Linux Patches Updated For Parallel CPU Bring-Up
While missing out on the recently-closed Linux 6.4 merge window, one of the patch series seeing a lot of activity in recent weeks/months has been around parallel CPU bring-up in order to speed-up the Linux kernel boot process for today's high-end desktops and many-core servers.
Thomas Gleixner on Monday posted the third iteration of his patches providing the parallel CPU bring-up. Just a few basic changes were made over the v2 patches from last week that succeeded Gleixner's patches from April that in turn were a reincarnation of prior CPU parallel bring-up patches floating on the Linux kernel mailing list for months previously.
The intent on this parallel CPU bring-up is on shortening in particular the Kexec reboot time for many-core cloud servers. Yes, POST'ing modern servers and memory initialization can still be very time consuming on large servers, which leads many public cloud providers, hyperscalers, and others to just rely on Kexec rebooting Linux to avoid that more costly downtime of a full system shutdown/reboot. For that Kexec reboot time with these parallel CPU bring-up patches able to shave off hundreds of milliseconds, that can be beneficial to large production server deployments.
The new v3 patches can be found on the LKML. With the pace the patches have been going out, hopefully this CPU parallel bring-up support could be ready for the v6.5 merge window in early July.
Thomas Gleixner on Monday posted the third iteration of his patches providing the parallel CPU bring-up. Just a few basic changes were made over the v2 patches from last week that succeeded Gleixner's patches from April that in turn were a reincarnation of prior CPU parallel bring-up patches floating on the Linux kernel mailing list for months previously.
The intent on this parallel CPU bring-up is on shortening in particular the Kexec reboot time for many-core cloud servers. Yes, POST'ing modern servers and memory initialization can still be very time consuming on large servers, which leads many public cloud providers, hyperscalers, and others to just rely on Kexec rebooting Linux to avoid that more costly downtime of a full system shutdown/reboot. For that Kexec reboot time with these parallel CPU bring-up patches able to shave off hundreds of milliseconds, that can be beneficial to large production server deployments.
The new v3 patches can be found on the LKML. With the pace the patches have been going out, hopefully this CPU parallel bring-up support could be ready for the v6.5 merge window in early July.
Add A Comment