Unclear output by riscv objdump -d - linux

Now I am trying to understand the RISC-V ISA but I have an unclear point about the machine code and assembly.
I have written a C code like this:
int main() {
return 42;
}
Then, I produced the .s file by this command:
$ /opt/riscv/bin/riscv64-unknown-linux-gnu-gcc -S 42.c
The output was:
.file "42.c"
.option nopic
.text
.align 1
.globl main
.type main, #function
main:
addi sp,sp,-16
sd s0,8(sp)
addi s0,sp,16
li a5,42
mv a0,a5
ld s0,8(sp)
addi sp,sp,16
jr ra
.size main, .-main
.ident "GCC: (g5964b5cd727) 11.1.0"
.section .note.GNU-stack,"",#progbits
Now, I run following command to produce an elf.
$ /opt/riscv/bin/riscv64-unknown-linux-gnu-gcc -nostdlib -o 42 42.s
So, a binary file is produced. I tried to read that by objdump like this:
$ /opt/riscv/bin/riscv64-unknown-linux-gnu-objdump -d 42
So the output was like this:
42: file format elf64-littleriscv
Disassembly of section .text:
00000000000100b0 <main>:
100b0: 1141 addi sp,sp,-16
100b2: e422 sd s0,8(sp)
100b4: 0800 addi s0,sp,16
100b6: 02a00793 li a5,42
100ba: 853e mv a0,a5
100bc: 6422 ld s0,8(sp)
100be: 0141 addi sp,sp,16
100c0: 8082 ret
What I don't understand is the meaning of the machine code in objdump output.
For example, the first instruction addi is translated into .....0010011 according to this page, (while this is not an official spec). However, the dumped hex is 1141. 1141 can only represent 2 bytes, but the instruction should be 32-bit, 4bytes.
I guess I am missing some points, but how should I read the output of objdump for riscv?

You can tell objdump to show compressed (16-bit) instructions by using -M no-aliases in this way
riscv64-unknown-elf-objdump -d -M no-aliases
In that case, instructions starting with c. are compressed ones.
Unfortunately that will also disable some other aliases, making the asm less nice to read if you're used to them. You can just look at the number of bytes (2 vs. 4) in the hexdump to see if it's a compressed instruction or not.

Related

How to print value of a register using spike?

I have an assembly code for RISCV machine.
I have added an instruction to access floating point control and status register and store floating point flags in register a3. I want to print its value to demonstrate that flag gets set when floating point exception occurs.
I tried using spike. There is an instruction in spike(in debug mode) to print value of a register:
: reg 0 a3
to print value of a3.
But first i have to reach my desired point.
I do not know how will i be able to reach that point.
.file "learn_Assembly.c"
.option nopic
.text
.comm a,4,4
.comm b,4,4
.align 1
.globl main
.type main, #function
main:
addi sp,sp,-32
sd s0,24(sp)
addi s0,sp,32
lui a5,%hi(a)
lui a4,%hi(.LC0)
flw fa5,%lo(.LC0)(a4)
fsw fa5,%lo(a)(a5)
lui a5,%hi(b)
lui a4,%hi(.LC1)
flw fa5,%lo(.LC1)(a4)
fsw fa5,%lo(b)(a5)
lui a5,%hi(a)
flw fa4,%lo(a)(a5)
lui a5,%hi(b)
flw fa5,%lo(b)(a5)
fmul.s fa5,fa4,fa5
frflags a3
fsw fa5,-20(s0)
li a5,0
mv a0,a5
ld s0,24(sp)
addi sp,sp,32
jr ra
.size main, .-main
.section .rodata
.align 2
.LC0:
.word 1082130432
.align 2
.LC1:
.word 1077936128
.ident "GCC: (GNU) 8.2.0"
The other option is to somehow write print it using assembly instruction which i am not sure how to do.
To understand the flow of your program , you could create object dump of your program from compiled elf .
To create elf :-
riscv64-unknown-elf-gcc assmebly_code.s -o executable.elf
Then you could create the object dump by :-
riscv64-unknown-elf-objdump -d executable.elf > executable.dump
executable.dump will contains the program flow like this :-
executable.elf: file format elf64-littleriscv
Disassembly of section .text:
00000000000100b0 <_start>:
100b0: 00002197 auipc gp,0x2
100b4: 35018193 addi gp,gp,848 # 12400 <__global_pointer$>
100b8: 81818513 addi a0,gp,-2024 # 11c18 <_edata>
100bc: 85818613 addi a2,gp,-1960 # 11c58 <_end>
100c0: 8e09 sub a2,a2,a0
100c2: 4581 li a1,0
100c4: 1e6000ef jal ra,102aa <memset>
100c8: 00000517 auipc a0,0x0
100cc: 13850513 addi a0,a0,312 # 10200 <__libc_fini_array>
100d0: 104000ef jal ra,101d4 <atexit>
100d4: 174000ef jal ra,10248 <__libc_init_array>
100d8: 4502 lw a0,0(sp)
100da: 002c addi a1,sp,8
100dc: 4601 li a2,0
100de: 0be000ef jal ra,1019c <main>
100e2: 0fe0006f j 101e0 <exit>
....... ........ .................
....... ........ .................
....... ........ .................
Recognize the required pc with required a3 value .
then on spike use command until to run till that pc value :
: until pc 0 <*required pc*>
Note : Your compiler and assembler names may vary.
You can use until spike instruction to execute until a desired equality is reached:
: until pc 0 2020 (stop when pc=2020)
As explain here (interactive debug).
Once value reached you can use reg to read value you want.

