Masahiro Yamada [Thu, 12 Jun 2014 03:26:15 +0000 (12:26 +0900)]
sandbox: change local_irq_save() to macro
local_irq_save() should be a macro, not a function
because local_irq_save() saves flag to the given argument.
GCC is silent about this issue, but Clang warns:
In file included from lib/asm-offsets.c:15:
In file included from include/common.h:20:
In file included from include/linux/bitops.h:110:
arch/sandbox/include/asm/bitops.h:59:17:
warning: variable 'flags' is uninitialized when used here
[-Wuninitialized]
local_irq_save(flags);
^~~~~
That change causes another warning:
In file included from include/linux/bitops.h:110:0,
from include/common.h:20,
from lib/asm-offsets.c:15:
arch/sandbox/include/asm/bitops.h: In function ‘test_and_set_bit’:
arch/sandbox/include/asm/bitops.h:56:16: warning: unused variable ‘flags’ [-Wunused-variable]
So, flags should be set to __always_unused.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com> Cc: Simon Glass <sjg@chromium.org> Cc: Jeroen Hofstee <jeroen@myspectrum.nl> Acked-by: Simon Glass <sjg@chromium.org> Acked-by: Jeroen Hofstee <jeroen@myspectrum.nl>
Stephen Warren [Wed, 11 Jun 2014 16:26:23 +0000 (10:26 -0600)]
sandbox: terminate os_dirent_ls() result list
Each node in the linked-list that os_dirent_ls() returns has its next
pointer set only when the next node is created. For the last node in the
list, there is no next node, so this never happens, and the next pointer
is never initialized. Explicitly initialize the next pointer so that it
isn't dangling. Without this, "sb ls" might crash.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Acked-by: Simon Glass <sjg@chromium.org>
Stephen Warren [Thu, 12 Jun 2014 16:28:32 +0000 (10:28 -0600)]
sandbox: restore ability to access host fs through standard commands
Commit 95fac6ab4589 "sandbox: Use os functions to read host device tree"
removed the ability for get_device_and_partition() to handle the "host"
device type, and redirect accesses to it to the host filesystem. This
broke some unit tests that use this feature. So, revert that change. The
code added back by this patch is slightly different to pacify checkpatch.
However, we're then left with "host" being both:
- A pseudo device that accesses the hosts real filesystem.
- An emulated block device, which accesses "sectors" inside a file stored
on the host.
In order to resolve this discrepancy, rename the pseudo device from host
to hostfs, and adjust the unit-tests for this change.
The "help sb" output is modified to reflect this rename, and state where
the host and hostfs devices should be used.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Tested-by: Josh Wu <josh.wu@atmel.com> Acked-by: Simon Glass <sjg@chromium.org> Tested-by: Simon Glass <sjg@chromium.org>
Vasili Galka [Tue, 10 Jun 2014 13:14:56 +0000 (16:14 +0300)]
x86: Enable 32-bit build using x86_64 multilib toolchain
Until now building the x86 arch boards required 32-bit toolchain. As
many x86_64 toolchains come with 32-bit support (multilib) that's a
good idea to enable build with such toolchains.
The change required was to specify the usage of 32-bit explicitly to
the compiler and the linker (-m32 and -m elf_i386 flags) and locate
the right libgcc path.
Signed-off-by: Vasili Galka <vvv444@gmail.com> Acked-by: Simon Glass <sjg@chromium.org>
Heiko Schocher [Sun, 22 Jun 2014 04:33:29 +0000 (06:33 +0200)]
lib, fdt: move fdtdec_get_int() out of lib/fdtdec.c
move fdtdec_get_int() out of lib/fdtdec.c into lib/fdtdec_common.c
as this function is also used, if CONFIG_OF_CONTROL is not
used. Poped up on the ids8313 board using signed FIT images,
and activating CONFIG_SYS_GENERIC_BOARD. Without this patch
it shows on boot:
No valid FDT found - please append one to U-Boot binary, use u-boot-dtb.bin or define CONFIG_OF_EMBED. For sandbox, use -d <file.dtb>
With this patch, it boots again with CONFIG_SYS_GENERIC_BOARD
enabled.
