Hi, > After a bit of a break (moving, new job, etc.) I'm getting (hopefully) > pretty close to having the Sparc disassembler complete. So far, the > bugs have mostly been me fat fingering data as I entered the > instruction data into tables.
Just out of curiosity, why are you doing Sparc disassembler? dis is
standard part of Solaris.
$ uname -a
SunOS box 5.9 Generic_112233-11 sun4u sparc SUNW,Sun-Fire-480R
$ /usr/ccs/bin/dis /bin/ls | head -20
**** DISASSEMBLER ****
disassembly for /bin/ls
section .text
_start()
10f00: bc 10 20 00 clr %fp
10f04: e0 03 a0 40 ld [%sp + 0x40], %l0
10f08: 13 00 00 93 sethi %hi(0x24c00), %o1
10f0c: e0 22 60 3c st %l0, [%o1 + 0x3c]
10f10: a2 03 a0 44 add %sp, 0x44, %l1
10f14: 13 00 00 93 sethi %hi(0x24c00), %o1
10f18: e2 22 60 38 st %l1, [%o1 + 0x38]
10f1c: 13 00 00 90 sethi %hi(_GLOBAL_OFFSET_TABLE_), %o1
10f20: e2 22 63 a4 st %l1, [%o1 + 0x3a4]
10f24: a5 2c 20 02 sll %l0, 0x2, %l2
10f28: a4 04 a0 04 add %l2, 0x4, %l2
10f2c: a4 04 40 12 add %l1, %l2, %l2
10f30: 27 00 00 90 sethi %hi(_GLOBAL_OFFSET_TABLE_), %l3
> For a given instruction, there might be multiple correct strings that
> it disassembles to. Are there any requirements for a specific output
> when there might be multiple correct values (such as 'add %g0, 0x5,
> %l1' vs. 'inc 0x5, %l1') ?
Who would pose such requirements?
> The only one I can think of would be to supress display of superfluous
> 0 / %g0 values. Any others?
I would say, that if the output is compilable again, it's good enough.
--
Vlad
pgpaMWi91ZQlS.pgp
Description: PGP signature
_______________________________________________ tools-discuss mailing list tools-discuss@opensolaris.org