Where is segment %fs for static elf images setup?

I'm trying to figure out how the %fs register is initialized
when creating a elf image by hand.
The simple snippet I'd like to run is:
.text
nop
movq %fs:0x28, %rax;
1: jmp 1b
Which should read at offset 0x28 in the %fs segment. Normally this is where the stack canary is stored. Because I create the elf image by hand the %fs segment is not setup at all by my code this fails expectedly(?) .
Here is how I create the elf image:
0000000000000000 <.text>:
0: 90 nop
1: 64 48 8b 04 25 28 00 mov %fs:0x28,%rax
8: 00 00
a: eb fe jmp 0xa
I create the .text segment via
echo 9064488b042528000000ebfe | xxd -r -p > r2.bin
Then I convert to elf:
ld -b binary -r -o raw.elf r2.bin
objcopy --rename-section .data=.text --set-section-flags .data=alloc,code,load raw.elf
At that point raw.elf contains my instructions. I then link with
ld -T raw.ld -o out.elf -M --verbose where raw.ld is:
OUTPUT_FORMAT("elf64-x86-64", "elf64-x86-64", "elf64-x86-64")
OUTPUT_ARCH(i386:x86-64)
ENTRY(_entry)
PHDRS {
phdr4000000 PT_LOAD;
}
SECTIONS
{
_entry = 0x4000000;
.text 0x4000000 : { raw.elf (.text) } :phdr4000000
}
I can now start out.elf with gdb:
gdb --args out.elf
and set a breakpoint at 0x4000000:
(gdb)break *0x4000000
(gdb)run
The first nop can be stepped via stepi, however the stack canary read mov %fs:0x28,%rax segfaults.
I suppose that is expected given that maybe the OS is not setting up %fs.
For a simple m.c: int main() { return 0; } program compiled with gcc --static m.c -o m I can read from %fs. Adding:
long can()
{
long v = 0;
__asm__("movq %%fs:0x28, %0;"
: "=r"(val)::);
return v;
}
lets me read from %fs - even though I doubt that %fs:28 is setup because ld.so is not run (it is a static image).
Question:
Can anyone point out where %fs is setup in the c runtime for static images?
You need to call arch_prctl with an ARCH_SET_FS argument before you can use the %fs segment prefix. You will have to allocate the backing store somewhere (brk, mmap, or an otherwise unused part of the stack).
glibc does this in __libc_setup_tls in csu/libc-tls.c for statically linked binaries, hidden behind the TLS_INIT_TP macro.

Where const strings are saved in assembly?

