These two instructions generate a bit mask using operands NB and ME. The operands are defined in the ppc_common.sinc file as "ME=(1,5)" and "NB= (11,15)". ME looks correct but NB is incorrectly set to the same location as the bit bit shift field "SHL=(11,15)".
I think the definition of NB in ppc_common.since should read "NB=(6,10)" instead.
As an example "74 63 37 7F" decodes using as "e_rlwinm A,S,SHL,NB,ME" with fields:
OP=(26,31) = 011101
S=(21,25) = 00011 gives r3
A=(16,20) = 00011 gives r3
SHL=(11,15) = 00110 0x06 , But NB=(11,15) so gets 0x06
(6,10) = 11101 NB should (6,10) to go here and get 0x1D
ME=(1,5) = 11111 0x1F
BIT_0=(0,0) = 1
Which gives the incorrect current behavior:
08fd4168 74 63 37 7f e_rlwinm r3,r3,0x6,0x6,0x1f
Whereas the expected correct behavior taking NB=(6,10) would be:
08fd4168 74 63 37 7f e_rlwinm r3,r3,0x6,0x1d,0x1f
For reference the VLE reference manual e_rlwinm instruction is shown below. Note that Ghidra uses NB in place of MB and that the manual's bit ordering needs to be reversed. So operand MB at bits 21..25 would correspond to NB(6,10).
GhidorahRex
commented
5 years ago
These two instructions generate a bit mask using operands NB and ME. The operands are defined in the ppc_common.sinc file as "ME=(1,5)" and "NB= (11,15)". ME looks correct but NB is incorrectly set to the same location as the bit bit shift field "SHL=(11,15)".
I think the definition of NB in ppc_common.since should read "NB=(6,10)" instead.
As an example "74 63 37 7F" decodes using as "e_rlwinm A,S,SHL,NB,ME" with fields:
Which gives the incorrect current behavior: 08fd4168 74 63 37 7f e_rlwinm r3,r3,0x6,0x6,0x1f
Whereas the expected correct behavior taking NB=(6,10) would be: 08fd4168 74 63 37 7f e_rlwinm r3,r3,0x6,0x1d,0x1f
For reference the VLE reference manual e_rlwinm instruction is shown below. Note that Ghidra uses NB in place of MB and that the manual's bit ordering needs to be reversed. So operand MB at bits 21..25 would correspond to NB(6,10).