commit | 55c97947531d88af6cf9f56ccccb41e26c7f8cd5 | [log] [tgz] |
---|---|---|
author | Paul Gortmaker <paul.gortmaker@windriver.com> | Wed Feb 20 19:05:38 2013 -0500 |
committer | Paul Gortmaker <paul.gortmaker@windriver.com> | Wed Feb 20 19:05:38 2013 -0500 |
tree | b9d8ca328a6a742bdf3b1cb2fab1a9b193f0eca6 | |
parent | 8d3f338a630c95d5eac844c4d779645feed9ea72 [diff] |
preempt-rcu: untangle kconfig RT_FULL/PREEMPT_RCU relationship The intent of the rcu-force-preempt-rcu-for-rt.patch was to ensure you weren't even asked about PREEMPT_RCU when RT_FULL was =y. However it tried to achieve this with a dependency, and that just masks the option completely. This wasn't noticed since the patch was accidentally reverted, and misplaced (see prev two commits) and hence the misuse of a dependency causing the exact opposite of the desired effect went un-noticed. By using a select PREEMPT_RCU in the RT_FULL block, we get the desired effect, i.e. Kconfig will not ask the user whether they want PREEMPT_RCU. It will simply set it unconditionally for anyone with RT_FULL=y Since RT_FULL Kconfig control block doesn't even appear until the kconfig-preempt-rt-full.patch - we insert the select line into that patch. Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>