If an application registers rseq, and ever switches to another pkey
protection (such that the rseq becomes inaccessible), then any
context switch will cause failure in __rseq_handle_notify_resume()
attempting to read/write struct rseq and/or rseq_cs. Since context
switches are asynchronous and are outside of the application control
(not part of the restricted code scope), temporarily enable access
to 0 (default) PKEY to read/write rseq/rseq_cs.
0 is the only PKEY supported for rseq for now.
Theoretically other PKEYs can be supported, but it's unclear
how/if that can work. So for now we don't support that to simplify
code.
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Aruna Ramakrishna <aruna.ramakrishna@oracle.com>
Cc: x86@kernel.org
Cc: linux-kernel@vger.kernel.org
Dmitry Vyukov (4):
pkeys: add API to switch to permissive/zero pkey register
x86/signal: Use write_permissive_pkey_val() helper
rseq: Make rseq work with protection keys
selftests/rseq: Add test for rseq+pkeys
arch/x86/Kconfig | 1 +
arch/x86/include/asm/pkeys.h | 30 +++++++
arch/x86/include/asm/pkru.h | 10 ++-
arch/x86/kernel/signal.c | 6 +-
include/linux/pkeys.h | 31 +++++++
include/uapi/linux/rseq.h | 4 +
kernel/rseq.c | 11 +++
mm/Kconfig | 2 +
tools/testing/selftests/rseq/Makefile | 2 +-
tools/testing/selftests/rseq/pkey_test.c | 101 +++++++++++++++++++++++
tools/testing/selftests/rseq/rseq.h | 1 +
11 files changed, 191 insertions(+), 8 deletions(-)
create mode 100644 tools/testing/selftests/rseq/pkey_test.c
base-commit: 4a95bc121ccdaee04c4d72f84dbfa6b880a514b6
--
2.49.0.1143.g0be31eac6b-goog