Signed-off-by: Heiko Schocher <hs@denx.de> Acked-by: Simon Glass <sjg@chromium.org> Cc: Tom Rini <trini@ti.com>
Simon Glass [Thu, 12 Jun 2014 05:29:54 +0000 (23:29 -0600)]
dm: Tidy up four minor code nits
There is a spelling mistake and two functions are missing comments
altogether. Also the flags declaration is correct, but doesn't follow
style. Finally, the uclass_get_device() function has some errors in
its documentation.
Fix these problems.
Signed-off-by: Simon Glass <sjg@chromium.org> Acked-by: Marek Vasut <marex@denx.de>
Simon Glass [Thu, 12 Jun 2014 05:29:41 +0000 (23:29 -0600)]
Add an I/O tracing feature
When debugging drivers it is useful to see what I/O accesses were done
and in what order.
Even if the individual accesses are of little interest it can be useful to
verify that the access pattern is consistent each time an operation is
performed. In this case a checksum can be used to characterise the operation
of a driver. The checksum can be compared across different runs of the
operation to verify that the driver is working properly.
In particular, when performing major refactoring of the driver, where the
access pattern should not change, the checksum provides assurance that the
refactoring work has not broken the driver.
Add an I/O tracing feature and associated commands to provide this facility.
It works by sneaking into the io.h heder for an architecture and redirecting
I/O accesses through its tracing mechanism.
For now no commands are provided to examine the trace buffer. The format is
fairly simple, so 'md' is a reasonable substitute.
Note: The checksum feature is only useful for I/O regions where the contents
do not change outside of software control. Where this is not suitable you can
fall back to manually comparing the addresses. It might be useful to enhance
tracing to only checksum the accesses and not the data read/written.
Jeroen Hofstee [Sun, 15 Jun 2014 22:10:42 +0000 (00:10 +0200)]
cmd_md5sum.c: remove dead code / fix warning
commit ecd729500 "Add parameter to md5sum to save the md5 sum"
adds support to build a string to be saved in the env and tries
to zero end it with str_ptr = '\0'; This does actually set the
pointer to the end of the buffer itself to zero. Since the
string was already zero terminated by the sprintf before it,
just remove the line, preventing a clang warning.
cc: Joe Hershberger <joe.hershberger@ni.com> Signed-off-by: Jeroen Hofstee <jeroen@myspectrum.nl>
Vasili Galka [Sun, 15 Jun 2014 15:42:09 +0000 (18:42 +0300)]
Remove nios-32 arch remnants
nios-32 arch was removed back in 2010 (1117cbf). Code depending on
its headers (nios.h, nios-io.h) can't possibly compile since then.
As it wasn't fixed till now it is safe to remove.
Vasili Galka [Sun, 15 Jun 2014 15:41:16 +0000 (18:41 +0300)]
m68k: Remove CONFIG_CMD_BEDBUG related code
This flag does not compile on m68k since 2003 (8bde7f7) when a
required "cmd_bedbug.h" header was removed. Eleven years passed,
lets clean up a little...
Jeroen Hofstee [Thu, 12 Jun 2014 20:27:12 +0000 (22:27 +0200)]
includes: move openssl headers to include/u-boot
commit 18b06652cd "tools: include u-boot version of sha256.h"
unconditionally forced the sha256.h from u-boot to be used
for tools instead of the host version. This is fragile though
as it will also include the host version. Therefore move it
to include/u-boot to join u-boot/md5.h etc which were renamed
for the same reason.
cc: Simon Glass <sjg@chromium.org> Signed-off-by: Jeroen Hofstee <jeroen@myspectrum.nl>
Tom Rini [Thu, 19 Jun 2014 12:46:21 +0000 (08:46 -0400)]
am335x_evm: Only enable OF_CONTROL/OF_SEPARATE on VBOOT for now
We don't make use of the device tree otherwise yet (and will need to
think how to not break the current multi-board support) and this causes
further breakage with additional changes.
