add BinExport tests #2340
add BinExport tests #2340
Conversation
|
alternative idea, we use a tiny ELF ARM binary like this instead of writing a manual BinExport proto definition
|
There was a problem hiding this comment.
I think this is a good start, I like how BE2_DICT visualizes our expectations for the breakdown expression, operand, etc.. My concern here is that these tests won't catch if an underlying tool (Ghidra, IDA, etc.) changes how it stores expressions, operands, etc. because we our using our expectations to model the data, which as we've learned, isn't consistent across the tools. Your suggestion of generating BinExport files for smaller programs to use for these tests may be a better fit.
|
I'm working on something using a small ELF file for the tests now. |
tests/test_binexport_accessors.py
Outdated
| ) | ||
|
|
||
|
|
||
| def _TODO_test_is_stack_register_expression(): |
There was a problem hiding this comment.
no stack expressions in the small ELF so using the hardcoded BE2 data here
mr-tz
force-pushed
the
tests/add-binexport
branch
from
62b6019 to
c662176
Compare
mr-tz
force-pushed
the
tests/add-binexport
branch
from
a97afb7 to
7142bf7
Compare
| ), | ||
| ), | ||
| ), | ||
| # 00210184 02 68 61 38 ldrb w2,[x0, x1, LSL ] |
There was a problem hiding this comment.
BinaryNinja also sees this as ldrb w2, [x0, x1]. so perhaps when Ghidra sees lsl with no sub expressions (as is the case here), we can omit it?
There was a problem hiding this comment.
hm, still a bug in Ghidra? anyway, we can ignore LSL #0
02 68 61 38 ldrb w2, [x0, x1]
02 78 61 38 ldrb w2, [x0, x1, lsl #0]There was a problem hiding this comment.
yeah ghidra is definitely missing the #0 constant, at least the BinExport is. locally i have a fix, but of course it would be nice for this to be fixed upstream.
|
sounds good to me |
WIP to discuss the approach
Is this to brittle / confusing or does it make sense?
Checklist