Make Hello World with libc and ld work

IF you want to understand how amazing this is, watch https://www.twitch.tv/videos/152115137
emkey1 · Jun 15, 2017 · 559ae4c · 559ae4c
1 parent c90e717
commit 559ae4c
Show file tree

Hide file tree

Showing 14 changed files with 109 additions and 22 deletions.
diff --git a/emu/cpu.c b/emu/cpu.c
@@ -183,6 +183,9 @@ int cpu_step(struct cpu_state *cpu) {
                     do_cpuid(&cpu->eax, &cpu->ebx, &cpu->ecx, &cpu->edx);
                     break;
 
+                case 0xa3: TRACEI("bt reg, modrm\t");
+                           READMODRM; BT(modrm_reg, modrm_val); break;
+
                 case 0xa5: TRACEI("shld cl, reg, modrm");
                            READMODRM; SHLD(cpu->cl, modrm_reg, modrm_val); break;
 
@@ -259,8 +262,13 @@ int cpu_step(struct cpu_state *cpu) {
             }
             break;
 
+        case 0x11: TRACEI("adc reg, modrm");
+                   READMODRM; ADC(modrm_reg, modrm_val_w); break;
+
         case 0x19: TRACEI("sbb reg, modrm");
                    READMODRM; SBB(modrm_reg, modrm_val); break;
+        case 0x1b: TRACEI("sbb modrm, reg");
+                   READMODRM; SBB(modrm_val, modrm_reg); break;
 
         case 0x21: TRACEI("and reg, modrm");
                    READMODRM; AND(modrm_reg, modrm_val_w); break;
@@ -379,7 +387,7 @@ int cpu_step(struct cpu_state *cpu) {
         case 0x6a: TRACEI("push imm8\t");
                    READIMM8; PUSH((int8_t) imm8); break;
         case 0x6b: TRACEI("imul imm8\t");
-                   READMODRM; READIMM8; MUL3((int8_t) imm8, (int8_t) modrm_val, modrm_reg); break;
+                   READMODRM; READIMM8; MUL3((int8_t) imm8, (int32_t) modrm_val, modrm_reg); break;
 
         case 0x70: TRACEI("jo rel8\t");
                    READIMM8; J_REL(O, (int8_t) imm8); break;

diff --git a/emu/instructions.h b/emu/instructions.h
@@ -115,6 +115,8 @@
                 OR(src, dst##_w); break; \
         case 2: TRACE("adc"); \
                 ADC(src, dst##_w); break; \
+        case 3: TRACE("sbb"); \
+                SBB(src, dst##_w); break; \
         case 4: TRACE("and"); \
                 AND(src, dst##_w); break; \
         case 5: TRACE("sub"); \
@@ -203,6 +205,9 @@
         case 7: TRACE("undefined"); return INT_UNDEFINED; \
     }
 
+#define BT(bit, val) \
+    cpu->cf = (val) & (1 << bit)
+
 #define BUMP_SI(size) \
     if (!cpu->df) \
         cpu->esi += size; \

diff --git a/emu/memory.c b/emu/memory.c
@@ -44,6 +44,7 @@ int pt_map(struct mem *mem, page_t start, pages_t pages, void *memory, unsigned
             pt_unmap(mem, page, 1);
         }
         struct pt_entry *entry = malloc(sizeof(struct pt_entry));
+        // FIXME this could allocate some of the memory and then abort
         if (entry == NULL) return _ENOMEM;
         entry->data = memory;
         entry->refcount = 1;
@@ -57,7 +58,10 @@ int pt_map(struct mem *mem, page_t start, pages_t pages, void *memory, unsigned
     return 0;
 }
 
