Skip to content

Fix problems with XMVectorSwizzle/XMVectorPermute for GCC #306

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
walbourn merged 3 commits into from
May 14, 2026
+30 −16
Merged

Fix problems with XMVectorSwizzle/XMVectorPermute for GCC #306

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c298d5d
Fix problems with XMVectorSwizzle/XMVectorPermute for MinGW
walbourn May 13, 2026
83965ea
Fix applies to all GCC
walbourn May 13, 2026
feaeaf5
Comment update
walbourn May 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
46 changes: 30 additions & 16 deletions Inc/DirectXMathVector.inl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1239,18 +1239,23 @@ inline XMVECTOR XM_CALLCONV XMVectorSwizzle
unsigned int elem[4] = { E0, E1, E2, E3 };
__m128i vControl = _mm_loadu_si128(reinterpret_cast<const __m128i*>(&elem[0]));
return _mm_permutevar_ps(V, vControl);
#else
#if defined(__GNUC__) && !defined(__clang__)
// workaround some GCC optimization behavior that breaks this function
XMVECTORU32 T;
T.v = V;
auto aPtr = reinterpret_cast<const uint32_t*>(&T);
#else
auto aPtr = reinterpret_cast<const uint32_t*>(&V);
#endif

XMVECTOR Result;
auto pWork = reinterpret_cast<uint32_t*>(&Result);

pWork[0] = aPtr[E0];
pWork[1] = aPtr[E1];
pWork[2] = aPtr[E2];
pWork[3] = aPtr[E3];
XMVECTORU32 vResult;
vResult.u[0] = aPtr[E0];
vResult.u[1] = aPtr[E1];
vResult.u[2] = aPtr[E2];
vResult.u[3] = aPtr[E3];

return Result;
return vResult.v;
#endif
}

Expand Down Expand Up @@ -1313,29 +1318,38 @@ inline XMVECTOR XM_CALLCONV XMVectorPermute
#else

const uint32_t* aPtr[2];

#if defined(__GNUC__) && !defined(__clang__)
// workaround some GCC optimization behavior that breaks this function
XMVECTORU32 T1;
T1.v = V1;
XMVECTORU32 T2;
T2.v = V2;
aPtr[0] = reinterpret_cast<const uint32_t*>(&T1);
aPtr[1] = reinterpret_cast<const uint32_t*>(&T2);
#else
aPtr[0] = reinterpret_cast<const uint32_t*>(&V1);
aPtr[1] = reinterpret_cast<const uint32_t*>(&V2);
#endif

XMVECTOR Result;
auto pWork = reinterpret_cast<uint32_t*>(&Result);

XMVECTORU32 vResult;
const uint32_t i0 = PermuteX & 3;
const uint32_t vi0 = PermuteX >> 2;
pWork[0] = aPtr[vi0][i0];
vResult.u[0] = aPtr[vi0][i0];

const uint32_t i1 = PermuteY & 3;
const uint32_t vi1 = PermuteY >> 2;
pWork[1] = aPtr[vi1][i1];
vResult.u[1] = aPtr[vi1][i1];

const uint32_t i2 = PermuteZ & 3;
const uint32_t vi2 = PermuteZ >> 2;
pWork[2] = aPtr[vi2][i2];
vResult.u[2] = aPtr[vi2][i2];

const uint32_t i3 = PermuteW & 3;
const uint32_t vi3 = PermuteW >> 2;
pWork[3] = aPtr[vi3][i3];
vResult.u[3] = aPtr[vi3][i3];

return Result;
return vResult.v;
#endif
}

Expand Down
Loading