When i declare a string in assembly like that:
string DB "My string", 0
where is the string saved?
Can i determine where it will be saved when declaring it?
db assembles output bytes to the current position in the output file. You control exactly where they go.
There is no indirection or reference to any other location, it's like char string[] = "blah blah", not char *string = "blah blah" (but without the implicit zero byte at the end, that's why you have to use ,0 to add one explicitly.)
When targeting a modern OS (i.e. not making a boot-sector or something), your code + data will end up in an object file and then be linked into an executable or library.
On Linux (or other ELF platforms), put read-only constant data including strings in section .rodata. This section (along with section .text where you put code) becomes part of the text segment after linking.
Windows apparently uses section .rdata.
Different assemblers have different syntax for changing sections, but I think section .whatever works in most of the one that use DB for data bytes.
;; NASM source for the x86-64 System V ABI.
section .rodata ; use section .rdata on Windows
string DB "My string", 0
section .data
static_storage_for_something: dd 123 ; one dword with value = 123
;; usually you don't need .data and can just use registers or the stack
section .bss ; zero-initialized memory, bytes not stored in the executable, just size
static_array: resd 12300000 ;; 12300000 dwords with value = 0
section .text
extern puts ; defined in libc
global main
main:
mov edi, string ; RDI = address of string = first function arg
;mov [rdi], 1234 ; would segfault because .rodata is mapped read-only
jmp puts ; tail-call puts(string)
peter#volta:/tmp$ cat > string.asm
(and paste the above, then press control-D)
peter#volta:/tmp$ nasm -f elf64 string.asm && gcc -no-pie string.o && ./a.out
My string
peter#volta:/tmp$ echo $?
10
10 characters is the return value from puts, which is the return value from main because we tail-called it, which becomes the exit status of our program. (Linux glibc puts apparently returns the character count in this case. But the manual just says it returns non-negative number on success, so don't count on this)
I used -no-pie because I used an absolute address for string with mov instead of a RIP-relative LEA.
You can use readelf -a a.out or nm to look at what went where in your executable.

cross compiling mips: Linker error: cannot find entry symbol __start;

So Im trying to cross compile simple program written in assembly (using MARS) to mips executable.
# Purpose: First program, Hello World
.text # Define the program instructions.
main: # Label to define the main program.
li $v0,4 #Load 4into $v0 to indicate a print string.
la $a0, greeting #Load the address of the greeting into $a0.
syscall #Print greeting. The print is indicated by
# $v0 having a value of 4, and the string to
# print is stored at the address in $a0.
li $v0,10 #Load a 10 (halt) into $v0.
syscall # The program ends.
.data # Define the program data.
greeting: .asciiz "Hello World" #The string to print.
first I use
mips-linux-as ./HelloWorld.asm -o HelloWorld.o
then
mips-linux-ld HelloWorld.o -o HelloWorld
and I get an error
mips-linux-xld: warning: cannot find entry symbol __start; defaulting to 00000000004000b0
Simply adding _start section does not work I get same error.So I written simple program in C (HelloWorld.c) and generated assembly using gcc with following command
mips-linux-gcc -O2 -S -c HelloWorld.c
and there is no _start section
.file 1 "HelloWorld.c"
.section .mdebug.abi32
.previous
.gnu_attribute 4, 3
.abicalls
.section .rodata.str1.4,"aMS",#progbits,1
.align 2
$LC0:
.ascii "Hello Worldn\000"
.text
.align 2
.globl main
.set nomips16
.ent main
.type main, #function
main:
.frame $sp,32,$31 # vars= 0, regs= 1/0, args= 16, gp= 8
.mask 0x80000000,-4
.fmask 0x00000000,0
.set noreorder
.set nomacro
addiu $sp,$sp,-32
sw $31,28($sp)
lui $28,%hi(__gnu_local_gp)
addiu $28,$28,%lo(__gnu_local_gp)
.cprestore 16
lui $4,%hi($LC0)
lw $25,%call16(printf)($28)
jalr $25
addiu $4,$4,%lo($LC0)
move $2,$0
lw $31,28($sp)
j $31
addiu $sp,$sp,32
.set macro
.set reorder
.end main
.size main, .-main
.ident "GCC: (GNU) 4.4.5-1.5.5p4"
So what Im missing ? I would like to run this program on my mips box.
Try adding this to your asm file
.globl main
.ent main
.type main, #function
However, you will need to use MIPS Linux syscalls, and not MARS syscalls.
Here is a working hello-world executable for MIPS on Linux (not on MARS):
# file hello.s
.data
msg: .ascii "Hello, World!\n"
len = . - msg
.text
.globl __start
__start:
li $v0, 4004 # __NR_write == 4004
li $a0, 1 # argument: STDOUT_FILENO
la $a1, msg # argument: message to write
li $a2, len # argument: size of message
syscall
li $v0, 4001 # __NR_exit == 4001
li $a0, 0 # argument: EXIT_SUCCESS
syscall
On Debian and Ubuntu Linux, install the assembler and linker (for compilation) like this:
$ sudo apt-get install binutils-mips-linux-gnu
On Debian and Ubuntu Linux, install the emulator like this:
$ sudo apt-get install qemu-user
Compile to big endian executable like this:
$ mips-linux-gnu-as -EB -mips1 -o hello.o hello.s
$ mips-linux-gnu-ld -s -m elf32btsmip -o hello hello.o
Run the big endian executable in the emulator like this:
$ qemu-mips ./hello; echo $?
Hello, World!
0
Compile to little endian executable like this:
$ mips-linux-gnu-as -EL -mips1 -o hellol.o hello.s
$ mips-linux-gnu-ld -s -m elf32ltsmip -o hellol hellol.o
Run the little endian executable in the emulator like this:
$ qemu-mipsel ./hellol; echo $?
Hello, World!
0

