, so I refactored it on the way to make it,
IMHO, considerably more comprehensible.

Being left as dead code, get_mmc_blk_size() was removed.

Tested on Samsung Goni.

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
Acked-by: Tom Rini <trini@ti.com>
Cc: Minkyu Kang <mk7.kang@samsung.com>
10 years agousb: dfu: fix boards wo USB cable detection
Mateusz Zalega [Wed, 30 Apr 2014 11:07:48 +0000 (13:07 +0200)]
usb: dfu: fix boards wo USB cable detection

Former usb_cable_connected() patch broke compilation of boards which do
not support this feature.

I've renamed usb_cable_connected() to g_dnl_usb_cable_connected() and added
its default implementation to gadget downloader driver code. There's
only one driver of this kind and it's unlikely there'll be another, so
there's no point in keeping it in /common.

Previously this function was declared in usb.h. I've moved it, since
it's more appropriate to keep it in g_dnl.h - usb.h seems to be intended
for USB host implementation.

Existing code, confronted with default -EOPNOTSUPP return value,
continues as if the cable was connected.

CONFIG_USB_CABLE_CHECK was removed.

Change-Id: Ib9198621adee2811b391c64512f14646cefd0369
Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>
10 years agopart: header fix
Mateusz Zalega [Mon, 28 Apr 2014 19:13:22 +0000 (21:13 +0200)]
part: header fix

Implementation made use of types defined in common.h, even though it
wasn't #included. It worked in circumstances when .c files included
every needed header (all).

Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Cc: Tom Rini <trini@ti.com>
Cc: Minkyu Kang <mk7.kang@samsung.com>
10 years agommc: mmc header fix
Mateusz Zalega [Wed, 30 Apr 2014 11:04:15 +0000 (13:04 +0200)]
mmc: mmc header fix

Structure definition used type block_dev_desc_t, defined in part.h, which
wasn't included in mmc.h. It worked only in circumstances when common.h,
or another header using part.h was incuded in implementation files.

Change-Id: I5b203928b689887e3e78beb00a378955e0553eb7
Signed-off-by: Mateusz Zalega <m.zalega@samsung.com>
Acked-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Minkyu Kang <mk7.kang@samsung.com>
10 years agoMerge branch 'next' of git://git.denx.de/u-boot-sh
Tom Rini [Fri, 2 May 2014 15:48:07 +0000 (11:48 -0400)]
Merge branch 'next' of git://git.denx.de/u-boot-sh

10 years agoARM: highbank: use default prompt
Rob Herring [Thu, 10 Apr 2014 21:17:31 +0000 (16:17 -0500)]
ARM: highbank: use default prompt

Since highbank is actually shared between Highbank and Midway platforms,
remove the Highbank name from the prompt and use the default.

Signed-off-by: Rob Herring <robh@kernel.org>
10 years agoARM: highbank: use config_distro_defaults.h
Rob Herring [Thu, 10 Apr 2014 21:17:30 +0000 (16:17 -0500)]
ARM: highbank: use config_distro_defaults.h

Adapt highbank to use config_distro_defaults.h and remove the redundant
defines.

Signed-off-by: Rob Herring <robh@kernel.org>
10 years agousb: gadget: allow ci_udc to build with new gadget framework
Stephen Warren [Mon, 28 Apr 2014 21:42:00 +0000 (15:42 -0600)]
usb: gadget: allow ci_udc to build with new gadget framework

Allow ci_udc.o to be built when using the new(?) USB gadget framework,
as enabled by CONFIG_USB_GADGET.

Note that this duplicates the Makefile entry for ci_udc.o, since it's
also included inside #ifdef CONFIG_USB_ETHER. I'm not sure what that
define means; perhaps an old style of Ethernet-specific USB gadget
implementation?

I wonder if the line that this patch adds shouldn't be outside all of
the ifdefs, so it stands on its own, similar to how e.g. epautoconf.o
is shared between the two?

Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agovideo: mxc_ipuv3_fb: stash frame buffer pointer in global data.
Eric Nelson [Tue, 29 Apr 2014 21:37:56 +0000 (14:37 -0700)]
video: mxc_ipuv3_fb: stash frame buffer pointer in global data.

This patch updates the i.MX video driver to store the
frame-buffer address in the fb_base field of the global
data structure *gd.

By doing this, you can find the frame buffer address
using the 'bdinfo' command:

U-Boot > bdinfo
arch_number = 0x00000EB9
...
FB base     = 0x4F35F1C0

This is very useful when debugging display connections.

Signed-off-by: Eric Nelson <eric.nelson@boundarydevices.com>
Acked-by: Otavio Salvador <otavio@ossystems.com.br>
Acked-by: Marek Vasut <marex@denx.de>
10 years agousb: ums: use only 1 buffer for CI_UDC
Stephen Warren [Thu, 24 Apr 2014 23:52:40 +0000 (17:52 -0600)]
usb: ums: use only 1 buffer for CI_UDC

ci_udc.c allocates only a single buffer for each endpoint, which
ci_ep_alloc_request() returns as a hard-coded value rather than
dynamically allocating. Consequently, storage_common.c must limit
itself to using a single buffer at a time. Add a special case
to the definition of FSG_NUM_BUFFERS for this.

Another option would be to fix ci_ep_alloc_request() to dynamically
allocate the buffers like some/all(?) other device mode drivers do.
However, I don't think that ci_ep_queue() supports queueing up
multiple buffers either yet, and I'm not familiar enough with the
controller yet to implement that. As such, any attempt to use multiple
buffers simply results in data corruption and other errors.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agousb: ci_udc: support variants with hostpc register
Stephen Warren [Thu, 24 Apr 2014 23:52:39 +0000 (17:52 -0600)]
usb: ci_udc: support variants with hostpc register

Tegra's USB controller appears to be a variant of the ChipIdea
controller; perhaps derived from it, or simply a different version of
the IP core to what U-Boot supports today.

