https://bugs.kde.org/show_bug.cgi?id=517974

Paul Floyd <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[email protected]

--- Comment #1 from Paul Floyd <[email protected]> ---
Most of my epxerience is with FreeBSD (amd64, x86 and arm64) and Darwin (amd64
and x86, a but with arm64 using Louis Brunner's macOS fork of Valgrind). On all
those platforms clang is the system compiler (since 2012 and 2007
respectively). Apple clang has a lot of changes compared to upstream clang. In
particular that means that Apple clang DWARF is a bit different to upstream
clang DWARF. FreeBSD clang usually just has a small patch set that eventually
gets accepted upstream.

As far as I know most of these failures are due to debuginfo. I don't know
DWARF well enough to say whether this is because LLVM DWARF lacks elements that
are present GCC or if the information is there but our DWARF reader is skipping
over it. I suspect the latter.

The main platforms that I see that have issues are
Linux arm and arm64 GCC
Darwin

On FreeBSD my policy has been to accept a small number of tests that don't give
as good results as GCC. I'd rather have the test passing with less than perfect
results than failing and then not being able to tell if it ever has any further
regressions. Examples of that are the memcheck varinfo tests where I've added
extra expecteds. Two of them are very minor (paths and frame numbers).

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to