diff options
| author | Ard Biesheuvel | 2018-03-13 15:09:21 +0100 |
|---|---|---|
| committer | Ingo Molnar | 2018-03-13 15:28:29 +0100 |
| commit | 79832f0b5f718e0023d9dd73e6845310609a564d (patch) | |
| tree | 676d06e2e0b97673702dcc7ac8e5a0bfd41028c5 /fs/btrfs/hash.c | |
| parent | Merge tag 'nfs-for-4.16-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs (diff) | |
| download | kernel-qcow2-linux-79832f0b5f718e0023d9dd73e6845310609a564d.tar.gz kernel-qcow2-linux-79832f0b5f718e0023d9dd73e6845310609a564d.tar.xz kernel-qcow2-linux-79832f0b5f718e0023d9dd73e6845310609a564d.zip | |
efi/libstub/tpm: Initialize pointer variables to zero for mixed mode
As reported by Jeremy Cline, running the new TPM libstub code in mixed
mode (i.e., 64-bit kernel on 32-bit UEFI) results in hangs when invoking
the TCG2 protocol, or when accessing the log_tbl pool allocation.
The reason turns out to be that in both cases, the 64-bit pointer
variables are not fully initialized by the 32-bit EFI code, and so
we should take care to zero initialize these variables beforehand,
or we'll end up dereferencing bogus pointers.
Reported-by: Jeremy Cline <jeremy@jcline.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: hdegoede@redhat.com
Cc: jarkko.sakkinen@linux.intel.com
Cc: javierm@redhat.com
Cc: linux-efi@vger.kernel.org
Cc: tweek@google.com
Link: http://lkml.kernel.org/r/20180313140922.17266-2-ard.biesheuvel@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Diffstat (limited to 'fs/btrfs/hash.c')
0 files changed, 0 insertions, 0 deletions