CVE-2022-49760

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> mm/hugetlb: fix PTE marker handling in hugetlb_change_protection()<br /> <br /> Patch series "mm/hugetlb: uffd-wp fixes for hugetlb_change_protection()".<br /> <br /> Playing with virtio-mem and background snapshots (using uffd-wp) on<br /> hugetlb in QEMU, I managed to trigger a VM_BUG_ON(). Looking into the<br /> details, hugetlb_change_protection() seems to not handle uffd-wp correctly<br /> in all cases.<br /> <br /> Patch #1 fixes my test case. I don&amp;#39;t have reproducers for patch #2, as it<br /> requires running into migration entries.<br /> <br /> I did not yet check in detail yet if !hugetlb code requires similar care.<br /> <br /> <br /> This patch (of 2):<br /> <br /> There are two problematic cases when stumbling over a PTE marker in<br /> hugetlb_change_protection():<br /> <br /> (1) We protect an uffd-wp PTE marker a second time using uffd-wp: we will<br /> end up in the "!huge_pte_none(pte)" case and mess up the PTE marker.<br /> <br /> (2) We unprotect a uffd-wp PTE marker: we will similarly end up in the<br /> "!huge_pte_none(pte)" case even though we cleared the PTE, because<br /> the "pte" variable is stale. We&amp;#39;ll mess up the PTE marker.<br /> <br /> For example, if we later stumble over such a "wrongly modified" PTE marker,<br /> we&amp;#39;ll treat it like a present PTE that maps some garbage page.<br /> <br /> This can, for example, be triggered by mapping a memfd backed by huge<br /> pages, registering uffd-wp, uffd-wp&amp;#39;ing an unmapped page and (a)<br /> uffd-wp&amp;#39;ing it a second time; or (b) uffd-unprotecting it; or (c)<br /> unregistering uffd-wp. Then, ff we trigger fallocate(FALLOC_FL_PUNCH_HOLE)<br /> on that file range, we will run into a VM_BUG_ON:<br /> <br /> [ 195.039560] page:00000000ba1f2987 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x0<br /> [ 195.039565] flags: 0x7ffffc0001000(reserved|node=0|zone=0|lastcpupid=0x1fffff)<br /> [ 195.039568] raw: 0007ffffc0001000 ffffe742c0000008 ffffe742c0000008 0000000000000000<br /> [ 195.039569] raw: 0000000000000000 0000000000000000 00000001ffffffff 0000000000000000<br /> [ 195.039569] page dumped because: VM_BUG_ON_PAGE(compound &amp;&amp; !PageHead(page))<br /> [ 195.039573] ------------[ cut here ]------------<br /> [ 195.039574] kernel BUG at mm/rmap.c:1346!<br /> [ 195.039579] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI<br /> [ 195.039581] CPU: 7 PID: 4777 Comm: qemu-system-x86 Not tainted 6.0.12-200.fc36.x86_64 #1<br /> [ 195.039583] Hardware name: LENOVO 20WNS1F81N/20WNS1F81N, BIOS N35ET50W (1.50 ) 09/15/2022<br /> [ 195.039584] RIP: 0010:page_remove_rmap+0x45b/0x550<br /> [ 195.039588] Code: [...]<br /> [ 195.039589] RSP: 0018:ffffbc03c3633ba8 EFLAGS: 00010292<br /> [ 195.039591] RAX: 0000000000000040 RBX: ffffe742c0000000 RCX: 0000000000000000<br /> [ 195.039592] RDX: 0000000000000002 RSI: ffffffff8e7aac1a RDI: 00000000ffffffff<br /> [ 195.039592] RBP: 0000000000000001 R08: 0000000000000000 R09: ffffbc03c3633a08<br /> [ 195.039593] R10: 0000000000000003 R11: ffffffff8f146328 R12: ffff9b04c42754b0<br /> [ 195.039594] R13: ffffffff8fcc6328 R14: ffffbc03c3633c80 R15: ffff9b0484ab9100<br /> [ 195.039595] FS: 00007fc7aaf68640(0000) GS:ffff9b0bbf7c0000(0000) knlGS:0000000000000000<br /> [ 195.039596] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> [ 195.039597] CR2: 000055d402c49110 CR3: 0000000159392003 CR4: 0000000000772ee0<br /> [ 195.039598] PKRU: 55555554<br /> [ 195.039599] Call Trace:<br /> [ 195.039600] <br /> [ 195.039602] __unmap_hugepage_range+0x33b/0x7d0<br /> [ 195.039605] unmap_hugepage_range+0x55/0x70<br /> [ 195.039608] hugetlb_vmdelete_list+0x77/0xa0<br /> [ 195.039611] hugetlbfs_fallocate+0x410/0x550<br /> [ 195.039612] ? _raw_spin_unlock_irqrestore+0x23/0x40<br /> [ 195.039616] vfs_fallocate+0x12e/0x360<br /> [ 195.039618] __x64_sys_fallocate+0x40/0x70<br /> [ 195.039620] do_syscall_64+0x58/0x80<br /> [ 195.039623] ? syscall_exit_to_user_mode+0x17/0x40<br /> [ 195.039624] ? do_syscall_64+0x67/0x80<br /> [ 195.039626] entry_SYSCALL_64_after_hwframe+0x63/0xcd<br /> [ 195.039628] RIP: 0033:0x7fc7b590651f<br /> [ 195.039653] Code: [...]<br /> [ 195.039654] RSP: 002b:00007fc7aaf66e70 EFLAGS: 00000293 ORIG_RAX: 000000000000011d<br /> [ 195.039655] RAX: ffffffffffffffda RBX: 0000558ef4b7f370 RCX: 00007fc7b590651f<br /> ---truncated---