hi,
On 1/26/26 07:31, Uwe Kleine-K nig wrote:
i rarely turn that box on, the one before 6.18 was 6.17.9 and that
worked....
I intend to look into the detail here later this week, but if you come
around to test the earlier 6.18.y kernels and report back which is the
first broken one that would help me to know where to look for the
regression.
i'll try my best to bisect this at debian kernel version level, but
don't have high hopes to do so at commit level; the device i have
(asterfusion x312p with 2 dpu-s, and those latter are the aarch64s)
revolves around the intel tofino asic and the two arm64 socs are just
sidecars, moreover its a development box, and as such, to power cycle
them needs about 10 minutes manual process from my side bring up all the
stuff to have the arm64s some networking again... it also adds to the
mix that the 6.17 had about 50% chance of not booting up and getting to
the (initramfs) because the kernel failed to bring up the emmc.... and
if it even booted up, had tons of complains about the nics, which,
later, i use with dpdk, thats the reason i did not bugged about this
before....
tldr; i'll get back to you with the latest known working kernel version
if i find some spare hours, or n/a if i've given up in the middle of the
process...:)
all the best,
csaba
