The USB boot code is too fat and complicated to be included in SPL
(at least for now). So, it was implemented as a separate project
(what we call USB-loader).
The expected boot sequence is as follows:
Boot ROM -> USB-loader -> SPL -> U-Boot proper
The USB-loader loads the SPL and U-Boot proper from a USB memory
onto the locked L2 cache. Then, SPL needs to copy the U-Boot proper
to DRAM, so this mode looks like a NOR boot from the view of SPL.
However, we want to distinguish between (genuine) NOR boot and USB
boot in some places.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
UniPhier SoCs are not equipped with dedicated on-chip SRAM. Instead,
locked outer cache is used as RAM area during the early boot stage
where DRAM is not ready yet. This effectively means MMU must be
always enabled while we are in SPL.
Currently, the SPL image for UniPhier SoCs contains the page table
statically defined at compile time. It has been a burden because the
16KB page table occupies a quarter memory footprint of the 64KB SPL
image.
Finally, there is no more room to implement new features in SPL.
Setting aside the NOR boot mode, this issue can be solved by creating
the page table onto RAM at run time.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
No special reason for the current stack address 0x0ff08000.
Change it to 0x00100000 to simplify the init_page_table.
There are two types of SoCs in terms of the load address of SPL.
[1] PH1-sLD3, PH1-LD4, PH1-sLD8
SPL is loaded at 0x00040000-0x0004ffff
[2] PH1-Pro4, PH1-Pro5, ProXstream2, PH1-LD6b
SPL is loaded at 0x00100000-0x0010ffff
The new stack area (0x000f8000-0x00100000) should be safe for all the
cases.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Since commit ad1ecd2063 ("fdt: Build a U-Boot binary without device
tree") and commit 03c25bcd26 ("fdt: Build an SPL binary without
device tree"), we can use shorter file names for the output images.
The default configuration for UniPhier SoCs enables CONFIG_OF_SEPARATE
and CONFIG_SPL_OF_CONTROL. In this case, spl/u-boot-spl.bin is the
same as spl/u-boot-spl-dtb.bin. Likewise, u-boot.img is the same as
as u-boot-dtb.img. So, this change of the flow has no impact.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Some of PH1-Pro4 boards are equipped with larger amount of DRAM than
the reference board. Add UMC settings to support them.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The environment "bootm_low" is updated before the "bootz" command.
This is common for all the boot modes (NOR, NAND, TFTP, etc.), so
can be factored out.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Commit 25d4eb8091 ("ARM: uniphier: add bootm_low environment")
missed to add "bootm_low" for FIT boot. Set "bootm_low" to the
DRAM base address.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Commit 89835b3557 ("ARM: uniphier: allow to run zImage rather than
uImage") changed the kernel boot commands. Unlike "bootm", "bootz"
does not relocate the kernel image. When the boot device is a NOR
flash, the zImage should be copied from the NOR onto the DRAM before
it is passed to the "bootz" command.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Boards with a saved environment may use 'lcd' in their stdout environment
variable, expecting that this will enable output to the LCD. When the board
moves to use driver model for video, this will no-longer work. Add a
work-around to fix this. A warning messages is printed, and we will remove
the work-around at the end of 2016.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: Anatolij Gustschin <agust@denx.de>
Tested-by: Stephen Warren <swarren@nvidia.com>
This commit breaks bootup on sunxi boards, the get stuck
when running the main u-boot binary at:
CPU: Allwinner H3 (SUN8I)
I2C: ready
DRAM:
This reverts commit 8e7cba048b.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
The existing regex simply ensures that the captured version string doesn't
go past the end of a line. We really want to grab as much as possible. Do
this by explicitly including a ) character at the end of the regex to
match the last character of the version test.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
A regex match object's .end() value is already the index after the match,
not the index of the last character in the match, so there's no need to
add 1 to point past the match.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Implement three improvements to the HTML log file:
- Ability to expand/contract sections. All passing sections are contracted
at file load time so the user can concentrate on issues requiring
action.
- The overall status report is copied to the top of the log for easy
access.
- Add links from the status report to the test logs, for easy navigation.
This all relies on Javascript and the jquery library. If the user doesn't
have Javascript enabled, or jquery can't be downloaded, the log should
look and behave identically to how it did before this patch.
A few notes on the diff:
- A few more 'with log.section("xxx")' were added, so that all stream
blocks are kept within a section block for consistent HTML entity
nesting structure. This changed indentation in a few places, making
the diff look slightly larger.
