| Age | Commit message (Collapse) | Author |
|---|
|
Add the option to load the bootscript with the tftp command (static IP)
instead of the dhcp command (dynamic IP). For this a new function
tftpb_run similar to dhcp_run, is needed. The selection of which command
to use can be done with the ip_dyn environment variable, which can be
set to yes or no. The ip_dyn variable was chosen as it is already in use
on the imx platforms.
Also edit the bootstd doc.
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Benjamin Hahn <[email protected]>
|
|
CVE-2024-42040 describes a possible buffer overflow when calling
bootp_process_vendor() in bootp_handler() since the total length
of the packet is passed to bootp_process_vendor() without being
reduced to len-(offsetof(struct bootp_hdr,bp_vend)+4).
The packet length is also checked against its minimum size to avoid
reading data from struct bootp_hdr outside of the packet length.
Signed-off-by: Paul HENRYS <[email protected]>
Signed-off-by: Philippe Reynes <[email protected]>
|
|
There are currently two implementations of dhcp_run(): one in cmd/net.c
for NET and one in net/lwip/dhcp.c for NET_LWIP. There is no
justification for that. Therefore, move the NET version into
net/net-common.c to be used by both stacks, and drop the NET_LWIP
version which by the way does not look totally correct.
Signed-off-by: Jerome Forissier <[email protected]>
Suggested-by: Tom Rini <[email protected]>
Acked-by: Benjamin Hahn <[email protected]>
|
|
It has been over a decade since we had a platform that implemented the
bootp vendor extension support hook. Remove this option due to lack of
use.
Signed-off-by: Tom Rini <[email protected]>
Acked-by: Jerome Forissier <[email protected]>
|
|
The support found under SYS_FAULT_ECHO_LINK_DOWN requires that the
SYS_FAULT_MII_ADDR symbol also be set. This wasn't previously found in
Kconfig, so add it now.
Signed-off-by: Tom Rini <[email protected]>
Acked-by: Jerome Forissier <[email protected]>
|
|
A few lines of code being guarded by the CONFIG_TFTP_PORT option seems
an unnecessary restriction on the TFTP support provided by a vanilla
U-Boot image. In cases where the TFTP server cannot run as superuser -
and hence cannot run on the well-known port 69 - this quirk incurs a
full reconfiguration and rebuild of the bootloader only in order to
select the appropriate destination port.
Remove the CONFIG_TFTP_PORT option entirely and make the tftpdstp and
tftpsrcp variables always have an effect. Their being unset will mean
that U-Boot behaves the same as if CONFIG_TFTP_PORT was unset. Update
the documentation accordingly. And fix up the single board which was
originally enabling this option.
Signed-off-by: Alvin Šipraga <[email protected]>
Reviewed-by: Quentin Schulz <[email protected]>
|
|
dm_eth_phy_connect_interface() is a variant of dm_eth_phy_connect() that
allows to set the used PHY mode, in case the MAC driver needs to fix it
up. The previously static dm_eth_connect_phy_handle() is renamed and
extended for this purpose.
Signed-off-by: Matthias Schiffer <[email protected]>
|
|
Network packet buffers should be aligned to PKTALIGN. Add a compiler
attribute to ensure this.
Signed-off-by: Tim Harvey <[email protected]>
[jf: use __aligned(x) from <linux/compiler_attributes.h>]
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Once the DHCP exchange is complete, if we have an IP address for an NTP
server, set the ntpserverip environment variable accordingly. Although
not necessary to make the sntp command use that server (since it is
known internally to the lwIP stack), this makes the behavior in line
with the legacy NET stack. This is also consistent with exporting the
DNS servers etc.
Signed-off-by: Jerome Forissier <[email protected]>
Suggested-by: Michal Simek <[email protected]>
|
|
When both CMD_DHCP and CMD_SNTP are enabled, one would expect the NTP
servers received by DHCP to be used by the sntp command by default. Fix
dhcp_loop() so that it is indeed the case.
Signed-off-by: Jerome Forissier <[email protected]>
Reported-by: Michal Simek <[email protected]>
|
|
The legacy network stack sets net_try_count to 1 at the beginning of the
net_loop() function. This is required for net_start_again() to work
properly. Therefore, set the variable accordingly in the do_ping()
function when NET_LWIP=y. This fixes an issue where a ping to an
unreachable destination would run twice on the same network device. For
example with qemu_arm64_lwip_defconfig:
=> dhcp
DHCP client bound to address 10.0.2.15 (3 ms)
=> ping 10.0.0.1
Using virtio-net#32 device
ping failed; host 10.0.0.1 is not alive
Using virtio-net#32 device
ping failed; host 10.0.0.1 is not alive
=> QEMU: Terminated
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Introduce the DNS Kconfig symbol so that various network commands may
use host names without the dns command (CMD_DNS) being selected.
