CVE-2022-48631

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ext4: fix bug in extents parsing when eh_entries == 0 and eh_depth &gt; 0<br /> <br /> When walking through an inode extents, the ext4_ext_binsearch_idx() function<br /> assumes that the extent header has been previously validated. However, there<br /> are no checks that verify that the number of entries (eh-&gt;eh_entries) is<br /> non-zero when depth is &gt; 0. And this will lead to problems because the<br /> EXT_FIRST_INDEX() and EXT_LAST_INDEX() will return garbage and result in this:<br /> <br /> [ 135.245946] ------------[ cut here ]------------<br /> [ 135.247579] kernel BUG at fs/ext4/extents.c:2258!<br /> [ 135.249045] invalid opcode: 0000 [#1] PREEMPT SMP<br /> [ 135.250320] CPU: 2 PID: 238 Comm: tmp118 Not tainted 5.19.0-rc8+ #4<br /> [ 135.252067] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b-rebuilt.opensuse.org 04/01/2014<br /> [ 135.255065] RIP: 0010:ext4_ext_map_blocks+0xc20/0xcb0<br /> [ 135.256475] Code:<br /> [ 135.261433] RSP: 0018:ffffc900005939f8 EFLAGS: 00010246<br /> [ 135.262847] RAX: 0000000000000024 RBX: ffffc90000593b70 RCX: 0000000000000023<br /> [ 135.264765] RDX: ffff8880038e5f10 RSI: 0000000000000003 RDI: ffff8880046e922c<br /> [ 135.266670] RBP: ffff8880046e9348 R08: 0000000000000001 R09: ffff888002ca580c<br /> [ 135.268576] R10: 0000000000002602 R11: 0000000000000000 R12: 0000000000000024<br /> [ 135.270477] R13: 0000000000000000 R14: 0000000000000024 R15: 0000000000000000<br /> [ 135.272394] FS: 00007fdabdc56740(0000) GS:ffff88807dd00000(0000) knlGS:0000000000000000<br /> [ 135.274510] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> [ 135.276075] CR2: 00007ffc26bd4f00 CR3: 0000000006261004 CR4: 0000000000170ea0<br /> [ 135.277952] Call Trace:<br /> [ 135.278635] <br /> [ 135.279247] ? preempt_count_add+0x6d/0xa0<br /> [ 135.280358] ? percpu_counter_add_batch+0x55/0xb0<br /> [ 135.281612] ? _raw_read_unlock+0x18/0x30<br /> [ 135.282704] ext4_map_blocks+0x294/0x5a0<br /> [ 135.283745] ? xa_load+0x6f/0xa0<br /> [ 135.284562] ext4_mpage_readpages+0x3d6/0x770<br /> [ 135.285646] read_pages+0x67/0x1d0<br /> [ 135.286492] ? folio_add_lru+0x51/0x80<br /> [ 135.287441] page_cache_ra_unbounded+0x124/0x170<br /> [ 135.288510] filemap_get_pages+0x23d/0x5a0<br /> [ 135.289457] ? path_openat+0xa72/0xdd0<br /> [ 135.290332] filemap_read+0xbf/0x300<br /> [ 135.291158] ? _raw_spin_lock_irqsave+0x17/0x40<br /> [ 135.292192] new_sync_read+0x103/0x170<br /> [ 135.293014] vfs_read+0x15d/0x180<br /> [ 135.293745] ksys_read+0xa1/0xe0<br /> [ 135.294461] do_syscall_64+0x3c/0x80<br /> [ 135.295284] entry_SYSCALL_64_after_hwframe+0x46/0xb0<br /> <br /> This patch simply adds an extra check in __ext4_ext_check(), verifying that<br /> eh_entries is not 0 when eh_depth is &gt; 0.