Sphinx 4.x is way past EOL and due to it not pinning its dependencies,
it's effectively broken. See
https://github.com/sphinx-doc/sphinx/issues/11890 The recommended fix,
although not ideal in the context of an LTS branch, is to update to
Sphinx 5.0.2, which should have minimal impact of how the rendered
documentation looks.
Signed-off-by: Benjamin Cabé <benjamin@zephyrproject.org>
Since K_SYSCALL_MEMORY can be called with signed/unsigned size types, if
we check if size >= 0, static anlysis will complain about it when
size in unsigned.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
This macros needed additional checks before invoking
arch_buffer_validate.
- size can not be less then 0. Some functions invoke this macro
using signed type which will be promote to unsigned when invoking
arch_buffer_validate. We need to do an early check.
- We need to check for possible overflow, since a malicious user
application could use a negative number that would be promoted
to a big value that would cause a integer overflow when adding it
to the buffer address, leading to invalid checks.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
The Z_DETECT_POINTER_OVERFLOW() macro is intended detect whether
or not a buffer spans a region of memory that goes beyond the
highest possible address (thereby overflowing the pointer).
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
Always use k_thread_foreach_unlocked with callbacks which print
something out to the shell, as they might call arch_irq_unlock.
Fixes#66660.
Signed-off-by: Benedikt Schmidt <benedikt.schmidt@embedded-solutions.at>
(cherry picked from commit 4c731f27c6)
call k_thread_foreach_unlocked to avoid assertions caused
by calling shell_print while holding a global lock
Signed-off-by: Maxim Adelman <imax@meta.com>
(cherry picked from commit ecf2cb5932)
The controller can implement a reception FIFO as deep as 256 bytes.
However, the computation made by the driver code to determine how many
bytes can be asked is stored in a signed 8-bit variable called rx_empty.
If the reception FIFO depth is greater or equal to 128 bytes and the FIFO
is currently empty, the rx_empty value will be 128 (or more), which
stands for a negative value as the variable is signed.
Thus, the later code checking if the FIFO is full will run while it should
not and exit from the i2c_dw_data_ask() function too early.
This hangs the controller in an infinite loop of interrupt storm because
the interrupt flags are never cleared.
Storing the rx_empty empty on a signed 32-bit variable instead of a 8-bit
one solves the issue and is compliant with the controller hardware
specifications of a maximum FIFO depth of 256 bytes.
It has been agreed with upstream maintainers to change the type of the
variables tx_empty, rx_empty, cnt, rx_buffer_depth and tx_buffer_depth to
plain int because it is most effectively handled by the CPUs. Using 8-bit
or 16-bit variables had no meaning here.
Signed-off-by: Adrien Ricciardi <aricciardi@baylibre.com>
(cherry picked from commit 4824e405cf)
We would drop the received packet if the source address is our
address so tweak the test and make source address different.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
(cherry picked from commit 155e2149f2)
Drop received packet if the source address is the same as
the device address.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
(cherry picked from commit 8d3d48e057)
If we receive a packet where the source address is our own
address, then we should drop it.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
(cherry picked from commit 19392a6d2b)
If we receive a packet from non localhost interface, then
drop it if either source or destination address is a localhost
address.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
(cherry picked from commit 6d41e68352)
GCC_VERSION is defined in a few modules, and those headers are often
included first, so replace the one used in zephyr with
TOOLCHAIN_GCC_VERSION. Do the same with CLANG_VERSION, replacing it with
TOOLCHAIN_CLANG_VERSION.
BUILD_ASSERT is also defined in include/toolchain/common.h, which might
get included before gcc.h. We want to use the gcc-specific one instead
of the general one.
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit c58c76ef0a)
Move extra warning option from generic twister script into
compiler-dependent config files.
ARCMWDT compiler doesn't support extra warning options ex.
"-Wl,--fatal-warnings". To avoid build fails flag
"disable_warnings_as_errors" should be passed to twister.
This allows all warning messages and make atomatic test useles.
Signed-off-by: Nikolay Agishev <agishev@synopsys.com>
(cherry picked from commit 0dec4cf927)
Fixes an issue whereby a board revision is 0 and the overlay file
exists but would not be included
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
the sam xdmac driver does not yet implement the
get_status() function. with this commit the function
will be implemented. Fixes#62003
Signed-off-by: Sven Ginka <sven.ginka@gmail.com>
(cherry picked from commit bc695c6df5)
The spi_nor erase op selection was based on the alignment of the end of
the region to be erased. This prevented larger erase operations being
selected in many cases
Closes#60904
Signed-off-by: Joshua Crawford <joshua.crawford@levno.com>
(cherry picked from commit ea2dd9fc65)
The code is designed to handle RF0L and RF1L in
line 1, but they were being sent to line 0. Becuase
they weren't handled, the interrupts would never
be handled which locked up the chip.
Signed-off-by: Abram Early <abram.early@gmail.com>
Change the CAN controller driver implementations for the
can_remove_rx_filter() API call to be consistent in their
validation of the supplied filter_id.
Fixes: #64398
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
Many releases ago, specifying to block indefinitely in the log
processing thread would do just that.
However, a subtle bug was introduced such that specifying -1
for `CONFIG_LOG_BLOCK_IN_THREAD_TIMEOUT_MS` would have the
exact opposite effect than what was intended.
As per Kconfig, a value of -1 should translate to a timeout of
`K_FOREVER`. However, conversion via `K_MSEC(-1)` results in
a `k_timeout_t` that is equal to `K_NO_WAIT` rather than the
intent which is `K_FOREVER`.
Add a dedicated check to to ensure that a value of -1 is
correctly interpreted as `K_FOREVER` in `log_core.c`.
For reference, the blocking feature was described in #15196,
added in #16194, and it would appear that the regression
happened in c5f2cdef09.
Signed-off-by: Christopher Friedt <cfriedt@meta.com>
(cherry picked from commit 137097f5c3)
Return EMSGSIZE if trying to copy too much data into
user supplied buffer.
Signed-off-by: Jukka Rissanen <jukka.rissanen@nordicsemi.no>
(cherry picked from commit 0a16d5c7c3)
(cherry picked from commit b13b4eb38a50ee02d599332eb99752e814340487)
In case upper layer does not follow the convention, and the net_pkt
provided to the nRF 15.4 driver had a payload larger than the maximum
payload size of an individual 15.4 frame, the driver would end up with
buffer overflow.
Fix this by adding an extra payload_len check before attempting to copy
the payload to the internal buffer.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Rework the can_dlc_to_bytes table lookup code in a way that allow the
compiler to guess the resulting output and somehow fix the build
warning:
zephyr/drivers/can/can_nxp_s32_canxl.c:757:9: warning:
'__builtin___memcpy_chk' forming offset [16, 71] is out of the bounds
[0, 16] of object 'frame' with type 'struct can_frame' [-Warray-bounds]
757 | memcpy(frame->data, msg_data.data, can_dlc_to_bytes(frame->dlc));
where the compiler detects that frame->data is 8 bytes long but
can_dlc_to_bytes could return more than that.
Can be reproduced with:
west build -p -b s32z270dc2_rtu1_r52 \
-T samples/net/sockets/can/sample.net.sockets.can.one_socket
Suggested-by: Martin Jäger <martin@libre.solar>
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
While the maximum sizes were already correctly checked by the code, the
minimum sizes of the PDUs were not. This meant that PDUs smaller than
the minimum required length (typically 6 bytes for AdvA) were
incorrectly forwarded up to the Host.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
(cherry picked from commit 3f0d7012a6)
The timing_max parameters defined in the stm32 bxcan driver don't match the
register description in the reference manuals.
- sjw does have only 2 bits representing 1 to 4 tq.
- phase_seg1 and phase_seg2 max is one tq higher.
I have checked the following reference manuals and all match:
- RM0090: STM32F405, F415, F407, F417, F427, F437 AND F429
- RM0008: STM32F101, F102, F103, F105, F107 advanced arm-based mcus
- RM0351, RM0394: all STM32L4
- RM0091: all STM32F0 with CAN support
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
(cherry picked from commit cec279b5b6)
Convert the ISO-TP SF length check in send_sf() from __ASSERT() to a
runtime check.
Fixes: #61501
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
(cherry picked from commit 1b3d1e01de)
In the tickless kernel mode, the nRF system timer does not schedule
any timeout on initialization. This can lead to a situation that
for certain applications no timeout is scheduled at all (for example,
when an application does not create any threads, it exits `main()`
without any sleeping and only handles interrupts) and in consequence
`sys_clock_announce()` is never called (the nRF system timer calls
this function only from the timeout handler). This in turn causes that
uptime is reported correctly only until the RTC used as the system
timer overflows (what happens after 512 seconds).
Fix this by setting a maximum allowed timeout when initializing
the nRF system timer for the tickless kernel mode.
Signed-off-by: Andrzej Głąbek <andrzej.glabek@nordicsemi.no>
Increase `NET_BUF_USER_DATA_SIZE` value to 8 if `BT_CONN` is enabled.
This is necessary because one of the backported commits adds one struct
member into `struct tx_meta`, and that will be stored in the buffer
user data.
On main, this Kconfig option has been deprecated, hence why this change
was not present in the original patch set.
Signed-off-by: Théo Battrel <theo.battrel@nordicsemi.no>
This commit removes illegal use of NET_BUF_FRAG in friend.c, which is an
internal flag.
Now `struct bt_mesh_friend_seg` keeps pointer to a first received
segment of a segmented message. The rest segments are added as fragments
using net_buf API. Friend Queue keeps only head of the fragments.
When one segment (currently head of fragments) is removed from Friend
Queue, the next segment is added to the queue. Head has always 2
references: one when allocated, another one when added as fragments
head.
Signed-off-by: Pavel Vasilyev <pavel.vasilyev@nordicsemi.no>
(cherry picked from commit 5d059117fd)
This patch reworks how fragments are handled in the net_buf
infrastructure.
In particular, it removes the union around the node and frags members
in the main net_buf structure. This is done so that both can be used at
the same time, at a cost of 4 bytes per net_buf instance.
This implies that the layout of net_buf instances changes whenever
being inserted into a queue (fifo or lifo) or a linked list (slist).
Until now, this is what happened when enqueueing a net_buf with frags
in a queue or linked list:
1.1 Before enqueueing:
+--------+ +--------+ +--------+
|#1 node|\ |#2 node|\ |#3 node|\
| | \ | | \ | | \
| frags |------| frags |------| frags |------NULL
+--------+ +--------+ +--------+
net_buf #1 has 2 fragments, net_bufs #2 and #3. Both the node and frags
pointers (they are the same, since they are unioned) point to the next
fragment.
1.2 After enqueueing:
+--------+ +--------+ +--------+ +--------+ +--------+
|q/slist |-----|#1 node|-----|#2 node|-----|#3 node|-----|q/slist |
|node | | *flag | / | *flag | / | | / |node |
| | | frags |/ | frags |/ | frags |/ | |
+--------+ +--------+ +--------+ +--------+ +--------+
When enqueing a net_buf (in this case #1) that contains fragments, the
current net_buf implementation actually enqueues all the fragments (in
this case #2 and #3) as actual queue/slist items, since node and frags
are one and the same in memory. This makes the enqueuing operation
expensive and it makes it impossible to atomically dequeue. The `*flag`
notation here means that the `flags` member has been set to
`NET_BUF_FRAGS` in order to be able to reconstruct the frags pointers
when dequeuing.
After this patch, the layout changes considerably:
2.1 Before enqueueing:
+--------+ +--------+ +--------+
|#1 node|--NULL |#2 node|--NULL |#3 node|--NULL
| | | | | |
| frags |-------| frags |-------| frags |------NULL
+--------+ +--------+ +--------+
This is very similar to 1.1, except that now node and frags are
different pointers, so node is just set to NULL.
2.2 After enqueueing:
+--------+ +--------+ +--------+
|q/slist |------|#1 node|------|q/slist |
|node | | | |node |
| | | frags | | |
+--------+ +--------+ +--------+
| +--------+ +--------+
| |#2 node|--NULL |#3 node|--NULL
| | | | |
+-----------| frags |-------| frags |------NULL
+--------+ +--------+
When enqueuing net_buf #1, now we only enqueue that very item, instead
of enqueing the frags as well, since now node and frags are separate
pointers. This simplifies the operation and makes it atomic.