In this variant, at least the following difference are present:
- Some registers are moved about.
- Setup transaction completion is reported in a separate 'epsetupstat'
  register, rather than in 'epstat' (which still exists, perhaps for
  other transaction types).
- USB connection speed is reported in a separate 'hostpc1_devlc'
  register, rather than 'portsc'.
- The registers used by ci_udc.c begin at offset 0x130 from the USB
  register base, rather than offset 0x140. However, this is handled
  by the associated EHCI controller driver, since the register address
  is stored in controller.ctrl->hcor.

Introduce define CONFIG_CI_UDC_HAS_HOSTPC to indicate which variant of
the controller should be supported. The "HAS_HOSTPC" part of this name
mirrors the similar "has_hostpc" field used by the Linux EHCI controller
core to represent the presence/absence of the hostpc1_devlc register.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agousb: ci_udc: make PHY initialization conditional
Stephen Warren [Thu, 24 Apr 2014 23:52:38 +0000 (17:52 -0600)]
usb: ci_udc: make PHY initialization conditional

usb_gadget_register_driver() currently unconditionally programs PORTSC
to select a ULPI PHY. This is incorrect on at least the Tegra boards I
am testing with, which use a UTMI PHY for the OTG ports. Make the PHY
selection code conditional upon the specific EHCI controller that is in
use.

Ideally, I believe that the PHY initialization code should be part of
ehci_hcd_init() in the relevant EHCI controller driver, or some board-
specific function that ehci_hcd_init() calls.

For MX6, I'm not sure this PHY initialization code is correct even before
this patch, since ehci-mx6's ehci_hcd_init() already configures PORTSC to
a board-specific value, and it seems likely that the code in ci_udc.c is
incorrectly undoing this. Perhaps this is not an issue if the PHY
selection register bits aren't implemented on this instance of the MX6
USB controller?

ehci-mxs.c doens't appear to touch PORTSC, so this code is likely still
required there.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agousb: ci_udc: set ep->req.actual after transfer
Stephen Warren [Thu, 24 Apr 2014 23:52:37 +0000 (17:52 -0600)]
usb: ci_udc: set ep->req.actual after transfer

At least drivers/usb/gadget/storage_common.c expects that ep->req.actual
contain the number of bytes actually transferred. (At least in practice,
I observed it failing to work correctly unless this was the case).

However, ci_udc.c modifies ep->req.length instead. I assume that .length
 is supposed to represent the allocated buffer size, whereas .actual is
supposed to represent the actual number of bytes transferred. In the OUT
transaction case, this may happen simply because the host sends a smaller
 packet than the max possible size, which is quite legal. In the IN case,
transferring fewer bytes than requested could presumably happen as an
error.

Modify handle_ep_complete() to write to .actual rather than modifying
.length.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agousb: ci_udc: Support larger packets
Stephen Warren [Thu, 24 Apr 2014 23:52:36 +0000 (17:52 -0600)]
usb: ci_udc: Support larger packets

ci_ep_queue() currently only fills in the page0/page1 fields in the
queue item. If the buffer is larger than 4KiB (unaligned) or 8KiB
(page-aligned), then this prevents the HW from knowing where to write
the balance of the data.

Fix this by initializing all 5 pageN pointers, which allows up to
16KiB (potentially non-page-aligned) buffers.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
10 years agodfu:fix: Replace wrong return value with proper one
Lukasz Majewski [Thu, 24 Apr 2014 08:24:53 +0000 (10:24 +0200)]
dfu:fix: Replace wrong return value with proper one

This patch remove always false (since we tested ret = 0) ternary operator
with ret value returned.

Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
10 years agoexynos: usb: Fix data abort on boards w/o vbus-gpio node in the DT
andrey.konovalov@linaro.org [Tue, 22 Apr 2014 17:23:49 +0000 (21:23 +0400)]
exynos: usb: Fix data abort on boards w/o vbus-gpio node in the DT

Commit 4a271cb1b4ff doesn't take into account that fdtdec_setup_gpio()
returns success when the gpio passed to it is FDT_GPIO_NONE (no
gpio node found in the fdtdec_decode_gpio() call). This results in
calling gpio_direction_output() on invalid gpio. For this reason
executing "usb start" command on Arndale causes data abort in the
ehci-exynos driver.

Add the fdt_gpio_isvalid() check to fix that problem.

Signed-off-by: Andrey Konovalov <andrey.konovalov@linaro.org>
Cc: Julius Werner <jwerner@chromium.org>
Cc: Simon Glass <sjg@chromium.org>
Cc: Minkyu Kang <mk7.kang@samsung.com>
Cc: Marek Vasut <marex@denx.de>
10 years agousb: musb: fill in usb_gadget_unregister_driver
Rob Herring [Fri, 18 Apr 2014 13:54:30 +0000 (08:54 -0500)]
usb: musb: fill in usb_gadget_unregister_driver

Add missing missing disconnect and unbind calls to the musb gadget driver's
usb_gadget_unregister_driver function. Otherwise, any gadget drivers fail
to uninitialize and run a 2nd time.

Signed-off-by: Rob Herring <robh@kernel.org>
10 years agousb: handle NULL table in usb_gadget_get_string
Rob Herring [Fri, 18 Apr 2014 13:54:28 +0000 (08:54 -0500)]
usb: handle NULL table in usb_gadget_get_string

Allow a NULL table to be passed to usb_gadget_get_string for cases
when a string table may not be populated.

Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Tom Rini <trini@ti.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Lukasz Majewski <l.majewski@samsung.com>