Re: [PATCH RESEND] checkpatch: don't flag ACQUIRE_ERR() assignments in if conditions

[Alison Schofield <alison.schofield@intel.com>]

On Wed, Sep 16, 2026 at 10:09:21AM +0800, Guixin Liu wrote:
> ACQUIRE_ERR() and its wrappers, PM_RUNTIME_ACQUIRE_ERR() and
> IIO_DEV_ACQUIRE_FAILED(), report whether a conditional cleanup.h guard
> was acquired, and drivers consume the result directly in an if
> condition:
> 
> 	if ((rc = ACQUIRE_ERR(mutex_intr, &lock)))
> 		return rc;
> 
> That combined form is the established style at the 49 in-tree call
> sites under drivers/cxl and drivers/pci/tsm.c, so ASSIGN_IN_IF fires
> there only as a false positive, and every patch touching those lines
> carries noise that reviewers have to wave off manually.
> 
> Skip the check when the assigned expression is a call to a macro from
> that family, matched by the *_ACQUIRE_ERR() / *_ACQUIRE_FAILED() naming
> convention of its wrappers. Plain assignments and near-miss identifiers
> still get flagged.

Hi Guixin,

Glad to see you picking this up!

I tried to address the same checkpatch false positive last year. Joe Perches
gave some useful feedback on my v1, including a suggestion to use checkpatch's
$Lval, and I incorporated that into v2 along with some test cases. That may
be useful for your next version.

More inline -


> 
> Signed-off-by: Guixin Liu <kanie@linux.alibaba.com>
> ---
> Discussion leading to this patch, where the false positive showed up on
> a cxl patch and covering the wrapper macros was requested as well:
> 
> https://lore.kernel.org/linux-cxl/20260910094017.4032170-3-kanie@linux.alibaba.com/
> 
>  scripts/checkpatch.pl | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> index 8a7787d228a6..fbc500bcfaa1 100755
> --- a/scripts/checkpatch.pl
> +++ b/scripts/checkpatch.pl
> @@ -5777,7 +5777,12 @@ sub process {
>  			my ($s, $c) = ($stat, $cond);
>  			my $fixed_assign_in_if = 0;
>  
> -			if ($c =~ /\bif\s*\(.*[^<>!=]=[^=].*/s) {
> +			# ACQUIRE_ERR() and its wrappers, e.g. PM_RUNTIME_ACQUIRE_ERR()
> +			# and IIO_DEV_ACQUIRE_FAILED(), are intended to be evaluated in
> +			# an if condition, with the error assigned in the condition:
> +			#	if ((rc = ACQUIRE_ERR(name, &lock)))
> +			if ($c =~ /\bif\s*\(.*[^<>!=]=[^=].*/s &&
> +			    $c !~ /=\s*\w*ACQUIRE_(?:ERR|FAILED)\s*\(/) {

What happens if there is also an unrelated assignment in the condition? For example:
	if ((rc = regular_function()) || (ret = ACQUIRE_ERR(lock_type, &lock)))

It looks like the ACQUIRE_ERR() match suppresses ASSIGN_IN_IF for the whole condition,
including the unrelated assignment.

Could you use $Lval, as Joe suggested on my v1, to make this more precise?

I also suggest running this against a small test program with the expected-pass cases,
ordinary assignments, and mixed cases like the one above. Feel free to reuse or extend
the test cases from my v2 [1].

Could you also Cc linux-cxl on the next version? There are several users of this form
there, so it may help get some review and testing tags on the patch.

-- Alison


[1] https://lore.kernel.org/linux-cxl/20250815010645.2980846-1-alison.schofield@intel.com/




>  				if (ERROR("ASSIGN_IN_IF",
>  					  "do not use assignment in if condition\n" . $herecurr) &&
>  				    $fix && $perl_version_ok) {
> 
> base-commit: 50d05c7c76c96b90462f24debacca971d2e86713
> -- 
> 2.43.7
>