Resolves#52718.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
(cherry picked from commit 3d306c181f)
Do these things:
- use the proper macros for reserving the SDU header
- make every L2CAP PDU fragment into 3 ACL packets
- set tx data and verify rx matches the pattern
- measure segment pool usage
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit 8bc094610c)
Only copy the data from the 'parent' buf into the segment buf if we
successfully acquired the HCI semaphore (ie there are available
controller buffers).
This way we avoid pulling and pushing back the data in the case where
there aren't any controller buffers available.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit ca51439cd1)
When there are no buffers, it doesn't make sense to repeatedly try to
send the host TX queue.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit ef19c64f1b)
Make the ACL fragmentation asynchronous, freeing the TX thread for
sending commands when the ACL buffers are full.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit c3e5fabbf1)
This callback allows use-cases where the SDU is much larger than the
l2cap MPS. The stack will then try to allocate using this callback if
specified, and fall-back on using the buffer's pool (previous
behavior).
This way one can define two buffer pools, one with a very large buffer
size, and one with a buffer size >= MPS, and the stack will allocate
from that instead.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit 77e1a9dcad)
See the code comments.
SDUs might enter a state where they will be blocked forever, as a
workaround, we nudge them when another SDU has been sent.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit 8e207fefad)
There was an edge-case where we were sending back too much credits, add
a check so we can't do that anymore.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit 3c1ca93fe8)
This test reproduces more-or-less #34600.
It has a central that connects to multiple peripherals, opens one l2cap
CoC channel per connection, and transmits a few SDUs largely exceeding
the MPS of the channel.
In this commit, the test doesn't pass, but when it passes (after the
subsequent commits), error and warning messages are expected from the
stack, as this is not the happy path.
We can later debate on whether these particular error messages should
be downgraded to debug.
Signed-off-by: Jonathan Rico <jonathan.rico@nordicsemi.no>
(cherry picked from commit 7a6872d837)
Logging module data (including filters) are not accessible by
the user space. Macro for creating logs where creating local
variable with filters before checking is we are in the user
context. It was not used in that case but creating variable
was violating access writes that resulted in failure.
Removing variable creation and using filters directly in the
if clause but after checking condition that it is not the
user context. With this approach data is accessed only in
the kernel mode.
Cherry-picked with modifications from
4ee59e2cdb.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Zephyr mbedTLS was updated to 2.28.x which is a LTS release and
address several vulnerabilities affecting 2.26 (version that used to be
used on Zephyr LTS).
Unfortunately this mbedTLS version is not compatible with TF-M and
backporting mbedTLS fixes was not a viable solution. Due this problem
we are removing TF-M module from Zephyr's LTS. One still can go and add
it to this manifest if needed, but this is no longer "officially"
supported.
More information in:
https://github.com/zephyrproject-rtos/zephyr/issues/56071https://github.com/zephyrproject-rtos/zephyr/pull/54084
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Fix errors like:
inlined from ‘test_mbedtls’ at
zephyrproject/zephyr/tests/crypto/mbedtls/src/mbedtls.c:172:6:
zephyrproject/zephyr/tests/crypto/mbedtls/src/mbedtls.c:96:17: error:
‘test_snprintf’ reading 10 bytes from a region of size 1
[-Werror=stringop-overread]
96 | test_snprintf(1, "", -1) != 0 ||
| ^~~~~~~~~~~~~~~~~~~~~~~~
In GCC >= 11 because `ret_buf` in some calls are shorter literals
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
Fixes: #55490
Follow-up: #53124
Prefix local version of the return variable before calling
`zephyr_check_compiler_flag_hardcoded()`.
This ensures that there will never be any naming collision between named
return argument and the variable name used in later functions when
PARENT_SCOPE is used.
The issue #55490 provided description of situation where the double
de-referencing was not working correctly.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
(cherry picked from commit 599886a9d3)
This commit updates the CI workflows to use the `zephyrproject-rtos`
organisation-level AWS secrets instead of the repository-level secrets.
Using organisation-level secrets allows more centralised management of
the access keys used throughout the GitHub Actions CI infrastructure.
Note that the `AWS_*_ACCESS_KEY_ID` is now stored in plaintext as a
variable instead of a secret because it is equivalent to username and
needs to be identifiable for management and audit purposes.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The test doesn't use eventfd so we can disable it to save some space.
Signed-off-by: Kumar Gala <kumar.gala@intel.com>
(cherry picked from commit 70e921dbc7)
Have eventfd Kconfig select POLL is the code utilizes the polling
API. We get a link error for tests/lib/fdtable/libraries.os.fdtable
when building on arm-clang without this.
Signed-off-by: Kumar Gala <kumar.gala@intel.com>
(cherry picked from commit f215e4494c)
The test does not appear to support 64-bit DMA
* mitigate compiler warning
* support 64-bit addressing mode with `CONFIG_DMA_64BIT`
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 8c6c96715f)
The test does not appear to support 64-bit DMA
* mitigate compiler warning
* support 64-bit addressing mode with `CONFIG_DMA_64BIT`
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 7f6d976916)
The test does not appear to support 64-bit DMA
* mitigate compiler warning
* support 64-bit addressing mode with `CONFIG_DMA_64BIT`
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 5afcac5e14)
Verify that threads are actually schedulable for min and max
scheduler priority for both `SCHED_RR` (preemptive) and
`SCHED_FIFO` (cooperative).
Fixes#56729
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit ad71b78770)
Previously, there was an off-by-one error for SCHED_RR.
Fixes#56729
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 2b2cbf8107)
Compile-out `iproc_pcie_pl330_dma_xfer()` if there are no active
DMA users in devicetree.
Signed-off-by: Christopher Friedt <cfriedt@meta.com>
(cherry picked from commit 9ad78eb60c)
The `config` and `api` members of `struct device` are expected
to be `const`. This also improves reliability, as `config`
and `api` are stored in rom rather than ram, which has the
potential to be corrupted at runtime in the absense of an MMU.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 7212792295)
There are use cases for the pcie_ep driver where we don't
necessarily need the dma functionality. Added ifdef's around
the dma functionality so that it's only available if we
specify the dma engines in the device tree similar to
```
dmas = <&pl330 0>, <&pl330 1>;
dma-names = "txdma", "rxdma";
```
Signed-off-by: Tarun Karuturi <tkaruturi@meta.com>
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 9d95f69a87)
This commit updates the labeler workflow to use the labeler action v4,
which is based on node.js 16 and @actions/core 1.10.0, in preparation
for the upcoming removal of the deprecated GitHub features.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the labeler workflow to use the labeler action v4,
which is based on node.js 16 and @actions/core 1.10.0, in preparation
for the upcoming removal of the deprecated GitHub features.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The previous method used to calculate seconds in `clock_gettime()`
seemed to have an inaccuracy that grew with time causing the
seconds to be off by an order of magnitude when ticks would roll
over.
This change fixes the method used to calculate seconds.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
Add maximum timeout used for conversion to Kconfig. Option is used
to determine which conversion algorithm to use: faster but overflowing
earlier or slower without early overflow.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
(cherry picked from commit 50c7c7b1e4)
Prior to #41602, due to the ordering of operations (first mul,
then div), an intermediate value would overflow, resulting in
a time non-linearity.
This test ensures that time rolls-over properly.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 74c9c0e7a3)
Algorithm was converting uptime to nanoseconds which can easily
lead to overflows. Changed algorithm to use milliseconds and
nanoseconds for remainder only.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
The header can't be fully used in standalone mode: toolchain.h has to be
included first, otherwise the ALWAYS_INLINE attribute is not defined.
Headers that can be directly included and are not self-contained should
be considered a bad practice.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
This commit updates the pull request backport issue check workflow to
use the Ubuntu 22.04 virtual environment.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit cadd6e6fa4)
This commit updates the manifest workflow to use the Ubuntu 22.04
virtual environment.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit af6d77f7a7)
Fixes: #53124
Fix de-referencing of check and exists function arguments by correctly
de-referencing the argument references using `${<var>}`.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
(cherry picked from commit 04a27651ea)
Follow-up: #53124
The PR#53124 fixed an issue where the variable `check` was not properly
dereferenced into the correct variable name for return value storage.
This was corrected in 04a27651ea.
However, some code was passing a return argument as:
`zephyr_check_compiler_flag(... ${check})`
but checking the result like:
`if(${check})`
thus relying on a faulty behavior of code updating `check` and not the
`${check}` variable.
Fix this by updating to use `${${check}}` as that will point to the
correct return value.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
(cherry picked from commit 45b25e5508)
In case the IPv6 context pointer was not set on an interface (for
instance due to IPv6 context shortage), processing the RA message could
lead to a crash (i. e. NULL pointer dereference). Protect against this
by adding NULL pointer check, similarly to other functions in this area.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
(cherry picked from commit c6c2098255)
The net_pkt pointer provided to net pkt commands was not validated in
any way. Therefore it was fairly easy to crash an application by
providing invalid address.
This commit adds the pointer validation. It's checked whether the
pointer provided belongs to any net_pkt pools known to the net stack,
and if the pointer offset within the slab actually points to the
beginning of the net_pkt structure.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
(cherry picked from commit e540a98331)
New LaTeX Docker image (Debian based) uses Python 3.11. On Debian
systems, this version does not allow to install packages to the system
environment using pip. Use a virtual environment instead.
Signed-off-by: Gerard Marull-Paretas <gerard@teslabs.com>
(cherry picked from commit e6d9ff2948)
`rp->le_max_num` was passed unchecked into `k_sem_init()`, this could
lead to the value being uninitialized and an unknown behavior.
To fix that issue, the `rp->le_max_num` value is checked the same way as
`bt_dev.le.acl_mtu` was already checked. The same things has been done
for `rp->acl_max_num` and `rp->iso_max_num` in
`read_buffer_size_v2_complete()` function.
Signed-off-by: Théo Battrel <theo.battrel@nordicsemi.no>
(cherry picked from commit ac3dec5212)
These two test cases both are fault injection test cases,
and there are designed for testing some negative branches
to improve code coverage. But I find that this branch
shouldn't be tested, because the spinlock will be locked
before a procedure performs here, and then it will trigger
an assert error and the process will be rescheduled to the
handler function, and terminated the current test case,
so spinlock will never be unlocked. And it will impact
the next test case in the same test suite(the next testcase
will be never get spinlock).
Signed-off-by: NingX Zhao <ningx.zhao@intel.com>
(cherry picked from commit cb4a629bc8)
This adds the internal function z_work_submit_to_queue(), which
submits the work item to the queue but doesn't force the thread to yield,
compared to the public function k_work_submit_to_queue().
When called from poll.c in the context of k_work_poll events, it ensures
that the thread does not yield in the context of the spinlock of object
that became available.
Fixes#45267
Signed-off-by: Lucas Dietrich <ld.adecy@gmail.com>
(cherry picked from commit 9a848b3ad4)
The work queue has a semi/non-standard reschedule point implemented
using k_yield(), with a check to see if the current thread is
preemptible. Just call z_reschedule_unlocked(), it has this check
internally and is the intended API for this.
Really, this is only a half fix. Ideally the schedule point and the
lock release should be atomic[1] via the more idiomatic
z_reschedule(). But that would take some surgery, so let's go with
the simpler cleanup first.
This also avoids having to duplicate logic that gets added to
reschedule points by an upcoming patch.
[1] So that they represent a condition variable and don't race at the
end. In this case the race is present but benign, since the only thing
we really want to know is that the queue thread gets a chance to run.
The only cost is an occasional duplicated/needless context switch if
two threads are racing on a submit.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
(cherry picked from commit 8d94967ec4)
Condition of work == NULL is checked before, so there is no need to
check it again.
Signed-off-by: Lixin Guo <lixinx.guo@intel.com>
(cherry picked from commit d4826d874e)
Fixes an issue in sys_clock_tick_get() that could lead to drift in
a k_timer handler. The handler is invoked in the timer ISR as a
callback in sys_tick_announce().
1. The handler invokes k_uptime_ticks().
2. k_uptime_ticks() invokes sys_clock_tick_get().
