Low-Level C Programming

CSEE W4840
Prof. Stephen A. Edwards

Columbia University
Spring 2011

Low-Level C Programming p.
Goals

Function is correct
Source code is concise, readable, maintainable
Time-critical sections of program run fast enough
Object code is small and efficient

Basically, optimize the use of three resources:

Execution time
Memory
Development/maintenance time

Low-Level C Programming p.
Like Writing English

You can say the same thing many different ways

and mean the same thing.
There are many different ways to say the same
thing.
The same thing may be said different ways.
There is more than one way to say it.
Many sentences are equivalent.
Be succinct.

Low-Level C Programming p.
Arithmetic

Integer Arithmetic Fastest

Floating-point arithmetic in hardware Slower

Floating-point arithmetic in software Very slow

+,



slower



sqrt, sin, log, etc. y

Low-Level C Programming p.
Simple benchmarks

for (i = 0 ; i < 10000 ; ++i)

/* arithmetic operation */

On my desktop Pentium 4 with good hardware

floating-point support,
Operator Time Operator Time
+ (int) 1 + (double) 5
* (int) 5 * (double) 5
/ (int) 12 / (double) 10
(int) 2 sqrt 28
sin 48
pow 275

Low-Level C Programming p.
Simple benchmarks

On my Zaurus SL 5600, a 400 MHz Intel PXA250

Xscale (ARM) processor:
Operator Time
+ (int) 1 + (double) 140
* (int) 1 * (double) 110
/ (int) 7 / (double) 220
(int) 1 sqrt 500
sin 3300
pow 820

Low-Level C Programming p.
C Arithmetic Trivia

Operations on char, short, int, and long

probably run at the same speed (same ALU).
Same for unsigned variants
int or long slower when they exceed machines
word size.

Low-Level C Programming p.
Arithmetic Lessons

Try to use integer addition/subtraction

Avoid multiplication unless you have hardware
Avoid division
Avoid floating-point, unless you have hardware
Really avoid math library functions

Low-Level C Programming p.
Bit Manipulation

C has many bit-manipulation operators.

& Bit-wise AND
| Bit-wise OR
^ Bit-wise XOR
~ Negate (ones complement)
>> Right-shift
<< Left-shift
Plus assignment versions of each.

Low-Level C Programming p.
Bit-manipulation basics

a |= 0x4; /* Set bit 2 */

b &= ~0x4; /* Clear bit 2 */

c &= ~(1 << 3); /* Clear bit 3 */

d ^= (1 << 5); /* Toggle bit 5 */

e >>= 2; /* Divide e by 4 */

Low-Level C Programming p. 1
Advanced bit manipulation

/* Set b to the rightmost 1 in a */

b = a & (a ^ (a - 1));

/* Set d to the number of 1s in c */

char c, d;
d = (c & 0x55) + ((c & 0xaa) >> 1);
d = (d & 0x33) + ((d & 0xcc) >> 2);
d = (d & 0x0f) + ((d & 0xf0) >> 4);

Low-Level C Programming p. 1
Faking Multiplication

Addition, subtraction, and shifting are fast. Can

sometimes supplant multiplication.
Like floating-point, not all processors have a
dedicated hardware multiplier.
Recall the multiplication algorithm from
elementary school, but think binary:

101011
1101
101011
= 43 + 43 << 2 + 43 << 3 = 559
10101100
+101011000
1000101111
Low-Level C Programming p. 1
Faking Multiplication

Even more clever if you include subtraction:

101011
1110
= 43 << 1 + 43 << 2 + 43 << 3
1010110
= 43 << 4 - 43 << 2
10101100
= 602
+101011000
1001011010

Only useful

for multiplication by a constant

for simple multiplicands
when hardware multiplier not available Low-Level C Programming p. 1
Faking Division

