Linux Works Towards True CPU Hotplug Support
One of the first pull requests sent in following the the Linux 3.7 kernel release is for getting closer to "true CPU hotplug support" with the Linux 3.8 kernel.
There isn't yet "true" CPU hotplug support for Linux 3.8, but developers are getting closer. Ingo Molnar sent in this morning the x86 generic pull request for 3.8. While the Linux kernel has had CPU hotplugging support, up to this point you haven't been able to hotplug the first (boot) processor.
With this pull for the Linux 3.8 kernel, it will become possible to online/offline the boot processor just like any other x86 CPUs. This feature right now will be supported first by Intel CPUs. For 3.8, this feature is behind a configuration option that is disabled by default.
There isn't yet "true" CPU hotplug support for Linux 3.8, but developers are getting closer. Ingo Molnar sent in this morning the x86 generic pull request for 3.8. While the Linux kernel has had CPU hotplugging support, up to this point you haven't been able to hotplug the first (boot) processor.
With this pull for the Linux 3.8 kernel, it will become possible to online/offline the boot processor just like any other x86 CPUs. This feature right now will be supported first by Intel CPUs. For 3.8, this feature is behind a configuration option that is disabled by default.
This tree enables CPU#0 (the boot processor) to be onlined/offlined on x86, just like any other CPU. Enabled on Intel CPUs for now.
Allowing this required the identification and fixing of latent CPU#0 assumptions (such as CPU#0 initializations, etc.) in the x86 architecture code, plus the identification of barriers to BSP-offlining, such as active PIC interrupts which can only be serviced on the BSP.
It's behind a default-off option, and there's a debug option that allows the automatic testing of this feature.
The motivation of this feature is to allow and prepare for true CPU-hotplug hardware support: recent changes to MCE support enable us to detect a deteriorating but not yet hard-failing L1/L2 cache on a CPU that could be soft-unplugged - or a failing L3 cache on a multi-socket system.
Note that true hardware hot-plug is not yet fully enabled by this, because that requires a special platform wakeup sequence to be sent to the freshly powered up CPU#0. Future patches for this are planned, once such a platform exists. Chicken and egg
15 Comments