Printk Changes For Linux 6.4 Are Light With Still Waiting For Threaded/Atomic Console
The printk code changes were merged last week for the ongoing Linux 6.4 merge window and it's notable not for what is in the pull request but rather what is still outstanding.
With the printk updates for Linux 6.4 it amounts to code cleanups and dead code removal... Namely just deleting around three dozen lines of code.
What unfortunately isn't ready for Linux 6.4 and what we've been waiting on when it comes to printk is the much anticipated threaded/atomic console support.
With the in-development Linux threaded/atomic consoles, each console has its own KThread and different consoles do not contend with each other and do not use the global console lock, each console is able to write any context, atomic printing is performed until console threads are brought up and allows many-core CPUs to boot at full-speed without waiting on console printing, and various other features are supported. But it's not ready for mainlining yet.
Besides threaded/atomic console support being nifty on its own, it's also the last major blocker before the real-time "PREEMPT_RT" patches can be upstreamed in full to the mainline Linux kernel. The current console code is not compatible with the current RT code.
Here's to hoping that the threaded/atomic console support and in turn the RT mainlining manages to get squared away soon -- now waiting for at least the v6.5 cycle.
With the printk updates for Linux 6.4 it amounts to code cleanups and dead code removal... Namely just deleting around three dozen lines of code.
What unfortunately isn't ready for Linux 6.4 and what we've been waiting on when it comes to printk is the much anticipated threaded/atomic console support.
With the in-development Linux threaded/atomic consoles, each console has its own KThread and different consoles do not contend with each other and do not use the global console lock, each console is able to write any context, atomic printing is performed until console threads are brought up and allows many-core CPUs to boot at full-speed without waiting on console printing, and various other features are supported. But it's not ready for mainlining yet.
Besides threaded/atomic console support being nifty on its own, it's also the last major blocker before the real-time "PREEMPT_RT" patches can be upstreamed in full to the mainline Linux kernel. The current console code is not compatible with the current RT code.
Here's to hoping that the threaded/atomic console support and in turn the RT mainlining manages to get squared away soon -- now waiting for at least the v6.5 cycle.
1 Comment