Re: [PATCH 3/4] sched_ext: Add scx_cmask_test() and scx_cmask_for_each_cid()

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The `scx_cmask_for_each_cid` macro expands into two nested `for` loops, breaking standard C iteration semantics for `break` statements. If a caller uses a `break` statement to terminate the iteration early, it will only exit the inner loop (terminating the scan for the current 64-bit word). The outer loop will then increment, fetch the next word, and continue iterating. This creates a trap where developers intending to completely abort the loop will inadvertently only skip the rest of the current word and silently resume on the next one.
--

commit e9a4307d205f9de4f9da9f425dd9329d36d47094
Author: Tejun Heo <tj@kernel.org>

sched_ext: Add scx_cmask_test() and scx_cmask_for_each_cid()

Adds a single-bit test and an iterator over set cids in an scx_cmask.
The iterator is bounded to the active span.

> diff --git a/kernel/sched/ext_cid.h b/kernel/sched/ext_cid.h
> index 46fd8eda04435..5745e5785e897 100644
> --- a/kernel/sched/ext_cid.h
> +++ b/kernel/sched/ext_cid.h

[ ... ]

> +/**
> + * scx_cmask_for_each_cid - iterate set cids in @m
> + * @cid: s32 loop var that receives each set cid in turn
> + * @m: cmask to iterate
> + *
> + * Visits set bits within @m's active range in ascending order. Scans only the
> + * words the active range spans, where head and tail padding is kept zero, so
> + * no per-cid range check is needed.
> + */
> +#define scx_cmask_for_each_cid(cid, m)						\
> +	for (u64 __bs = (m)->base & ~63u, __wi = 0,				\
> +		     __nw = scx_cmask_nr_used_words(m);				\
> +	     __wi < __nw; __wi++)						\
> +		for (u64 __w = READ_ONCE((m)->bits[__wi]);			\
> +		     __w && ((cid) = __bs + __wi * 64 + __ffs64(__w), true);	\
> +		     __w &= __w - 1)

[Severity: High]
Since this macro expands into two nested for loops, what happens if a caller
uses a break statement to terminate the iteration early?

It appears a break will only exit the inner loop, terminating the scan for
the current 64-bit word. The outer loop will then increment, fetch the next
word, and continue iterating.

Could this create a trap where a standard break statement inadvertently skips
the rest of the current word but silently resumes yielding items on the next
word, violating standard C iteration semantics?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260604020032.3536466-1-tj@kernel.org?part=3