-void pt_unmap(struct mem *mem, page_t start, pages_t pages) {
+int pt_unmap(struct mem *mem, page_t start, pages_t pages) {
+    for (page_t page = start; page < start + pages; page++)
+        if (mem->pt[page] == NULL)
+            return -1;
     for (page_t page = start; page < start + pages; page++) {
         struct pt_entry *entry = mem->pt[page];
         mem->pt[page] = NULL;
@@ -67,6 +71,7 @@ void pt_unmap(struct mem *mem, page_t start, pages_t pages) {
             free(entry);
         }
     }
+    return 0;
 }
 
 int pt_map_nothing(struct mem *mem, page_t start, pages_t pages, unsigned flags) {
@@ -84,6 +89,17 @@ int pt_map_file(struct mem *mem, page_t start, pages_t pages, int fd, off_t off,
     return pt_map(mem, start, pages, memory, flags);
 }
 
+// FIXME this can overwrite P_GROWSDOWN or P_GUARD
+int pt_set_flags(struct mem *mem, page_t start, pages_t pages, int flags) {
+    for (page_t page = start; page < start + pages; page++)
+        if (mem->pt[page] == NULL)
+            return _ENOMEM;
+    for (page_t page = start; page < start + pages; page++) {
+        mem->pt[page]->flags = flags;
+    }
+    return 0;
+}
+
 void pt_dump(struct mem *mem) {
     for (unsigned i = 0; i < PT_SIZE; i++) {
         if (mem->pt[i] != NULL) {

diff --git a/emu/memory.h b/emu/memory.h
@@ -50,8 +50,10 @@ int pt_map(struct mem *mem, page_t start, pages_t pages, void *memory, unsigned
 int pt_map_file(struct mem *mem, page_t start, pages_t pages, int fd, off_t off, unsigned flags);
 // Map empty space into fake memory
 int pt_map_nothing(struct mem *mem, page_t page, pages_t pages, unsigned flags);
-// Unmap fake memory
-void pt_unmap(struct mem *mem, page_t page, pages_t pages);
+// Unmap fake memory, return -1 if any part of the range isn't mapped and 0 otherwise
+int pt_unmap(struct mem *mem, page_t start, pages_t pages);
+// Set the flags on memory
+int pt_set_flags(struct mem *mem, page_t start, pages_t pages, int flags);
 
 void pt_dump(struct mem *mem);
 

diff --git a/fs/real.c b/fs/real.c
@@ -32,7 +32,7 @@ static void copy_stat(struct statbuf *fake_stat, struct stat *real_stat) {
     fake_stat->nlink = real_stat->st_nlink;
     fake_stat->uid = real_stat->st_uid;
     fake_stat->gid = real_stat->st_gid;
-    fake_stat->rdev = real_stat->st_dev;
+    fake_stat->rdev = real_stat->st_rdev;
     fake_stat->size = real_stat->st_size;
     fake_stat->blksize = real_stat->st_blksize;
     fake_stat->blocks = real_stat->st_blocks;
@@ -91,7 +91,7 @@ static int realfs_mmap(struct fd *fd, off_t offset, size_t len, int prot, int fl
     int mmap_flags = 0;
     if (flags & MMAP_PRIVATE) mmap_flags |= MAP_PRIVATE;
     // TODO more flags are probably needed
-    void *mem = mmap(NULL, len, prot, flags, fd->real_fd, offset);
+    void *mem = mmap(NULL, len, prot, mmap_flags, fd->real_fd, offset);
     if (mem == MAP_FAILED)
         return err_map(errno);
     *mem_out = mem;

diff --git a/meson.build b/meson.build
@@ -45,4 +45,4 @@ subdir('tests')
 # ptraceomatic et al
 subdir('tools')
 
-executable('thingy', ['main.c'], dependencies: [ish])
+executable('ish', ['main.c'], dependencies: [ish])
diff --git a/misc.h b/misc.h
@@ -9,7 +9,7 @@
 #include <sys/types.h>
 