Signed-off-by: Jerome Forissier <[email protected]>
CC: E Shattow <[email protected]>
|
|
The header of LiMon imported files reference a License file which
does not exist in U-Boot. Some files were forgotten when adding the
SPDX-License-Identifier.
The LiMon files were originally licensed under GPLv2 as can be seen in
commit [2ea91039].
Based on this commit, add the correct SPDX license identifier.
While at it drop the reference to the non-existing License file from all
LiMon files and update the SPDX-License-Identifier to SPDX version 3.
Signed-off-by: Max Merchel <[email protected]>
|
|
Add Kconfig symbol LWIP_ICMP_SHOW_UNREACH which, when enabled, prints a
message to the console upon reception of ICMP unreachable messages. For
example:
$ make qemu_arm64_lwip_defconfig
$ qemu-system-aarch64 -M virt -cpu max -nographic -bios u-boot.bin
[...]
=> dhcp
DHCP client bound to address 10.0.2.15 (0 ms)
=> tftp 192.168.0.100:69:Image
Using virtio-net#32 device
TFTP from server 192.168.0.100; our IP address is 10.0.2.15
Filename 'Image'.
Load address: 0x40200000
Loading: ICMP destination unreachable (host unreachable) from 192.168.0.16
Timeout!
=> tftp 192.168.0.16:69:Image
Using virtio-net#32 device
TFTP from server 192.168.0.16; our IP address is 10.0.2.15
Filename 'Image'.
Load address: 0x40200000
Loading: ICMP destination unreachable (port unreachable) from 192.168.0.16
Timeout!
=>
Submitted upstream as https://github.com/lwip-tcpip/lwip/pull/73.
Signed-off-by: Jerome Forissier <[email protected]>
|
|
The call to eth_init within net_init causes the network interface to
start, stop, start again which can cause issues with certain network
device drivers. Remove it to make it behave like the legacy network
path.
Fixes: 5666865decb8 ("net: lwip: fix initialization sequence before a command")
Signed-off-by: Tim Harvey <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
|
For NET_LWIP eth_is_on_demand_init() is always 1 so remove the check and
simplify the code.
Suggested-by: Jerome Forissier <[email protected]>
Signed-off-by: Tim Harvey <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
|
content_length is an unsigned long and so testing that it is >= 0 will
always be true. Instead test that it is != -1 as that is the condition
set on error.
This issue found by Smatch.
Signed-off-by: Andrew Goodbody <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
|
In tcp_parse_options the uchar p[0] is attempted to test for a match
with the 32bit macro TCP_0_NOP which can never be true. Instead test
against the 8bit macro TCP_1_NOP.
This issue found by Smatch.
Signed-off-by: Andrew Goodbody <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
|
Extract the code that sets the RTC clock from sntp_handler() in
net/sntp.c and make it a new function net_sntp_set_rtc() in
net/net-common.c. This will allow re-use with NET_LWIP.
According to [1] it is safe to assume that all devices have been
converted to DM_RTC so drop the useless code.
[1] https://lists.denx.de/pipermail/u-boot/2025-June/591376.html
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Add a helper fonction to convert an IP address (supplied as a text
string) or a host name to an ip_addr_t.
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Split net/lwip/wget.c in two: one part which implements CONFIG_WGET
stays in net/ while the part that implements CONFIG_CMD_WGET is moved
into cmd/.
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Prepare to split the ping command from cmd/net-lwip.c by moving the
implementation from net/lwip/dns.c to cmd/lwip.
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Prepare to split the dns command from cmd/net-lwip.c by moving the
implementation from net/lwip/dns.c to cmd/lwip.
Signed-off-by: Jerome Forissier <[email protected]>
|
|
net/lwip/eth_internal.h is not used. Remove it.
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Initialize dns servers if a hostname (vs ipaddr) is used in a URL.
Otherwise the wget will fail without displaying an error due to
dns_gethostbyname failing silently when no DNS servers are set unless
you have previously performed a 'dns <arg>' command.
Signed-off-by: Tim Harvey <[email protected]>
Acked-by: Jerome Forissier <[email protected]>
|
|
move the dns init including setting the dns servers from env vars to a
common function as other commands that support hostname lookups will
need this.
Signed-off-by: Tim Harvey <[email protected]>
[jf: add CMD_DNS conditional to support NET_LWIP && !CMD_DNS]
Reviewed-by: Jerome Forissier <[email protected]>
|
|
Add support for setting the blocksize at runtime via the tftpblocksize env
variable as such is done with the legacy stack (CONFIG_NET).
Signed-off-by: Tim Harvey <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
|
Call schedule() in net_lwip_rx() to service U-Boot tasks and
actions during packet rx.
As a cleanup also move sys_check_timeouts() here and remove it from the
functions that call net_lwip_rx().
