On 10/07/2025 15:33, Thomas Lamprecht wrote: > On Thu, 10 Jul 2025 14:50:29 +0200, Friedrich Weber wrote: >> On host CPUs with the split_lock_detect flag (newer Intel CPUs), >> booting an OVMF VM with more than one core may trigger the host >> kernel's split lock detection, as reported in [1]. >> >> With default settings, a kernel >= 5.19 slows down the corresponding >> thread for 10ms when it detects a split lock operation, as documented >> in [2]. A warning is logged, e.g.: >> >> [...] > > Applied, thanks!
Thanks for the merge! > But I had to re-export the patch to get the \r\n line endings back, the reason > for that is that you did not explicitly passed `--transfer-encoding=base64` to > git send-email as the "auto" default value is seemingly not smart enough to go > for that encoding when it sees such line endings.. Oops, sorry about that, will keep this in mind for next time with a CRLF patch. > btw. you git note suggested to me that you tried this on PVE 9, but you > at least did not try to build there, as I had to fix a bit of stuff (mostly > python 3.13 related). I did mention I only got it to build on bookworm, but forgot to mention I only tested on bookworm too. My workstation is still on bookworm and I didn't have any luck getting split-lock detection to work in a nested setup (even with CPU type 'host', the PVE VM doesn't see the 'split_lock_detect' CPU flag). Should have made this more clear though, sorry for the confusion. > > [1/1] fix #6430: backport patch to fix split lock detection warnings > commit: 47055b2c7b56fecd84c6441c4f4c419738cc1df8 _______________________________________________ pve-devel mailing list pve-devel@lists.proxmox.com https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
