On 09/10/15 10:17, Ramana Radhakrishnan wrote:
This started as a Friday afternoon project ...
It turned out enabling VTV for AArch64 and ARM was a matter of fixing PR67868
which essentially comes from building libvtv with section anchors turned on.
The problem was that the flow of control from output_object_block through to
switch_section did not have the same special casing for the vtable section that
exists in assemble_variable.
Once this was done, I managed to build and test aarch64-none-linux-gnu with
--enable-vtable-verify, a similar test was done for armhf.
Testing showed no regressions in the gcc/ g++ testsuites for aarch64 and armhf
Testing showed no failures in libvtv testsuite for aarch64 but a few more
failures - see below.
Testing showed 2 failures in libstdc++-v3 testsuite compared to without vtable
verification.
FAIL: libstdc++-abi/abi_check
FAIL: experimental/filesystem/iterators/directory_iterator.cc execution test
However both these failures also occur on x86_64 - so I'm content to declare
victory on AArch64
as far as basic enablement goes.
On ARM I see the following failures that I still need to debug - I can see that
the failure is because the write to _ZN4_VTVI1BE12__vtable_mapE does not elicit
a SEGV but I need to go further than that.
FAIL: libvtv.cc/thunk_vtable_map_attack.cc -O0 -fvtable-verify=std execution
test
FAIL: libvtv.cc/thunk_vtable_map_attack.cc -O2 -fvtable-verify=std execution
test
FAIL: libvtv.cc/thunk_vtable_map_attack.cc -O0 -fvtable-verify=preinit
execution test
FAIL: libvtv.cc/thunk_vtable_map_attack.cc -O2 -fvtable-verify=preinit
execution test
Questions -
1. Are the generic changes to varasm.c ok ?
2. Can we take the AArch64 support in now, given this amount of testing ?
Marcus / Caroline ?
+ aarch64*-*-linux*)
+ VTV_SUPPORTED=yes
;;
Ramana, Go ahead an add the aarch64 enable once you have the
pre-requisite varasm changes approved.
Cheers
/Marcus
3. Any suggestions / helpful debug hints for VTV debugging (other than turning
VTV_DEBUG on and inspecting trace) ?
There's an arm*-*-* hunk there but I'm easy about applying that right now and
figuring out the issues
over time. In case we don't fix it for 6.0 we can rip the support out before
release.
Thanks,
Ramana
P.S. (Yes, I'll provide a Changelog :) )