This resolves the issue of an active watchdog resetting the board on
long network activities.
Suggested-by: Jerome Forissier <[email protected]>
Signed-off-by: Tim Harvey <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
|
DHCP and DHCPv6 use the same value defined in
https://www.iana.org/assignments/dhcpv6-parameters#processor-architecture
to encode the processor architecture type. We should only use a single
Kconfig symbol for both protocols.
Furthermore we should make the value customizable. This allows for instance
to choose between "x86 BIOS" or "x64 UEFI".
As "x86 BIOS" is encoded as 0, we should not use this value to switch
off transmission of the DHCP option. Use 0xFF instead.
Signed-off-by: Heinrich Schuchardt <[email protected]>
Acked-by: Ilias Apalodimas <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
|
|
Tom Rini <[email protected]> says:
Hey all,
This is a v3 of Simon's series[1] and depends on the series[2] I posted
the other day that removes <env.h> from <command.h>. With this series
done, I believe we've tackled all of the current cases of headers which
include <env.h> without directly needing it. Much of this series is in
fact Simon's v2 with the main differneces being:
- Removing <env.h> from <net.h> at the end
- Removing env_to_ip() given how little it's used rather than shuffling
around where it's declared and un-inline'ing it. For a rarely used
helper, this ends up being cleaner I think. Especially looking at some
of the users (which called env_get repeatedly). If there's strong
opinion here about using the other method[3] we can do that instead.
- Setting aside for now how to handle CMD_ELF=y and NO_NET=y because
today it's actually fine as we unconditionally build lib/net_utils.c
where string_to_ip() is defined. I'm unsure if a further series is
warranted here or not. We rely on link-time optimization to keep code
readable too.
[1]: https://lore.kernel.org/all/[email protected]
[2]: https://lore.kernel.org/all/[email protected]
[3]: https://lore.kernel.org/all/[email protected]
Link: https://lore.kernel.org/r/[email protected]
|
|
Now that env_get_ip() has been removed, the include file <net.h> does
not need anything from <env.h>. Furthermore, include/env.h itself
includes other headers which can lead to longer indirect inclusion
paths. To prepare to remove <env.h> from <net.h> fix all of the
remaining places which had relied on this indirect inclusion to instead
include <env.h> directly.
Reviewed-by: Jerome Forissier <[email protected]> # net/lwip
Acked-by: Ilias Apalodimas <[email protected]>
Reviewed-by: Wolfgang Wallner <[email protected]>
Reviewed-by: Martyn Welch <[email protected]>
Signed-off-by: Tom Rini <[email protected]>
|
|
This file uses the environment but does not include the header file.
Update it.
Signed-off-by: Simon Glass <[email protected]>
|
|
Currently, we have the function env_get_ip which takes an IP address
in string form and returns a struct in_addr representation of that
address. It is however little used and means that a number of places
indirectly (and unclearly) get <env.h> via <net.h>. To clean this up
start by replacing env_get_ip() calls with string_to_ip() calls. This is
generally a no-op as env_get_ip(str) is an inline of
string_to_ip(env_get(str)) but in a few cases we can or already have
stored the result of env_get(str) and can save the additional call.
Signed-off-by: Tom Rini <[email protected]>
|
|
Prepare v2025.07-rc3
|
|
Contrary to doc/develop/driver-model/ethernet.rst contract, eth_ops
.free_pkt can be called after .stop, there are several error paths in TFTP,
for instance:
eth_halt() <= tftp_handler() <= net_process_received_packet() <= eth_rx()
...
am65_cpsw_free_pkt() <= eth_rx()
Which results in (deliberately "tftpboot"ing non-existing file):
TFTP error: 'File not found' (1)
Not retrying...
am65_cpsw_nuss_port ethernet@8000000port@1: RX dma free_pkt failed -22
Avoid the DMA error message (and follow the documentation) by deferring
eth_halt() until net_loop() calls net_start_again() and only do
eth_halt_state_only() instead.
Fixes: aafda38fb266 ("Add error codes/handling for TFTP-server")
Signed-off-by: Alexander Sverdlin <[email protected]>
|
|
Sean Edmond <[email protected]> says:
In our datacenter application, a single DHCP server is servicing 36000+ clients.
Improvements are required to the DHCPv4 retransmission behavior to align with
RFC and ensure less pressure is exerted on the server:
- retransmission backoff interval maximum is configurable
(environment variable bootpretransmitperiodmax)
- initial retransmission backoff interval is configurable
(environment variable bootpretransmitperiodinit)
- transaction ID is kept the same for each BOOTP/DHCPv4 request
(not recreated on each retry)
For our application we'll use:
- bootpretransmitperiodmax=16000
- bootpretransmitperiodinit=2000
A new configuration BOOTP_RANDOM_XID has been added to enable a randomized
BOOTP/DHCPv4 transaction ID.