Division is a much more complicated algorithm

that generally involves decisions.
However, division by a power of two is just a shift:
a / 2 = a >> 1
a / 4 = a >> 2
a / 8 = a >> 3
There is no general shift-and-add replacement for
division, but sometimes you can turn it into
multiplication:
a / 1.33333333
= a * 0.75
= a * 0.5 + a * 0.25
= a >> 1 + a >> 2
Low-Level C Programming p. 1
Multi-way branches

switch (a) {
if (a == 1) case 1:
foo(); foo(); break;
else if (a == 2) case 2:
bar(); bar(); break;
else if (a == 3) case 3:
baz(); baz(); break;
else if (a == 4) case 4:
qux(); qux(); break;
else if (a == 5) case 5:
quux(); quux(); break;
else if (a == 6) case 6:
corge(); corge(); break;
}
Low-Level C Programming p. 1
Nios code for if-then-else

ldw r2, 0(fp) # Fetch a from stack

cmpnei r2, r2, 1 # Compare with 1
bne r2, zero, .L2 # If not 1, jump to L2
call foo # Call foo()
br .L3 # branch out
.L2:
ldw r2, 0(fp) # Fetch a from stack (again!)
cmpnei r2, r2, 2 # Compare with 2
bne r2, zero, .L4 # If not 1, jump to L4
call bar # Call bar()
br .L3 # branch out
.L4:

Low-Level C Programming p. 1
Nios code for switch (1)
ldw r2, 0(fp) # Fetch a
cmpgeui r2, r2, 7 # Compare with 7
bne r2, zero, .L2 # Branch if greater or equal
ldw r2, 0(fp) # Fetch a
muli r3, r2, 4 # Multiply by 4
movhi r2, %hiadj(.L9) # Load address .L9
addi r2, r2, %lo(.L9)
add r2, r3, r2 # = a * 4 + .L9
ldw r2, 0(r2) # Fetch from jump table
jmp r2 # Jump to label
.section .rodata
.align 2
.L9:
.long .L2 # Branch table
.long .L3
.long .L4
.long .L5
.long .L6
.long .L7
.long .L8
Low-Level C Programming p. 1
Nios code for switch (2)
.section .text
.L3:
call foo
br .L2
.L4:
call bar
br .L2
.L5:
call baz
br .L2
.L6:
call qux
br .L2
.L7:
call quux
br .L2
.L8:
call corge
.L2:
Low-Level C Programming p. 1
Computing Discrete Functions

There are many ways to compute a random

function of one variable:
/* OK, especially for sparse domain */
if (a == 0) x = 0;
else if (a == 1) x = 4;
else if (a == 2) x = 7;
else if (a == 3) x = 2;
else if (a == 4) x = 8;
else if (a == 5) x = 9;

Low-Level C Programming p. 1
Computing Discrete Functions

/* Better for large, dense domains */

switch (a) {
case 0: x = 0; break;
case 1: x = 4; break;
case 2: x = 7; break;
case 3: x = 2; break;
case 4: x = 8; break;
case 5: x = 9; break;
}

/* Best: constant-time lookup table */

int f[] = {0, 4, 7, 2, 8, 9};
x = f[a]; /* assumes 0 <= a <= 5 */

Low-Level C Programming p. 2
Function calls

Modern processors, especially RISC, strive to

make this cheap. Arguments passed through
registers. Still has noticable overhead.
Calling, entering, and returning:
int foo(int a, int b) {
int c = bar(b, a);
return c;
}

Low-Level C Programming p. 2
Code for foo() (unoptimized)
foo:
addi sp, sp, -20 # Allocate space on stack
stw ra, 16(sp) # Store return address
stw fp, 12(sp) # Store frame pointer
mov fp, sp # Frame pointer is new SP
stw r4, 0(fp) # Save a on stack
stw r5, 4(fp) # Save b on stack

ldw r4, 4(fp) # Fetch b

ldw r5, 0(fp) # Fetch a
call bar # Call bar()
stw r2, 8(fp) # Store result in c

ldw r2, 8(fp) # Return value in r2 = c

ldw ra, 16(sp) # Restore return address
ldw fp, 12(sp) # Restore frame pointer
addi sp, sp, 20 # Release stack space
ret # Return from subroutine

