whisper : guard null source in buffer loader read callback (#3982)

* whisper : guard null source in buffer loader read callback

whisper_init_from_buffer_with_params_no_state installs a read callback that
copies from buf->buffer + current_offset. When the buffer is exhausted (or the
supplied buffer is empty), size_to_copy is 0 and the source pointer can be null;
passing a null pointer to memcpy is undefined behavior even for a zero-length
copy (UBSan: 'null pointer passed as argument 2' at the memcpy). Loading a
crafted/short model through the buffer loader could hit this.

Skip the memcpy when there is nothing to copy. Loading from a null/empty or
truncated buffer now fails gracefully (returns NULL) with no UB.

This addresses bug 1 of #3879. Bug 2 (integer overflow when sizing the mel
filter buffer) is covered by the open PR #3780.

* fixup! whisper : guard null source in buffer loader read callback

---------

Co-authored-by: Ben Younes <2910651+ousamabenyounes@users.noreply.github.com>

Author: ousamabenyounes

src/parakeet.cpp

@@ -3109,7 +3109,9 @@ struct parakeet_context * parakeet_init_from_buffer_with_params_no_state(void *
 
         size_t size_to_copy = buf->current_offset + read_size < buf->size ? read_size : buf->size - buf->current_offset;
 
-        memcpy(output, buf->buffer + buf->current_offset, size_to_copy);
+        if (size_to_copy > 0 && buf->buffer != nullptr) {
+            memcpy(output, buf->buffer + buf->current_offset, size_to_copy);
+        }
         buf->current_offset += size_to_copy;
 
         return size_to_copy;

src/whisper.cpp

@@ -3766,7 +3766,9 @@ struct whisper_context * whisper_init_from_buffer_with_params_no_state(void * bu
 
         size_t size_to_copy = buf->current_offset + read_size < buf->size ? read_size : buf->size - buf->current_offset;
 
-        memcpy(output, buf->buffer + buf->current_offset, size_to_copy);
+        if (size_to_copy > 0 && buf->buffer != nullptr) {
+            memcpy(output, buf->buffer + buf->current_offset, size_to_copy);
+        }
         buf->current_offset += size_to_copy;
 
         return size_to_copy;

tests/CMakeLists.txt

@@ -96,6 +96,13 @@ target_link_libraries(${UTF8_TEST} PRIVATE common)
 add_test(NAME ${UTF8_TEST} COMMAND ${UTF8_TEST})
 set_tests_properties(${UTF8_TEST} PROPERTIES LABELS "unit")
 
+set(BUFFER_LOADER_TEST test-whisper-buffer-loader)
+add_executable(${BUFFER_LOADER_TEST} ${BUFFER_LOADER_TEST}.cpp)
+target_include_directories(${BUFFER_LOADER_TEST} PRIVATE ../include ../ggml/include ../examples)
+target_link_libraries(${BUFFER_LOADER_TEST} PRIVATE common)
+add_test(NAME ${BUFFER_LOADER_TEST} COMMAND ${BUFFER_LOADER_TEST})
+set_tests_properties(${BUFFER_LOADER_TEST} PROPERTIES LABELS "unit;gh")
+
 # VAD test tests VAD in isolation
 set(VAD_TEST test-vad)
 add_executable(${VAD_TEST} ${VAD_TEST}.cpp)
@@ -177,4 +184,3 @@ add_parakeet_transcription_test(
     samples/diffusion2023-07-03.flac
     tests/parakeet-expected-diffusion-output.txt
     0.95)
-

tests/test-whisper-buffer-loader.cpp

@@ -0,0 +1,24 @@
+#include "whisper.h"
+
+#include <cstdint>
+#include <cstdio>
+
+#ifdef NDEBUG
+#undef NDEBUG
+#endif
+#include <cassert>
+
+int main() {
+    struct whisper_context_params cparams = whisper_context_default_params();
+    cparams.use_gpu = false;
+
+    struct whisper_context * ctx_empty = whisper_init_from_buffer_with_params(nullptr, 1, cparams);
+    assert(ctx_empty == nullptr);
+
+    uint8_t truncated[8] = { 0 };
+    struct whisper_context * ctx_trunc = whisper_init_from_buffer_with_params(truncated, sizeof(truncated), cparams);
+    assert(ctx_trunc == nullptr);
+
+    printf("test-whisper-buffer-loader: OK\n");
+    return 0;
+}