It's possible that the data cache for the buffer still holds data
to be flushed to memory, since the buffer was probably used as stack
before. Thus we need to make sure to flush it also on reads, since
it's possible that the cache is automatically flused to memory after
the NVMe DMA transfer happened, thus overwriting the NVMe transfer's
data. Also add a missing dcache flush for the prp list.
Signed-off-by: Patrick Wildt <patrick@blueri.se>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Update from upstream.
Just minor changes like checking that the author has also done a sign-off.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Profiling the EHCI driver shows a significant performance problem in
ehci_submit_async(). Specifically, this function keeps enabling and
disabling async schedule back and forth for every single transaction.
However, enabling/disabling the async schedule does not take effect
immediatelly, but instead may take up to 1 mS (8 uFrames) to complete.
This impacts USB storage significantly, esp. since the recent reduction
of maximum transfer size to support more USB storage devices. This in
turn results in sharp increase in the number of ehci_submit_async()
calls. Since one USB storage BBB transfer does three such calls and
the maximum transfer size is 120 kiB, the overhead is 6 mS per 120 kiB,
which is unacceptable.
However, this overhead can be removed simply by keeping the async
schedule running. Specifically, the first transfer starts the async
schedule and then each and every subsequent transfer only adds a new
QH into that schedule, waits until the QH is completed and does NOT
disable the async schedule. The async schedule is stopped only by
shutting down the controller, which must happen before moving out
of U-Boot, otherwise the controller will corrupt memory.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Clear the USB_READY flag in the storage driver only in case there
is an error, otherwise usb_stor_BBB_transport() waits 5 mS before
doing anything every single time.
This is because the USB_READY flag is only ever set in
usb_test_unit_ready(), which is called only upon storage device
probe, not between each and every request. However, the device
cannot move out of USB_READY state once it was initialized.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Due to constant influx of more and more weird and broken USB sticks,
do as Linux does in commit 779b457f66e10de3471479373463b27fd308dc85
usb: storage: scsiglue: further describe our 240 sector limit
Just so we have some sort of documentation as to why
we limit our Mass Storage transfers to 240 sectors,
let's update the comment to make clearer that
devices were found that would choke with larger
transfers.
While at that, also make sure to clarify that other
operating systems have similar, albeit different,
limits on mass storage transfers.
And reduce the maximum transfer length of USB storage to 120 kiB.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
Add weak callback to allow board specific behavior
- flush
- initiated
This patch prepare usage of DFU back end for communication with
STM32CubeProgrammer on stm32mp1 platform with stm32prog command.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Add a virtual DFU backend to allow board specific read and write
(for OTP update for example).
Acked-by: Lukasz Majewski <lukma@denx.de>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Add the support of MTD partition for the MTD backend.
The expected dfu_alt_info for one alternate on the mtd device :
<name> part <part_id>
<name> partubi <part_id>
"partubi" also erase up to the end of the partition after write operation.
For example: dfu_alt_info = "spl part 1;u-boot part 2; UBI partubi 3"
U-Boot> dfu 0 mtd nand0
Acked-by: Lukasz Majewski <lukma@denx.de>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Add DFU backend for MTD device: allow to read
and write on all MTD device (NAND, SPI-NOR,
SPI-NAND,...)
For example :
> set dfu_alt_info "nand_raw raw 0x0 0x100000"
> dfu 0 mtd nand0
This MTD backend provides the same level than dfu nand
backend for NAND and dfu sf backend for SPI-NOR;
So it can replace booth of them but it also
add support of spi-nand.
> set dfu_alt_info "nand_raw raw 0x0 0x100000"
> dfu 0 mtd spi-nand0
The backend code is based on the "mtd" command
introduced by commit 5db66b3aee ("cmd: mtd:
add 'mtd' command")
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
This patch allows the DFU backend to indicate that that it can't
provide no more data to fill the DFU buffer, by setting b_left =0
without error, even if the size of received data is lower of the
expected total size indicated by get_medium_size.