3. sys_clock_tick_get() must call elapsed() and not
sys_clock_elapsed() as we do not want to count any
unannounced ticks that may have elapsed while
processing the timer ISR.
Fixes#46378
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
(cherry picked from commit 71ef669ea4)
Updates sys_clock_announce() such that the <announce_remaining> update
calculation is done after the callback. This prevents another core from
entering the timeout processing loop before the first core leaves it.
Signed-off-by: Peter Mitsis <peter.mitsis@intel.com>
(cherry picked from commit 3e2f30a7ef)
Commit b1182bf83b ("kernel/timeout: Serialize handler callbacks on
SMP") introduced an important fix to timeout handling on
multiprocessor systems, but it did it in a clumsy way by holding a
spinlock across the entire timeout process on all cores (everything
would have to spin until one core finished the list). The lock also
delays any nested interrupts that might otherwise be delivered, which
breaks our nested_irq_offload case on xtensa+SMP (where contra x86,
the "synchronous" interrupt is sensitive to mask state).
Doing this right turns out not to be so hard: take the timeout lock,
check to see if someone is already iterating
(i.e. "announce_remaining" is non-zero), and if so just increment the
ticks to announce and exit. The original cpu will then complete the
full timeout list without blocking any others longer than needed to
check the timeout state.
Fixes#44758
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
(cherry picked from commit 0b2ed3818d)
On multiprocessor systems, it's routine to enter sys_clock_announce()
in parallel (the driver will generally announce zero ticks on all but
one cpu).
When that happens, each call will independently enter the loop over
the timeout list. The access is correctly synchronized, so the list
handling is correct. But the lock is RELEASED around the invocation
of the callback, which means that the individual callbacks may
interleave between cpus. That means that individual
application-provided callbacks may be executed in parallel, which to
the app is indistinguishable from "out of order".
That's surprising and error-prone. Don't do it. Place a secondary
outer spinlock around the announce loop (but not the timeslicing
handling) to correctly serialize the timeout handling on a single cpu.
(It should be noted that this was discovered not because of a timeout
callback race, but because the resulting simultaneous calls to
sys_clock_set_timeout from separate cores seems to cause extremely
high latency excursions on intel_adsp hardware using the cavs_timer
driver. That hardware issue is still poorly understood, but this fix
is desirable regardless.)
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
(cherry picked from commit b1182bf83b)
We can't simply use CLAMP to set the next timeout because
when CONFIG_SYSTEM_CLOCK_SLOPPY_IDLE is set, MAX_WAIT is
a negative number and then CLAMP will be called with
the higher boundary lower the lower boundary.
Fixes#41422
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
(cherry picked from commit 47b7c2e931)
The original design intent with arch_sched_ipi() was that
interprocessor interrupts were fast and easily sent, so to reduce
latency the scheduler should notify other CPUs synchronously when
scheduler state changes.
This tends to result in "storms" of IPIs in some use cases, though.
For example, SOF will enumerate over all cores doing a k_sem_give() to
notify a worker thread pinned to each, each call causing a separate
IPI. Add to that the fact that unlike x86's IO-APIC, the intel_adsp
architecture has targeted/non-broadcast IPIs that need to be repeated
for each core, and suddenly we have an O(N^2) scaling problem in the
number of CPUs.
Instead, batch the "pending" IPIs and send them only at known
scheduling points (end-of-interrupt and swap). This semantically
matches the locations where application code will "expect" to see
other threads run, so arguably is a better choice anyway.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
(cherry picked from commit b4e9ef0691)
Minor cleanup, we had a bunch of duplicated #if logic to send IPIs,
put it all in one place.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
(cherry picked from commit 3267cd327e)
Move coredump_backend_api struct to public header so that custom backends
for coredump can be defined out of tree. Create simple backend in test
directory for verification.
Signed-off-by: Mark Holden <mholden@fb.com>
(cherry picked from commit 7b2b283677)
Let's consider this (simplified) compilation result of a debug build
using -O0 for riscv64:
|__pinned_func
|static inline int k_sem_init(struct k_sem * sem,
| unsigned int initial_count,
| unsigned int limit)
|{
| 80000ad0: 6105 addi sp,sp,32
| 80000ad2: ec06 sd ra,24(sp)
| 80000ad4: e42a sd a0,8(sp)
| 80000ad6: c22e sw a1,4(sp)
| 80000ad8: c032 sw a2,0(sp)
| ret = arch_is_user_context();
| 80000ada: b39ff0ef jal ra,80000612
| if (z_syscall_trap()) {
| 80000ade: c911 beqz a0,80000af2
| return (int) arch_syscall_invoke3(*(uintptr_t *)&sem,
| *(uintptr_t *)&initial_count,
| *(uintptr_t *)&limit,
| K_SYSCALL_K_SEM_INIT);
| 80000ae0: 6522 ld a0,8(sp)
| 80000ae2: 00413583 ld a1,4(sp)
| 80000ae6: 6602 ld a2,0(sp)
| 80000ae8: 0b700693 li a3,183
| [...]
We clearly see the 32-bit values `initial_count` (a1) and `limit` (a2)
being stored in memory with the `sw` (store word) instruction. Then,
according to the source code, the address of those values is casted
as a pointer to uintptr_t values, and that pointer is dereferenced to
get back those values with the `ld` (load double) instruction this time.
In other words, the assembly does exactly what the C code indicates.
This is wrong for 2 reasons:
- The top half of a1 and a2 will contain garbage due to the `ld` used
to retrieve them. Whether or not the top bits will be cleared
eventually depends on the architecture and compiler.
- Regardless of the above, a1 and a2 would be plain wrong on a big
endian system.
- The load of a1 will cause a misaligned trap as it is 4-byte aligned
while `ld` expects a 8-byte alignment.
The above code happens to work properly when compiling with
optimizations enabled as the compiler simplifies the cast and
dereference away, and register content is used as is in that case.
That doesn't make the code any more "correct" though.
The reason for taking the address of an argument and dereference it as an
uintptr_t pointer is most likely done to work around the fact that the
compiler refuses to cast an aggregate value to an integer, even if that
aggregate value is in fact a simple structure wrapping an integer.
So let's fix this code by:
- Removing the pointer dereference roundtrip and associated casts. This
gets rid of all the issues listed above.
- Using a union to perform the type transition which deals with
aggregates perfectly well. The compiler does optimize things to the
same assembly output in the end.
This also makes the compiler happier as those pragmas to shut up warnings
are no longer needed. It should be the same about coverity.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
(cherry picked from commit 1db5c8b948)
With CONFIG_TIMEOUT_64BIT it is both k_timeout_t and k_ticks_t that
need to be split, otherwise many syscalls returning a number of ticks
are being truncated to 32 bits.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
(cherry picked from commit 2cdac33d39)
Some architectures (e.g. RISC-V) has .sdata/.sbss section for small
data/bss. Memory partition should also manage the permission of these
sections in library so they should be put into app_smem.
(For example, newlib _impure_ptr is in .sdata section and
__malloc_top_pad is in .sbss section in RISC-V.)
Signed-off-by: Jim Shu <cwshu@andestech.com>
(cherry picked from commit 46eb3e5fce)
Using the same memory as a user data pointer and FIFO reserved space
could lead to a crash in certain circumstances, those two use cases were
not completely separate.
The crash could happen for example, if an incoming TCP connection was
abruptly closed just after being established. As TCP uses the user data
to notify error condition to the upper layer, the user data pointer
could've been used while the newly allocated context could still be
waiting on the accept queue. This damaged the data area used by the FIFO
and eventually could lead to a crash.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
(cherry picked from commit 2ab11953e3)
k_current_get() may rely on TLS which might not yet be initialized
when those tracing functions are called, resulting in a crash.
This is different from the main branch as in that case the implementation
was completely revamped and neither k_current_get() nor z_current_get()
are used anymore. This is a much simpler fix than a backport of that
code, similar to the implication in commit commit f07df42d49 ("kernel:
make k_current_get() work without syscall").
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
When active, z_libc_partition consumes an MPU region which leaves too
few for some MPU tests. Free up one by disabling HW stack protection.
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit 19c8956946)
K_OBJ_MSGQ, K_OBJ_PIPE, and K_OBJ_STACK objects have pointers
to additional memory that can be allocated. The k_obj_alloc()
returns these objects as uninitialized so when they are freed
there are random opportunities for freeing invalid memory
and causing random faults.
Signed-off-by: David Leach <david.leach@nxp.com>
(cherry picked from commit fdea2a628b)
The K_OBJ_MSGQ object is unitialized so when the thread cleanup occurs
after an expected fault for invalid access the test case can randomly
fault again because the cleanup of the thread will sometimes attempt
to free invalid buffer_start pointer in the msgq object.
Fixes#42705
Signed-off-by: David Leach <david.leach@nxp.com>
(cherry picked from commit a0737e687c)
Because k_thread size in RISCV64 is near 512 bytes, (num_of_thread *
256) bytes heap size is not enough. Enlarge heap size in RISCV64
to the (num_of_thread * 1024) bytes like x86_64 and ARM64.
Signed-off-by: Jim Shu <cwshu09@gmail.com>
(cherry picked from commit e2d67d60ba)
When using THREAD_LOCAL_STORAGE the thread_userspace_local_data stuff
isn't used, so these tests wouldn't build.
Signed-off-by: Keith Packard <keithp@keithp.com>
(cherry picked from commit b03b2e0403)
On SMP, and especially using qemu on a busy system, it is possible for
a thread with a later timeout to get ahead of another one with an
earlier timeout. The tight timeout value difference (10ms) makes it
possible albeit difficult to reproduce. The result is something like:
|START - test_timeout_threads_pend_on_lifo
| thread (q order: 2, t/o: 0, lifo 0x4001d350)
|
| Assertion failed at main.c:140:
|test_multiple_threads_pending: (data->timeout_order not equal to ii)
| *** thread 2 woke up, expected 1
Let's make timeout values 10 times larger to make this unlikely race
even less likely.
While at it... The timeout field in struct timeout_order_data is some ms
value and not a number of ticks, so change the type accordingly.
And leverage k_cyc_to_ms_floor32() to simplify computation in
is_timeout_in_range().
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
(cherry picked from commit a1ce2fb990)
This adds a bunch of k_thread_join() to make sure threads spawned
for a test are no longer running between exiting that test. This
prevents interference between tests if some threads are still
running when assumed not.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
(cherry picked from commit dbe3874079)
Formalize and rework the issue reproducer for #40795 and add it to the
SMP test suite.
Signed-off-by: Carlo Caione <ccaione@baylibre.com>
(cherry picked from commit 8edf9817c0)
This test uses k_yield() to "sync" between threads, so it's implicitly
supposed to run on a single CPU. Make it explicit, to avoid issues on
platforms with more cores.
Signed-off-by: Ederson de Souza <ederson.desouza@intel.com>
FIXKFLOATDISABLE
(cherry picked from commit ab17f69a72)
For historical reasons[1] suspending threads would release the
scheduler lock between pend() (which places the current thread onto a
wait queue) and z_swap() (which effects the context swtich). This
process happens with the caller's lock held, so local interrupts are
masked. But on SMP this opens a tiny race where another CPU could
grab the pended thread and switch to it while we were still executing
on its stack!
Fix this by elevating the "lock swap" code that already exists in the
(portable/switch-based) z_swap() code one level so that it happens in
z_pend_curr() also. Now we hold the scheduler lock between pend and
the final context switch.
Note that this technique can't work for the older z_swap_irqlock()
implementation, which exists to vestigially support a few bits of arch
code (mostly direct interrupts) that don't work on SMP anyway.
Address with an assert to prevent future misuse.
[1] z_swap() is a historical API implemented in per-arch assembly for
older architectures (like ARM32!). It was designed to be called
with what at the time was a global IRQ lock, so it doesn't
understand the idea of a separate scheduler lock. When we finally
get all archictures on arch_switch() this design can be cleaned up
quite a bit.
Signed-off-by: Andy Ross <andyross@google.com>
(cherry picked from commit c32f376e99)
Users may want to do some configuration after the kernel is up, but
before initializing the log_core. Making the log_core's init priority
configurable makes that possible.