Enhance DHCPv4 sending/parsing option 209 (PXE config file). A previous
patch was accepted. A new patch fixes a possible double free() and
addresses latest review comments.
Link: https://lore.kernel.org/r/[email protected]
|
|
The new config option BOOTP_RANDOM_XID will randomize the transaction ID
for each new BOOT/DHCPv4 exchange.
Signed-off-by: Sean Edmond <[email protected]>
Reviewed-by: Tom Rini <[email protected]>
|
|
This patch introduces 3 improvements to align with RFC 951:
- retransmission backoff interval maximum is configurable
- initial retranmission backoff interval is configurable
- transaction ID is kept the same for each BOOTP/DHCPv4 request
In applications where thousands of nodes are serviced by a single DHCP
server, maximizing the retransmission backoff interval at 2 seconds (the
current u-boot default) exerts high pressure on the DHCP server and
network layer.
RFC 951 “7.2. Client Retransmission Strategy” states that the
retransmission backoff interval should be limited to 60 seconds. This
patch allows the interval to be configurable using the environment
variable "bootpretransmitperiodmax"
The initial retranmission backoff period defaults to 250ms, which is
also too small for these scenarios with many clients. This patch makes
the initial retransmission interval to be configurable using the
environment variable "bootpretransmitperiodinit".
Also, on a retransmission it is not expected for the transaction ID to
change (only the 'secs' field should be updated). Let's save the
transaction ID and use the same transaction ID for each BOOTP/DHCPv4
exchange.
Signed-off-by: Sean Edmond <[email protected]>
|
|
- Enable option 209 by default
- Set pxelinux_configfile to NULL to avoid potential double free
- change hardcoded 209 to a define
Signed-off-by: Sean Edmond <[email protected]>
|
|
When there is no reply from the TFTP server, do_tftpb() should
eventually time out. Add a 10 second timer for that purpose.
Reported-by: Heinrich Schuchardt <[email protected]>
Signed-off-by: Jerome Forissier <[email protected]>
|
|
Legacy NET tftp invokes a store_block() function which performs buffer
validation (LMB, address wrapping). Do the same with NET_LWIP.
Signed-off-by: Jerome Forissier <[email protected]>
Suggested-by: Sughosh Ganu <[email protected]>
Acked-by: Sughosh Ganu <[email protected]>
Tested-by: Sughosh Ganu <[email protected]>
|
|
Legacy NET wget invokes a store_block() function which performs buffer
validation (LMB, address wrapping). Do the same with NET_LWIP.
Signed-off-by: Jerome Forissier <[email protected]>
Suggested-by: Sughosh Ganu <[email protected]>
Acked-by: Sughosh Ganu <[email protected]>
Tested-by: Sughosh Ganu <[email protected]>
|
|
Functions called from EFI applications should not do console output.
Refactor the wget code to implement this requirement. The wget_http_info
struct is used to hold the boolean that signifies whether the output is
allowed or not.
Signed-off-by: Jerome Forissier <[email protected]>
Reported-by: Heinrich Schuchardt <[email protected]>
|
|
wget_do_request() currently does so little before calling the static
function wget_loop() that we may as well rename wget_loop() to
wget_do_request() and put everything in one function.
Signed-off-by: Jerome Forissier <[email protected]>
Acked-by: Sughosh Ganu <[email protected]>
|
|
Now that the DSA tests in test/dm/dsa.c are compatible with NET_LWIP,
remove the dependency of DM_DSA on NET.
Signed-off-by: Jerome Forissier <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
When the sandbox timer is available, use it. This allows skipping
time in the tests (sandbox_eth_skip_timeout()).
Signed-off-by: Jerome Forissier <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Use net_start_again() in do_ping() to determine if a failed ping should
be restarted on a different interface.
Signed-off-by: Jerome Forissier <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
Implement net_start_again() when NET_LWIP=y in a very similar way to
NET. This will be used in a future commit to determine if a failed
ping needs to be tried again on a different interface.
Signed-off-by: Jerome Forissier <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
The things that are done prior to executing a network command with
NET_LWIP are not consistent with what is done with NET. It impacts the
selection of the current device, and more precisely if the active device
is invalid NET would return an error while NET_LWIP would try to pick a
new device. This incorrect behavior was detected thanks to the eth_rotate
sandbox test (dm_test_eth_rotate()).
Fix it by re-using a sequence similar to what NET has in net_loop().
This piece of code is inserted in a function called net_lwip_eth_start()
renamed from net_lwip_eth_set_current().
Signed-off-by: Jerome Forissier <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
|
|
!A || (A && B) is equivalent to !A || B
Drop the middle expression from the statement
Signed-off-by: Bryan Brattlof <[email protected]>
Reviewed-by: Jerome Forissier <[email protected]>
|