For USB DFU stack point of view, it is acceptable:
the read length < requested size in DFU_UPLOAD and the
transaction is stopped.
That avoid infinite loop issue in dfu_read_buffer_fill because the
size for the DFU read is limited by get_medium_size = r_left
and the DFU stack expects that read is allowed up to this size.
This issue never occurs for current flash device (where chunk are
always completely read, and b_left will be never 0) but it is useful for
virtual partition when the backend only know the max size of this
alternate, the real size of the data are only known in the read
treatment.
PS: for file access on mmc, EOF is never reached as
dfu_get_medium_size_mmc returns the exact size of the file.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Add support of DFU for several interface/device
with one command.
The format for "dfu_alt_info" in this case is :
- <interface> <dev>'='alternate list (';' separated)
- each interface is separated by '&'
The previous behavior is always supported.
One example for NOR (bootloaders) + NAND (rootfs in UBI):
U-Boot> env set dfu_alt_info \
"sf 0:0:10000000:0=spl part 0 1;u-boot part 0 2; \
u-boot-env part 0 3&nand 0=UBI partubi 0,3"
U-Boot> dfu 0 list
DFU alt settings list:
dev: SF alt: 0 name: spl layout: RAW_ADDR
dev: SF alt: 1 name: ssbl layout: RAW_ADDR
dev: SF alt: 2 name: u-boot-env layout: RAW_ADDR
dev: NAND alt: 3 name: UBI layout: RAW_ADDR
U-Boot> dfu 0
$> dfu-util -l
Found DFU: [0483:5720] ver=9999, devnum=96, cfg=1,\
intf=0, alt=3, name="UBI", serial="002700333338511934383330"
Found DFU: [0483:5720] ver=9999, devnum=96, cfg=1,\
intf=0, alt=2, name="u-boot-env", serial="002700333338511934383330"
Found DFU: [0483:5720] ver=9999, devnum=96, cfg=1,\
intf=0, alt=1, name="u-boot", serial="002700333338511934383330"
Found DFU: [0483:5720] ver=9999, devnum=96, cfg=1,\
intf=0, alt=0, name="spl", serial="002700333338511934383330"
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Copy the partition support from NAND backend to SF,
support part and partubi option.
In case of ubi partition, erase the rest of the
partition as it is mandatory for UBI.
The added code is under compilation flag CONFIG_DFU_SF_PART
activated by default.
for example:
U-Boot> env set dfu_alt_info "spl part 0 1;\
u-boot part 0 2;u-boot-env part 0 3;UBI partubi 0 4"
U-Boot> dfu 0 sf 0:0:10000000:0
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Calling cache flush on invalid buffer, even with zero length might cause
an exception on certain platforms.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Add possibility to define a part of partition as a separate DFU entity.
This allows to have more than one items on the given partition.
The real use case for this option is TM2 board. It can use u-boot stored
as Linux kernel on the defined partition (as RAW data) and load the real
kernel from the same partition, but stored under the certain offset.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Lukasz Majewski <lukma@denx.de>
'THOR' download command requires DFU infrastructure to properly flash
board images. It can be used without enabling DFU command, so add such
dependency to Kconfig.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Lukasz Majewski <lukma@denx.de>
This is needed to make Windows THOR flash tool happy, because it
starts sending data only when interrupt packet is received on the 3rd
endpoint.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Lukasz Majewski <lukma@denx.de>
In combination with multiple partitions in NAND, this printf() ends up
being more noise than helpful. Change it to debug() instead.
Signed-off-by: Ralph Siemsen <ralph.siemsen@linaro.org>
Acked-by: Lukasz Majewski <lukma@denx.de>
Some boards don't populate clk/reset entries as these are are optional
as per binding documentation. Therefore, don't fail driver probe if
clk/reset entries are absent in DT.