Signed-off-by: Ian Oliver <io@amperecomputing.com>
(cherry picked from commit 1675d49b4c)
Previously, there was no test coverage for `sleep()` and
`usleep()`.
This change adds full test coverage.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 027b79ecc4)
The original implementation of `usleep()` was not compliant
to the POSIX spec in 3 ways.
- calling thread may not be suspended (because `k_busy_wait()`
was previously used for short durations)
- if `usecs` > 1000000, previously we did not return -1 or set
`errno` to `EINVAL`
- if interrupted, previously we did not return -1 or set
`errno` to `EINTR`
This change addresses those issues to make `usleep()` more
POSIX-compliant.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 7b95428fa0)
In the case that `sleep()` is interrupted, the POSIX spec requires
it to return the number of "unslept" seconds (i.e. the number of
seconds requested minus the number of seconds actually slept).
Since `k_sleep()` already returns the amount of "unslept" time
in ms, we can simply use that.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit dcfcc6454b)
Using `usleep()` for >= 10000000 microseconds results
in an error, so this test was kind of defective, having
explicitly called `usleep()` for seconds.
Also, check the return values of `clock_gettime()`.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit 23a1f0a672)
Updates mcumgr to resolve an issue with the state of a firmware
update not being reset if an error occurs or if the underlying
area is erased.
Fixes#52247
Backporting commit 4c48b4f21a
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
Pin the types-PyYAML version to 6.0.7. Version 6.0.8 is causing CI
errors for other pull requests, so we need this in to get other PRs
moving.
Fixes: #46286
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
In the new ephemeral Zephyr runners, the cached repository files are
located in a foreign file system and Git clone operation cannot create
hard-links to the cached repository objects, which forces the Git clone
operation to copy the objects from the cache file system to the runner
container file system.
This commit updates the CI workflows to instead perform a "shared
clone" of the cached repository, which allows the cloned repository to
utilise the object database of the cached repository.
While "shared clone" can be often dangerous because the source
repository objects can be deleted, in this case, the source repository
(i.e. cached repository) is mounted as read-only and immutable.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the codecov workflow to pre-clone the Zephyr
repository from the runner repository cache.
Note that the `origin` remote URL is reconfigured to that of the GitHub
Zephyr repository because the checkout action attempts to delete
everything and re-clone otherwise.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The repository clean-up steps are no longer necessary because the new
zephyr-runner is ephemeral and does not contain any files from the
previous runs.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the clang workflow to pre-clone the Zephyr
repository from the runner repository cache.
Note that the `origin` remote URL is reconfigured to that of the GitHub
Zephyr repository because the checkout action attempts to delete
everything and re-clone otherwise.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the clang workflow to use the new Kubernetes-based
zephyr-runner.
Note that the repository cache directory path has been changed for the
new runner.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The repository clean-up steps are no longer necessary because the new
zephyr-runner is ephemeral and does not contain any files from the
previous runs.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the twister workflow to pre-clone the Zephyr
repository from the runner repository cache.
Note that the `origin` remote URL is reconfigured to that of the GitHub
Zephyr repository because the checkout action attempts to delete
everything and re-clone otherwise.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the twister workflow to use the new Kubernetes-
based zephyr-runner.
Note that the repository cache directory path has been changed for the
new runner.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the CI workflows to use the latest "cache" action
v3, which is based on Node.js 16.
Note that Node.js 12-based actions are now deprecated by GitHub and may
stop working in the near future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the CI workflows to use the latest "setup-python"
action v4, which is based on Node.js 16.
Note that Node.js 12-based actions are now deprecated by GitHub and may
stop working in the near future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the CI workflows to use the latest
"upload-artifact" action v3, which is based on Node.js 16.
Note that Node.js 12-based actions are now deprecated by GitHub and may
stop working in the near future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the CI workflows to use the latest "checkout"
action v3, which is based on Node.js 16.
Note that Node.js 12-based actions are now deprecated by GitHub and may
stop working in the near future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the workflow to use the output parameter file
(`GITHUB_OUTPUT`) instead of the stdout-based output parameter setting,
which is now deprecated by GitHub and will be removed in the near
future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the workflow to use the output parameter file
(`GITHUB_OUTPUT`) instead of the stdout-based output parameter setting,
which is now deprecated by GitHub and will be removed in the near
future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the workflow to use the output parameter file
(`GITHUB_OUTPUT`) instead of the stdout-based output parameter setting,
which is now deprecated by GitHub and will be removed in the near
future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the workflow to use the output parameter file
(`GITHUB_OUTPUT`) instead of the stdout-based output parameter setting,
which is now deprecated by GitHub and will be removed in the near
future.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds a concurrency group to the workflow in order to ensure
that only one instance of the workflow runs for an event-ref
combination.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds a concurrency group to the workflow in order to ensure
that only one instance of the workflow runs for an event-ref
combination.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds a concurrency group to the workflow in order to ensure
that only one instance of the workflow runs for an event-ref
combination.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds a concurrency group to the workflow in order to ensure
that only one instance of the workflow runs for an event-ref
combination.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds a concurrency group to the workflow in order to ensure
that only one instance of the workflow runs for an event-ref
combination.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds a concurrency group to the workflow in order to ensure
that only one instance of the workflow runs for an event-ref
combination.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the "Create a release" workflow to use a specific
upload-artifact action version, v3, instead of the latest master branch
in order to prevent any potential breakages due to the newer commits.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the issue count tracker workflow to use a specific
upload-artifact action version, v3, instead of the latest master branch
in order to prevent any potential breakages due to the newer commits.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the documentation build workflow to use a specific
upload-artifact action version, v3, instead of the latest master branch
in order to prevent any potential breakages due to the newer commits.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the compliance check workflow to use a specific
upload-artifact action version, v3, instead of the latest master branch
in order to prevent any potential breakages due to the newer commits.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the backport workflow to use the ubuntu-20.04
runner image because the ubuntu-18.04 image is deprecated and will
become unsupported by December 1, 2022.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the "West Command Tests" workflow to use a specific
runner image version (ubuntu-20.04, macos-11, windows-2022) instead of
the latest version in order to prevent any potential breakages due to
the 'latest' version change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the "Devicetree script tests" workflow to use a
specific runner image version (ubuntu-20.04, macos-11, windows-2022)
instead of the latest version in order to prevent any potential
breakages due to the 'latest' version change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the "Run tests with twister" workflow to use a
specific runner image version, ubuntu-20.04, instead of the latest
version in order to prevent any potential breakages due to the 'latest'
version change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the Twister Testsuite workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the stale issue workflow to use a specific runner
image version, ubuntu-20.04, instead of the latest version in order to
prevent any potential breakages due to the 'latest' version change by
GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the "Create a Release" workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the manifest check workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the license check workflow to use a specific runner
image version, ubuntu-20.04, instead of the latest version in order to
prevent any potential breakages due to the 'latest' version change by
GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the issue count tracker workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the footprint delta workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the footprint tracking workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the error number check workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the documentation build and publish workflows to
use a specific runner image version, ubuntu-20.04, instead of the
latest version in order to prevent any potential breakages due to the
'latest' version change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the daily test version workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the compliance check workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the coding guidelines workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the Clang workflow to use a specific runner image
version, ubuntu-20.04, instead of the latest version in order to
prevent any potential breakages due to the 'latest' version change by
GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the backport issue check workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit updates the Bluetooth tests workflow to use a specific
runner image version, ubuntu-20.04, instead of the latest version in
order to prevent any potential breakages due to the 'latest' version
change by GitHub.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit fixes a stale reference to the 'master' branch when
downloading the issue report configuration file.
Note that the 'master' branch is no longer the default
branch -- 'main' is.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The documentation supports a special target named "html-fast" that skips
generation of all Kconfig pages. Instead, it creates a single dummy page
where a reference to all existing Kconfig options is placed. This means
that references are resolved, but content is not rendered. Since Kconfig
help is rendered as a literal, chances of breaking documentation
build due to Kconfig changes should be low. The change proposed in this
patch should speed up documentation build on pull requests while a
proper solution is found for the Kconfig docs.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Add the documentation build jobs to a concurrency group so that branch
force pushes will automatically cancel in progress jobs.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
When an error happens during the Sphinx build (e.g. due to a broken
reference), the process hangs on CI when run with both `-j auto`
(parallel build) and `-W` (warnings as errors) options. The root cause
of the issue is unknown, and does not seem to happen locally. Parallel
builds are experimental on Sphinx, so they have been disabled on CI for
now. Because CI runner is a single core machine, the build time should
remain equal or similar. The option is still left as default, so local
builds will continue to benefit from parallelization.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Give documentation build up to 30 minutes to finish. This should improve
the user experience when Sphinx hangs due to broken references, a
problem that needs some investigation.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Previously, this change was added to `mutex_error_case`.
That worked fine in `main`, but once the change was backported to
`v2.7-branch`, the test would fail because it *did not* cause a
failure. The reason for that, was that the `mutex_error_case`
suite has `CONFIG_ZTEST_FATAL_HOOK=y`.
With the newer ztest API, it allowed a separate suite to be used,
allowing the test to pass (although it did not really fit in with
the rest of the testsuite).
The solution is to simply merge it with the `mutex_api` suite
which uses non-inverted success logic.
This change will also have to be cherry-picked for the backport
in #49031.
Fixes#48056.
Signed-off-by: Chris Friedt <cfriedt@meta.com>
Say threadA holds a mutex and threadB tries
to lock it with a timeout, a race would occur
if threadA unlock that mutex after threadB
got unpended by sys_clock and before it gets
scheduled and calls k_spin_lock.
This patch fixes this issue by checking the
mutex's status again after k_spin_lock calls.
Fixes#48056
Signed-off-by: Qi Yang <qi.yang@cmind-semi.com>
(cherry picked from commit 89c4a074dc)
This fixes issues with the Bluetooth SMP transport whereby deadlocks
could arise from connection references being held in long-lasting
mcumgr command processing functions.
Note: Heavily modified from original PR due to differences in MCUmgr
operation since the Zephyr 2.7 release.
Fixes#51846
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
(cherry picked from commit 2aca1242f7)
This commit updates the CI documentation build workflow to upload the
HTML pull request documentation builds to the S3 builds.zephyrproject.io
bucket so that they are directly accessible from the web.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit 1dd92ec865)
Using a socketpair for communication in an ISR is not a great
solution, but the implementation should be robust in that case
as well.
It is not acceptible to block in ISR context, so robustness here
means to return -1 to indicate an error, setting errno to `EAGAIN`
(which is synonymous with `EWOULDBLOCK`).
Fixes#25417
Signed-off-by: Chris Friedt <cfriedt@meta.com>
(cherry picked from commit d832b04e96)
This commit updates the CI workflows to limit their event trigger scope
to the v2.7-branch in order to prevent the workflows from running when
the backport branches are pushed.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Fix for issue #41012 to allow compiler to treat
IRQn_Type to be more than 8-bit. This will ensure NVIC numbers
more than 127 (required for MEC172x device) will work
correctly with irq_enable() API
Signed-off-by: Jay Vasanth <jay.vasanth@microchip.com>
(cherry picked from commit 4495f43dca)
Adds the -EBUSY return code to the documentation of the
counter_set_channel_alarm function which is returned when an alarm
is already active.
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
(cherry picked from commit c607599068)
The builtin list function `.sort()` sorts the list in-place and returns
None. As this is an invalid type for iteration, use the builtin `sorted`
function, which returns a sorted copy of the list, which we can iterate
over.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
(cherry picked from commit 06aae61019)
If the is no update from the server, the _links will be NULL.
Check if it is NULL before trying to LOG these strings.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
Use Z_KERNEL_STACK_BUFFER instead of
Z_THREAD_STACK_BUFFER for initial stack.
Fixes#50467
Signed-off-by: Ruud Derwig <Ruud.Derwig@synopsys.com>
(cherry picked from commit 9bccb5cc4b)
...Z_THREAD_STACK_BUFFER.
This is currently a symbolic change as Z_THREAD_STACK_BUFFER
is simply an alias to Z_KERNEL_STACK_BUFFER without userspace,
and Xtensa does not support userspace at the moment.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
(cherry picked from commit b820cde7a9)
...Z_THREAD_STACK_BUFFER.
