user: implement mlibc as the libc, finally.

It's finally done..

Signed-off-by: kaguya <vpshinomiya@protonmail.com>

user/include/mlibc/options/rtld/aarch64/elf.hpp

@@ -0,0 +1,54 @@
+#pragma once
+
+#include <elf.h>
+
+#define ELF_CLASS ELFCLASS64
+#define ELF_MACHINE EM_AARCH64
+
+using elf_ehdr = Elf64_Ehdr;
+using elf_phdr = Elf64_Phdr;
+using elf_dyn = Elf64_Dyn;
+using elf_rel = Elf64_Rel;
+using elf_rela = Elf64_Rela;
+using elf_relr = Elf64_Relr;
+using elf_sym = Elf64_Sym;
+using elf_addr = Elf64_Addr;
+
+using elf_info = Elf64_Xword;
+using elf_addend = Elf64_Sxword;
+
+using elf_version = Elf64_Half;
+using elf_verdef = Elf64_Verdef;
+using elf_verdaux = Elf64_Verdaux;
+using elf_verneed = Elf64_Verneed;
+using elf_vernaux = Elf64_Vernaux;
+
+#define ELF_R_SYM ELF64_R_SYM
+#define ELF_R_TYPE ELF64_R_TYPE
+#define ELF_ST_BIND ELF64_ST_BIND
+#define ELF_ST_TYPE ELF64_ST_TYPE
+
+#define R_NONE R_AARCH64_NONE
+#define R_JUMP_SLOT R_AARCH64_JUMP_SLOT
+#define R_ABSOLUTE R_AARCH64_ABS64
+#define R_GLOB_DAT R_AARCH64_GLOB_DAT
+#define R_RELATIVE R_AARCH64_RELATIVE
+#define R_IRELATIVE R_AARCH64_IRELATIVE
+// #define R_OFFSET
+#define R_COPY R_AARCH64_COPY
+#define R_TLS_DTPMOD R_AARCH64_TLS_DTPMOD
+#define R_TLS_DTPREL R_AARCH64_TLS_DTPREL
+#define R_TLS_TPREL R_AARCH64_TLS_TPREL
+#define R_TLSDESC R_AARCH64_TLSDESC
+
+#define TP_TCB_OFFSET (16)
+
+struct ifunc_arg {
+	unsigned long _size;
+	unsigned long _hwcap;
+	unsigned long _hwcap2;
+	unsigned long _hwcap3;
+	unsigned long _hwcap4;
+};
+
+using ifunc_handler = elf_addr (*)(uint64_t, ifunc_arg *);

user/include/mlibc/options/rtld/aarch64/entry.S

@@ -0,0 +1,13 @@
+#include "mlibc-asm/helpers.h"
+
+PROC_START(_start)
+	bl relocateSelf
+
+	mov x0, sp
+	bl interpreterMain
+
+	br x0
+PROC_END(_start)
+
+GNU_STACK_NOTE()
+

user/include/mlibc/options/rtld/aarch64/runtime.S

@@ -0,0 +1,62 @@
+
+.global __mlibcTlsdescStatic
+.hidden __mlibcTlsdescStatic
+.type __mlibcTlsdescStatic,@function
+__mlibcTlsdescStatic:
+	ldr x0, [x0, #8]
+	ret
+
+// This function depends on the Tcb layout, since it pulls out the dtv pointer
+// out of the thread control block
+.global __mlibcTlsdescDynamic
+.hidden __mlibcTlsdescDynamic
+.type __mlibcTlsdescDynamic,@function
+__mlibcTlsdescDynamic:
+	stp x1, x2, [sp, #-16]!
+	ldr x0, [x0, #8]
+	ldp x1, x2, [x0] // tlsIndex, addend
+	mrs x0, tpidr_el0 // tp
+	ldr x0, [x0, #-104] // tp->dtvPointers
+	ldr x0, [x0, x1, lsl 3] // [tlsIndex]
+	add x0, x0, x2 // + addend
+	mrs x1, tpidr_el0 // tp
+	sub x0, x0, x1 // result - tp
+	ldp x1, x2, [sp], #16
+	ret
+
+.global pltRelocateStub
+pltRelocateStub:
+	// we need to save / restore all registers than can hold function arguments
+	// we do not need to save callee-saved registers as they will not be trashed by lazyRelocate
+	// TODO: save floating point argument registers
+
+	stp x0, x1, [sp, #-16]!
+
+	// pointer to PLT entry
+	ldr x1, [sp, #24]
+	ldr x0, [x16]
+	sub x1, x1, x0
+	asr x0, x0, #3
+
+	// pointer GOT
+	sub x0, x16, #8 // &PLTGOT[1]
+
+	stp x2, x3, [sp, #-16]!
+	stp x4, x5, [sp, #-16]!
+	stp x6, x7, [sp, #-16]!
+	stp x8, x30, [sp, #-16]!
+
+	bl lazyRelocate
+	mov x9, x0
+
+	ldp x8, x30, [sp], #16
+	ldp x6, x7, [sp], #16
+	ldp x4, x5, [sp], #16
+	ldp x2, x1, [sp], #16
+
+	ldp x0, x1, [sp], #16
+	add sp, sp, #16
+	br x9
+
+.section .note.GNU-stack,"",%progbits
+