CVE-2024-36943

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fs/proc/task_mmu: fix loss of young/dirty bits during pagemap scan<br /> <br /> make_uffd_wp_pte() was previously doing:<br /> <br /> pte = ptep_get(ptep);<br /> ptep_modify_prot_start(ptep);<br /> pte = pte_mkuffd_wp(pte);<br /> ptep_modify_prot_commit(ptep, pte);<br /> <br /> But if another thread accessed or dirtied the pte between the first 2<br /> calls, this could lead to loss of that information. Since<br /> ptep_modify_prot_start() gets and clears atomically, the following is the<br /> correct pattern and prevents any possible race. Any access after the<br /> first call would see an invalid pte and cause a fault:<br /> <br /> pte = ptep_modify_prot_start(ptep);<br /> pte = pte_mkuffd_wp(pte);<br /> ptep_modify_prot_commit(ptep, pte);