sanitizers.supp

## Every suppression in this file should have a comment pointing to
## the corresponding ticket!
##
## Notes:
##
##  - If the issue is in Swift code, please use the `no_sanitize`
##    attribute if possible.  This allows you to be more precise about
##    both what checks and what code to exclude than masking a whole
##    file or a common function name, and it documents the issue right
##    where it is most relevant.
##
##    - https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html
##    - https://clang.llvm.org/docs/AttributeReference.html#no-sanitize
##
##  - Sanitizer suppressions in this list can be applied to templated
##    functions if bracketed by * wildcards.  I (@peddie) am not aware
##    of a way to apply suppressions to templated functions via
##    compiler attributes.  You can be specific to class names and
##    namespaces using `*`.
##
##  - `src` type suppressions (as opposed to `fun` ones) must be given
##    a relative path to the `build` directory (the working directory
##    of the compiler when it's invoked), which is assumed everywhere
##    to be just below the root of the repository.  They do not appear
##    to work for header files!
##
##  - It seems that undefined behavior suppressions cannot be
##    annotated with a specific check type at compile time.  Do not
##    confuse this file with the runtime suppression format hinted at
##    in the clang docs:
##    https://clang.llvm.org/docs/UndefinedBehaviorSanitizer.html#runtime-suppressions

# https://github.com/swift-nav/albatross/issues/382
# in ../third_party/cereal/include/cereal/external/rapidjson/internal/stack.h
fun:*rapidjson*internal*Stack*Reserve*

# https://gitlab.com/libeigen/eigen/-/issues/2482
#
# This is not the exact same sanitizer failure, but the response here
# and in the linked ticket is basically "we are not going to go out of
# our way to satisfy the sanitizer when we know the value is never
# used."  (Maybe they don't know that GCC takes the stance that once
# undefined behavior occurs anywhere in the program, the rest of the
# program may do anything.)
#
# The reporter provided a patch for 3.3 if we really care.
fun:*Eigen*internal*blas_data_mapper*()*