Regenerate readme with new tweaks
This commit is contained in:
parent
6cbd672a3e
commit
27afe49339
21
readme.md
21
readme.md
@ -42,12 +42,10 @@ to dig into source code or read the [ASAN paper](https://static.googleuserconten
|
||||
|
||||
We use a stack-allocated 16 byte array and test un/poisoning
|
||||
various ranges of bytes from different alignments to clarify the
|
||||
poisoning behaviour of the API.
|
||||
|
||||
This reveals that calling the API haphazardly, unaligned or
|
||||
straddling boundaries can lead to gaps in poisoned memory and hide
|
||||
potential leaks (as also demonstrated in [Manual ASAN poisoning and
|
||||
alignment](https://github.com/mcgov/asan_alignment_example)).
|
||||
poisoning behaviour of the API. This reveals that calling the API
|
||||
haphazardly, unaligned or straddling boundaries can lead to gaps in
|
||||
poisoned memory and hide potential leaks (as also demonstrated in
|
||||
[Manual ASAN poisoning and alignment](https://github.com/mcgov/asan_alignment_example)).
|
||||
|
||||
## References
|
||||
|
||||
@ -56,11 +54,12 @@ alignment](https://github.com/mcgov/asan_alignment_example)).
|
||||
- [sanitizer/asan_interface.h](https://github.com/llvm-mirror/compiler-rt/blob/master/include/sanitizer/asan_interface.h)
|
||||
|
||||
## Raw Test Results
|
||||
Here we demonstrate that ASAN poison-ing will only poison the
|
||||
byte region if the region meets an 8 byte boundary. It will only
|
||||
poison bytes upto the 8 byte boundary, any bytes that straddle
|
||||
the boundary that do not hit the next 8 byte boundary are not
|
||||
poison-ed.
|
||||
|
||||
Here we poison a sliding window of 7 bytes to demonstrate that ASAN
|
||||
poisoning will only poison the byte region if the region meets an 8
|
||||
byte aligned address. It will only poison bytes up to the boundary,
|
||||
any bytes that straddle the boundary that do not hit the next 8 byte
|
||||
boundary are not poisoned.
|
||||
|
||||
```
|
||||
Byte Array 00 01 02 03 04 05 06 07 | 08 09 10 11 12 13 14 15
|
||||
|
Loading…
Reference in New Issue
Block a user