Only call aspeednic_init when preparing to use network #4
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Tell the ethernet framework how to set the MAC address on the aspeed nic so it can be set at the expected points in the code. Rename set_mac_address to aspeed_write_hwaddr. Drop the unused argument, change the prototype to int, and return 0. Assign the device write_hwaddr method to this new function. Signed-off-by: Milton Miller <miltonm@us.ibm.com>
Delete the call to aspeenic_init from the aspeenic_initialize function. Instead rely on the framework to call the write_hwaddr method to set the MAC address for the operating system to use. The aspeednic driver was calling init before registering the hardware, causing the dma engine to be left running writing incoming ethernet packets to memory when handing control to the operating system. The initialize function is specifically documented to not touch the hardware or probe in doc/README.drivers.net. Also, the init method can take a significant amount of time especially when NC-SI and retries are involved. In addition extra work is being performed to find the MAC address which the framework will do again later. Signed-off-by: Milton Miller <miltonm@us.ibm.com>
The device method assignments had extra spaces before the = signs but all prior methods were 4 characters and so they were aligned. Since write_hwaddr is significantly longer, the resulting spacing is not pleasing. Since only a few other sites had simiar whitespace replace all multi-space sequences before an assignemnt with one space. Signed-off-by: Milton Miller <miltonm@us.ibm.com>
Commit bef1014 upstream. Backport changes CONFIG_NET_RANDOM_ETHADDR to CONFIG_RANDOM_MACADDR, net_random_ethaddr to eth_random_enetaddr, and is_zero_ethaddr to is_zero_ether_addr, and keeps existing function order. Upstream log: Implement the random ethaddr fallback in eth.c so it is in a common place and not reimplemented in each board or driver that wants this behavior. Signed-off-by: Joe Hershberger <joe.hershberger@ni.com> Reviewed-by: Simon Glass <sjg@chromium.org> Signed-off-by: Milton D. Miller II <miltonm@us.ibm.com>
Do not check the u-boot environemnt or fill in a random address in the write_hwaddr hook, instead rely on the framework to do so. The doc/README.ethaddr specifically states that random addresses are only to be assigned as part of a emergency such as a netboot recovery command. The upstream commit created a config variable to assign a random mac when none is set leaving it zero and that has now been backported. Note: The hardware address is reset to 0 as part of the ethernet reset performed at boot. If no valid MAC address is found in the environment the hardware will contain zeros and the operating system will assign a valid random MAC address if u-boot is configured not to. The net effect is an attempt to use the network will result in the ethernet address not set warning being printed if the ethaddr variable is not set, and a warning iwth the random mac address if the config is set. If a valid ethernet address is set in the environment it will be programmed in the hardware and used by the operating system. Signed-off-by: Milton Miller <miltonm@us.ibm.com>
mdmillerii
force-pushed
the
aspeednic-init
branch
from
a2f9a64
to
b39e67e
Compare
|
Merged to v2013.07-aspeed-openbmc branch |
eddiejames
pushed a commit
to eddiejames/u-boot
that referenced
this pull request
Jul 10, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The aspeednic driver was calling init from its initialization function and
calling halt later. This left the DMA engine running and writing to
memory when Ethernet traffic arrived when control was handed off
to the operating system.
Remove this call to init and rely on the framework to call
write_hwaddr with the assigned MAC address.
In addition upstream discourages setting the random address in
drivers. Later upstream created a config for this behavior and I
back-ported that patch.
In my testing, the random mac address is about 90% 1 value after a
u-boot reset command but after a kernel reboot it does not repeat.
The code uses rand with timer as the seed source; the network stack
then uses the mac as the seed for random delays. We may prefer
to only configure u-boot setting a random mac only after a better
source of entropy, letting the kernel use its sources to set it when
we do not net-boot.
milton
v2: update README and README.ethaddr with backported CONFIG var name.
This change is