How to find the main function's entry point of elf executable file without any symbolic information?

I developed a small cpp program on platform of Ubuntu-Linux 11.10.
Now I want to reverse engineer it. I am beginner. I use such tools: GDB 7.0, hte editor, hexeditor.
For the first time I made it pretty easy. With help of symbolic information I founded the address of main function and made everything I needed.
Then I striped (--strip-all) executable elf-file and I have some problems.
I know that main function starts from 0x8960 in this program.
But I haven't any idea how should I find this point without this knowledge.
I tried debug my program step by step with gdb but it goes into __libc_start_main
then into the ld-linux.so.3 (so, it finds and loads the shared libraries needed by a program). I debugged it about 10 minutes. Of course, may be in 20 minutes I can reach the main function's entry point, but, it seems, that more easy way has to exist.
What should I do to find the main function's entry point without any symbolic info?
Could you advise me some good books/sites/other_sources from reverse engineering of elf-files with help of gdb?
Any help would be appreciated.
Locating main() in a stripped Linux ELF binary is straightforward. No symbol information is required.
The prototype for __libc_start_main is
int __libc_start_main(int (*main) (int, char**, char**),
int argc,
char *__unbounded *__unbounded ubp_av,
void (*init) (void),
void (*fini) (void),
void (*rtld_fini) (void),
void (*__unbounded stack_end));
The runtime memory address of main() is the argument corresponding to the first parameter, int (*main) (int, char**, char**). This means that the last memory address saved on the runtime stack prior to calling __libc_start_main is the memory address of main(), since arguments are pushed onto the runtime stack in the reverse order of their corresponding parameters in the function definition.
One can enter main() in gdb in 4 steps:
Find the program entry point
Find where __libc_start_main is called
Set a break point to the address last saved on stack prior to the call to _libc_start_main
Let program execution continue until the break point for main() is hit
The process is the same for both 32-bit and 64-bit ELF binaries.
Entering main() in an example stripped 32-bit ELF binary called "test_32":
$ gdb -q -nh test_32
Reading symbols from test_32...(no debugging symbols found)...done.
(gdb) info file #step 1
Symbols from "/home/c/test_32".
Local exec file:
`/home/c/test_32', file type elf32-i386.
Entry point: 0x8048310
< output snipped >
(gdb) break *0x8048310
Breakpoint 1 at 0x8048310
(gdb) run
Starting program: /home/c/test_32
Breakpoint 1, 0x08048310 in ?? ()
(gdb) x/13i $eip #step 2
=> 0x8048310: xor %ebp,%ebp
0x8048312: pop %esi
0x8048313: mov %esp,%ecx
0x8048315: and $0xfffffff0,%esp
0x8048318: push %eax
0x8048319: push %esp
0x804831a: push %edx
0x804831b: push $0x80484a0
0x8048320: push $0x8048440
0x8048325: push %ecx
0x8048326: push %esi
0x8048327: push $0x804840b # address of main()
0x804832c: call 0x80482f0 <__libc_start_main#plt>
(gdb) break *0x804840b # step 3