This is currently a symbolic change as Z_THREAD_STACK_BUFFER
is simply an alias to Z_KERNEL_STACK_BUFFER without userspace,
and Xtensa does not support userspace at the moment.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
(cherry picked from commit 74df88d8f5)
The coredump tests output quite a large amount of data into
the console. However, the ACRN console only has very limited
history (comparatively), such that twister is unable to
match the necessary strings to consider the tests passed.
So skip those tests on acrn_ehl_crb.
Fixes#40887
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
(cherry picked from commit c1ac125068)
Remove a yaml monkeypatch. It is no longer needed since we support 3.6
or later on Zephyr v2.7 LTS and 3.8 or later on what will become v3.2.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
(cherry picked from commit 7ef9c4b20e)
Fixes: #49569
Kconfig requires quoted strings in its configuration files, like this:
> CONFIG_A_STRING="foo bar"
But CMake requires expects that strings are without additional qoutes,
and therefore qoutes are stripped when loading Kconfig config filers
into CMake.
This is particular important when the string in Kconfig is a path to a
file. In this case, not stripping the quotes leads to an error as the
file cannot be found.
When users pass a string to Kconfig through CMake, they are expected to
pass it so that qoutes are correct seen from Kconfig, that is:
> cmake -DCONFIG_A_STRING=\"foo bar\"
In CMake, those qoutes are written as-is to Kconfig extra config file,
and then removed in the CMake cache.
After Kconfig processing, the Kconfig settings are read back to CMake
but without quotes. Settings that was passed through the CMake cache,
for example using `-D` are written back to the cache, but this time
without the qoutes. This results in Kconfig errors on sub-sequent CMake
runs.
Instead of writing the Kconfig value setting back to the CMake cache,
introduce an internal shadow symbol in the cache prefixed with `CLI_`.
This allows the CMake cache to keep the value correctly formatted for
Kconfig extra config creation, while at the same time keep existing
behavior for CONFIG_ symbols read from Kconfig.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
It looks like the latest release, 0.14.2, changed the contents of
JLINK_SDK_NAME as it was before 0.14.0 release. That means that the
previous fix is only applicable to a couple of releases: 0.14.0/0.14.1.
Fixes#49564
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
(cherry picked from commit 2d712c6c55)
pylink 0.14.0 changed the class variable where JLink DLL library name
(libjlinkarm) is stored. This patch adds support for new pylink
libraries while keeping backwards compatibility.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
(cherry picked from commit a57001347f)
Follows the implementation of updatehub and set the
`ai_socktype` only if IPV4/IPV6
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
(cherry picked from commit dd9d6bbb44)
Initialize the `hints` struct to a known value so that it won't
cause undetermined behavior when used in `getaddrinfo()`.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
(cherry picked from commit 2ed88e998a)
This fixes a bug in the sm351lt driver whereby global triggering will
cause an MPU fault due to an unset pointer.
Signed-off-by: Jamie McCrae <jamie.mccrae@nordicsemi.no>
When an L2CAP_CREDIT_BASED_RECONFIGURE_REQ packet is received with
invalid parameters, the recipient shall send an
L2CAP_CREDIT_BASED_RECONFIGURE_RSP PDU with a non-zero Result field
and not change any MTU and MPS values.
This fix incorrectly reconfiguring valid channels while responding with
0x003 (Reconfiguration failed - one or more Destination CIDs invalid)
result code.
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
(cherry picked from commit 253070b76b)
TSE18813 clarified IUT behavior and rejecting reconfiguration which
would result in MTU decrease is enough. There is no need to disconnect
L2CAP channel(s).
This was affecting L2CAP/ECFC/BI-03-C qualification test case
(TCRL 2022-2).
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
(cherry picked from commit 266394dea4)
If the call to net_send_data() fails, for example if the forwading
interface is down, then the pkt will leak. The reference taken by
net_pkt_shallow_clone() will never be released. Fix the problem
by dropping the rerefence count in the error path.
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>
(cherry picked from commit a3cdb2102c)
Rather than stating a version information that will get out of date
at each release, refer to the source of information located in hal_stm32
module.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
(cherry picked from commit 6656607d02)
Check the frame ID type and RTR bit when comparing loopback CAN frames
against installed RX filters.
Fixes: #47904
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
When installing a RX filter, the driver uses "filter->rtr &
filter->rtr_mask" for setting the filter mask. It should just be using
filter->rtr_mask, otherwise filters for non-RTR frames will match RTR
frames as well.
When transmitting a RTR frame, the hardware automatically switches the
mailbox used for TX to RX in order to receive the reply. This, however,
does not match the Zephyr CAN driver model, where mailboxes are
dedicated to either RX or TX. Attempting to reuse the TX mailbox (which
was automatically switched to an RX mailbox by the hardware) fails on
the first call, after which the mailbox is reset and can be reused for
TX. To overcome this, the driver must abort the RX mailbox operation
when the hardware performs the TX to RX switch.
Fixes: #47902
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
The Bosch M_CAN IP does not support RX filtering of the RTR bit, so the
driver handles this bit in software.
If a recevied frame matches a filter with RTR enabled, the RTR bit of
the frame must match that of the filter in order to be passed to the RX
callback function. If the RTR bits do not match the frame must be
dropped.
Improve the readability of the the logic for determining if a frame
should be dropped and add a missing FIFO acknowledge write for dropped
frames.
Fixes: #47204
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
In Python versions >= 3.9, dicts can be merged with the `|` operator.
This is not the case for python versions < 3.9, and the simplest way
is to use `dict_c = {**dict_a, **dict_b}`.
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
(cherry picked from commit 3783cf8353)
This is an automated check for the Backports project to
require one or more `Fixes #<issue>` items in the body
of the pull request.
Fixes#46164
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
(cherry picked from commit aa4e437573)
Created list_backports.py to examine prs applied to a backport
branch and extract associated issues. This is helpful for
adding to release notes.
The script may also be used to ensure that backported changes
also have one or more associated issues.
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
(cherry picked from commit 57762ca12c)
Updated bug_bash.py and list_issues.py to use the GITHUB_TOKEN
environment variable for consistency with other scripts.
Updated bug_bash.py to use `-s / --start-date` instead of
`-b / --begin-date`.
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
(cherry picked from commit 3b3fc27860)
Add a simple test to ensure that we can create and join a
single thread `CONFIG_MAX_PTHREAD_COUNT` * 2 times. If
there are leaks, then `pthread_create()` should
eventually return `EAGAIN`. If there are no leaks, then
the test should pass.
Fixes#47609
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
(cherry picked from commit d37350bc19)
If a thread is joined using `pthread_join()`, then the
internal state would be set to `PTHREAD_EXITED`.
Previously, `pthread_create()` would only consider pthreads
with internal state `PTHREAD_TERMINATED` as candidates for new
threads. However, that causes a descriptor leak.
We should be able to reuse a single thread an infinite number
of times.
Here, we also consider threads with internal state
`PTHREAD_EXITED` as candiates in `pthread_create()`.
Fixes#47609
Signed-off-by: Christopher Friedt <cfriedt@fb.com>
(cherry picked from commit da0398d198)
This commit adds the `qemu_cortex_a53`, which is an MMU-based platform,
as an integration platform for the C++ subsystem tests.
This ensures that the `test_global_static_ctor_dynmem` test, which
verifies that the dynamic memory allocation service is functional
during the global static object constructor invocation, is tested on
an MMU-based platform, which may have a different libc heap
initialisation path.
In addition to the above, this increases the overall test coverage
ensuring that the C++ subsystem is functional on an MMU-based platform
in general.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit 8e6322a21a)
This commit changes the C++ subsystem test, which previously was only
being run with the minimal libc, to be run with all the mainstream C
libraries (minimal libc, newlib, newlib-nano, picolibc).
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit 03f0693125)
This commit adds a test to verify that the dynamic memory allocation
service (the `new` operator) is available and functional when the C++
static global object constructors are invoked called during the system
initialisation.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit dc4895b876)
This commit adds a test to verify that the C++ static global object
constructors are invoked called during the system initialisation.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit 6e0063af29)
This commit changes the invocation of the newlib malloc heap
initialisation function such that it is executed during the POST_KERNEL
phase instead of the APPLICATION phase.
This is necessary in order to ensure that the application
initialisation functions (i.e. the functions called during the
APPLICATIION phase) can make use of the libc heap.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit 43e1c28a25)
This commit changes the invocation of the minimal libc malloc
initialisation function such that it is executed during the POST_KERNEL
phase instead of the APPLICATION phase.
This is necessary in order to ensure that the application
initialisation functions (i.e. the functions called during the
APPLICATIION phase) can make use of the libc heap.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit db0748c462)
In the Zephyr implementation, `sem_timedwait()` uses a
potentially wildly different timebase for comparison via
`k_uptime_get()` (uptime in ms).
The standard specifies `CLOCK_REALTIME`. However, the real-time
clock can be modified to an arbitrary value via clock_settime()
and there is no guarantee that it will always reflect uptime.
This change ensures that `sem_timedwait()` uses a more
consistent timebase for comparison.
Fixes#46807
Signed-off-by: Christopher Friedt <chrisfriedt@gmail.com>
(cherry picked from commit 9d433c89a2)
Remove xtensa specific workaround as this code is now present in Zephyr
SDK cmake code.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
(cherry picked from commit 92a1ca61eb)
With the revert of commit 820d327b46 then
some additional code can be cleaned up.
This removes the final left-overs from Zephyr SDK 0.11.1 support and
older.
It further aligns message printing when including Zephyr SDK toolchain
to other toolchain message printing.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
(cherry picked from commit fb3a113eb8)
This reverts commit 820d327b46.
Commit b973cdc9e8 updated the minimum
required Zephyr SDK version to 0.13.
Therefore revert commit 820d327b46 as
backward support for 0.11.1 and 0.11.2 is no longer required.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
(cherry picked from commit e747fe73cd)
Fix PHY update procedure to handle unsupported PHY requested
by peer central device. PHY update complete will not be
generated to Host, connection is maintained on the old
PHY and the Controller will not respond to PDUs received on
the unsupported PHY.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
(cherry picked from commit 620a5524a5)
When using DMA to transfer over the spi, the spi_stm32_cs_control
is done after enabling the SPI. The same sequence applies
in the transceive_dma function as in transceive function
Signed-off-by: Francois Ramu <francois.ramu@st.com>
This commit fixes an infinite recusion in the
`z_vrfy_i2c_slave_driver_unregister` function.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit 745b7d202e)
Previous check in the if-statement would never allow to send last
segment if msg->len + 2 == MTU * x.
Signed-off-by: Pavel Vasilyev <pavel.vasilyev@nordicsemi.no>
(cherry picked from commit 1efce43a00)
When receiving Transaction Start PDU, assure that number of segments
needed to send a Provisioning PDU with TotalLength size is equal to SegN
value provided in the Transaction Start PDU.
Signed-off-by: Pavel Vasilyev <pavel.vasilyev@nordicsemi.no>
(cherry picked from commit a63c515679)
There is potential buffer overflow in pb adv.
If Transaction Continuation PDU comes before
Transaction Start PDU the last segment number is set to 0xff.
The current implementation has a strictly limited buffer size.
It is possible to receive malformed frame with wrong segment
number. All segments with number 2 and above will be stored
in the memory behind Rx buffer.
Signed-off-by: Aleksandr Khromykh <Aleksandr.Khromykh@nordicsemi.no>
(cherry picked from commit 6896075b62)
If CONFIG_SHELL_STATS is disabled, shell->stats is NULL and
must not be dereferenced. Guard against it.
Fixes#44089
Signed-off-by: Alexej Rempel <Alexej.Rempel@de.eckerle-gruppe.com>
(cherry picked from commit 476d199752)
ATT_READ_BY_TYPE_RSP returns Attribute Data List so handle it in the
application.
This affects GATT/CL/GAR/BV-03-C.
Signed-off-by: Michał Narajowski <michal.narajowski@codecoup.pl>
(cherry picked from commit 54cd46ac68)
Adjust get_mem_region to not return region when address == end
as there will be nothing to read there. Also, a subsequent region
may have that address as a start address and would be a more appropriate
selection.
Signed-off-by: Mark Holden <mholden@fb.com>
(cherry picked from commit d04ab82943)
sw_switch implementation uses two parallel groups of
PPIs connecting radio and timer tasks and events.
