> The (proposed) solution is to export store_constructor (and it's helper
> function int_expr_size) from expr.cc, by removing their static qualifier
> and prototyping both functions in expr.h, so they can be called directly
> from load_register_parameters in calls.cc. This cures both ICEs, but
> almost as important produces much better code generation than GCC 12.
>
> For PR 105853, GCC 12 generates:
>
> compose_nd_na_ipv6_src:
> movzx eax, WORD PTR eth_addr_zero[rip+2]
> movzx edx, WORD PTR eth_addr_zero[rip]
> movzx edi, WORD PTR eth_addr_zero[rip+4]
> sal rax, 16
> or rax, rdx
> sal rdi, 32
> or rdi, rax
> xor eax, eax
> jmp packet_set_nd
> eth_addr_zero: .zero 6
>
> where now (with this fix) GCC 13 generates:
> compose_nd_na_ipv6_src:
> xorl %edi, %edi
> xorl %eax, %eax
> jmp packet_set_nd
>
> Likewise, for PR 105856 on ARM, we'd previously generate:
> g_329_3:
> movw r3, #:lower16:.LANCHOR0
> movt r3, #:upper16:.LANCHOR0
> ldr r0, [r3]
> b func_19
>
> but with this optimization we now generate:
> g_329_3:
> mov r0, #6
> b func_19
Nice work! Fiddling with RTL expansion like this is never easy...
> This patch has been tested on x86_64-pc-linux-gnu with make bootstrap
> and make -k check with no new failures. I've also confirmed that on a
> cross-compiler to arm-linux-gnueabihf --with-arch=armv6 this fixes the
> target specific ICE in PR105856. The make check is currently running
> with --target_board=unix{-m32}, OK for mainline if that also passes?
Yes, it looks quite uncontroversial if you ask me and will unblock people.
--
Eric Botcazou