 // debug output utilities
-#if 1
+#if 0
 #define TRACE(msg, ...) printf(msg, ##__VA_ARGS__)
 #else
 #define TRACE(msg, ...) (void)NULL

diff --git a/sys/calls.c b/sys/calls.c
@@ -22,8 +22,12 @@ syscall_t syscall_table[] = {
     [41] = (syscall_t) sys_dup,
     [45] = (syscall_t) sys_brk,
     [85] = (syscall_t) sys_readlink,
+    [90] = (syscall_t) sys_mmap,
+    [91] = (syscall_t) sys_munmap,
     [122] = (syscall_t) _sys_uname,
-    [192] = (syscall_t) sys_mmap,
+    [125] = (syscall_t) sys_mprotect,
+    [146] = (syscall_t) sys_writev,
+    [192] = (syscall_t) sys_mmap2,
     [197] = (syscall_t) sys_fstat64,
     [243] = (syscall_t) sys_set_thread_area,
     [252] = (syscall_t) sys_exit_group,

diff --git a/sys/calls.h b/sys/calls.h
@@ -27,8 +27,13 @@ dword_t sys_fstat64(fd_t fd_no, addr_t statbuf_addr);
 dword_t sys_access(addr_t pathname_addr, dword_t mode);
 dword_t sys_readlink(addr_t pathname, addr_t buf, dword_t bufsize);
 
+struct io_vec {
+    addr_t base;
+    uint_t len;
+};
 dword_t sys_read(fd_t fd_no, addr_t buf_addr, dword_t size);
 dword_t sys_write(fd_t fd_no, addr_t buf_addr, dword_t size);
+dword_t sys_writev(fd_t fd_no, addr_t iovec_addr, dword_t iovec_count);
 
 dword_t sys_dup(fd_t fd);
 
@@ -37,8 +42,12 @@ int handle_pagefault(addr_t addr);
 
 #define MMAP_SHARED 0x1
 #define MMAP_PRIVATE 0x2
+#define MMAP_FIXED 0x10
 #define MMAP_ANONYMOUS 0x20
 addr_t sys_mmap(addr_t addr, dword_t len, dword_t prot, dword_t flags, fd_t fd_no, dword_t offset);
+addr_t sys_mmap2(addr_t addr, dword_t len, dword_t prot, dword_t flags, fd_t fd_no, dword_t offset);
+int_t sys_munmap(addr_t addr, uint_t len);
+int_t sys_mprotect(addr_t addr, uint_t len, int_t prot);
 
 #define UNAME_LENGTH 65
 struct uname {

diff --git a/sys/fs.c b/sys/fs.c
@@ -69,6 +69,20 @@ dword_t sys_write(fd_t fd_no, addr_t buf_addr, dword_t size) {
     return fd->ops->write(fd, buf, size);
 }
 
+dword_t sys_writev(fd_t fd_no, addr_t iovec_addr, dword_t iovec_count) {
+    struct io_vec iovecs[iovec_count];
+    user_get_count(iovec_addr, iovecs, sizeof(iovecs));
+    int res;
+    dword_t count = 0;
+    for (unsigned i = 0; i < iovec_count; i++) {
+        res = sys_write(fd_no, iovecs[i].base, iovecs[i].len);
+        if (res < 0)
+            return res;
+        count += res;
+    }
+    return count;
+}
+
 dword_t sys_fstat64(fd_t fd_no, addr_t statbuf_addr) {
     struct fd *fd = current->files[fd_no];
     struct statbuf stat;

diff --git a/sys/mmap.c b/sys/mmap.c
@@ -3,23 +3,33 @@
 #include "emu/process.h"
 #include "emu/memory.h"
 
+addr_t sys_mmap2(addr_t addr, dword_t len, dword_t prot, dword_t flags, fd_t fd_no, dword_t offset) {
+    return sys_mmap(addr, len, prot, flags, fd_no, offset << PAGE_BITS);
+}
+
 addr_t sys_mmap(addr_t addr, dword_t len, dword_t prot, dword_t flags, fd_t fd_no, dword_t offset) {
     int err;
 
     if (len == 0)
         return _EINVAL;
     if (prot & ~(P_READ | P_WRITE | P_EXEC))
         return _EINVAL;
-    if (!(flags & MMAP_PRIVATE))
-        // TODO MMAP_SHARED
-        return _EINVAL;
-    if (addr != 0)
+    if (!(flags & MMAP_PRIVATE)) {
+        TODO("MMAP_SHARED");
         return _EINVAL;
+    }
 