The groups are used interchaneably, one is set for
following radio TX/RX event while the other is in use
(enabled).
The group should be disabled by timer compare event that
starts Radio to TX/RX a PDU. The timer is responsible for
maintenance of TIFS/TMAFS. The disabled group collects
all PPIs required to maintain the TIFS/TMASF. After
the time is reached Radio is started and the group is
disabled. It will be enabled again by software radio
swich during next call.
If the group is not disabled then it will work in parallel
to other one. That causes issues in correct maintenance of
instant when radio shoudl be started for next TX/RX event
e.g. radio may be enabled to early.
In case the PHY CODED was enabled and periodic advertising
included chained PDUs, that are transmitted back-to-back,
there was missing group delay disable. The missing case was
sw_switch function called with dir_curr and dir_next set
to SW_SWITCH_TX.
Signed-off-by: Piotr Pryga <piotr.pryga@nordicsemi.no>
Add support for missing EHL SKUs. The information about SKUs is
already public and available in Linux kernel:
https://github.com/torvalds/linux/blob/
38f80f42147ff658aff218edb0a88c37e58bf44f/drivers/edac/
igen6_edac.c#L197-L208
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
(cherry picked from commit f6069aa8fa)
When "eager FPU sharing" mode is enabled, FPU registers must be
initialised at the time of thread creation because the floating-point
context is always active and no further FPU initialisation is performed
later.
Note that, in case of the "lazy FPU sharing" mode, floating-point
context is inactive by default and the FPU is initialised when the
first floating-point instruction is executed.
Refer to the issue #44902 for more details.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
(cherry picked from commit f9a3f02b86)
Virtual MSD J-Link support on some development boards has caused an
issue with SMP due to limiting the maximum size of UART data via the CDC
endpoint, add a link to the SMP documentation and smp_svr sample
application on how to disable MSD functionality and resolve the issue.
Signed-off-by: Jamie McCrae <jamie.mccrae@lairdconnect.com>
Increase CONFIG_MAX_THREAD_BYTES to 3 to fix the build issue on
tests/drivers/build_all/modem/ test case.
Signed-off-by: Jaxson Han <jaxson.han@arm.com>
Change-Id: I6a26955bdd7e8b176894fa8246aec63a3a3db05f
(cherry picked from commit a483828d46)
The current minimum required version of "Armv8-R AEM FVP" is 11.16.16.
Signed-off-by: Jaxson Han <jaxson.han@arm.com>
Change-Id: Iaaf1ec7fd432753371b58d13fdd29b1278f6c997
(cherry picked from commit 3875a1e787)
After the fix of FVP_BaseR_AEMv8R booting issue, the minimum required
version of FVP will be 11.16.16. Add an FVP minimal required version
check in building time.
When the ARMFVP_MIN_VERSION is set in board cmake file, the version
check will be enabled and print a warning.
Signed-off-by: Jaxson Han <jaxson.han@arm.com>
Change-Id: Ibbade0c328b5e91b8830fb35cba6917f08aabbda
(cherry picked from commit 3c1f3197e2)
In the Armv8R AArch64 profile[1], the Armv8R AArch64 is always in secure
mode. But the FVP_BaseR_AEMv8R before version 11.16.16 doesn't strictly
follow this rule. It still has some non-secure registers
(e.g. CNTHP_CTL_EL2).
Since version 11.16.16, the FVP_BaseR_AEMv8R has fixed this issue. The
CNTHP_XXX_EL2 registers have been changed to CNTHPS_XXX_EL2. So the
FVP_BaseR_AEMv8R (version >= 11.16.16) cannot boot Zephyr. This patch
will fix it.
[1] https://developer.arm.com/documentation/ddi0600/latest/
Signed-off-by: Jaxson Han <jaxson.han@arm.com>
Change-Id: If986f34dc080ae7a8b226bba589b6fe616a4260b
(cherry picked from commit fd231e32e9)
Reason why the prority was at its lowest is unknown, but now that it may
be used to send local packets (which used to be sent right away),
it seems to affect TCP scheduling in loopback mode. Raising the prority
so it matches how it was previously (i.e. sent right away) should fix
things. (Note however that this issue was not broadly present, only
sockets.tls test seemed to be affected.)
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Client thread might run before the server gets to put itself on accept.
Leading to the server waiting forever.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
TCP work queue is of higher priority so k_yield should do the trick, and
the test will not be affected by any timing.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Due to the previous change on when to send TCP packet on local IP, pkt
may be held in a queue which is to run on a k_work. This changes the
scheduling, and due to that one of the test is failing to allocate a
RX net_pkt at the time it wants to. (previous TCP connection is not yet
fully closed and still own PKT that new connection cannot get then).
Of course all those waiting paquets require buffers so raising them.
It was verified that there is no leak, adding net_pkt_print() at
tcp_conn_unref() shows that when all tcp connection are finally unrefed:
all net_pkt get freed as well.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Closing a connection, thus calling net_context_put() will not close a
TCP connection properly, and will leak tcp connection memory.
This is because: net_context_put calls net_context_unref which calls
net_tcp_unref which leads to unref tcp connection and thus sets
ctx->tcp to NULL. Back to net_context_put, that one finally calls
net_tcp_put: but that bails out directly since ctx->tcp is NULL.
Fixing it by inverting net_tcp_put() and net_context_unref() calls
within net_context_put().
Fixes#38598
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
On any target, running a TCP server and a net shell can show the issue:
net tcp connect local_ip port
will fail. Usally it ends up by consumming all tcp connection memory.
This is because in tcp_in(), state changes will most of the time lead to
sending SYN/ACK/etc... packets under the same thread, which will run all
through net_send_data(), back to tcp_in(). Thus a forever loop on SYN ->
SYN|ACK -> SYN -> SYN|ACK until tcp connection cannot be allocated
anymore.
Fixing it by scheduling any local packet to be sent on the queue.
Fixes#38576
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
This limits non-mcuboot builds to have a maximum size of 892KB to
prevent code being placed over the top of the bootloader's flash area.
Signed-off-by: Jamie McCrae <jamie.mccrae@lairdconnect.com>
- Supporting multiple instances of ivShMem virtual devices.
- Introduces DT based configuration for ivShMem devices.
- Add DTS overlay file to test new multiple ivshmem instance capability.
- Enable BDF unspecified device initialization.
(limited to one instance. An improved version of pcie_bdf_lookup()
will come soon that fixes this limitation)
- Make PCIE DTS file macros available for a proper ivshmem device
properties parsing.
Sample for dts file:
pcie0 {
label = "PCIE_0";
#address-cells = <1>;
#size-cells = <1>;
compatible = "intel,pcie";
ranges;
ivshmem0: ivshmem@800 {
compatible = "qemu,ivshmem";
reg = <PCIE_BDF_NONE PCIE_ID(0x1af4,0x1110)>;
label = "IVSHMEM";
status = "okay";
};
};
Signed-off-by: Michael Schmidt <michael1.schmidt@intel.com>
This fixes an issue with the filesystem mcumgr being registered twice
in the sample application which resolves an issue with an endless loop
if a mcumgr handler is used which is not registered.
Signed-off-by: Jamie McCrae <jamie.mccrae@lairdconnect.com>
New Sphinx version (or docutils) has slightly changed the output format
for code documentation directives. These changes try to mimic previous
behavior, even though it does not achieve 100% equal result. In some
cases the new default style does not require further tweaks, and in some
others styling as before is not possible.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
breathe: for simplicity, require versions > 4.30 (lower versions have
known issues, so do not take risks).
Sphinx: start requiring versions >=4.x. Keep with compatible versions,
since Sphinx major updrages can easily break extensions, themes, etc.
sphinx_rtd_theme: upgrade to >=1.x. Again, keep with compatible versions
since we have style customizations that can likely break on major
upgrades.
pygments: Allow any version >=2.9 (version that introduced DT support).
We do not have strong compatibility requirements here.
sphinx-notfound-page: Remove any requirements, we do not have strong
requirements for this one.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Use of stm32cubeprogrammer runner makes life easier
on test bench for this device.
Though, keep openocd as default.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
In order to enable use of stm32cubeprogrammer runner with twister,
add "sn" tool specific option which allows to provide target serial
number and hence select the target to flash when multiple ones are
connected to the host.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
autopts was updated to properly require support for Accept Filter List
in Auto Connection Establishment Procedure related tests. This patch
enabled support for it and adds required BTP support.
This was affecting following qualification test cases:
GAP/CONN/ACEP/BV-03-C
GAP/CONN/ACEP/BV-04-C
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
NSEC_PER_SEC is an unsigned integer macro. Thus, -NSEC_PER_SEC will be
treated as unsigned integer as well which lead to calculation error on
64bits integer variables. Added the correct type casting into the formula
to fix the calculation error.
Signed-off-by: Kweh Hock Leong <hock.leong.kweh@intel.com>
The test is using device and device runtime power management. Just
including them to the test build.
Signed-off-by: Flavio Ceolin <flavio.ceolin@intel.com>
We don't set core numbers for mdb-hw runner for nSIM
board, so it defaults to 1, so mdb-hw runner doesn't work with
SMP boards.
Fix that.
Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
Signed-off-by: Evgeniy Paltsev <PaltsevEvgeniy@gmail.com>
Closing a listening socket will set the accept callback to NULL.
This could lead to a crash, in case an already received packet,
finalizing the connection handshake, was processed after the socket was
closed. Thereby, it's needed to verify if the callback is actually set
before processing it.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
The verification function for accept() did not take into account that
addr and addrlen pointers provided could be NULL.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Claim the net_context mutext associated with a socket before claiming
the socket mutex. The receive callback claims the net_context mutex
internally, which will now always succeed immediately.
The TX path claims the net_context mutex before the socket mutex, and if
we don't use the same order, we can end up in a deadlock.
Fixes#43470.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
TCP module can report EAGAIN in case TX window is full. This should not
be forwarded to the application, as blocking socket is not supposed to
return EAGAIN.
Fix this for sendmsg by implementing the same mechanism for handling TX
errors as for regular send/sendto operations.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
This was affecting following qualification test cases:
GAP/SEC/SEM/BV-56-C
GAP/SEC/SEM/BV-57-C
GAP/SEC/SEM/BV-58-C
GAP/SEC/SEM/BV-59-C
GAP/SEC/SEM/BV-60-C
GAP/SEC/SEM/BV-61-C
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
Core Specification 5.3 clarified security requirements for GATT client
when handling incoming notifications and indications.
Vol 3: Part C: 10.3.2.2:
"...Since the configuration is persistent across a disconnection and
reconnection, the client shall check the security requirements against
the configuration upon a reconnection before processing any indications
or notifications from the server. Any notifications received before
the security requirements are met shall be ignored. Any indications
received before the security requirements are met shall be confirmed
and then discarded. ..."
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
Only run the `uninit` function if the SPI instance has previously been
configured. This stops an assertion in the HAL drivers from triggering
due to running `uninit` without a previous `init`.
Fixes#42299.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
ll_adv_set stores poitner to direction finding TX configuration.
When ll_reset is executed the pointer was not NULL assigned.
That lead to erroneous behavior e.g. df_cfg->is_enabled was set
to TRUE even the functionality was not enabled.
DF configuration is stored in memory pool. The memory pool uses
free elements to store its internal data. On reset whole pool is
expected to be free, so ll_adv_set->df_cfg may not point to any
element allocated from memory pool.
Signed-off-by: Piotr Pryga <piotr.pryga@nordicsemi.no>
When Zephyr runs directly on actual hardware, it will be always
directing MSI messages to BSP (BootStrap Processor). This was fine until
Zephyr could be ran on virtualizor that may NOT run it on BSP.
So directing MSI messages on current processor. If Zephyr runs on actual
hardware, it will be BSP since such setup is always made at boot time by
the BSP. On other use case it will be whatever is relevant at that time.
Fixes#43853
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Depending on whether X2APIC is enabled or not, it will be safer to grab
such ID from the right place.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
This will centralize CPUID related accessors. There was no need for it
so far, but this is going to change.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
We have single server but it can accept multiple connections so
keep same requirements for all connection attempts.
