On 13/10/2026 00:45, Tom Rini wrote:
> On Mon, Oct 12, 2026 at 04:43:06PM -0600, James Hilliard wrote:
>> On Mon, Oct 12, 2026 at 3:03 PM Tom Rini <[email protected]> wrote:
>>>
>>> On Mon, Oct 12, 2026 at 02:41:42PM -0600, James Hilliard wrote:
>>>> On Mon, Oct 12, 2026 at 2:08 PM Tom Rini <[email protected]> wrote:
>>>>>
>>>>> On Mon, Oct 12, 2026 at 09:47:05AM +0200, Jerome Forissier wrote:
>>>>>
>>>>>> Hi Tom,
>>>>>>
>>>>>> The following changes since commit
>>>>>> ed869863c7e2aca50d57a05d843830dd843138c4:
>>>>>>
>>>>>> Merge tag 'u-boot-marvell-next-20261005' of
>>>>>> https://git.u-boot-project.org/u-boot/custodians/u-boot-marvell into
>>>>>> next (2026-10-05 10:40:38 -0600)
>>>>>>
>>>>>> are available in the Git repository at:
>>>>>>
>>>>>> https://git.u-boot-project.org/u-boot/custodians/u-boot-net.git/
>>>>>> tags/net-next-20261009
>>>>>>
>>>>>> for you to fetch changes up to 28ddd369e2f7199539cdefe22ce014c592b890bd:
>>>>>>
>>>>>> net: dwc_eth_qos: spacemit: add support for K3 SoC (2026-10-09
>>>>>> 18:58:10 +0200)
>>>>>>
>>>>>> *** Please note ***
>>>>>>
>>>>>> Commit "qemu-xtensa: Increase the U-Boot memory reservation" fixes a
>>>>>> build
>>>>>> error detected by CI that would otherwise be caused by "net: e1000: use
>>>>>> all
>>>>>> RX descriptors". I did NOT send that one out for review to avoid delaying
>>>>>> this pull request, but if you'd like me to do otherwise just let me know.
>>>>>>
>>>>>> Similarly, the test added by "test: net: cover lwIP DHCP autoload
>>>>>> handoff"
>>>>>> initally failed CI and I incorporated my own fix (see the commit
>>>>>> description).
>>>>>>
>>>>>> ----------------------------------------------------------------
>>>>>> Pull request net-next 20261009.
>>>>>
>>>>> Now that I've fixed the sage lab in CI[1] I've found a run-time failure.
>>>>>
>>>>> [snip]
>>>>>> James Hilliard (20):
>>>>> [snip]
>>>>>> net: lwip: finish DHCP before starting autoload
>>>>>
>>>>> Fails on BeagleBone Black with lwIP enabled, none of the network tests
>>>>> run and the log output is:
>>>>> => setenv autoload no
>>>>> => dhcp
>>>>> link up on port 0, speed 100, full duplex
>>>>> =>
>>>>> and we never get an IP and so can't network.
>>>>
>>>> I think this should fix it:
>>>> https://lore.kernel.org/all/[email protected]/
>>>
>>> Are there likely other drivers with this problem?
>>
>> Yeah, this should fix them:
>> https://lore.kernel.org/u-boot/[email protected]/T/
>
> Thanks. Lets drop this patch from the PR for now (so please send a new
> one Jerome) and take that change along with all the driver fixes in the
> next one, once there's been time for people to review.
>
Alright, new PR is coming. I had to drop the whole series ("net: lwip: restore
BOOTP,
DHCPv4 boot policy and PXE parity") because removing only "net: lwip: finish
DHCP
before starting autoload" resulted in too many conflicts and also implied
changing
the tests. For my next PR I'll pick up the driver fixes first ("[PATCH v2 0/8]
net:
return -EAGAIN when no RX packet is ready") then apply the BOOTP/DHCPv4/PXE
series
on top.
Thanks,
--
Jerome
