try to document c_stdlib_version in knowledge base

docs/maintainer/knowledge_base.md

@@ -1928,6 +1928,34 @@ os_version:
 
 Finally, note that the `aarch64` and `ppc64le` platforms already use CentOS 7.
 
+<a id="consequences-of-newer-c-stdlib-version"></a>
+
+### Consequences of newer c_stdlib_version
+
+If a library links a newer version of `c_stdlib_version`,
+the _runtime_ requirement on `__glibc` is applied,
+but it does not mean downstream dependencies will (or should) be built against the same version.
+The default `c_stdlib_version` is still `2.12` on linux-64.
+**A dependency on a newer c_stdlib_version does not mean dependents must also use that version**.
+However, to build a package with `c_stdlib_version=2.12` that links a library built with `c_stdlib_version=2.17`,
+the linker flag `-Wl,--allow-shlib-undefined` must be specified.
+Without `--allow-shlib-undefined`, you may see errors like:
+
+```
+ld: $PREFIX/lib/libc++abi.so: undefined reference to `memcpy@GLIBC_2.14'
+ld: $PREFIX/lib/libpmix.so.2: undefined reference to `clock_gettime@GLIBC_2.17'
+```
+
+This flag is in conda-forge's default $LDFLAGS, so most recipes should not have any issue.
+If they do, it means $LDFLAGS is not being handled consistently,
+which is probably something feedstock and/or package maintainers should consider fixing.
+
+There is more nuance in runtime compilation in user environments,
+which are also affected by the default version of `sysroot_linux-64` being 2.12.
+The `gcc` package does not set environment variables $CC, $LDFLAGS, etc..
+To set these variables on environment activation, install the `c-compiler` package (or `gcc_linux-64` on linux-64),
+or set them by hand in the environment, or avoid the issue by specifying `sysroot_linux-64>=2.17` in the environment.
+
 <a id="cuda"></a>
 
 <a id="cuda-builds"></a>