On Mon, Jan 16, 2017 at 3:24 AM, Stefan Liebler <stli@linux.vnet.ibm.com> wrote:
On 01/10/2017 09:22 AM, Stefan Liebler wrote:
In setup_explicit_clear, the buffer is filled with the test_pattern.
On s390x the memcpy in prepare_test_buffer is done by loading
r4 / r5 with the test_pattern and using store multiple instruction
to store r4 / r5 to buf.
If explicit_bzero is resolved in setup_explicit_clear, r4 / r5 is
stored to stack by _dl_runtime_resolve and the call to memmem in
count_test_patterns finds a hit of the test_pattern on the stack.
The attached patch resolves all symbols at program startup by linking
with -z now. This omits the call of _dl_runtime_resolve within
setup_explicit_clear and the test passes.
If this is okay, I'll commit this patch and clear this bug in the
release blockers list in the release-wiki.
This seems like a reasonable workaround to me. Please commit.
(Guess we better add "spill slots for callee-save registers, including
registers saved only by dynamic linker stubs" to the list of things to
worry about when adding explicit_bzero to the compiler...)
zw