Re: [PATCH] gpio / ACPI: Don't crash on NULL chip->dev

[Sabrina Dubroca]

2014-03-31, 15:16:49 +0300, Mika Westerberg wrote:
> Commit aa92b6f689ac (gpio / ACPI: Allocate ACPI specific data directly in
> acpi_gpiochip_add()) moved ACPI handle checking to acpi_gpiochip_add() but
> forgot to check whether chip->dev is NULL before dereferencing it.
> 
> Since chip->dev pointer is optional we can end up with crash like following:
> 
>  BUG: unable to handle kernel NULL pointer dereference at 00000138
>  IP: [<c126c2b3>] acpi_gpiochip_add+0x13/0x190
>  *pde = 00000000
>  Oops: 0000 [#1] PREEMPT SMP
>  Modules linked in: ssb(+) ...
>  CPU: 0 PID: 512 Comm: modprobe Tainted: G        W     3.14.0-rc7-next-20140324-t1 #24
>  Hardware name: Dell Inc. Latitude D830                   /0UY141, BIOS A02 06/07/2007
>  task: f5799900 ti: f543e000 task.ti: f543e000
>  EIP: 0060:[<c126c2b3>] EFLAGS: 00010282 CPU: 0
>  EIP is at acpi_gpiochip_add+0x13/0x190
>  EAX: 00000000 EBX: f57824c4 ECX: 00000000 EDX: 00000000
>  ESI: f57824c4 EDI: 00000010 EBP: f543fc54 ESP: f543fc40
>   DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
>  CR0: 8005003b CR2: 00000138 CR3: 355f8000 CR4: 000007d0
>  Stack:
>   f543fc5c fd1f7790 f57824c4 000000be 00000010 f543fc84 c1269f4e f543fc74
>   fd1f78bd 00008002 f57822b0 f5782090 fd1f8400 00000286 fd1f9994 00000000
>   f5782000 f543fc8c fd1f7e39 f543fcc8 fd1f0bd8 000000c0 00000000 00000000
>  Call Trace:
>   [<fd1f7790>] ? ssb_pcie_mdio_write+0xa0/0xd0 [ssb]
>   [<c1269f4e>] gpiochip_add+0xee/0x300
>   [<fd1f78bd>] ? ssb_pcicore_serdes_workaround+0xfd/0x140 [ssb]
>   [<fd1f7e39>] ssb_gpio_init+0x89/0xa0 [ssb]
>   [<fd1f0bd8>] ssb_attach_queued_buses+0xc8/0x2d0 [ssb]
>   [<fd1f0f65>] ssb_bus_register+0x185/0x1f0 [ssb]
>   [<fd1f3120>] ? ssb_pci_xtal+0x220/0x220 [ssb]
>   [<fd1f106c>] ssb_bus_pcibus_register+0x2c/0x80 [ssb]
>   [<fd1f40dc>] ssb_pcihost_probe+0x9c/0x110 [ssb]
>   [<c1276c8f>] pci_device_probe+0x6f/0xc0
>   [<c11bdb55>] ? sysfs_create_link+0x25/0x40
>   [<c131d8b9>] driver_probe_device+0x79/0x360
>   [<c1276512>] ? pci_match_device+0xb2/0xc0
>   [<c131dc51>] __driver_attach+0x71/0x80
>   [<c131dbe0>] ? __device_attach+0x40/0x40
>   [<c131bd87>] bus_for_each_dev+0x47/0x80
>   [<c131d3ae>] driver_attach+0x1e/0x20
>   [<c131dbe0>] ? __device_attach+0x40/0x40
>   [<c131d007>] bus_add_driver+0x157/0x230
>   [<c131e219>] driver_register+0x59/0xe0
>   ...
> 
> Fix this by checking chip->dev pointer against NULL first. Also we can now
> remove redundant check in acpi_gpiochip_request/free_interrupts().
> 
> Reported-by: Sabrina Dubroca <sd@xxxxxxxxxxxxxxx>
> Signed-off-by: Mika Westerberg <mika.westerberg@xxxxxxxxxxxxxxx>
> ---
> Sabrina,
> 
> Can you please re-test this and provide your tested-by? I changed the patch
> a bit to remove redundant checks. Just to be sure that I don't accidentally
> break something.
> 
> Thanks.

Everything looks good.

Tested-by: Sabrina Dubroca <sd@xxxxxxxxxxxxxxx>


Thanks,
-- 
Sabrina
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/