As a followup to JDK-8325881, we should be able to clean up some old code. I've searched the code base for references to GCC which seem to indicate that a piece of code is inferior to the ideal due to limitations in GCC. Maybe not all these are yet possible to fix, and maybe the comment do no longer apply and should be removed instead.

Here is what I found:

* src/java.base/linux/native/libsimdsort/simdsort-support.hpp:34
// GCC >= 9.1 is needed to build AVX2 portions of libsimdsort using C++17 features

* src/hotspot/os_cpu/linux_aarch64/atomic_linux_aarch64.hpp:42
// This really ought to be a template definition, but see GCC Bug
// 33661, template methods forget explicit local register asm
// vars. The problem is that register specifiers attached to local
// variables are ignored in any template function.

* src/hotspot/cpu/s390/assembler_s390.inline.hpp:1455
// This assignment is there to make gcc8 happy.

* src/hotspot/cpu/s390/copy_s390.hpp:307
// Waiting for gcc to support EXRL.

* src/hotspot/share/classfile/stackMapTableFormat.hpp:48 (and also line 178, 909 and 941)
// No constructors - should be 'private', but GCC issues a warning if it is

* src/hotspot/os_cpu/linux_ppc/orderAccess_linux_ppc.hpp:35
// Compiler version last used for testing: gcc 4.1.2

* src/hotspot/os_cpu/linux_x86/orderAccess_linux_x86.hpp:30
// Compiler version last used for testing: gcc 4.8.2