- HTML entity IDs are cleaned up. We assign simple incrementing integer
IDs now, rather than using mangled test names which were possibly
invalid.
- Sections and streams now use common CSS class names (in addition to the
current separate class names) to more easily share the new behaviour.
This also reduces the CSS file size since rules don't need to be
duplicated.
- An "OK" status is logged after some external command executions so that
make and flash steps are auto-contracted at log file load time, assuming
they passed.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
The test/py/test.py wrapper script catches exceptions thrown when
exec()ing py.test in order to print a helpful error message. However,
the exception handling code squashes the exception and so the script
exits with a non-zero exit code, leading callers to believe that it
passed. Fix this.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
On some systems, RAM starts at address 0. If the user executes itest
against address 0 on such a system, it will call map_physmem(0, ...)
which will return 0 back; mapping only changes the address on sandbox.
This causes itest to believe map_physmem() has failed, and hence fails
the comparison.
Fix itest so that it allows map_physmem() to return 0 /if/ the orignal
address passed to it was also 0.
This fixes "tegra-uboot-flasher flash" on Tegra20.
This has the disadvantage that on sandbox, failed mapping attempts for
address 0 are not detected. Instead, should the code only call
map_physmem() on sandbox? Or, should map_physmem() return its error status
some other way. Or, should the special case only be allowed on systems
where the base of RAM is 0 somehow?
Fixes: 7861204c9a ("itest: make memory access work under sandbox")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Current, the following passes:
./u-boot -d arch/sandbox/dts/test.dtb -c 'ut_image_decomp'
but the following fails:
./u-boot -d arch/sandbox/dts/test.dtb -c 'ut dm; ut_image_decomp'
This is because the gunzip code reads input data beyond the end of its
input buffer. In the first case above, this data just happens to be 0,
which just happens to trigger gzip to signal the error the decompression
unit test expects. In the second case above, the "ut dm" test has written
data to the accidentally-read memory, which causes the gzip code to take a
different path and so return a different value, which triggers the test
failure.
The cause of gunzip reading past its input buffer is the re-calculation of
s.avail_in in zunzip(), since it can underflow. Not only is the formula
non-sensical (it uses the delta between two output buffer pointers to
calculate available input buffer size), it also appears to be unnecessary,
since the gunzip code already maintains this value itself. This patch
removes this re-calculation to avoid the underflow and redundant work.
The loop exit condition is also adjusted so that if inflate() has consumed
the entire input buffer, without indicating returning Z_STREAM_END (i.e.
decompression complete without error), an error is raised. There is still
opportunity to simplify the code here by splitting up the loop exit
condition into separate tests. However, this patch makes the minimum
modifications required to solve the problem at hand, in order to keep the
Acked-by: Kees Cook <keescook@chromium.org>
diff simple.
I am not entirely convinced that the loop in zunzip() is necessary at all.
It could only be useful if inflate() can return Z_BUF_ERROR (which
typically means that it needs more data in the input buffer, or more space
in the output buffer), even though Z_FINISH is set /and/ the full input is
available in the input buffer /and/ there is enough space to store the
decompressed output in the output buffer. The comment in zlib.h after the
prototype of inflate() implies this is never the case. However, I assume
there must have been some reason for introducing this loop in the first
place, as part of commit "Fix gunzip to work for any gziped uImage size".
This patch is similar to the earlier b75650d84d "gzip: correctly
bounds-check output buffer", which corrected a similar issue for
s.avail_out.
Cc: Catalin Radu <Catalin@VirtualMetrix.com>
Cc: Kees Cook <keescook@chromium.org>
Fixes: f039ada5c1 ("Fix gunzip to work for any gziped uImage size")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
There are various different boards with the same hardware sold as LG Optimus
Black, such as P970, P970g and KU5900. Since this port is functional for all
variants, it doesn't make sense to keep references to P970.
Signed-off-by: Paul Kocialkowski <contact@paulk.fr>
Reviewed-by: Tom Rini <trini@konsulko.com>
The ARM Linux kernel requires the DT to be in memory accessible early
during the boot process. This always happens naturally on the RPi 1,
since the maximum memory size of 512MiB, and additionally some of that
is reserved for use by the GPU. The RPi 2 has 1GiB of RAM (minus some
GPU usage), and so if the DT is relocated to the top of RAM, Linux cannot
access it. Prevent this from happening by setting fdt_high.
