diff options
| author | Andrey Kleshchev <117672381+akleshchev@users.noreply.github.com> | 2026-01-09 23:03:50 +0200 |
|---|---|---|
| committer | Andrey Kleshchev <117672381+akleshchev@users.noreply.github.com> | 2026-01-10 00:08:12 +0200 |
| commit | d541eb9c300a5e362e059442e60ab83d97050d82 (patch) | |
| tree | 970744da5486171047b466b6df989c9b3589f9a9 /indra/llmath | |
| parent | 03abf07421d393b4bb4ccb24cbc021a60e11cd47 (diff) | |
#5249 Crash at LLVolumeFace::remap()
Crash happens inside generateRemapMultiwith, no clear cause. Ideally we need the crashing model, but untill that, zero-initing remap. meshopt wants zero initialized paddings and since we are providing more space than nessesary, that might count as padding.
Diffstat (limited to 'indra/llmath')
| -rw-r--r-- | indra/llmath/llvolume.cpp | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/indra/llmath/llvolume.cpp b/indra/llmath/llvolume.cpp index 85172ddbf8..ecc5a7b0c5 100644 --- a/indra/llmath/llvolume.cpp +++ b/indra/llmath/llvolume.cpp @@ -5171,7 +5171,7 @@ void LLVolumeFace::remap() // Documentation for meshopt_generateVertexRemapMulti claims that remap should use vertice count // but all examples use indice count. There are out of bounds crashes when using vertice count. // To be on the safe side use bigger of the two. - std::vector<unsigned int> remap(llmax(mNumIndices, mNumVertices)); + std::vector<unsigned int> remap(llmax(mNumIndices, mNumVertices), 0); S32 remap_vertices_count = static_cast<S32>(LLMeshOptimizer::generateRemapMultiU16(&remap[0], mIndices, mNumIndices, |
