Revert "ARM64: unwind: Fix PC calculation"
authorWill Deacon <will.deacon@arm.com>
Wed, 28 Oct 2015 16:56:13 +0000 (16:56 +0000)
committerWill Deacon <will.deacon@arm.com>
Wed, 28 Oct 2015 17:07:07 +0000 (17:07 +0000)
commit9702970c7bd3e2d6fecb642a190269131d4ac16c
treeaa2ce8d5878c8ba7072e368062f9872437561e79
parente13d918a19a7b6cba62b32884f5e336e764c2cc6
Revert "ARM64: unwind: Fix PC calculation"

This reverts commit e306dfd06fcb44d21c80acb8e5a88d55f3d1cf63.

With this patch applied, we were the only architecture making this sort
of adjustment to the PC calculation in the unwinder. This causes
problems for ftrace, where the PC values are matched against the
contents of the stack frames in the callchain and fail to match any
records after the address adjustment.

Whilst there has been some effort to change ftrace to workaround this,
those patches are not yet ready for mainline and, since we're the odd
architecture in this regard, let's just step in line with other
architectures (like arch/arm/) for now.

Cc: <stable@vger.kernel.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
arch/arm64/kernel/stacktrace.c