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
The problem with append is that when doing incremental (non-clean)
build, the content of these files are appended each time to the previous
leftover content.
Example:
# west build -p -b qemu_cortex_m3 \
samples/application_development/code_relocation/
# // disable for example CONFIG_XIP
# west build -b qemu_cortex_m3 \
samples/application_development/code_relocation/
at this point the content of the generated files (i.e.
linker_relocate.ld) is wrong.
Signed-off-by: Carlo Caione <ccaione@baylibre.com>
.eh_frame section should not be removed directly in the
hex format and bin format output, it should be based on
whether we need exception handler feature.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
The response of a KTCPSND has two phases. According to documentation by
wireless the timeout is 60 seconds. The fix respects the timeout on the
second phase, too (waiting for OK or errors from modem). Previously only
the first phase used 60 seconds and the second phase used 5 seconds.
Without this fix the hl7800 will lock the tcp stack for the current
socket indefinitely if another socket operation is performed before the
response from the modem is received.
Additionally all timeouts are adjusted to be at least one second longer
as the documented timeout from wireless. This avoids races between the
hl7800 and the driver.
Signed-off-by: Rene Bredlau <git@unrelated.de>
If HCI packet length is greater than endpoint MPS or currently
received data block (over USB), next block could be larger
than allocated net_buf buffer.
Check buffer tailroom before copying data using net_buf_add_mem().
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
Function bt_buf_get_tx(), which is used to allocate buffer from
fixed-size pool, does not check size argument before copying
the data with the length size into fixed-size buffer, wich may
not be large enough.
Check immediately before copying if the tailroom of the buffer
is large enough.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
It was noted that `pthread_cond_wait()` would always return
ETIMEDOUT, even when successful (and no timeout should ever
occur with `K_FOREVER`).
The z_sched_wake() / z_sched_wake_all() / z_sched_wait() API
are used here with a swap return value of 0 to indicate
success.
Fixes#41284
Signed-off-by: Christopher Friedt <chrisfriedt@gmail.com>
Split Bluetooth tests workflow into 2 steps:
- One that runs the actual tests and stored results
- A second one that fecthes and uploads tests results
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
With this setting enabled, Git credentials are not kept after checkout.
Credentials are not necessary after the checkout step since we do not
do any further manual push/pull operations.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Correct DT property to set correct STM32_PLL_XTPRE value.
The driver bindings defined `xtpre` instead of used `xtre`
in the `DT_PROP` macro.
That allows to use F1 PLL clock with division by 2.
Signed-off-by: Georgij Cernysiov <geo.cgv@gmail.com>
The current driver doesn't handle the LBD flag, this leads
uart_stm32_err_check to return always true if a Line Break
is detected.
This PR adds Line Break Detection and the related flag clearing,
F0 series it's excluded from the changes.
Fixes zephyrproject-rtos#41339
Signed-off-by: Andrea Campanella <andrea.campanella@helvar.com>
drivers.flash.nrf_qspi_nor and drivers.flash.soc_flash_nrf:
keep nrf52840dk_nrf52840 as integration_platform
drivers.flash.default:
Use mimxrt1060_evk instead as integration_platform,
this is the only platform allowed.
drivers.flash.stm32:
Add a bunch of boards as integration_platforms for
this test configuration.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
This commit fixes dual bank flash handling on stm32g0 targets.
In contrast to other Series (G4, L5) the flash page size does not change
in single bank configuration (2KiB in both configurations).
nSWAP_BANK:
While the reference manual(RM) only documents:
"This bit selects the bank that is the subject of empty check upon boot"
as expected, this behaves similar to BFB2 on G4 and SWAP_BANK on L5.
It has been observed that this bit swaps the address mapping of bank1
and bank2, regardless of DUAL_BANK bit being set or not.
As documented in the RM the nSWAP_BANK bit is ignored when the BOOT_LOOK
bit is set. This applies to the empty check as well as the address
mapping.
On this Series FLASH_CR_BKER must be set in single-bank as well as
dual-bank configuration for erase operations on bank2 regardless of
the swap status.
On a G0B1RE (dev-id: 0x467) I could not observe a difference between
DUAL_BANK flash option bit set and not.
It this may be different on 256KiB Flash targets.
The HAL indicates that "FLASH_SALES_TYPE_0" only uses a single bank if
OB_DUAL_BANK_VALUE is not set, but as I don't know which SoC this is
and I can't test the behaviour and the driver does not take this into
account.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
This commit makes no functional changes, it only refactors the
driver such that dual bank flash handling can be easily added.
Instead of using HAL macros directly in the code, new macros
with STM32G0 prefix are defined.
The erase_page function gets passed the offset instead of the page,
and the FLASH CR reg is written once with all erase parameters.
flash_stm32_wait_flash_idle is already called before each
write to CR, consequently it is also made sure CFGBSY flag
is not set.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
Some series (namely g0, u5, wb, wl, ?) use CFGBSY to indicate
that FLASH_CR is not ready to be modfied.
This commit adds this flag additionally to other the flash busy flags,
in flash_stm32_wait_flash_idle such that the driver waits before
trying to modify PG, PNB[6:0], PER, and MER bits in FLASH_CR.
Additionally, dual bank variants of STM32G0 have a seperarate BSY2 flag
for flash bank two.
Until now this was not yet checked in flash_stm32_wait_flash_idle.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
In STM32G0 HAL FLASH_FLAG_xxx defines don't follow the pattern of
other Series to simply redefine the FLASH_SR_xxx Msk.
Instead an ID for the SR reg and the position of the Error flag
are defined.
As a result error checking in flash_stm32_check_status was not working
until this fix on stm32g0 series.
In order to avoid complexity in the driver, the ifdef-ery of the flags
was moved to the header file.
Other series except g0 use FLASH_FLAG_xxx defines, because those
are valid for both cores in dual core(wl) and in secure/non-secure
targets(l5,u5).
FLASH_STM32_SR_ERRORS mask is introduced to check for any active error
in the SR.
The flags for SIZERR, MISERR, FASTERR are newly introduced.
the latter two are only required once fast programming is used,
which is not yet the case for any series.
The FLASH_SR_OPTVERR flag (option validity flag) is also present
in the SR, but is not added.
Also ecc errors are generally not checked, but these are in a different
register.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
An attempt to simplify the ifdef-ery around FLASH_SR is made.
Define Registers and flags in the header file instead of including
several individual operations in the driver.
FLASH_FLAG_BSY is not only defined for STM32L5, but also other series.
Therefore use this flag instead of FLASH_SR_BSY.
Only the g0 series definition is not valid in our context,
therefore use FLASH_SR_BSY1 instead.
No functional changes, only refactoring.
Signed-off-by: Thomas Stranger <thomas.stranger@outlook.com>
Current driver set a fixed prescaler value for the lpuart
that caused certain baudrate configurations to fail due to
LPUARTDIV overflow the LPUART_BRR register.
This PR attempt to calculate a suitable PRESCALER for the
selected baudrate, throws error and return if it couldn't get
an optimal one.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
Some tests like:
tests/kernel/sched/metairq/kernel.scheduler.metairq
tests/kernel/profiling/profiling_api/kernel.common.profiling
tests/kernel/sched/schedule_api/kernel.scheduler
tests/kernel/sched/schedule_api/kernel.scheduler.multiq
tests/kernel/profiling/profiling_api/kernel.common.profiling
tests/kernel/workq/work_queue/kernel.workqueue
don't support that the current thread change when writing a message with
printk (which uses poll_out). So, we remove the call to k_yield which is
useful only for optimizing cpu usage by forcing a thread change if the
usart send stack is full.
Signed-off-by: Julien D'ascenzio <julien.dascenzio@paratronic.fr>
A dead lock could happen if 2 threads with differents priorities use
poll_out. In fact, the lock data->tx_lock could be lock by a thread with
lower priority and then a thread with higher priority can't take the
lock. There was a race condition here:
/* Wait for TXE flag to be raised */
while (1) {
if (atomic_cas(&data->tx_lock, 0, 1)) {
/* !!!!!!!! RACE CONDITION !!!!!!!!!!!!!!
if (LL_USART_IsActiveFlag_TXE(UartInstance)) {
break;
}
atomic_set(&data->tx_lock, 0);
}
}
To fix race condition, the interrupts are locked in poll_out.
Signed-off-by: Julien D'ascenzio <julien.dascenzio@paratronic.fr>
A lock was added to manage situation where the API poll_out and irq API
are used in same time.
Signed-off-by: Julien D'ascenzio <julien.dascenzio@paratronic.fr>
The strtol() function use errno to return error code.
However, it is not being initialized in the parser_arg()
function before calling the strtol(). Thus, hitting error
when performing net ping command with -c / -i parameters.
Signed-off-by: Kweh Hock Leong <hock.leong.kweh@intel.com>
Transports should close the socket in case of `setsockopt()` failure,
otherwise we end up with a leaked socket, as it won't be closed
elsewhere.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
There was a problem with source address selection for ARP
retransmissions, when an ARP entry was already pending. In such case,
the `entry` value passed to `arp_prepare()` is NULL, which in result
caused the `current_ip` variable being used as the source value. The
problem with this approach is, that the `current_ip` is only set in
IPv4 autoconf, the Ethernet L2 does not set this variable. In result,
every retransmission of an ARP packet was sent with unspecified source
address, preventing the response from being handled.
Fix this by partially restoring the behaviour of the ARP source address
assignment from before IPv4 autoconf was introduced. If the ARP is sent
by the IPv4 autoconf, use the `current_ip` value provided. If entry is
not set, use the source IPv4 address set in the actual data packet.
Otherwise, search for a source address on the interface corresponding to
the `entry`.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Fix net_pkt leak by increasing net_context the reference count earlier
in the zsock_accepted_cb() with instalment of the
zsock_received_cb() callback.
And consequently flushing recv_q and decrement net_context
reference count if zsock_accept_ctx() fails.
Signed-off-by: Daniel Nejezchleb <dnejezchleb@hwg.cz>
Added locking to posix read(), write(), close()
for additional protection.
In read() missing lock would create uneven calls to locking
mechanism in sockets.c after k_condvar_wait().
That results in socket lock not ever being unlocked
Signed-off-by: Daniel Nejezchleb <dnejezchleb@hwg.cz>
This reverts commit e7489d8de7.
And fixes the deadlock by allowing only 1 thread to actualy clean up
the connection when the ref_count is 0.
Signed-off-by: Daniel Nejezchleb <dnejezchleb@hwg.cz>
Increments send retry every time
after the tcp_send_data when resending.
That way unhandled return values can time
out after set amount of tcp_retries.
Signed-off-by: Daniel Nejezchleb <dnejezchleb@hwg.cz>
Log_strdup was used in NET_ASSERT macro which is not logging.
As a result linking fails when logging is disabled but asserts
are enabled.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
For instance, DHCP (UDP protocol) can send broadcasted packet and if we
no not serve the requested destination port, let's not send an error
back.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Invert src/dst strings, the icmpv4 error is sent from the dst (us) to
src (sender of the ipv4 packet that generated the error).
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
When multiple shields are defined, only the shield last in the -DSHIELD
list gets defined in `.config`. This is due to too many backslashes
used defining it for an env setting.
Signed-off-by: Ryan McClelland <ryanmcclelland@fb.com>
Calling gettimeofday() from _gettimeofday() in a non-Posix build
environment can result in a recursive call loop, causing a stack
overflow. Modify _gettimeofday() to return -1 for non-posix systems
(the previous behaviour that was added in #22508).
Fixes#41095
Signed-off-by: Binu Jacob <bjj@planetinnovation.com.au>
This fix a typo where incorrect member of bt_keys was used for
checking if LTK is present. This was resulting in bogus results
depending on connection role and current identity used.
This was affecting L2CAP/LE/CFC/BV-25-C qualification test case.
Fixes: #42862
Signed-off-by: Szymon Janc <szymon.janc@codecoup.pl>
There are two macros for declaring stack arrays:
K_KERNEL_STACK_ARRAY_DEFINE:
Defines the array, allocating storage and setting the section name
K_KERNEL_STACK_ARRAY_EXTERN
Declares the name of a stack array allowing code to reference
the array which must be defined elsewhere
arch/arm/include/aarch32/cortex_m/stack.h was mis-using
K_KERNEL_STACK_ARRAY_DEFINE to declare z_interrupt_stacks by sticking
'extern' in front of the macro use. However, when this macro also set
the object file section for the symbol, having two of those caused a
conflict in the compiler due to the automatic unique name mechanism used
for sections to allow unused symbols to be discarded during linking.