Low-Level C Programming p. 2
Code for foo() (optimized)

foo:
addi sp, sp, -4 # Allocate stack space
stw ra, 0(sp) # Store return address
mov r2, r4 # Swap arguments (r4, r5)
mov r4, r5 # using r2 as temporary
mov r5, r2
call bar # Call bar() (return in r2)
ldw ra, 0(sp) # Restore return address
addi sp, sp, 4 # Release stack space
ret # Return from subroutine

Low-Level C Programming p. 2
Strength Reduction

Why multiply when you can add?

struct { struct {
int a; int a;
char b; char b;
int c; int c;
} foo[10]; } *fp, *fe, foo[10];
int i;
fe = foo + 10;
for (i=0 ; i<10 ; ++i) { for (fp = foo ; fp != fe ; ++fp) {
foo[i].a = 77; fp->a = 77;
foo[i].b = 88; fp->b = 88;
foo[i].c = 99; fp->c = 99;
} }
Good optimizing compilers do this automatically.

Low-Level C Programming p. 2
Unoptimized array code (fragment)

.L2:
ldw r2, 0(fp) # Fetch i
cmpgei r2, r2, 10 # i >= 10?
bne r2, zero, .L1 # exit if true
movhi r3, %hiadj(foo) # Get address of foo array
addi r3, r3, %lo(foo)
ldw r2, 0(fp) # Fetch i
muli r2, r2, 12 # i * 12
add r3, r2, r3 # foo[i]
movi r2, 77
stw r2, 0(r3) # foo[i].a = 77
movhi r3, %hiadj(foo)
addi r3, r3, %lo(foo)
ldw r2, 0(fp)
muli r2, r2, 12
add r2, r2, r3 # compute &foo[i]
addi r3, r2, 4 # offset for b field
movi r2, 88
stb r2, 0(r3) # foo[i].b = 88

Low-Level C Programming p. 2
Unoptimized pointer code (fragment)

.L2:
ldw r3, 0(fp) # fp
ldw r2, 4(fp) # fe
beq r3, r2, .L1 # fp == fe?
ldw r3, 0(fp)
movi r2, 77
stw r2, 0(r3) # fp->a = 77
ldw r3, 0(fp)
movi r2, 88
stb r2, 4(r3) # fp->b = 88
ldw r3, 0(fp)
movi r2, 99
stw r2, 8(r3) # fp->c = 99
ldw r2, 0(fp)
addi r2, r2, 12
stw r2, 0(fp) # ++fp
br .L2

Low-Level C Programming p. 2
Optimized (O2) array code

movi r6, 77 # Load constants

movi r5, 88
movi r4, 99
movhi r2, %hiadj(foo) # Load address of array
addi r2, r2, %lo(foo)
movi r3, 10 # iteration count
.L5:
addi r3, r3, -1 # decrement iterations
stw r6, 0(r2) # foo[i].a = 77
stb r5, 4(r2) # foo[i].b = 88
stw r4, 8(r2) # foo[i].c = 99
addi r2, r2, 12 # go to next array element
bne r3, zero, .L5 # if there are more to do
ret

Low-Level C Programming p. 2
Optimized (O2) pointer code

movhi r6, %hiadj(foo+120) # fe = foo + 10

addi r6, r6, %lo(foo+120)
addi r2, r6, -120 # fp = foo
movi r5, 77 # Constants
movi r4, 88
movi r3, 99
.L5:
stw r5, 0(r2) # fp->a = 77
stb r4, 4(r2) # fp->b = 88
stw r3, 8(r2) # fp->c = 99
addi r2, r2, 12 # ++fp
bne r2, r6, .L5 # fp == fe?
ret

Low-Level C Programming p. 2
How Rapid is Rapid?

