diff options
Diffstat (limited to 'miniasm/README')
| -rw-r--r-- | miniasm/README | 35 |
1 files changed, 35 insertions, 0 deletions
diff --git a/miniasm/README b/miniasm/README new file mode 100644 index 0000000..a7098f2 --- /dev/null +++ b/miniasm/README @@ -0,0 +1,35 @@ +Design +------ + +- one-pass: patching up addresses later or two-pass (scan location of all symbols first, + have them ready for the second pass), two-pass assemblers are easier to write and + relocation of addresses is possible in a deferred way, for instance at load time +- intermediate listing-format +- LC: location counter +- extreme approach: encode the specific assembly language (e.g. register names) into + the grammar, another extreme approach: hand-coded parser to parse the lines. One + line == one instruction is something we usually don't like in a grammar ('\n' having + meaning in the grammar), on the other hand keywords in assembly ('INSTR MOV AX, B END') + looks a little bit clumpsy. Definitely when we have complex address calcualations + we can benefit from a real parser (generated or not). +- one vs. two-pass +- labels and local labels (FUNC: and .loop and FUNC.loop) +- LC-relative addressing $+5 + +Building +-------- + +gcc -I../minilib -I../miniemu -g -O0 -m32 -march=i386 -ffreestanding -Werror -Wall -Wno-return-type -pedantic -std=c89 -o miniasm *.c ../minilib/*.c +clang -I../minilib -I../miniemu -g -O0 -march=i386 -fno-builtin -std=c89 -Werror -Wall -Wno-return-type -o miniasm *.c ../minilib/*.c + +Usage +----- + +./miniasm test1.asm ../miniemu/test1.bin +./miniasm test2.asm ../miniemu/test2.bin +./miniasm test3.asm ../miniemu/test3.bin +./miniasm test4.asm ../miniemu/test4.bin +./miniasm test5.asm ../miniemu/test5.bin +./miniasm test6.asm ../miniemu/test6.bin +./miniasm test7.asm ../miniemu/test7.bin +./miniasm test8.asm ../miniemu/test8.bin |