This fixes fastboot failures seen due to enabling of CONFIG_CLK on AM57xx
Fixes: e8e683d33b ("board: ti: am57xx-idk: Configure the CDCE913 clock synthesizer")
Reported-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
Some configurations (i.e. sandbox) result in unused
splash_display_banner() and we get build errors like:
common/splash.c:148:20: error: unused function 'splash_display_banner' [-Werror,-Wunused-function]
static inline void splash_display_banner(void) { }
Drop empty inline function and add guards around the call.
Fixes: d2a8271c88 ("splash: fix splash banner output")
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Reported-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
As initially this feature was implemented as a negative CONFIG and
later it was redesigned to be positive the help text should be
updated to reflect this change.
This commit updates the help text to match the current implementation.
Signed-off-by: Walter Lozano <wlozano@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
When we have an excess size growth, fix the "limit" printf call to pass
in just the limit variable rather than the string bytes to the format
character.
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
_gpio_direction_output function currently calls gpio_set_value
with the wrong gpio number. gpio_set_value in the uclass driver
expects a different gpio number and the _gpio_direction_output
is currently providing the number specific to the bank.
Hence fix it by calling the _gpio_set_value function instead.
Reported-by: Faiz Abbas <faiz_abbas@ti.com>
Fixes: 8e51c0f254 ("dm: gpio: Add DM compatibility to GPIO driver for Davinci")
Signed-off-by: Keerthy <j-keerthy@ti.com>
We have two further uses of raw string usage in the test/py codebase
that are used under CI. The first of which is under the bind test and
is a direct update. The second of which is to strip VT100 codes from
the match buffer. While switching this to a raw string is also a direct
update, the comment it notes that problems were encountered on Ubuntu
14.04 (and whatever Python 2 version that was) that required slight
tweaks to the regex. Replace that now that we're saying Python 3.5 is
the minimum.
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
As things stand today, we have tools that CI requires where "python"
must be "python2". We need to use a virtualenv and pip in order to
ensure that our pytest tests can be run. Rework things slightly so
that:
- On Travis-CI, we install python-pyelftools for the platforms that
require pyelftools to be installed.
- On GitLab-CI, we move to a newer base image that includes python3-pip
and continue to use a virtualenv per job that needs it, for the
correct set of packages.
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
To be more closely aligned with Python community best practices, we need
to better document our usage of pip and make use of a requirements.txt
file that shows the versions of the tools that we are using. This will
aide in ensuring reproducibility of our tests as well.
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
Now that we have moved to being based on pytest for python3 we need to
make our test.py wrapper more robust in terms of only calling python3
rather than possibly finding and using python2. To do this, change from
execvp()'ing pytest to invoking the package itself via python. In the
event that pytest is unavailable we still get a user-friendly error:
pkg_resources.DistributionNotFound: The 'pytest' distribution was not found and is required by the application
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
The check_output function from the subprocess Python module by default
returns data as encoded bytes and leaves decoding to the application.
Given our uses of the call, it makes the most sense to immediately
decode the results.
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
- In python 3 you must use raw strings for regex as other forms are
deprecated and would require further changes to the pattern here.
In one case this lets us have a simpler match pattern.
- As strings are now Unicode our complex tests (Euro symbol,
SHIFT+ALT+FN 5) we need to declare that as a bytes string and then
decode it for use.
Cc: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
In the case of some unit tests we are working with providing a fake
flash device that we have written some text strings in to. In this case
we want to tell Python to encode things to bytes for us.
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
- Modern pytest is more visible in telling us about parameters that we
had not described, so describe a few more.
- ConfigParser.readfp(...) is now configparser.read_file(...)
- As part of the "strings vs bytes" conversions in Python 3, we use the
default encoding/decoding of utf-8 but in some places tell Python to
replace problematic conversions rather than throw a fatal error.
- Fix a typo noticed while doing the above ("tot he" -> "to the").
