Verified Commit a7a94d35 authored by Levente Polyak's avatar Levente Polyak 🚀
Browse files

Merge remote-tracking branch 'allan/buildflags'

parents 33d8b5fd d1cf7187
======================
Updates to build flags
======================
- Date proposed: 2020-03-04
- RFC MR: https://gitlab.archlinux.org/archlinux/rfcs/-/merge_requests/0003
Summary
-------
Updating our buildflags will improve our packages security
Motivation
----------
Compiler security features have advanced since we last updated our buildflags.
While we have enabled some of these by default in our compilers, there is
further improvements to be made.
This RFC puts forward a set of compiler flags that have seen real world usage
in other distributions. e.g. [1] [2]
[1] https://src.fedoraproject.org/rpms/redhat-rpm-config/blob/master/f/buildflags.md.
[2] https://wiki.ubuntu.com/ToolChain/CompilerFlags
Specification
-------------
We will change the distributed makepkg.conf to the following:
::
#CPPFLAGS=""
CFLAGS="-march=x86-64 -mtune=generic -O2 -pipe -fno-plt -fexceptions \
-Wp,-D_FORTIFY_SOURCE=2,-D_GLIBCXX_ASSERTIONS -Werror=format-security \
-fstack-clash-protection -fcf-protection"
CXXFLAGS="$CFLAGS"
LDFLAGS="-Wl,-O1,--sort-common,--as-needed,-z,relro,-z,now"
DEBUG_CFLAGS="-g -fvar-tracking-assignments"
Details of new additions:
* Move ``-D_FORTIFY_SOURCE=2`` from CPPFLAGS to CFLAGS using -Wp
Unfortunately, there are still build systems which use CFLAGS but not CPPFLAGS.
Or configure tests that use CPPFLAGS and not CFLAGS which creates errors due to
fortify source needing optimisation. Ultimately, we can cover more code with
this workaround.
* Add ``-fexceptions``
Provide exception unwinding support for C programs. This also hardens cancellation
handling in C programs, and makes it possible to unwind the stack (using C++ throw
or Rust panics) from C callback functions
* Add ``-Wp,-D_GLIBCXX_ASSERTIONS``
Enable lightweight assertions in the C++ standard library. This flag is added to
both CFLAGS and CXXFLAGS; C compilations will simply ignore it.
* Add ``-Werror=format-security``
Turn on format string warnings and treat them as errors. This catches a set of
readily exploitable bugs. This can occasionally result in compilation errors,
but this should be minimal as it is widely used by other distributions.
Alternatives Considered
-----------------------
* Add ``-grecord-gcc-switches`` to our debug builds
This is the default in GCC.
* Add ``-fasynchronous-unwind-tables`` to our debug builds
This is the default in GCC, at least for x86_64.
* Add ``-fstack-clash-protection``
Ensures all variable length memory allocated from the stack (via alloca() or
gcc variable length arrays etc) are probed at the time they are allocated. This
mitigates stack-clash attacks by ensuring all stack memory allocations are
valid (or by raising a segmentation fault if they are not, and turning a
possible code-execution attack into a denial of service). Without this flag,
vulnerabilities can result where the stack overlaps with the heap, or thread
stacks spill into other regions of memory.
* Add ``-fcf-protection``
Generates instructions to support Intel's Control-flow Enforcement Technology
(CET). Instrument binaries to guard against ROP/JOP attacks. Used on i686 and
x86_64.
Drawbacks
---------
Moving ``-D_FORTIFY_SOURCE=2`` is zero cost.
Adding ``-Werror=format-security`` may cause limited build issues, but patches are
readily available.
There is a minimal performance overhead of adding ``-Wp,-D_GLIBCXX_ASSERTIONS``,
though many of the added checks are optimised away by the compiler. Adding
-fstack-clash-protection also has very little run-time overhead.
Adding ``-fexceptions`` can produce some data size overhead in C programs, though
does not affect execution. GCC enables it by default for C++.
Using -fcf-protection is incompatible with -mindirect-branch (which is used
to implement retpoline). In such cases it is recommended to disable
-fcf-protection. Disabled with -fcf-protection=none in CFLAGS / CXXFLAGS.
Unresolved Questions
--------------------
None.
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment