Thanks a lot Leo! A good finding! Both of a them looks good.
The bug was introduced by PSCI. So it's clear in LTS.
Thanks Alex
On 03/25/2016 11:22 PM, Leo Yan wrote:
Hi Alex,
Below two patches are important for CPU's suspend/resume flow; These two patches have been merged into mainline kernel already.
If without Sudeep's patch, cpu will get wrong context when it returns back and finally introduce system hang issue. So patch 1 is fatal.
For Will Deacon's patch, I have not seen obvious issue if without it on Hikey. But for more safe, also ported this patch to fetch correct instructions from PoU but discard potential stale instructions which fetched from PoC.
Sudeep Holla (1): arm64: restore cpu suspend/resume functionality
Will Deacon (1): arm64: mm: ensure patched kernel text is fetched from PoU
arch/arm64/kernel/head.S | 8 ++++++++ arch/arm64/kernel/sleep.S | 9 ++++++++- arch/arm64/mm/proc.S | 1 - 3 files changed, 16 insertions(+), 2 deletions(-)