     pages_t pages = PAGE_ROUND_UP(len);
-    page_t page = pt_find_hole(&curmem, pages);
-    if (page == BAD_PAGE)
-        return _ENOMEM;
+    page_t page;
+    if (addr == 0) {
+        page = pt_find_hole(&curmem, pages);
+        if (page == BAD_PAGE)
+            return _ENOMEM;
+    } else {
+        if (OFFSET(addr) != 0)
+            return _EINVAL;
+        page = PAGE(addr);
+    }
     if (flags & MMAP_ANONYMOUS) {
         if ((err = pt_map_nothing(&curmem, page, pages, prot)) < 0)
             return err;
@@ -39,3 +49,21 @@ addr_t sys_mmap(addr_t addr, dword_t len, dword_t prot, dword_t flags, fd_t fd_n
     return page << PAGE_BITS;
 }
 
+int_t sys_munmap(addr_t addr, uint_t len) {
+    if (OFFSET(addr) != 0)
+        return _EINVAL;
+    if (len == 0)
+        return _EINVAL;
+    if (pt_unmap(&curmem, PAGE(addr), PAGE_ROUND_UP(len)) < 0)
+        return _EINVAL;
+    return 0;
+}
+
+int_t sys_mprotect(addr_t addr, uint_t len, int_t prot) {
+    if (OFFSET(addr) != 0)
+        return _EINVAL;
+    if (prot & ~(P_READ | P_WRITE | P_EXEC))
+        return _EINVAL;
+    pages_t pages = PAGE_ROUND_UP(len);
+    return pt_set_flags(&curmem, PAGE(addr), pages, prot);
+}
diff --git a/tests/meson.build b/tests/meson.build
@@ -4,9 +4,9 @@ executable('hello-libc-static', ['hello-clib.c'], c_args: ['-m32'], link_args: [
 executable('hello-libc', ['hello-clib.c'], c_args: ['-m32'], link_args: ['-m32'])
 
 # simple benchmark
-executable('looper', ['looper.c', 'nothing.c'], c_args: ['-m32'], link_args: ['-m32', '-static'])
-executable('fibbonaci', ['fibbonaci.c'], c_args: ['-m32'], link_args: ['-m32', '-static'])
+executable('looper', ['looper.c', 'nothing.c'], c_args: ['-m32'], link_args: ['-m32'])
+executable('fibbonaci', ['fibbonaci.c'], c_args: ['-m32'], link_args: ['-m32'])
 
 # filesystem motivation
-executable('cat', ['cat.c'], c_args: ['-m32'], link_args: ['-m32', '-static'])
-executable('stat', ['stat.c'], c_args: ['-m32', '-D_FILE_OFFSET_BITS=64'], link_args: ['-m32', '-static'])
+executable('cat', ['cat.c'], c_args: ['-m32'], link_args: ['-m32'])
+executable('stat', ['stat.c'], c_args: ['-m32', '-D_FILE_OFFSET_BITS=64'], link_args: ['-m32'])
diff --git a/tools/ptraceomatic.c b/tools/ptraceomatic.c
@@ -210,6 +210,7 @@ void step_tracing(struct cpu_state *cpu, int pid, int sender, int receiver) {
                 pt_copy(pid, regs.rcx, sizeof(struct newstat64));
                 break;
 
+            case 90: // mmap
             case 192: // mmap2
                 if (cpu->eax < 0xfffff000 && cpu->edi != (dword_t) -1) {
                     // fake mmap didn't fail, change fd
@@ -220,7 +221,7 @@ void step_tracing(struct cpu_state *cpu, int pid, int sender, int receiver) {
 
             // some syscalls need to just happen
             case 45: // brk
-            case 90: // mmap
+            case 125: // mprotect
             case 243: // set_thread_area
                 goto do_step;
         }

diff --git a/vdso/note.S b/vdso/note.S
@@ -8,7 +8,7 @@ name:
 after_name:
     .balign 4
 note:
-    .long 0x010000 // let's pretend we're linux 2.0.0
+    .long 0x020620 // let's pretend we're linux 2.6.32
 after_note:
     .balign 4
 .popsection