BE-32 is supported by ARM cores up to the ARM11 family (v6) (for example ARM7TDMI, ARM926EJ-S, ARM1136JF-S). It is enabled by setting a bit in the CP15 system control coprocessor.

BE-8 is supported by the ARM11 family and later (for example ARM1136JF-S, Cortex-R4, Cortex-A8). Architecture v7 cores do not support BE-32.

It is controlled by setting a bit in the CPSR.

Setting both bits is reserved (not a valid configuration).

In terms of data access:

BE-8 is byte invariant endianness

BE-32 is word invariant endianness

This is easiest to see with examples. I’ve starred **** the important ones:

Basic endianness:

Consider a word stored 0×11223344 where 11 is the most significant byte.

Little endian:

Address 0 1 2 3

Data 44 33 22 11

Big endian:

Address 0 1 2 3

Data 11 22 33 44

BE-32 and BE-8

Now consider data stored like this:

Address 0 1 2 3

Data 11 22 33 44

Core in little-endian mode makes word access to address 0:

LDR r0, [0]

r0 contains 0×44332211

Data loaded to register as little endian

Core in little-endian mode makes byte access to address 0:

LDRB r0, [0]

r0 contains 0×00000011

Data loaded from 0

Core in little-endian mode makes byte access to address 3:

LDRB r0, [3]

r0 contains 0×00000044

Data loaded from 3

Core in BE-32 mode makes word access to address 0:

LDR r0, [0]

r0 contains 0×44332211

Word accesses are endianness-invariant

****Core in BE-32 mode makes word access to address 0:

LDRB r0, [0]

r0 contains 0×00000044

Byte access in BE-32 reads the word as if it was stored big-endian

Core in BE-32 mode makes word access to address 3:

LDRB r0, [1]

r0 contains 0×00000011

As above

****Core in BE-8 mode makes word access to address 0:

LDR r0, [0]

r0 contains 0×11223344

Data loaded to register as big endian

Core in BE-8 mode makes byte access to address 0:

LDRB r0, [0]

r0 contains 0×00000011

Byte at address 0 is loaded, NOT the byte at address 3

Core in BE-8 mode makes byte access to address 3:

LDRB r0, [3]

r0 contains 0×00000044

Byte at address 3 is loaded.

Essentially BE-32 operates by altering the addresses of memory accesses when accessing subword quantities. This gives the appearances of big endian.

我是使用 debian for ARM , gcc 4.4.2
[C]# cat f.c
int main()
{

float f1=1.2, f2=1.3;
f1 = f2*f1;
}[/C]
# gcc -mfpu=vfp -mfloat-abi=softfp -c f.c
# objdump -d f.o

f.o: file format elf32-littlearm

Disassembly of section .text:

00000000 :
0: e52db004 push {fp} ; (str fp, [sp, #-4]!)
4: e28db000 add fp, sp, #0
8: e24dd00c sub sp, sp, #12
c: eddf7a09 vldr s15, [pc, #36] ; 0×24
10: ed4b7a03 vstr s15, [fp, #-12]
14: eddf7a08 vldr s15, [pc, #32]
18: ed4b7a02 vstr s15, [fp, #-8]
1c: ed1b7a03 vldr s14, [fp, #-12]
20: ed5b7a02 vldr s15, [fp, #-8]
24: ee677a27 vmul.f32 s15, s14, s15
28: ed4b7a03 vstr s15, [fp, #-12]
2c: e28bd000 add sp, fp, #0
30: e8bd0800 pop {fp}
34: e12fff1e bx lr
38: 3f99999a .word 0×3f99999a
3c: 3fa66666 .word 0×3fa66666
有 vldr, vstr, vmul.f32 等 instruction .

# cat test2.c
[C]
#include <unistd.h>
#include <stdio.h>
void vfp_regs_load(float arrays[32])
{
asm volatile(『fldmias %0, {s0-s31}\n』
:
:』r』(arrays));
}
void vfp_regs_save(float arrays[32])
{
asm volatile (『fstmias %0, {s0-s31}』
:
:』r』(arrays));
}
void print_array(float array[32])
{
int i;
for(i=0; i<32; i++)
{
if(i%8==0)
printf(『\n』);
printf(『%f 『,i, array[i]);
}
printf(『\n』);
}
int main()
{
unsigned int fpscr;
float f1=1.0, f2=1.0;
float farrays[32], farrays2[32];
int i;
fpscr = 0×130000;
asm volatile (『fmxr fpscr, %0\n』
:
:』r』(fpscr));
asm volatile (『fmrx %0, fpscr\n』
:』=r』(fpscr));
vfp_regs_save(farrays2);
for(i=0; i<32; i++) farrays[i] = f1+f2*(float) i; vfp_regs_load(farrays); vfp_regs_save(farrays2); printf(『\n1:ScalarA op ScalarB->ScalarD』);
vfp_regs_load(farrays);
asm volatile(『fadds s0, s1, s2″);
vfp_regs_save(farrays2);
print_array(farrays2);
printf(『\n2:VectorA[?] op ScalarB->VectorD[?]『);
vfp_regs_load(farrays);
asm volatile(『fadds s8, s24, s0″);
vfp_regs_save(farrays2);
print_array(farrays2);
printf(『\n3:VectorA[?] op VectorB[?]->VectorD[?]『);
vfp_regs_load(farrays);
asm volatile(『fadds s8, s16, s24″);
vfp_regs_save(farrays2);
print_array(farrays2);
}[/C]