- As suggested by Stephen, re-alphabetize the import list
- Per Heinrich, replace how we write contents in test_fit.py
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
Use the 2to3 tool to perform numerous automatic conversions from Python
2 syntax to Python 3. Also fix whitespace problems that Python 3
catches that Python 2 did not.
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
Fix the following spit from pytest:
u-boot/test/py/conftest.py:438: RemovedInPytest4Warning: MarkInfo objects are deprecated as they contain merged marks which are hard to deal with correctly.
Please use node.get_closest_marker(name) or node.iter_markers(name).
Docs: https://docs.pytest.org/en/latest/mark.html#updating-code
for board in mark.args:
In both cases, the later suggestion is applicable.
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Igor Opaniuk <igor.opaniuk@gmail.com>
[trini: Update for current file with a few more cases, un-pin pytest in CI]
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
We inconsistently note multiple dependencies today in our tests,
sometimes with a single line that declares multiple and sometimes
multiple single lines. Current pytest seems to fail on the single line
format so change to multiple declarations.
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Simon Glass <sjg@chromium.org> [on sandbox]
Signed-off-by: Tom Rini <trini@konsulko.com>
Currently we set the entire PATH rather than prepend the new paths that
we need to have searched. This however breaks parts of the "virtualenv"
that was have set up and need to use as that also will be modifying
PATH. To fix this, prepend our new locations instead.
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
In a number of our stanzas we had multi-line commands that were one
space short of alignment, correct this.
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
- Add I2C clocks for i.MX6Q CCF driver
- Fix check in clk_set_default_parents()
- Managed API to get clock from device tree
- Fixes for core clock code (including sandbox regression tests)
'make tests' on an 32bit ARM system leads to
In file included from ../lib/efi_loader/efi_variable.c:9:
../include/malloc.h:364:7: error: conflicting types for ‘memset’
void* memset(void*, int, size_t);
^~~~~~
In file included from ../include/compiler.h:126,
from ../include/env.h:12,
from ../lib/efi_loader/efi_variable.c:8:
../include/linux/string.h:103:15:
note: previous declaration of ‘memset’ was here
extern void * memset(void *,int,__kernel_size_t);
^~~~~~
In file included from ../lib/efi_loader/efi_variable.c:9:
../include/malloc.h:365:7: error: conflicting types for ‘memcpy’
void* memcpy(void*, const void*, size_t);
^~~~~~
In file included from ../include/compiler.h:126,
from ../include/env.h:12,
from ../lib/efi_loader/efi_variable.c:8:
../include/linux/string.h:106:15:
note: previous declaration of ‘memcpy’ was here
extern void * memcpy(void *,const void *,__kernel_size_t);
^~~~~~
Use common.h as first include as recommended by the U-Boot coding style
guide.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
With this patch, when setting UEFI variable with "env set -e" command,
we will be able to
- specify vendor guid with "-guid guid",
- specify variable attributes, BOOTSERVICE_ACCESS, RUNTIME_ACCESS,
respectively with "-bs" and "-rt",
- append a value instead of overwriting with "-a",
- use memory as variable's value instead of explicit values given
at the command line with "-i address,size"
If guid is not explicitly given, default value will be used.
Meanwhile, "env print -e," will be modified so that it will NOT dump
a variable's value if '-n' is specified.
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
The ext4 file system requires log2blksz to be set. So when setting the
block size on the block descriptor we should fill this field too.
This fixes a problem with EFI block devices providing ext4 partitions, cf.
https://lists.denx.de/pipermail/u-boot/2019-October/387702.html.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
This expands current Azure Pipelines Windows host tools build
testing to cover all the CI testing in gitlab and travis CI.
Note for some unknown reason, the 'container' cannot be used for
any jobs that have buildman, for buildman does not exit properly
and hangs the job forever. As a workaround, we manually call
docker to run the image to perform the CI tasks.
A complete run on Azure Pipelines takes about 2 hours and 10
minutes.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
