Re: [RFC PATCH 0/2] Add cleanup_plugin for detecting problematic cleanup patterns

[Peter Zijlstra]

On Wed, Nov 05, 2025 at 05:04:02PM +0800, Li Chen wrote:
> +Peter, Dan, and Bjorn
> 
> (My apologies for the oversight)
> 
>  ---- On Wed, 05 Nov 2025 16:46:55 +0800  Li Chen <me@linux.beauty> wrote --- 
>  > From: Li Chen <chenl311@xxxxxxxxxxxxxxx>
>  > 
>  > Hello,
>  > 
>  > This patch series introduces a new GCC plugin called cleanup_plugin that
>  > warns developers about problematic patterns when using variables with
>  > __attribute__((cleanup(...))). The plugin addresses concerns documented
>  > in include/linux/cleanup.h regarding resource leaks and interdependency
>  > issues.
>  > 
>  > The cleanup attribute helpers (__free, DEFINE_FREE, etc.) are designed
>  > to automatically clean up resources when variables go out of scope,
>  > following LIFO (last in first out) ordering. However, certain patterns
>  > can lead to subtle bugs:
>  > 
>  > 1. Uninitialized cleanup variables: Variables declared with cleanup
>  >    attributes but not initialized can cause issues when cleanup functions
>  >    are called on undefined values.
>  > 
>  > 2. NULL-initialized cleanup variables: The "__free(...) = NULL" pattern
>  >    at function top can cause interdependency problems, especially when
>  >    combined with guards or multiple cleanup variables, as the cleanup
>  >    may run in unexpected contexts.
>  > 
>  > The plugin detects both of these problematic patterns and provides clear
>  > warnings to developers, helping prevent  incorrect cleanup ordering.
>  > Importantly, the plugin's warnings are not converted
>  > to errors by -Werror, allowing builds to continue while still alerting
>  > developers to potential issues.
>  > 
>  > The plugin is enabled by default as it provides valuable compile-time
>  > feedback without impacting build performance.

IIRC GCC also allow dumb stuff like gotos into the scope of a cleanup
variable, where clang will fail the compile. Does this plugin also fix
this?