How much time does the following loop take?

for ( i = 0 ; i < 1024 ; ++i) a += b[i];
Operation Cycles per iteration
Memory read 2 or 7
Addition 1
Loop overhead 4
Total 612
The Nios runs at 50 MHz, one instruction per
cycle, so this takes
1 1
6 1024 = 0.12s or 12 1024 = 0.24s
50MHz 50MHz

Low-Level C Programming p. 2
Double-checking

GCC generates good code with -O7:

movhi r4, %hiadj(b) # Load &b[0]
addi r4, r4, %lo(b)
movi r3, 1024 # Iteration count

.L5: # cycles
ldw r2, 0(r4) # Fetch b[i] 2-7
addi r3, r3, -1 # --i 1
addi r4, r4, 4 # next b element 1
add r5, r5, r2 # a += b[i] 1
bne r3, zero, .L5 # repeat if i > 0 3
mov r2, r5 # result
ret

Low-Level C Programming p. 3
Features in order of increasing cost

1. Integer arithmetic
2. Pointer access
3. Simple conditionals and loops
4. Static and automatic variable access
5. Array access
6. Floating-point with hardware support
7. Switch statements
8. Function calls
9. Floating-point emulation in software
10. Malloc() and free()
11. Library functions (sin, log, printf, etc.)
12. Operating system calls (open, sbrk, etc.)
Low-Level C Programming p. 3
Storage Classes in C

/* fixed address: visible to other files */

int global_static;
/* fixed address: only visible within file */
static int file_static;

/* parameters always stacked */

int foo(int auto_param)
{
/* fixed address: only visible to function */
static int func_static;
/* stacked: only visible to function */
int auto_i, auto_a[10];
/* array explicitly allocated on heap */
double *auto_d =
malloc(sizeof(double)*5);

/* return value in register or stacked */

return auto_i;
} Low-Level C Programming p. 3
Dynamic Storage Allocation

0
free()

0
malloc( )

Low-Level C Programming p. 3
Dynamic Storage Allocation

Rules:
Each allocated block contiguous (no holes)
Blocks stay fixed once allocated
malloc()
Find an area large enough for requested block
Mark memory as allocated
free()
Mark the block as unallocated

Low-Level C Programming p. 3
Simple Dynamic Storage Allocation

Maintaining information about free memory

Simplest: Linked list

The algorithm for locating a suitable block

Simplest: First-fit

The algorithm for freeing an allocated block

Simplest: Coalesce adjacent free blocks

Low-Level C Programming p. 3
Dynamic Storage Allocation

S N S S N

malloc( )

S S N S S N

Low-Level C Programming p. 3
Simple Dynamic Storage Allocation

S S N S S N

free()

S S N

Low-Level C Programming p. 3
Storage Classes Compared

On most processors, access to automatic

(stacked) data and globals is equally fast.
Automatic usually preferable since the memory is
reused when function terminates.
Danger of exhausting stack space with recursive
algorithms. Not used in most embedded systems.
The heap (malloc) should be avoided if possible:
Allocation/deallocation is unpredictably slow
Danger of exhausting memory
Danger of fragmentation
Best used sparingly in embedded systems Low-Level C Programming p. 3
Memory-Mapped I/O

Magical memory locations that, when written or

read, send or receive data from hardware.
Hardware that looks like memory to the
processor, i.e., addressable, bidirectional data
transfer, read and write operations.
Does not always behave like memory:
Act of reading or writing can be a trigger (data
irrelevant)
Often read- or write-only
Read data often different than last written
Low-Level C Programming p. 3
Memory-Mapped I/O Access in C

#define SWITCHES \
((volatile char *) 0x1800)
#define LEDS \
((volatile char *) 0x1810)

void main() {
for (;;) {
*LEDS = *SWITCHES;
}
}

Low-Level C Programming p. 4
Whats With the Volatile?