This patch makes the header use the correct macro.
Signed-off-by: Keith Packard <keithp@keithp.com>
Use sys_clock_hw_cycles_per_sec() instead of
CONFIG_SYS_CLOCK_HW_CYCLES_PER_SEC to determine clock cycles.
Signed-off-by: Michel Haber <michel-haber@hotmail.com>
Clear ADRDY before enabling ADC to ensure the subsequent
wait for ADRDY does not stop prematurely in case ADRDY
was already set.
The "ADC on-off control" sections of the following reference manuals
were consulted. That gives at least one instance per series affected
by this change, even if not every affected MCU is covered.
- RM0438 (STM32L552xx and STM32L562xx)
- RM0351 (STM32L47xxx, STM32L48xxx, STM32L49xxx and STM32L4Axxx)
- RM0434 (STM32WB55xx and STM32WB35xx)
- RM0454 (STM32G0x0)
- RM0440 (STM32G4 Series)
- RM0399 (STM32H745/755 and STM32H747/757)
- RM0433 (STM32H742, STM32H743/753 and STM32H750)
- RM0453 (STM32WL5x)
Signed-off-by: Alexander Mihajlovic <a@abxy.se>
Use a wrapper for LL_ADC_Enable that also waits for ADRDY if required
by the SoC to make sure it's properly enabled everywhere this is done.
Signed-off-by: Alexander Mihajlovic <a@abxy.se>
RM 0453: the sw is allowed to write the Oversampling
ratio or shift of the ADC Config.Reg.2 only when ADSTART = 0
(no conversion is on-going). So disabling it will be stopped.
Signed-off-by: Francois Ramu <francois.ramu@st.com>
RM 0453: the sw is allowed to write the Data Resolution bits
of the ADC Config.Reg.1 only when ADEN = 0 (ADC disable).
Signed-off-by: Francois Ramu <francois.ramu@st.com>
The STM32 docs state that the ADC may not be calibrated unless
the ADC is disabled (ADEN=0). This commit implements this constraint
Fixes#40936
Signed-off-by: Pete Dietl <petedietl@gmail.com>
The previous approach to detect if the underlying transport was closed
(by checking the return value of `mbedtls_ssl_read()` was not right,
since the function call does not request any data - therefore 0 as a
return value is perfectly fine.
Instead, rely on the underlying transport ZSOCK_POLLHUP event - if it
reports that the connection ended, forward the event to the application.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Report ZSOCK_POLLHUP event if peer closed the connection, and thus the
socket is in EOF state.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
When message is dropped then log_process is called with
bypass flag set and additionally z_log_dropped() is called.
In both functions counter of buffered messages was decremented.
That resulted in counter being decremented twice. It resulted
in logging misbehavior after messages being dropped (delayed
processing). Fixing it by decrementing the counter in log_process
only when bypass flag is not set.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Use value returned by atomic_inc to decide on action.
Previously direct value was used and that could lead to
delays in logging processing because thread waking up
could be mishandled.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
A common pattern here was to take the work item as the subfield of a
containing object. But the contained field is not a k_work, it's a
k_work_delayable.
Things were working only because the work field was first, so the
pointers had the same value. Do things right and fix things to
produce correct code if/when that field ever moves within delayable.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
Instead of redefining own `struct zsock_timeval` type at the socket
layer, use a standard type provided by libc. This prevents the
compliation errors when application includes both, `net/socket.h` and
standard C header defining `struct timeval` (sys/time.h).
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Verify the `ref` value of the neighbor entry in `net_route_foreach()`,
so that it only executes the callback on routes in use. Otherwise, the
function could call the callback for the route that has already been
deleted.
This could be encountered when executing `net route` shell command,
which printed the already deleted routes along the existing ones.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
When new route is allocated, a corresponding NBR entry (containing
`struct net_route_nexthop` data) is allocated from
`net_route_nexthop_pool`. When the route was deleted however, the entry
was not freed - only the "core" neighbor entry from the neighbor
management module (nbr.c) was dereferenced. This lead to a resource
leak, effecitevly leaking one `net_route_nexthop_pool` entry for each
deleted route.
Fix this, by defreferencing the NBR entry corresponding to the `struct
net_route_nexthop` data of the deleted route when route is removed.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Simplify common `getsockname()` implementation by using VTABLE_CALL()
macro, in the same way as other socket calls do. This additionally
allows to cover the case, when `getsockname()` is not implemnented by
particular socket implementation, preventing the crash.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
This patch is to fix an issue where the transceiver chip is
disabled before it has transmitted all data. This causes the
message to be corrupted because the last few bytes are missing.
The fix adds a check to make sure the transmission is completed
before disabling the transceiver.
Signed-off-by: Marius Scholtz <mariuss@ricelectronics.com>
CONFIG_UART_MUX_DEVICE_NAME is used as a prefix for the uart muxes
name. Pointer comparison will always return false
Fix#39774
Signed-off-by: Guillaume Lager <g.lager@innoseis.com>
In case system workqueue processing is delayed for any reason, and
resolver callback is executed after getaddrinfo() call already timed
out, the system would crash as the callback makes use of the user data
allocated on the stack within getaddrinfo() function.
Prevent that, by cancelling the DNS request explicitly from the
getaddrinfo() context, therefore preventing the resolver callback
from being executed after the getaddrinfo() call ends.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Set `body_start` pointer regardless of the body position in the recv
buffer. In result, the pointer shall indicate correctly position of the
body for each fragment, it's also explicit now that if the pointer is
not set for a fragment, there's no body in that particular fragment.
Signed-off-by: Robert Lubos <robert.lubos@nordicsemi.no>
Conceptually the net_mgmt_lock should be a mutex instead of a
semaphore. It is easier to identify the owner of a mutex and
debug when deadlock happens, so convert it.
Signed-off-by: Yong Cong Sin <yongcong.sin@gmail.com>
ESP32 build is failing after f-string is added into python
files. This sets esp32 build to use Zephyr's build system python
version.
Signed-off-by: Sylvio Alves <sylvio.alves@espressif.com>
As EVENT_OVERHEAD_START_US offset is used in ticks unit in
LLL, ULL scheduling using ticker should also use ticks unit
for EVENT_OVERHEAD_START_US when reducing the first Periodic
Advertising event preparation.
Relates to commit 858dc7fab4 ("Bluetooth: controller: Fix
EVENT_OVERHEAD_START_US jitter").
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Added test that validates CONFIG_LOG_PRINTK option where
printk is redirected to logging.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Backend implemented in the test did not implemented panic
function and test crashed when log_panic was used.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Pointer to int was passed to a function which expected pointer to
uint8_t. There was an unexpected content for big endian as the
highest byte was read instead of the lowest which contained valid char.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Fixed a dependency from printk.h to logging headers which in
certain configurations could lead to circular dependencies.
Cleaned up printk.c to call z_log_vprintk from vprintk.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Refactoring to remove code redundancy caused by splitted
handling based on USERSPACE enabled.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Document the existance of `zephyr.tag` by providing a link on the main
documentation guide page. Also include an example of how to use this
file in an external project.
Implements #41529.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
After the introduction of #41688, doxyrunner extension modifies the
fmt_vars content (from a variable defined in `conf.py`) when the output
directory variable is defined. This means that Sphinx will always get an
outdated environment and so will always perform a full build instead of
an incremental build. This patch makes a copy first to prevent this
situation.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
The doxyrunner_outdir_var option allows to specify which variable (if
any) is used to represent the output directory (as used by
OUTPUT_DIRECTORY). This makes sure that other options referencing it are
processed correctly, since the output directory is not passed as a
variable.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Don't re-enable the DIO1 interrupt when the modem is in sleep mode. This
fixes the power regression introduced in Zephyr v2.7.0 #41026.
Signed-off-by: Jordan Yates <jordan.yates@data61.csiro.au>
Set enable configuration change bit in can_mcan_set_timing() because
the NBTP register can only be changed if we're in init mode AND
configuration change bit is set, per MCU docs.
Signed-off-by: Jeremy Wood <jeremy@bcdevices.com>
Algorithm was failing in case when overflow mode was enabled
but allocation of new message failed. It could happen if message
size exceeded buffer size. Losing track of buffered messages
can lead to logging processing freeze.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
A variable is tracking number of buffered messages. This is used
to trigger processing thread in certain cases. Counter was not
handled correctly when message was dropped. In certain cases that
can lead to hanging of log processing.
Added counter decrementation in the callback called whenever
message is dropped.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
When LOG2 is enabled, timestamp func which was just set
according to various conditions, is overwritten using sysclock.
Since sysclock is very fast, it will cause uint32_t to wrap very
fast and cause the time to start over from 0 often.
This commit combines the two statements doing the same thing,
to one statement.
Signed-off-by: Emil Lindqvist <emil@lindq.gr>
Removing the 'U' to avoid the type of num_events changed.
And make sure it is meaningful Z_SYSCALL_VERIFY micro.
Fixed#40614
Signed-off-by: NingX Zhao <ningx.zhao@intel.com>
Avoid setting the switch_handler in the z_get_next_switch_handle() code
when the context is not fully saved yet to avoid a race against other
cores waiting on wait_for_switch().
See issue #40795 and discussion in #41840
Signed-off-by: Carlo Caione <ccaione@baylibre.com>
ssd16xx driver is not well designed and does not pass configuration
via device->config but via struct ssd16xx_data, this was not taken
into account in the commit 4d6d50e2bc
("display: ssd16xx: convert to spi_dt_spec").
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
It could be observed on native_posix_64 platform that
without alignment attribute the usb_cfg_data structures
are placed with a gap or padding in specified RAM section.
This breaks the possibility to iterate over the structures.
Signed-off-by: Johann Fischer <johann.fischer@nordicsemi.no>
The commit a28830b aligned the data and rename some symbols. However
there are two symbols at riscv linker script that were missing, which
causes below linker error:
kernel/xip.c:28: undefined reference to `__itcm_load_start'
kernel/xip.c:43: undefined reference to `__dtcm_data_load_start'
Rename below symbols to fix the issues.
__itcm_rom_start -> __itcm_load_start
__dtcm_data_rom_start -> __dtcm_data_load_start
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
The board is fitted with the SMPS components and have no path from
VDD_MCU to VDD_LDO.
The stm32h7_init code used to default to SMPS on supported SoC but has
been changed in 22186c7c51 to only use SMPS if explicitly configured to
do so. This is causing the board to become unresponsive after a power
cycle when flashed with the current defconfig, and needs to be started
pulling BOOT0 high to be recovered.
Setting CONFIG_POWER_SUPPLY_DIRECT_SMPS restores the intended behavior.
Signed-off-by: Fabio Baltieri <fabio.baltieri@gmail.com>
Change the can_tx_callback_t function signature to use an "int" (not an
uint32_t) for representing transmission errors.
The "error" callback function parameter is functionally equivalent to
the return value from can_send() and thus needs to use the same data
type and needs to be able to hold negative errno values.
Signed-off-by: Henrik Brix Andersen <hebad@vestas.com>
Direct SMPS is the default configuration out of the box of:
* stm32h474i_disco
* stm32h735g_disco
Fixes#34732
Signed-off-by: Alexandre Bourdiol <alexandre.bourdiol@st.com>
STM32H7 has different power supply modes but now Zephyr supports just
LDO and direct SMPS. This commit introduses POWER_SUPPLY_CHOICE
configuration parameter and add support for missed power supply modes.
Signed-off-by: Gennady Kovalev <gik@bigur.com>
Fixes#40730.
Use SMPS power supply only if enabled.
The default power supply configuration for the
NUCLEO board with -Q subfix is SMPS,
so it's essential to match with hardware configuration
to avoid deadlocks due to mismatch.
if a custom board with LDO configuration is in use,
then no need to enable `CONFIG_POWER_SUPPLY_SMPS`
Signed-off-by: Manojkumar Subramaniam <manoj@electrolance.com>
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