Cc: Daniel Stone <daniels@collabora.com>
Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
Update rpi-common.h's documentation that describes the rationale for
choosing various addresses for standardized variables used by boot
scripts. This comment was correct when written, but not updated when some
of the values were changed.
Fixes: 14006a5671 ("rpi: set fdt_addr_r to 0x00000100 to match default
...device_tree_address")
Cc: Jonathan Liu <net147@gmail.com>
Cc: Daniel Stone <daniels@collabora.com>
Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
The build command of u-boot-spl.dtb is not constant, but dependent
on CONFIG_OF_SPL_REMOVE_PROPS. Use $(call if_changed,...) so that
the change of CONFIG_OF_SPL_REMOVE_PROPS is detected.
Also, add tools/fdtgrep to the dependency to make sure u-boot-spl.dtb
is generated by the up-to-date fdtgrep in case the tool is modified.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
FORCE is needed for $(call if_changed,...) to be evaluated every time.
Otherwise, the command is not executed when the command line has
changed but any prerequisite has not been updated.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
These rules are only used for SOCFPGA, SUNXI, but no need to hide
them from other SoCs.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Commit 9c11135ce0 ("image: fix getenv_bootm_size() function") fixed
the case where "bootm_low" is defined, but "bootm_size" is not.
Instead, it broke the case where neither "bootm_low" nor "bootm_size"
is defined. Fix this function again.
Fixes: 9c11135ce0 ("image: fix getenv_bootm_size() function")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Tested-by: Matthias Weisser <m.weisser.m@gmail.com>
Tested-by: Hannes Schmelzer <oe5hpm@oevsv.at>
Implement command--line option --gdbserver COMM, which does two things:
a) Run the sandbox process under gdbserver, using COMM as gdbserver's
communication channel.
b) Disables all timeouts, so that if U-Boot is halted under the debugger,
tests don't fail. If the user gives up in the middle of a debugging
session, they can simply CTRL-C the test script to abort it.
This allows easy debugging of test failures without having to manually
re-create the failure conditions. Usage is:
Window 1:
./test/py/test.py --bd sandbox --gdbserver localhost:1234
Window 2:
gdb ./build-sandbox/u-boot -ex 'target remote localhost:1234'
When using this option, it likely makes sense to use pytest's -k option
to limit the set of tests that are executed.
Simply running U-Boot directly under gdb (rather than gdbserver) was
also considered. However, this was rejected because:
a) gdb's output would then be processed by the test script, and likely
confuse it causing false failures.
b) pytest by default hides stdout from tests, which would prevent the
user from interacting with gdb.
While gdb can be told to redirect the debugee's stdio to a separate
PTY, this would appear to leave gdb's stdio directed at the test
scripts and the debugee's stdio directed elsewhere, which is the
opposite of the desired effect. Perhaps some complicated PTY muxing
and process hierarchy could invert this. However, the current scheme
is simple to implement and use, so it doesn't seem worth complicating
matters.
c) Using gdbserver allows arbitrary debuggers to be used, even those with
a GUI. If the test scripts invoked the debugger themselves, they'd have
to know how to execute arbitary applications. While the user could hide
this all in a wrapper script, this feels like extra complication.
An interesting future idea might be a --gdb-screen option, which could
spawn both U-Boot and gdb separately, and spawn the screen into a newly
created window under screen. Similar options could be envisaged for
creating a new xterm/... too.
--gdbserver currently only supports sandbox, and not real hardware.
That's primarily because the test hooks are responsible for all aspects of
hardware control, so there's nothing for the test scripts themselves can
do to enable gdbserver on real hardware. We might consider introducing a
separate --disable-timeouts option to support use of debuggers on real
hardware, and having --gdbserver imply that option.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
The recent change to memalign() caused the allocation failure detection
code to be dead code; the "retry" logic is always activated under the same
condition that the original failure detection code is, and also fully
handles any possible failures. This patch solves the presence of dead
code.
Two alternatives are possible:
a) Delete the now-dead test, and rely on the "retry" path to handle any
allocation problems, as it does.
b) Make the "retry" path fall through to the existing (currently dead)
failure detection code, thus making it not-dead.
(b) was chosen since it reduces the diff between U-Boot's and the upstream
dlmalloc. This should make it marginally easier to import a new version of
dlmalloc in the future.
