The issue is related to the mm/vmalloc component of the Linux kernel, where the ` vmap pages range noflush()` function assumes that its argument `pages**` contains pages with the same page shift. However, if `gfp flags` includes ` GFP NOFAIL` with high order in `vm area alloc pages()` and page allocation fails for high order, `pages**` may contain two different page shifts (high order and order-0), leading to incorrect mappings and potentially resulting in memory corruption. Users might encounter this issue when using `kvmalloc()` with ` GFP NOFAIL` and `GFP X` flags, and `vmap allow huge` is set to true.