Stephen Warren [Wed, 11 Jun 2014 18:46:16 +0000 (12:46 -0600)]
fs: ext4: fix writing zero-length files
ext4fs_allocate_blocks() always allocates at least one block for a file.
If the file size is zero, this causes total_remaining_blocks to
underflow, which then causes an apparent hang while 2^32 blocks are
allocated.
To solve this, check that total_remaining_blocks is non-zero as part of
the loop condition (i.e. before each loop) rather than at the end of
the loop.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Data written to DTB must be converted to big endian order.
It is usually done by using cpu_to_fdt32(), cpu_to_fdt64(), etc.
fdt_initrd() invoked write_cell(), which always swaps byte order.
It means the function only worked on little endian architectures.
(On big endian architectures, the byte order should be kept as it is)
This commit uses cpu_to_fdt32() and cpu_to_fdt64()
and deletes write_cell().
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com> Acked-by: Simon Glass <sjg@chromium.org>
fdt_support: add 'const' qualifier for unchanged argument
In the next commit, I will add a new function, fdt_pack_reg()
which uses get_cells_len().
Beforehand, this commit adds 'const' qualifier to get_cells_len().
Otherwise, a warning message will appear:
warning: passing argument 1 of 'get_cells_len' discards 'const'
qualifier from pointer target type [enabled by default]
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com> Acked-by: Simon Glass <sjg@chromium.org>
- Do not use a deep indentation. We have only 80-character
on each line and 1 indentation consumes 8 spaces. Before the
code moves far to the right, you should consider to
fix your code. See Linux Documentation/CodingStyle.
- Add CONFIG_OF_STDOUT_VIA_ALIAS and OF_STDOUT_PATH macros
only to their definition. Do not add them to both
callee and caller. This is a tip to avoid using #ifdef
everywhere.
- OF_STDOUT_PATH and CONFIG_OF_STDOUT_VIA_ALIAS are exclusive.
If both are defined, the former takes precedence.
Do not try to fix-up "linux,stdout-path" property twice.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com> Acked-by: Simon Glass <sjg@chromium.org>
fdt_support: refactor with fdt_find_or_add_subnode helper func
Some functions in fdt_support.c do the same routine:
search a node with a given name ("chosen", "memory", etc.)
or newly create it if it does not exist.
So this commit makes that routine to a helper function.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com> Acked-by: Simon Glass <sjg@chromium.org>
In function 'flash_erase': 180:21:
warning: variable 'last' set but not used [-Wunused-but-set-variable]
In function 'write_buff': 322:10:
warning: variable 'port_width' set but not used [-Wunused-but-set-variable]
Jon Nalley [Wed, 26 Feb 2014 16:32:21 +0000 (11:32 -0500)]
libfdt: Fix segfault when calling fit_check_format() on corrupt FIT images
It has been observed that fit_check_format() will fail when passed a
corrupt FIT image. This was tracked down to _fdt_string_eq():
return (strlen(p) == len) && (memcmp(p, s, len) == 0);
In the case of a corrupt FIT image one can't depend on 'p' being NULL
terminated. I changed it to use strnlen() to fix the issue.
Dan Murphy [Wed, 2 Oct 2013 19:00:15 +0000 (14:00 -0500)]
ARM: fdt support: Add usbethaddr as an acceptable MAC
A board that has a USB ethernet device only may set the usbetheraddr
and not the ethaddr.
ethaddr will be the default MAC address that is chosen and if that
is not populated then the usbethaddr is looked at. If neither are set
then then device tree blob is not modified.
Vasili Galka [Mon, 16 Jun 2014 14:40:59 +0000 (17:40 +0300)]
Fix bug in io64 target (introduced by commit aba27ac)
From comparison of current logic and the logic that was prior to commit aba27ac, we see that first parameter of FPGA_GET_REG() shall be the
FPGA index and not channel number. The re-factoring in commit aba27ac
accidentally changed that.