Reported by: Coverity Scan
Fixes: 4f144a4164 ("malloc: work around some memalign fragmentation issues")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Without this, builds default to using new Travis CI infra-structure which
does no allow sudo. The builds need sudo in order to install the ELDK
compilers. Consequently, almost all builds fail without this.
I suspect that existing Travis CI users have not noticed this because
their accounts or builds have been grand-fathered into backwards-
compatible default settings, whereas I just set up a new build from
scratch and received new default settings.
Cc: Wolfgang Denk <wd@denx.de>
Cc: Heiko Schocher <hs@denx.de>
Cc: Tom Rini <trini@ti.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Andreas Färber <afaerber@suse.de>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Heiko Schocher <hs@denx.de>
Let's set "ethaddr" when we get the ethernet address too, so that
fdt_fixup_ethernet() sets the address in the device tree and the Linux
driver can pick it up.
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Tested-by: Stephen Warren <swarren@wwwdotorg.org>
The case of memory of size 0 is not that different from a memory of any other
size, so we remove the duplicate code and treat the small differences when it
is the case.
Signed-off-by: Eddy Petrișor <eddy.petrisor@gmail.com>
enable net driver model for k2g evm as keystone_net supports
driver model
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Add keystone net DT support for k2g evm.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
enable net driver model for k2e evm as keystone_net supports
driver model
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
enable net driver model for k2l evm as keystone_net supports
driver model
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
enable net driver model for k2hk evm as keystone_net supports
driver model
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When Micrel phy is selected without CONFIG_PHY_MICREL_KSZ9031 or
CONFIG_PHY_MICREL_KSZ9021 there is a build error. Fixing this
by adding proper ifdefs
drivers/net/phy/micrel.c:370:39: error: array type has incomplete element type
static const struct ksz90x1_reg_field ksz9031_ctl_grp[] =
^
drivers/net/phy/micrel.c:372:39: error: array type has incomplete element type
static const struct ksz90x1_reg_field ksz9031_clk_grp[] =
^
drivers/net/phy/micrel.c: In function ‘ksz9031_of_config’:
drivers/net/phy/micrel.c:377:23: error: array type has incomplete element type
struct ksz90x1_ofcfg ofcfg[] = {
^
drivers/net/phy/micrel.c:379:13: error: ‘ksz90x1_rxd_grp’ undeclared (first use in this function)
{ MII_KSZ9031_EXT_RGMII_RX_DATA_SKEW, 2, ksz90x1_rxd_grp, 4 },
^
drivers/net/phy/micrel.c:379:13: note: each undeclared identifier is reported only once for each function it appears in
drivers/net/phy/micrel.c:380:13: error: ‘ksz90x1_txd_grp’ undeclared (first use in this function)
{ MII_KSZ9031_EXT_RGMII_TX_DATA_SKEW, 2, ksz90x1_txd_grp, 4 },
^
drivers/net/phy/micrel.c:386:3: warning: implicit declaration of function ‘ksz90x1_of_config_group’ [-Wimplicit-function-declaration]
ret = ksz90x1_of_config_group(phydev, &(ofcfg[i]));
^
drivers/net/phy/micrel.c:377:23: warning: unused variable ‘ofcfg’ [-Wunused-variable]
struct ksz90x1_ofcfg ofcfg[] = {
^
drivers/net/phy/micrel.c: At top level:
drivers/net/phy/micrel.c:370:39: warning: ‘ksz9031_ctl_grp’ defined but not used [-Wunused-variable]
static const struct ksz90x1_reg_field ksz9031_ctl_grp[] =
^
drivers/net/phy/micrel.c:372:39: warning: ‘ksz9031_clk_grp’ defined but not used [-Wunused-variable]
static const struct ksz90x1_reg_field ksz9031_clk_grp[] =
^
scripts/Makefile.build:277: recipe for target 'drivers/net/phy/micrel.o' failed
make[1]: *** [drivers/net/phy/micrel.o] Error 1
Makefile:1201: recipe for target 'drivers/net/phy' failed
make: *** [drivers/net/phy] Error 2
make: *** Waiting for unfinished jobs....
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
remove board_eth_init when CONFIG_DM_ETH is defined
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
UBIFS is the preferred FS, and YAFFS isn't officially included in
Linux. Removing this feature reduces the code size.
Signed-off-by: Adam Ford <aford173@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
The interface automatically converts one 32-bit word into two 16 words.
The README said it is permissible to use this flag in that scenario.
Signed-off-by: Adam Ford <aford173@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
