From 1e503a586d9f3d45e27bc1dbb3b2093c1e5f7b6f Mon Sep 17 00:00:00 2001 From: hexcoder- Date: Sun, 22 Sep 2019 08:02:56 +0200 Subject: first realisation of splitting floating point compares activated with AFL_LLVM_LAF_SPLIT_COMPARES=1 needs testing on big endian machines A compare is split into sign integer comparison exponent integer comparison mantissa/fraction integer comparison These integer comparisons are further splitted if they are bigger than a byte. --- docs/env_variables.txt | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'docs') diff --git a/docs/env_variables.txt b/docs/env_variables.txt index d5688939..e763118e 100644 --- a/docs/env_variables.txt +++ b/docs/env_variables.txt @@ -97,7 +97,8 @@ Then there are a few specific features that are only available in llvm_mode: - Setting AFL_LLVM_LAF_TRANSFORM_COMPARES will split string compare functions - - Setting AFL_LLVM_LAF_SPLIT_COMPARES will split > 8 bit CMP instructions + - Setting AFL_LLVM_LAF_SPLIT_COMPARES will split all floating point and + 64, 32 and 16 bit integer CMP instructions See llvm_mode/README.laf-intel for more information. -- cgit 1.4.1