Breakpoint 2 at 0x804840b
(gdb) continue # step 4
Continuing.
Breakpoint 2, 0x0804840b in ?? () # now in main()
(gdb) x/x $esp+4
0xffffd110: 0x00000001 # argc = 1
(gdb) x/s **(char ***) ($esp+8)
0xffffd35c: "/home/c/test_32" # argv[0]
(gdb)
Entering main() in an example stripped 64-bit ELF binary called "test_64":
$ gdb -q -nh test_64
Reading symbols from test_64...(no debugging symbols found)...done.
(gdb) info file # step 1
Symbols from "/home/c/test_64".
Local exec file:
`/home/c/test_64', file type elf64-x86-64.
Entry point: 0x400430
< output snipped >
(gdb) break *0x400430
Breakpoint 1 at 0x400430
(gdb) run
Starting program: /home/c/test_64
Breakpoint 1, 0x0000000000400430 in ?? ()
(gdb) x/11i $rip # step 2
=> 0x400430: xor %ebp,%ebp
0x400432: mov %rdx,%r9
0x400435: pop %rsi
0x400436: mov %rsp,%rdx
0x400439: and $0xfffffffffffffff0,%rsp
0x40043d: push %rax
0x40043e: push %rsp
0x40043f: mov $0x4005c0,%r8
0x400446: mov $0x400550,%rcx
0x40044d: mov $0x400526,%rdi # address of main()
0x400454: callq 0x400410 <__libc_start_main#plt>
(gdb) break *0x400526 # step 3
Breakpoint 2 at 0x400526
(gdb) continue # step 4
Continuing.
Breakpoint 2, 0x0000000000400526 in ?? () # now in main()
(gdb) print $rdi
$3 = 1 # argc = 1
(gdb) x/s **(char ***) ($rsp+16)
0x7fffffffe35c: "/home/c/test_64" # argv[0]
(gdb)
A detailed treatment of program initialization and what occurs before main() is called and how to get to main() can be found be found in Patrick Horgan's tutorial "Linux x86 Program Start Up
or - How the heck do we get to main()?"
If you have a very stripped version, or even a binary that is packed, as using UPX, you can gdb on it in the tough way as:
$ readelf -h echo | grep Entry
Entry point address: 0x103120
And then you can break at it in GDB as:
$ gdb mybinary
(gdb) break * 0x103120
Breakpoint 1 at 0x103120gdb)
(gdb) r
Starting program: mybinary
Breakpoint 1, 0x0000000000103120 in ?? ()
and then, you can see the entry instructions:
(gdb) x/10i 0x0000000000103120
=> 0x103120: bl 0x103394
0x103124: dcbtst 0,r5
0x103128: mflr r13
0x10312c: cmplwi r7,2
0x103130: bne 0x103214
0x103134: stw r5,0(r6)
0x103138: add r4,r4,r3
0x10313c: lis r0,-32768
0x103140: lis r9,-32768
0x103144: addi r3,r3,-1
I hope it helps
As far as I know, once a program has been stripped, there is no straightforward way to locate the function that the symbol main would have otherwise referenced.
The value of the symbol main is not required for program start-up: in the ELF format, the start of the program is specified by the e_entry field of the ELF executable header. This field normally points to the C library's initialization code, and not directly to main.
While the C library's initialization code does call main() after it has set up the C run time environment, this call is a normal function call that gets fully resolved at link time.
In some cases, implementation-specific heuristics (i.e., the specific knowledge of the internals of the C runtime) could be used to determine the location of main in a stripped executable. However, I am not aware of a portable way to do so.

Resources