On Tue, 2020-12-15 at 16:38 +0300, Sergey Temerkhanov wrote:
> Avoid race condition at shutdown by shutting downn the TPM 2.0
> devices synchronously. This eliminates the condition when the
> shutdown sequence sets chip->ops to NULL leading to the following:
>
> [ 1586.593561][ T8669] tpm2_del_space+0x28/0x73
> [ 1586.598718][ T8669] tpmrm_release+0x27/0x33wq
> [ 1586.603774][ T8669] __fput+0x109/0x1d
> [ 1586.608380][ T8669] task_work_run+0x7c/0x90
> [ 1586.613414][ T8669] prepare_exit_to_usermode+0xb8/0x128
> [ 1586.619522][ T8669] entry_SYSCALL_64_after_hwframe+0x44/0xa9
> [ 1586.626068][ T8669] RIP: 0033:0x4cb4bb
An actual bug report would have been helpful. However, from this trace
it's easy to deduce that tpm2_del_space() didn't get converted to the
get/put of the chip ops ... it's still trying to do its own half arsed
thing with tpm_chip_start() and the mutex. So isn't a much simpler fix
simply to convert it as below? compile tested only, but if you can
test it out I'll send a proper patch.
James
---
diff --git a/drivers/char/tpm/tpm2-space.c b/drivers/char/tpm/tpm2-space.c
index 784b8b3cb903..0c0cd225046f 100644
--- a/drivers/char/tpm/tpm2-space.c
+++ b/drivers/char/tpm/tpm2-space.c
@@ -58,12 +58,12 @@ int tpm2_init_space(struct tpm_space *space, unsigned int buf_size)
void tpm2_del_space(struct tpm_chip *chip, struct tpm_space *space)
{
- mutex_lock(&chip->tpm_mutex);
- if (!tpm_chip_start(chip)) {
+
+ if (tpm_try_get_ops(chip) == 0) {
tpm2_flush_sessions(chip, space);
- tpm_chip_stop(chip);
+ tpm_put_ops(chip);
}
- mutex_unlock(&chip->tpm_mutex);
+
kfree(space->context_buf);
kfree(space->session_buf);
}
