The pattern searcher used by is_pattern_unique appears to be broken in BN 5.3.9434.
I was unable to generate signatures starting at some instructions, where nearby instructions would work.
I attached log excerpts and the assembly which I targeted in the txt attachment.
Referencing the attachment, it's very odd that as the signature is being built, uniqueness check match count can collapse to 0 (should decrease towards 1 if I understand the logic correctly).
I can say that the issue probably lies behind is_pattern_unique, as I was able to work around the issue by substituting the custom pattern searcher with bv.search.
p0lymeric@2c18a09
I did not debug much beyond is_pattern_unique, so I'm not sure about actual cause.
Thanks again for a very cool plugin, now I don't need to drop into Ghidra for this task. :)
Attachment:
coolsigmaker.txt
The pattern searcher used by is_pattern_unique appears to be broken in BN 5.3.9434.
I was unable to generate signatures starting at some instructions, where nearby instructions would work.
I attached log excerpts and the assembly which I targeted in the txt attachment.
Referencing the attachment, it's very odd that as the signature is being built, uniqueness check match count can collapse to 0 (should decrease towards 1 if I understand the logic correctly).
I can say that the issue probably lies behind is_pattern_unique, as I was able to work around the issue by substituting the custom pattern searcher with bv.search.
p0lymeric@2c18a09
I did not debug much beyond is_pattern_unique, so I'm not sure about actual cause.
Thanks again for a very cool plugin, now I don't need to drop into Ghidra for this task. :)
Attachment:
coolsigmaker.txt