thunderbolt: Do not enumerate more ports from DROM than the controller has
authorMika Westerberg <mika.westerberg@linux.intel.com>
Tue, 25 Jul 2017 14:41:58 +0000 (17:41 +0300)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 10 Aug 2017 21:25:35 +0000 (14:25 -0700)
Some Alpine Ridge LP DROMs (there might be others) erroneusly list more
ports than the controller actually has. Most probably because DROM of
the full Dual/Single port Thunderbolt controller was reused for LP
version. The current DROM parser does not check the upper bound thus it
leads to crash when sw->ports[] is accessed over bounds:

 BUG: unable to handle kernel NULL pointer dereference at 00000000000002ec
 IP: tb_drom_read+0x383/0x890 [thunderbolt]
 PGD 0
 P4D 0
 Oops: 0000 [#1] SMP
 CPU: 3 PID: 12248 Comm: systemd-udevd Not tainted 4.13.0-rc1-next-20170719 #1
 Hardware name: LENOVO 20HF000YGE/20HF000YGE, BIOS N1WET32W (1.11 ) 05/23/2017
 task: ffff8a293e4bcd80 task.stack: ffffa698027a8000
 RIP: 0010:tb_drom_read+0x383/0x890 [thunderbolt]
 RSP: 0018:ffffa698027ab990 EFLAGS: 00010246
 RAX: 0000000000000000 RBX: ffff8a2940af7800 RCX: 0000000000000000
 RDX: ffff8a2940ebb400 RSI: 0000000000000000 RDI: ffffa698027ab9a0
 RBP: ffffa698027ab9d0 R08: 0000000000000001 R09: 0000000000000002
 R10: ffff8a2940ebb5b0 R11: 0000000000000000 R12: ffff8a293bfa968c
 R13: 000000000000002c R14: 0000000000000056 R15: 0000000000000056
 FS:  00007f0a945a38c0(0000) GS:ffff8a2961580000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00000000000002ec CR3: 000000043e785000 CR4: 00000000003606e0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
 Call Trace:
  tb_switch_add+0x9d/0x730 [thunderbolt]
  ? tb_switch_alloc+0x3cd/0x4d0 [thunderbolt]
  icm_start+0x5a/0xa0 [thunderbolt]
  tb_domain_add+0xc3/0xf0 [thunderbolt]
  nhi_probe+0x19e/0x310 [thunderbolt]
  local_pci_probe+0x42/0xa0
  pci_device_probe+0x18d/0x1a0
  driver_probe_device+0x2ff/0x450
  __driver_attach+0xa4/0xe0
  ? driver_probe_device+0x450/0x450
  bus_for_each_dev+0x6e/0xb0
  driver_attach+0x1e/0x20
  bus_add_driver+0x1d0/0x270
  ? 0xffffffffc0bbb000
  driver_register+0x60/0xe0
  ? 0xffffffffc0bbb000
  __pci_register_driver+0x4c/0x50
  nhi_init+0x28/0x1000 [thunderbolt]
  do_one_initcall+0x50/0x190
  ? __vunmap+0x81/0xb0
  ? _cond_resched+0x1a/0x50
  ? kmem_cache_alloc_trace+0x15f/0x1c0
  ? do_init_module+0x27/0x1e9
  do_init_module+0x5f/0x1e9
  load_module+0x24e7/0x2a60
  ? vfs_read+0x115/0x130
  SYSC_finit_module+0xfc/0x120
  ? SYSC_finit_module+0xfc/0x120
  SyS_finit_module+0xe/0x10
  do_syscall_64+0x67/0x170
  entry_SYSCALL64_slow_path+0x25/0x25

Fix this by making sure we only enumerate DROM port entries the hardware
actually has.

Reported-by: Christian Kellner <ckellner@redhat.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Lukas Wunner <lukas@wunner.de>
Tested-by: Christian Kellner <ckellner@redhat.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/thunderbolt/eeprom.c

index 308b6e17c88aace0775b4ed0a5460097559d66ca..fe2f00ceafc5d7e1a3bbafea4224c838dc2158a4 100644 (file)
@@ -333,6 +333,15 @@ static int tb_drom_parse_entry_port(struct tb_switch *sw,
        int res;
        enum tb_port_type type;
 
+       /*
+        * Some DROMs list more ports than the controller actually has
+        * so we skip those but allow the parser to continue.
+        */
+       if (header->index > sw->config.max_port_number) {
+               dev_info_once(&sw->dev, "ignoring unnecessary extra entries in DROM\n");
+               return 0;
+       }
+
        port = &sw->ports[header->index];
        port->disabled = header->port_disabled;
        if (port->disabled)