Michal Simek [Mon, 16 Jun 2014 22:36:14 +0000 (00:36 +0200)]
configs: iocon: Enabling CONFIG_CMD_FPGAD again
This is a MIME GnuPG-signed message. If you see this text, it means that
your E-mail or Usenet software does not support MIME signed messages.
The Internet standard for MIME PGP messages, RFC 2015, was published in 1996.
To open this message correctly you will need to install E-mail or Usenet
software that supports modern Internet standards.
Shawn Guo [Wed, 11 Jun 2014 08:52:47 +0000 (16:52 +0800)]
mx6: drop ARM errata 742230
Commit e9fd66defd7e (ARM: mx6: define CONFIG_ARM_ERRATA_742230) enables
errata 742230 for imx6, because it helps remove one reboot issue.
However, this errata does not really apply on imx6, because Cortex-A9
on imx6 is r2p10 while the errata only applies to revisions r1p0..r2p2.
At a later time, commit f71cbfe3ca5d (ARM: Add workaround for Cortex-A9
errata 794072) adds support of errata 794072, which applies to all
Cortex-A9 revisions. As the workaround for both errata are exactly
same, it makes a lot more sense to select 794072 instead of 742230 for
imx6. Since we already enable 794072 for imx6, it's time to drop
errata 742230 to avoid confusion.
Fabio Estevam [Thu, 5 Jun 2014 22:31:30 +0000 (19:31 -0300)]
mx6: Fix definition of IOMUXC_GPR12_DEVICE_TYPE_RC
mx6 reference manual incorrectly states that the DEVICE_TYPE field of
IOMUXC_GPR12 register should be configured as '0010' for setting the PCI
controller in RC mode. The correct value should be '0100' instead.
This also aligns with the same value used in the mx6 pci kernel driver.
Masahiro Yamada [Thu, 5 Jun 2014 10:47:45 +0000 (19:47 +0900)]
arm: zynq: fix a bug in Zynq linker script
Commit 41623c91 moved exception handlers to ".vectores" section
but it missed to adjust Zynq linker script.
Zynq boards hang up after relocation because "_start" symbol
does not point to the correct address and gd->relocaddr gets insane.
Signed-off-by: Masahiro Yamada <yamada.m@jp.panasonic.com> Cc: Albert ARIBAUD <albert.u.boot@aribaud.net> Cc: Michal Simek <monstr@monstr.eu> Tested-by: Michal Simek <monstr@monstr.eu> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Wu, Josh [Tue, 27 May 2014 08:31:05 +0000 (16:31 +0800)]
net: macb: enable dcache in macb
Add to code to flush the dcache after we writing in DMA buffer.
Also we need invalidate the dcache before we check the status in the
DMA buffer.
Tested in SAMA5D3x-EK with gmac0. Tftp download speed shows in below:
Disable DCache: 1.1 MiB/s
Enable DCache: 1.6 MiB/s
Increase speed with about 40%.
The code should have no impact with the boards which are not
enable_dcache().
Tested in AT91SAM9M10G45EK.
Signed-off-by: Josh Wu <josh.wu@atmel.com> Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
Andreas Bießmann [Thu, 12 Jun 2014 20:07:52 +0000 (22:07 +0200)]
avr32: migrate cache functions
Unfortunately the avr32 cache implementation has another API than the one
described in common.h. Migrate the flush/invalidate dcache functions to the
common API to be usable in device drivers.
Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com> CC: Josh Wu <josh.wu@atmel.com>
Darwin Rambo [Mon, 26 May 2014 20:31:12 +0000 (13:31 -0700)]
mmc: free allocated memory on initialization errors
Cleanup to balance malloc/free calls.
Signed-off-by: Darwin Rambo <drambo@broadcom.com> Reviewed-by: Steve Rae <srae@broadcom.com> Acked-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Stephen Warren [Fri, 23 May 2014 19:24:47 +0000 (13:24 -0600)]
cmd_mmc: make mmc dev always re-probe the HW
Currently, U-Boot behaves as follows:
- Begin with no SD card inserted in "mmc 1"
- Execute: mmc dev 1
- This fails, since there is no card
- User plugs in an SD card
- Execute: mmc dev 1
- This still fails, since the HW isn't reprobed.
