CVE-2024-56677

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> powerpc/fadump: Move fadump_cma_init to setup_arch() after initmem_init()<br /> <br /> During early init CMA_MIN_ALIGNMENT_BYTES can be PAGE_SIZE,<br /> since pageblock_order is still zero and it gets initialized<br /> later during initmem_init() e.g.<br /> setup_arch() -&gt; initmem_init() -&gt; sparse_init() -&gt; set_pageblock_order()<br /> <br /> One such use case where this causes issue is -<br /> early_setup() -&gt; early_init_devtree() -&gt; fadump_reserve_mem() -&gt; fadump_cma_init()<br /> <br /> This causes CMA memory alignment check to be bypassed in<br /> cma_init_reserved_mem(). Then later cma_activate_area() can hit<br /> a VM_BUG_ON_PAGE(pfn &amp; ((1