#define ADDRESS \ Compiled with

((char *) 0x1800) optimization:
#define VADDRESS \
((volatile char *) 0x1800) foo:
movi r2, 6144
char foo() { ldbu r2, 0(r2)
char a = *ADDRESS; add r2, r2, r2
char b = *ADDRESS; andi r2, r2, 0xff
return a + b; ret
}
bar:
char bar() { movi r3, 6144
char a = *VADDRESS; ldbu r2, 0(r3)
char b = *VADDRESS; ldbu r3, 0(r3)
return a + b; add r2, r2, r3
} andi r2, r2, 0xff
ret
Low-Level C Programming p. 4
Altera I/O

/* Definitions of alt_u8, etc. */

#include "alt_types.h"

/* IORD_ALTERA_AVALON... for the PIO device */

#include "altera_avalon_pio_regs.h"

/* Auto-generated addresses for all peripherals */

#include "system.h"

int main() {
alt_u8 sw;
for (;;) {
sw = IORD_ALTERA_AVALON_PIO_DATA(SWITCHES_BASE);
IOWR_ALTERA_AVALON_PIO_DATA(LEDS_BASE, sw);
}
}

(From the Nios II Software Developers

Low-Level C Programming p. 4

Handbook)
HW/SW Communication Styles

Memory-mapped I/O puts the processor in

charge: only it may initiate communication.
Typical operation:
Check hardware conditions by reading status
registers
When ready, send next command by writing
control and data registers
Check status registers for completion, waiting
if necessary
Waiting for completion: polling
Are we there yet? No. Are we there yet? No
Low-Level C Programming p. 4

Are we there yet? No Are we there yet? No

HW/SW Communication: Interrupts

Idea: have hardware initiate communication

when it wants attention.
Processor responds by immediately calling an
interrupt handling routine, suspending the
currently-running program.

Low-Level C Programming p. 4
Unix Signals

The Unix environment provides signals, which

behave like interrupts.
#include <stdio.h>
#include <signal.h>

void handleint() {
printf("Got an INT\n");
/* some variants require this */
signal(SIGINT, handleint);
}

int main() {
/* Register signal handler */
signal(SIGINT, handleint);
/* Do nothing forever */
for (;;) { }
return 0; Low-Level C Programming p. 4
}
Interrupts under Altera (1)

#include "system.h"
#include "altera_avalon_pio_regs.h"
#include "alt_types.h"

static void button_isr(void* context, alt_u32 id)

{
/* Read and store the edge capture register */
*(volatile int *) context =
IORD_ALTERA_AVALON_PIO_EDGE_CAP(BUTTON_PIO_BASE);

/* Write to the edge capture register to reset it */

IOWR_ALTERA_AVALON_PIO_EDGE_CAP(BUTTON_PIO_BASE, 0);

/* Reset interrupt capability for the Button PIO */

IOWR_ALTERA_AVALON_PIO_IRQ_MASK(BUTTON_PIO_BASE, 0xf);
}

Low-Level C Programming p. 4
Interrupts under Altera (2)

#include "sys/alt_irq.h"
#include "system.h"

volatile int captured_edges;

static void init_button_pio()

{
/* Enable all 4 button interrupts. */
IOWR_ALTERA_AVALON_PIO_IRQ_MASK(BUTTON_PIO_BASE, 0xf);

/* Reset the edge capture register. */

IOWR_ALTERA_AVALON_PIO_EDGE_CAP(BUTTON_PIO_BASE, 0x0);

/* Register the ISR. */

alt_irq_register( BUTTON_PIO_IRQ,
(void *) &captured_edges,
button_isr );
}
Low-Level C Programming p. 4
Debugging Skills
If-
.3

'

ft
r
s-

V\
;
-
I
? i
V
0
!

*O

f* f
11 i

(y) ?

-a- ft
A

-- ;

-*.

Low-Level C Programming p. 4
The Edwards Way to Debug

1. Identify undesired behavior

2. Construct linear model for desired behavior
3. Pick a point along model
4. Form desired behavior hypothesis for point
5. Test
6. Move point toward failure if point working,
away otherwise
7. Repeat #4#6 until bug is found

Low-Level C Programming p. 4