lkml.org 
[lkml]   [2020]   [Jul]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Date
SubjectRe: [PATCH] tpm: avoid accessing cleared ops during shutdown
On Fri, Jul 10, 2020 at 4:40 AM Jarkko Sakkinen
<jarkko.sakkinen@linux.intel.com> wrote:
>
> On Thu, Jul 09, 2020 at 05:22:09PM -0700, Andrey Pronin wrote:
> > This patch prevents NULL dereferencing when using chip->ops while
> > sending TPM2_Shutdown command if both tpm_class_shutdown handler and
> > tpm_del_char_device are called during system shutdown.
> >
> > Both these handlers set chip->ops to NULL but don't check if it's
> > already NULL when they are called before using it.
> >
> > This issue was revealed in Chrome OS after a recent set of changes
> > to the unregister order for spi controllers, such as:
> > b4c6230bb0ba spi: Fix controller unregister order
> > f40913d2dca1 spi: pxa2xx: Fix controller unregister order
> > and similar for other controllers.
>
> I'm not sure I fully understand the scenario. When does thi happen?

This happens during system shutdown.
Here a sample stack trace from the panic:

BUG: unable to handle kernel NULL pointer dereference at 0000000000000058
...
Call Trace:
tpm_transmit_cmd+0x21/0x7f
tpm2_shutdown+0x84/0xc6
tpm_chip_unregister+0xa2/0xb9
cr50_spi_remove+0x19/0x26
spi_drv_remove+0x2a/0x42
device_release_driver_internal+0x123/0x1ec
bus_remove_device+0xe8/0x111
device_del+0x1bf/0x319
? spi_unregister_controller+0xfc/0xfc
device_unregister+0x12/0x28
__unregister+0xe/0x12
device_for_each_child+0x79/0xb8
spi_unregister_controller+0x27/0xfc
pxa2xx_spi_remove+0x45/0xb4
device_shutdown+0x181/0x1d3
kernel_restart+0x12/0x56
SyS_reboot+0x16a/0x1e7
do_syscall_64+0x6b/0xf7
entry_SYSCALL_64_after_hwframe+0x41/0xa6

> Why does not tpm_del_char_device need this?

"Not" is a typo in the sentence above, right? tpm_del_char_device *does*
need the fix. When tpm_class_shutdown is called it sets chip->ops to
NULL. If tpm_del_char_device is called after that, it doesn't check if
chip->ops is NULL (normal kernel API and char device API calls go
through tpm_try_get_ops, but tpm_del_char_device doesn't) and proceeds to
call tpm2_shutdown(), which tries sending the command and dereferences
chip->ops.

> The changes listed tell
> me nothing. Why they have this effect?

"spi: pxa2xx: Fix controller unregister order" adds a spi_unregister_controller
call to pxa2xx_spi_remove, which internally calls tpm_del_char_device, which
results in the stack trace leading to the panic above.

>
> I'm just trying to understand whether this could be a regression or
> not.
>
> I neither understand what you mean by "and similar for other
> controllers."

There was a series of spi unregister order changes for various
spi controllers, including the following:
1c6221b430a0 spi: bcm2835: Fix controller unregister order
f40913d2dca1 spi: pxa2xx: Fix controller unregister order
b4c6230bb0ba spi: Fix controller unregister order
54000d2e15e9 spi: dw: Fix controller unregister order
c8f309db490e spi: bcm2835aux: Fix controller unregister order

>
> NAK for the reason that I don't understand what I'm merging.
>
> /Jarkko
>
> >
> > Signed-off-by: Andrey Pronin <apronin@chromium.org>
> > ---
> > drivers/char/tpm/tpm-chip.c | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/char/tpm/tpm-chip.c b/drivers/char/tpm/tpm-chip.c
> > index 8c77e88012e9..a410ca40a3c5 100644
> > --- a/drivers/char/tpm/tpm-chip.c
> > +++ b/drivers/char/tpm/tpm-chip.c
> > @@ -296,7 +296,7 @@ static int tpm_class_shutdown(struct device *dev)
> > struct tpm_chip *chip = container_of(dev, struct tpm_chip, dev);
> >
> > down_write(&chip->ops_sem);
> > - if (chip->flags & TPM_CHIP_FLAG_TPM2) {
> > + if (chip->ops && (chip->flags & TPM_CHIP_FLAG_TPM2)) {
> > if (!tpm_chip_start(chip)) {
> > tpm2_shutdown(chip, TPM2_SU_CLEAR);
> > tpm_chip_stop(chip);
> > @@ -479,7 +479,7 @@ static void tpm_del_char_device(struct tpm_chip *chip)
> >
> > /* Make the driver uncallable. */
> > down_write(&chip->ops_sem);
> > - if (chip->flags & TPM_CHIP_FLAG_TPM2) {
> > + if (chip->ops && (chip->flags & TPM_CHIP_FLAG_TPM2)) {
> > if (!tpm_chip_start(chip)) {
> > tpm2_shutdown(chip, TPM2_SU_CLEAR);
> > tpm_chip_stop(chip);
> > --
> > 2.25.1
> >



--
Andrey

\
 
 \ /
  Last update: 2020-07-10 20:26    [W:0.435 / U:0.032 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site