Add information on generating new corpus files

test/fuzz/README.md

@@ -1,8 +1,32 @@
 # Corpora for fuzz tests with fixed API calls scenarios
 Corpora in 'corpus' directory contain UR API calls in a predefined order described below.
-All scenarios begin with single calls to urInit() and urAdapterGet().
+All such scenarios begin with single calls to urInit() and urAdapterGet().
+Corpus files are binary files containing ASCII characters which are interpreted by the test
+backwards, meaning that bytes are read from the end of the file to the beginning of the file.
+
+More corpora can be generated by fuzzer. Just run the test with a first positional parameter
+which provides the path where any new corpus will be saved. The path has to exist.
+It's worth running the test with tracing enabled while picking scenarios to be added to the repository
+for future short fuzz tests runs. Example of running the test with generating new corpus files:
+```
+UR_ADAPTERS_FORCE_LOAD=build/lib/libur_adapter_null.so \
+XPTI_TRACE_ENABLE=1 \
+XPTI_FRAMEWORK_DISPATCHER=build/lib/libxptifw.so \
+XPTI_SUBSCRIBERS=build/lib/libcollector.so \
+UR_ENABLE_LAYERS=UR_LAYER_TRACING \
+./build/bin/fuzztest-base test/fuzz/corpus -seed=1 -max_total_time=120 -verbosity=1
+```
+
+Pass path to a corpus file instead to run a single scenario:
+```
+UR_ADAPTERS_FORCE_LOAD=build/lib/libur_adapter_null.so \
+XPTI_TRACE_ENABLE=1 \
+XPTI_FRAMEWORK_DISPATCHER=build/lib/libxptifw.so \
+XPTI_SUBSCRIBERS=build/lib/libcollector.so \
+UR_ENABLE_LAYERS=UR_LAYER_TRACING \
+./build/bin/fuzztest-base test/fuzz/corpus/alloc -verbosity=1
+```
 
-<!--- TODO: extend this doc -->
 More details on seed corpora for fuzzer can be found
 [here](https://github.com/google/fuzzing/blob/master/tutorial/libFuzzerTutorial.md#seed-corpus).