Re: [PATCH v2 12/28] KVM: Introduce NR/NW/NX memory attributes

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] kvm: API Misuse in WARN_ON_ONCE with string literal
--

commit e4ff873cd187bd01c3fc362fdaed04c8d470baa7
Author: Nicolas Saenz Julienne <nsaenz@amazon.com>

KVM: Introduce NR/NW/NX memory attributes

This commit introduces memory attributes to map guest physical memory
regions as non-readable, non-writable, and/or non-executable. It also
adds an architecture-specific callback and a basic implementation for x86.

> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index 50e042c58246f..452ec3a99ba6f 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c

[ ... ]

> +
> +bool __weak kvm_arch_mem_attributes_supported_prot(struct kvm *kvm, unsigned long attrs)
> +{
> +	WARN_ON_ONCE("KVM_MEMORY_ATTRIBUTE_PROT requires kvm_arch_mem_attributes_supported_prot()");

[Severity: Low]
Does this code use the WARN_ON_ONCE macro correctly? Passing a string
literal as the condition evaluates to a non-NULL pointer, which is always
true.

When an architecture implements memory protection attributes without
overriding kvm_arch_mem_attributes_supported_prot(), the weak fallback
triggers this warning unconditionally. It will also fail to format or print
the intended string as a log message, producing a stack trace without the
descriptive message.

Should this perhaps use WARN_ONCE instead to print the descriptive warning,
or pass a proper boolean condition to WARN_ON_ONCE?

> +	return false;
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260918081543.139871-1-pbonzini@redhat.com?part=12