I'd like to revive the discussion and quickly summarize our options to
avoid a false-positive Lockdep-RCU splat when check_and_switch_context()
acquires the ASID spinlock:
1. Briefly report CPU as online again via
rcutree_report_cpu_{starting|dead}()
2. Replace raw_spin_lock*() by arch_spin_lock()
3. Remove ASID spinlock
Both 1. and 2. are workarounds with different pros and cons/proponents.
Regarding 3., the last comment form Russell is:
> For 32-bit non-LPAE, I can't recall any issues, nor can I think of any
> because module space is just another few entries in the L1 page tables
> below the direct mapping (which isn't a problem because we don't use
> anything in hardware to separate the kernel space from user space in
> the page tables.) TTBCR is set to 0.
>
> For LPAE, there may be issues there because TTBR0 and TTBR1 are both
> used, and TTBCR.T1SZ is set non-zero to:
>
> arch/arm/include/asm/pgtable-3level-hwdef.h:#define TTBR1_SIZE (((PAGE_OFFSET >> 30) - 1) << 16)
>
> so I suspect that's where the problems may lie - but then module
> mappings should also exist in init_mm (swapper_pg_dir) and should
> be global.
Unfortunately I'm don't feel qualified to contribute to the discussion
on option 3. Russell and Will, would you be able to spare some time to
drive this further? Otherwise I would propose to make a decision on
going for option 1 or 2.
Kind regards,
Stefan