Vector Instruciton 的範例

# ./a.out

1:ScalarA op ScalarB->ScalarD
5.000000 2.000000 3.000000 4.000000 5.000000 6.000000 7.000000 8.000000
9.000000 10.000000 11.000000 12.000000 13.000000 14.000000 15.000000 16.000000
17.000000 18.000000 19.000000 20.000000 21.000000 22.000000 23.000000 24.000000
25.000000 26.000000 27.000000 28.000000 29.000000 30.000000 31.000000 32.000000

2:VectorA[?] op ScalarB->VectorD[?]
1.000000 2.000000 3.000000 4.000000 5.000000 6.000000 7.000000 8.000000
26.000000 10.000000 28.000000 12.000000 30.000000 14.000000 32.000000 16.000000
17.000000 18.000000 19.000000 20.000000 21.000000 22.000000 23.000000 24.000000
25.000000 26.000000 27.000000 28.000000 29.000000 30.000000 31.000000 32.000000

3:VectorA[?] op VectorB[?]->VectorD[?]
1.000000 2.000000 3.000000 4.000000 5.000000 6.000000 7.000000 8.000000
42.000000 10.000000 46.000000 12.000000 50.000000 14.000000 54.000000 16.000000
17.000000 18.000000 19.000000 20.000000 21.000000 22.000000 23.000000 24.000000
25.000000 26.000000 27.000000 28.000000 29.000000 30.000000 31.000000 32.000000

1:ScalarA op ScalarB->ScalarD
單純的二個浮點運算
2:VectorA[?] op ScalarB->VectorD[?]
一個 Vector * Scalar 運算
3:VectorA[?] op VectorB[?]->VectorD[?]
Vector * Vector 運算

Ref.

ARM VFP的一点体会 寫的不錯, 範例很好, 就.. 照作一次就 OK 了
VFP11 ™ VectorFloating-point Coprocessor Technical Reference Manual
for ARM1136JF-S processorr1p5

像我將 debian 裝好之後, 就可以跑 wireshark 了

本圖中, 最左邊的是 console, 中間的視窗是 EVB 上的 wireshark, 右邊的視窗是 PC 端的 wireshark

真是超好用的……

首先先裝 debootstrap

# apt-get install debootstrap

再裝 sid 系統 (現在應該是 unstable)

# debootstrap –verbose –foreign –arch armel sid ./sid http://ftp.tw.debian.org/debian

在此一提, 我是用 ARM11MPCore 平台, 用 armel (使用 EABI) 比較好, 記得有些套件在 armel 才會有. arm 己經沒有了

抓完套件下來以後, 將 sid 打包起來, 丟到 target 上去, 再解壓到 /

接下來在 Target board 上下

# /debootstrap/debootstrap –second-stage

運氣好就可以解完, 如果解不完就重做一次看看
因為我是在 host PC 上做的, 所以在做的時候有一些 information 就會帶過去

修改 /etc/fstab
/dev/sda1 / ext3 defaults,noatime,check=none 0 0
proc /proc proc defaults 0 0
devpts /dev/pts devpts mode=0620,gid=5 0 0

因為我是要常常開關測試, 所以不希望 fsck disk,
在 format disk 後, 可以考慮下這種參數
# mke2fs -j /dev/sda
# tune2fs -c 0 -i 0 /dev/sda1

/etc/inittab 也要修改, getty 的部份可以全部關掉 (如果有 LCD 可以留 1,2 個下來)

T0:23:respawn:/sbin/getty -L ttyS0 38400 vt100
#1:2345:respawn:/sbin/getty 38400 tty1
#2:23:respawn:/sbin/getty 38400 tty2
#3:23:respawn:/sbin/getty 38400 tty3
#4:23:respawn:/sbin/getty 38400 tty4
#5:23:respawn:/sbin/getty 38400 tty5
#6:23:respawn:/sbin/getty 38400 tty6

產生 sources.list

# echo 『deb http://ftp.tw.debian.org/debian unstable main non-free contrib』 > /etc/apt/sources.list

設定環境變數

echo LANG=\』C\』 >> /etc/environment

接下來就網路設一設, 然後就可以開始安裝環境了

# apt-get update
# apt-get install openssh-server
# apt-get install rcconf

接下來有用過 ubuntu 應該就很熟了….

Ref.

http://emqbit.com/deboostrap-debian

這時請改用 timer interrupt

在 load oprofile modules 加上 timer=1
若是 static link 時, 就要在 boot_cmd 加上 oprofile.timer=1

似乎目前的 kernel oprofile 不支援 ARM11MPCore.

http://oprofile.sourceforge.net/doc/detailed-parameters.html#timer