bgra-rgba-swizzle-mismatch

Status: shipped (Phase 3) — see CHANGELOG.

(via ADR 0011)

What it detects

A shader that reads .rgba (or any subset like .r, .rgb, .gb) from a Texture2D<float4> (or analogous typed view) whose underlying resource format, as reported by Slang reflection, is DXGI_FORMAT_B8G8R8A8_UNORM, DXGI_FORMAT_B8G8R8A8_UNORM_SRGB, or DXGI_FORMAT_B8G8R8X8_UNORM, without a corresponding .bgra swizzle to compensate for the BGRA-to-RGBA channel order. The detector cross-references the sample call site's swizzle (or absence thereof) with the resource format from reflection. It does not fire on RGBA-format resources, nor on BGRA resources where the shader explicitly swizzles .bgra (or applies an equivalent component reorder).

Why it matters on a GPU

DXGI_FORMAT_B8G8R8A8_UNORM is the historical swap-chain format for D3D presentation; many older paths (especially UI / IMGUI overlays composited onto the swap-chain) still use BGRA8 throughout. The hardware texture sampler on AMD RDNA 2/3, NVIDIA Turing/Ada, and Intel Xe-HPG reads BGRA8 storage and presents it to the shader as a float4 whose .x lane carries the blue channel and .z lane carries the red channel — the storage order, not the conceptual RGBA order. There is no hardware swizzle on the load path on D3D12; the format-to-shader-view mapping is exactly the storage layout. (Vulkan exposes a per-view componentMapping swizzle that can compensate at descriptor creation; D3D12 does not.)

The bug surfaces when the shader treats the loaded float4 as if .r is conceptually "red", uses it in subsequent colour math, and writes it to an RGBA8 render target. Red and blue channels swap silently. Visually the regression is striking — sky becomes orange, skin becomes blue — but the cause is buried in the BGRA-vs-RGBA convention difference between the swap-chain texture and the asset textures. UI overlays that mix swap-chain BGRA with asset RGBA are the canonical source of this bug.

The rule pulls the format/access mismatch up to lint time. The fix is one of: swizzle .bgra (or .bgr, etc.) at the sample site to convert BGRA storage to conceptual RGBA, or change the resource format to RGBA8 if the BGRA storage is incidental rather than required. On Vulkan/Metal targets the equivalent is to set the descriptor's component swizzle once at view creation, but the shader-side bug surfaces identically when the descriptor swizzle is left at identity.

Examples

Bad

hlsl

Texture2D<float4> SwapChainCopy : register(t0);  // DXGI_FORMAT_B8G8R8A8_UNORM
SamplerState      LinearSampler : register(s0);

float4 ps_main(float2 uv : TEXCOORD0) : SV_Target {
    // .rgba on a BGRA8 storage delivers (B, G, R, A) — colours look swapped.
    return SwapChainCopy.Sample(LinearSampler, uv).rgba;
}

Good

hlsl

Texture2D<float4> SwapChainCopy : register(t0);  // DXGI_FORMAT_B8G8R8A8_UNORM
SamplerState      LinearSampler : register(s0);

float4 ps_main(float2 uv : TEXCOORD0) : SV_Target {
    // Swizzle to compensate for the BGRA storage order.
    return SwapChainCopy.Sample(LinearSampler, uv).bgra;
}

Options

none

Fix availability

suggestion — The fix is mechanical (.rgba -> .bgra at every load site against this resource) but its correctness depends on the application's intent for the BGRA format binding. The diagnostic identifies the mismatch and the candidate swizzle.

The v1.2 foundation (ADR 0019) wires this rule to the new ResourceBinding::dxgi_format field: when reflection surfaces a BGRA-suffixed format string (B8G8R8A8) for a texture binding, the rule fires anchored at that binding's declaration; otherwise it stays silent. The Slang 2026.7.1 ABI does not surface BGRA channel order through TypeReflection::getResourceResultType() (the typed-resource template arg is float4, indistinguishable from RGBA), so dxgi_format won't contain B8G8R8A8 in practice today. The probe is forward-compatible — when a future Slang surfaces channel order, this rule auto-fires with no further code change. The fix stays suggestion-only because the swizzle rewrite point is the read-site post-fix (.Sample(...).rgba), which the reflection-stage dispatch doesn't currently surface as a TextEdit.

bgra-rgba-swizzle-mismatch ​

What it detects ​

Why it matters on a GPU ​

Examples ​

Bad ​

Good ​

Options ​

Fix availability ​

See also ​