On Sunday 08/24 at 20:55 +0200, Willy Tarreau wrote:
Hello,
On Sun, Aug 24, 2025 at 11:31:01AM -0700, Kyle Sanderson wrote:
Thanks for maintaining these as always. For the first time in a long time, I booted the latest stable (6.15.x -> 6.16.3) and somehow lost my networking. It looks like there is a patch from Intel (reported by AMD) that did not make it into stable 6.16.
e67a0bc3ed4fd8ee1697cb6d937e2b294ec13b5e - ixgbe https://lore.kernel.org/all/94d7d5c0bb4fc171154ccff36e85261a9f186923.1755661...
- i40e
Your description is very confusing. First you indicate in the subject that there is a regression in 6.16.3 (implying from 6.16.2 then), and you describe a 6.15.x to 6.16.x upgrade, which then has nothing to do with 6.16.x alone but much more likely with 6.15 to 6.16. Or did a previous 6.16 work fine ? Also the patch you pointed above is neither in 6.15 nor in 6.16, so it's not just "missing from 6.16".
Based on your links and descriptions above I suspect that instead it's a 6.15 to 6.16 regression that was brought by a0285236ab93 and that commit e67a0bc3ed4fd above fixed it in 6.17-rc2, is that it ? If so, can you apply that patch to confirm that it works and is desired in 6.16.x ?
Yeah. 6.16-stable needs c5ec7f49b480 and e67a0bc3ed4f, I was going to bring this up next but Kyle beat me to it :)
Kyle, do you want to send them to stable@, or should I do it?
Thanks, Calvin
Regards, Willy