Support padding for the leaves of the merkle tree (#647)

* Did some organization in test_hash_api and fixed some edge case that can cause the test to seg-fault

Signed-off-by: Koren-Brand <[email protected]>

* basic test pass

* all tests pass without padding

* basic test pass

* Revert "basic test pass"

This reverts commit 8018c0d.

* basic tests pass

* Initial test for padding in merkle tree. currently crashes when trying to generate a proof of the last (and only) leaf index

Signed-off-by: Koren-Brand <[email protected]>

* bug fix at proof allocation

* output buffer overflow in hash (seems to launch hash twice)

Signed-off-by: Koren-Brand <[email protected]>

* bug fix for the case of 17 hases at the last layer

* formating

* Partial leaf test of zero padding has memory overflow when copying to padded_leaves (during build, init_padded_leaves, line 441)

Signed-off-by: Koren-Brand <[email protected]>

* bug fix T tHe padded vector

* Added interesting cases for zero padding and duplicated the zero padding test for last value padding

Signed-off-by: Koren-Brand <[email protected]>

* format

* input_default -> default_input

* assert changed

* spelling fix

* format

* Weird test with cuda and cpu MerkleTreeLarge - root doesn't match and cuda always has zeros in some parts

Signed-off-by: Koren-Brand <[email protected]>

* Prints now work (Decimal instead of hex)

Signed-off-by: Koren-Brand <[email protected]>

* Fixed stack-use-after-scope error (Accessing root after tree is deleted) and some additional memory leaks in Merkle tree cpu backend as well as the hash tests

Signed-off-by: Koren-Brand <[email protected]>

* formatting

Signed-off-by: Koren-Brand <[email protected]>

* Fixed bug that caused poseidon merkle tree to fail on rust (And added the same test to cpp test_hash_api)

Signed-off-by: Koren-Brand <[email protected]>

* Bug fix in poseidon_tree test in test-hash-api (CPP) where field elements weren't randomized properly

Signed-off-by: Koren-Brand <[email protected]>

* formatting

Signed-off-by: Koren-Brand <[email protected]>

* Fixed the documentation to match the actual function is_valid_tree

Signed-off-by: Koren-Brand <[email protected]>

* other PR changes

Signed-off-by: Koren-Brand <[email protected]>

* fix device that was not defined in scope to s_reference_target in padding tests

Signed-off-by: Koren-Brand <[email protected]>

* use smart pointers

* format

* Standardized the Merkle tree test in test_hash_api.cpp to all use the same test_merkle_tree function

Signed-off-by: Koren-Brand <[email protected]>

* spelling

Signed-off-by: Koren-Brand <[email protected]>

* Statistically possible valid wrong leaves during Merkle tree test has been addressed

Signed-off-by: Koren-Brand <[email protected]>

* formatting

Signed-off-by: Koren-Brand <[email protected]>

---------

Signed-off-by: Koren-Brand <[email protected]>
Co-authored-by: Koren-Brand <[email protected]>

icicle/CMakeLists.txt

@@ -18,6 +18,11 @@ endif()
 # Print the selected build type
 message(STATUS "Build type: ${CMAKE_BUILD_TYPE}")
 
+# Prevent build if both SANITIZE and CUDA_BACKEND are enabled
+if(SANITIZE AND CUDA_BACKEND) 
+  message(FATAL_ERROR "Address sanitizer and Cuda cannot be enabled at the same time.")
+endif()
+
 # Find the ccache program
 find_program(CCACHE_PROGRAM ccache)
 # If ccache is found, use it as the compiler launcher
@@ -132,5 +137,4 @@ endif()
 
 if (BUILD_TESTS)
   add_subdirectory(tests)
-endif()
-
+endif()