llava : fix memory management bug (#5491)

* Fix memory management in llava and server code

Fixes this error:

llama_new_context_with_model: graph splits (measure): 3
Available slots:
 -> Slot 0 - max context: 6000
{"timestamp":1707926446,"level":"INFO","function":"main","line":2623,"message":"model loaded"}
all slots are idle and system prompt is empty, clear the KV cache
slot 0 - loaded image
slot 0 is processing [task id: 0]
slot 0 : kv cache rm - [0, end)
slot 0 - encoding image [id: 1]
munmap_chunk(): invalid pointer
Aborted

* Make it cleaner by checking size in batch free wrapper

examples/llava/clip.cpp

@@ -1230,8 +1230,20 @@ struct clip_image_f32 * clip_image_f32_init() {
     return new clip_image_f32();
 }
 
-void clip_image_u8_free (struct clip_image_u8  * img) { delete img; }
+void clip_image_u8_free(struct clip_image_u8  * img) { delete img; }
 void clip_image_f32_free(struct clip_image_f32 * img) { delete img; }
+void clip_image_u8_batch_free(struct clip_image_u8_batch  & batch) {
+    if (batch.size > 0) {
+        delete[] batch.data;
+        batch.size = 0;
+    }
+}
+void clip_image_f32_batch_free(struct clip_image_f32_batch  & batch) {
+    if (batch.size > 0) {
+        delete[] batch.data;
+        batch.size = 0;
+    }
+}
 
 static void build_clip_img_from_data(const stbi_uc * data, int nx, int ny, clip_image_u8 * img) {
     img->nx = nx;
@@ -1494,11 +1506,8 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, cli
         pad_to_square = false;
     }
     // free the previous res_imgs if any set
-    if (res_imgs.size > 0 && res_imgs.size < 100) {
-        for (size_t i = 0; i < res_imgs.size; i++) {
-            clip_image_f32_free(&(res_imgs.data[i]));
-        }
-        delete[] res_imgs.data;
+    if (res_imgs.size > 0) {
+        clip_image_f32_batch_free(res_imgs);
     }
     res_imgs.data = nullptr;
     res_imgs.size = 0;
@@ -1650,7 +1659,8 @@ bool clip_image_preprocess(struct clip_ctx * ctx, const clip_image_u8 * img, cli
 
     res_imgs.size = 1;
     res_imgs.data = new clip_image_f32[res_imgs.size];
-    res_imgs.data[0] = std::move(*res);
+    res_imgs.data[0] = *res;
+    clip_image_f32_free(res);
 
     return true;
 }

examples/llava/clip.h

@@ -60,6 +60,8 @@ CLIP_API struct clip_image_f32 * clip_image_f32_init();
 
 CLIP_API void clip_image_u8_free (struct clip_image_u8  * img);
 CLIP_API void clip_image_f32_free(struct clip_image_f32 * img);
+CLIP_API void clip_image_u8_batch_free (struct clip_image_u8_batch  & batch);
+CLIP_API void clip_image_f32_batch_free(struct clip_image_f32_batch & batch);
 
 CLIP_API bool clip_image_load_from_file(const char * fname, struct clip_image_u8 * img);
 

examples/server/server.cpp

@@ -975,7 +975,12 @@ struct llama_server_context
             {
                 LOG_TEE("Error processing the given image");
                 clip_free(clp_ctx);
-                clip_image_f32_free(img_res_v.data);
+                clip_image_f32_batch_free(img_res_v);
+                return false;
+            }
+            if (img_res_v.size == 0)
+            {
+                LOG_TEE("Error processing the given image");
                 return false;
             }
 
@@ -987,6 +992,7 @@ struct llama_server_context
             if (!img.image_embedding)
             {
                 LOG_TEE("Unable to allocate memory for image embeddings\n");
+                clip_image_f32_batch_free(img_res_v);
                 clip_free(clp_ctx);
                 return false;
             }
@@ -994,10 +1000,11 @@ struct llama_server_context
             if (!clip_image_encode(clp_ctx, params.n_threads, img_res, img.image_embedding))
             {
                 LOG_TEE("Unable to encode image\n");
+                clip_image_f32_batch_free(img_res_v);
                 return false;
             }
 
-            clip_image_f32_free(img_res_v.data);
+            clip_image_f32_batch_free(img_res_v);
 
             img.request_encode_image = false;
         }