The current implementation checks repeatedly if `tc-is-lto` is true. In addition to running a small compile check redundantly, this has a
potential correctness downside. An ebuild could do this:
```
lto-guarantee-fat
tc-is-lto && myeconfargs+=( --enable-lto )
filter-lto
```
The idea is, the configure script has special handling and simply adding
-flto is too "simple" to work. e.g. not all sources build with LTO.
When this happens, strip-lto-bytecode "thinks" we didn't build with LTO
at all, and returns early. By tracking state across functions -- and
across phases -- we can reliably tell if `lto-guarantee-fat` actually "succeeded".
Bug:
https://bugs.gentoo.org/958412
Acked-by: Sam James <
sam@gentoo.org>
Signed-off-by: Eli Schwartz <
eschwartz@gentoo.org>
---
eclass/dot-a.eclass | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/eclass/dot-a.eclass b/eclass/dot-a.eclass
index ffd34d9e08aa..8d866e6dedb2 100644
--- a/eclass/dot-a.eclass
+++ b/eclass/dot-a.eclass
@@ -50,6 +50,12 @@ _DOT_A_ECLASS=1
inherit flag-o-matic toolchain-funcs
+# @VARIABLE: _DOT_A_IS_LTO
+# @INTERNAL
+# @DESCRIPTION:
+# Records the state of tc-is-lto across eclass function calls. +_DOT_A_IS_LTO=0
+
# TODO: QA check
# @FUNCTION: lto-guarantee-fat
@@ -59,6 +65,7 @@ inherit flag-o-matic toolchain-funcs
lto-guarantee-fat() {
tc-is-lto || return
+ _DOT_A_IS_LTO=1
# We add this for all languages as LTO obviously can't be done
# if different compilers are used for e.g. C vs C++ anyway.
append-flags $(test-flags-CC -ffat-lto-objects)
@@ -73,7 +80,9 @@ lto-guarantee-fat() {
# As an optimisation, if USE=static-libs exists for a package and is disabled,
# the default-searching behaviour with no arguments is suppressed.
s