Synchronize the RPL thread updating the RPL netstats with the RPL
shell command reading it by disabling IRQs. This will prevent printing
corrupted data on non-32bit platforms as well as printing inconsistent
data (e.g. TX count of old state in conjunction with TX bytes of new
state) for all platforms.
Co-authored-by: Martine Lenders <mail@martine-lenders.eu>
There is a repeating pattern in the struct that is split out into a
subtype in this commit. This makes handling the data easier, as now
done in the print routine.
Instead of retrieving a pointer with NETOPT_STATS, retrieve the current
data. This avoids data corruptions when reading from one thread (e.g.
the thread running the shell (ifconfig command)) while another thread
is updating it (e.g. the netif thread).
The issue affects all boards, as users typically expect the count of
TX packets and the number of TX bytes to refer to the same state. For
16 bit and 8 bit platforms even a single netstat entry can read back
corrupted.
This fixes the issue by just copying the whole netstat_t struct over
without requiring explicit locking on the user side. A multi-threaded
network stack still needs to synchronize the thread responding to
netopt_get with the thread writing to the netstat_t structure, but that
is an implementation detail no relevant to the user of the API.
The netif list is used like a stack, so it needs to be
iterated in reverse to keep the registration order.
Time complexity in O(n^2), but the the list is normally very short
(1-2 items).
Before:
```
> ifconfig
Iface 10 HWaddr: 24:0A:C4:E6:0E:9C Channel: 0 Link: down
[..]
Iface 7 HWaddr: 24:0A:C4:E6:0E:9F Link: down
[..]
```
Now they are in the increasing order:
```
> ifconfig
Iface 7 HWaddr: 24:0A:C4:E6:0E:9F Link: down
[..]
Iface 10 HWaddr: 24:0A:C4:E6:0E:9C Channel: 0 Link: down
[..]
```
When lwIP is hacked to use the same shell command, it also
lists it interfaces in the expected order (was ET1,ET0 before):
```
> ifconfig
Iface ET0 HWaddr: 24:0A:C4:E6:0E:9F Link: down
[..]
Iface ET1 HWaddr: 24:0A:C4:E6:0E:9C Channel: 0 Link: down
[..]
```
Module to lock the shell after a given timeout of time x. When the
shell did not receive any input within time x, then the shell is
locked automatically.
Module to lock the running shell with a password. Shell is proceeded only
when the valid password was entered by the user. After 3 failed attempts,
the input is blocked for a few seconds to slow down brute force attacks.
Does not make use of any cryptographic features yet.
This solves highly theoretical race conditions of file systems being
unmounted in an application while a shell `df` runs, fixes the previous
weird behavior that `/mountpoint/non-existant-path` could be df'd and
would even report that non-existant path as a file name, but more
practically ensures that an example of vfs_iter_mount_dirs is around.
The RSSI values reported by LoRa transceiver can be less than -127.
Therefore, `int8_t` is not enough. This commit defines the RSSI of
`netdev_lora_rx_info` as `int16_t` and adapt the drivers accordingly
(sx126x, sx127x).
`msg_avail()` will return -1 if the thread has no message queue.
Casting this to unsigned will result in the `ping` command to try
receiving 4294967295 messages, which hangs the shell.
Drop the cast to `unsigned` and the loop behaves as intended.
But then it's still wrong: If new messages become available, they
would be ignored.
So change the `for` loop to a `while` loop. The index variable is
not used at all.
Currently a valid netif name must be passed to show the usage
instructions:
```
> ifconfig help
error: invalid interface given
> ifconfig 6 help
usage: ifconfig
usage: ifconfig <if_id> [up|down]
[...]
```
`ifconfig --help` is also accepted.
The Linux ping utility has the nice feature that fills the ICMPv6 echo
request payload with a pattern `payload_index & 0xFF`.
Then the ICMPv6 echo response payload is checked to verify that the pattern
is still intact.
This way corrupted messages can be detected.
In the past that revealed some 6lo-fragmentation bugs in Linux when
corrupted replies arrived.
This feature is also useful for RIOT, so implement it in RIOTs `ping`
command.
The `udp` command is a valuable debugging tool that is also useful
outside of the gnrc_networking example.
To enable easy sending of udp messages in other applications during
development, move the `udp` command to the shell module and introduce
the `gnrc_udp_cmd` pseudo-module to enable it.
Lists state, link type, v4/v6 addresses.
Currently read-only.
Using lwIP debug system to print addresses, to limit dependencies
and work with dual stack setup. Most other code seems to only
allow either v4 or v6 networking. For that to compile I
had to change the `SZT_F` format string due to this error:
```
error: format '%lu' expects argument of type 'long unsigned int',
but argument 2 has type 'size_t {aka unsigned int}'
```
Switching to the lwIP default format string here.
Outputs the following on my ESP32 board with Ethernet,
when both v4 and v6 are enabled in examples/paho-mqtt:
```
> ifconfig
Iface ET0 HWaddr: 24:0a:c4:e6:0e:9f Link: up State: up
Link type: wired
inet addr: 10.4.4.81 mask: 255.255.254.0 gw: 10.4.4.1
inet6 addr: fe80:0:0:0:260a:c4ff:fee6:e9f scope: link
inet6 addr: 2001:db8:1000:0:260a:c4ff:fee6:e9f scope: global
Iface ET1 HWaddr: 24:0a:c4:e6:0e:9c Link: up State: up
Link type: wireless
inet addr: 10.4.4.82 mask: 255.255.254.0 gw: 10.4.4.1
inet6 addr: fe80:0:0:0:260a:c4ff:fee6:e9c scope: link
inet6 addr: 2001:db8:1000:0:260a:c4ff:fee6:e9c scope: global
>
```
Previously a value of 0 was used for the RSSI to signal that this value is not
present in `gnrc_netif_hdr_t`. However, an RSSI of 0 dBm is legal and even very
plausible data.
This commit defines `GNRC_NETIF_HDR_NO_RSSI` as `INT16_MIN`, which is below the
noise floor in the vacuum of outer space and hence impossible to receive.
For consistency, also GNRC_NETIF_HDR_NO_LQI is defined.
Ctrl-D was not caught in a special case so it was interpreted as
a standard character. Handle it now the same way like EOF and
terminate the shell instance.
Replace direct accesses to sched_active_thread and sched_active_pid with
the helper functions thread_getpid() and thread_get_active(). This serves
two purposes:
1. It makes accidental writes to those variable from outside core less likely.
2. Casting off the volatile qualifier is now well contained to those two
functions
In most places, picolibc and newlib are the same, so use
the existing newlib code when compiling with picolibc.
Signed-off-by: Keith Packard <keithp@keithp.com>
Introduce optional user shell_post_readline_hook, shell_pre_command_hook, shell_post_command_hook.
Enable with USEMODULE=shell_hooks.
Calls user implemented *_hook functions if defined.
If implementation does not exist, nothing happens.
The intent is to make profiling of the shell command timings easier.
Test provided in tests/shell with USEMODULE=shell_hooks.
Originally, the options and flags in the `netif` shell output were
separated by two spaces. For later added flags this is not the case,
making the parsing of those flags and options hard to impossible.
This change adds those missing spaces + comments so it might not happen
again in the future.