With this change, U-Boot behaves as follows:
- Begin with no SD card inserted in "mmc 1"
- Execute: mmc dev 1
- This fails, since there is no card
- User plugs in an SD card
- Execute: mmc dev 1
- The newly present SD card is detected
I know that "mmc rescan" will force the HW to be reprobed, but I feel it
makes more sense if "mmc dev" always reprobes the HW after selecting the
current MMC device. This allows scripts to just execute "mmc dev", and
not have to also execute "mmc rescan" to check for media presense.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Acked-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Stephen Warren [Fri, 23 May 2014 19:24:46 +0000 (13:24 -0600)]
cmd_mmc: Use init_mmc_device() from do_mmc_rescan()
The body of init_mmc_device() is now identical to that of do_mmc_rescan()
except for the error codes returned. Modify do_mmc_rescan() to simply
call init_mmc_device() and convert the error codes, to avoid code
duplication.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Acked-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Stephen Warren [Fri, 23 May 2014 19:24:45 +0000 (13:24 -0600)]
cmd_mmc: add force_init parameter to init_mmc_device()
This allows callers to inject mmc->has_init = 0 between finding the
MMC device, and calling mmc_init(), which forces mmc_init() to rescan
the HW. Future changes will use this feature.
Stephen Warren [Fri, 23 May 2014 18:48:11 +0000 (12:48 -0600)]
disk: default to HW partition 0 if not specified
Currently, get_device()/get_dev_hwpart() for MMC devices does not select
an explicit HW partition unless the user explicitly requests one, i.e. by
requesting device "mmc 0.0" rather than just "mmc 0". I think it makes
more sense if the default is to select HW partition 0 (main data area)
if the user didn't request a specific partition. Otherwise, the following
happens, which feels wrong:
Select HW partition 1 (boot0):
mmc dev 0 1
Attempts to access SW partition 1 on HW partition 1 (boot0), rather than
SW partition 1 on HW partition 0 (main data area):
ls mmc 0:1 /
With this patch, the second command above re-selects the main data area.
Many device types don't support HW partitions at all, so if HW partition
0 is selected (either explicitly or as the default) and there's no
select_hwpart function, we simply skip attempting to select a HW
partition.
Some MMC devices (i.e. SD cards) don't support HW partitions. However,
this patch still works, since mmc_start_init() sets the current
partition number to 0, and mmc_select_hwpart() succeeds if the requested
partition is already selected.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Acked-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Stephen Warren [Fri, 23 May 2014 18:48:10 +0000 (12:48 -0600)]
cmd_mmc: default to HW partition 0 if not specified
Currently, "mmc dev 0" does not change the selected HW partition. I
think it makes more sense if "mmc dev 0" is an alias for "mmc dev 0 0",
i.e. that HW partition 0 (main data area) is always selected by default
if the user didn't request a specific partition. Otherwise, the following
happens, which feels wrong:
Select HW partition 1 (boot0):
mmc dev 0 1
Doesn't change the HW partition, so it's still 1 (boot0):
mmc dev 0
With this patch, the second command above re-selects the main data area.
Note that some MMC devices (i.e. SD cards) don't support HW partitions.
However, this patch still works, since mmc_start_init() sets the current
partition number to 0, and mmc_select_hwpart() succeeds if the requested
partition is already selected.
Signed-off-by: Stephen Warren <swarren@nvidia.com> Acked-by: Pantelis Antoniou <panto@antoniou-consulting.com>
Stephen Warren [Fri, 23 May 2014 18:47:06 +0000 (12:47 -0600)]
mmc: return meaningful error codes from mmc_select_hwpart
Rather than just returning -1 everywhere, try to return something
meaningful from mmc_select_hwpart(). Note that most other MMC functions
don't do this, including functions called from mmc_select_hwpart(), so
I'm not sure how effective this will be. Still, it's one less place with
hard-coded -1.
projects / u-boot / log