[Bugfix] Early-reject requests with MM data longer than encode cache capacity (#33110)

Signed-off-by: YunzhuLu <lucia.yunzhu@gmail.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Roger Wang <hey@rogerw.io>

vllm/v1/engine/input_processor.py

@@ -35,6 +35,7 @@ from vllm.tokenizers import TokenizerLike
 from vllm.tokenizers.mistral import MistralTokenizer
 from vllm.utils import length_from_prompt_token_ids_or_embeds, random_uuid
 from vllm.utils.torch_utils import set_default_torch_num_threads
+from vllm.v1.core.encoder_cache_manager import compute_mm_encoder_budget
 from vllm.v1.engine import EngineCoreRequest
 from vllm.v1.metrics.stats import MultiModalCacheStats
 from vllm.v1.structured_output.backend_guidance import (
@@ -68,6 +69,17 @@ class InputProcessor:
 
         self.mm_registry = mm_registry
         self.mm_processor_cache = mm_registry.processor_cache_from_config(vllm_config)
+        self.mm_encoder_cache_size = None
+        if (
+            self.mm_registry.supports_multimodal_inputs(self.model_config)
+            and not self.model_config.skip_tokenizer_init
+        ):
+            max_tokens_by_modality = mm_registry.get_max_tokens_per_item_by_modality(
+                self.model_config
+            )
+            _, self.mm_encoder_cache_size = compute_mm_encoder_budget(
+                self.vllm_config.scheduler_config, max_tokens_by_modality
+            )
 
         self.input_preprocessor = InputPreprocessor(
             self.model_config,
@@ -743,6 +755,25 @@ class InputProcessor:
                 f"model length of {max_prompt_len}. {suggestion}"
             )
 
+        if (
+            prompt_type == "decoder"
+            and prompt_inputs["type"] == "multimodal"
+            and self.mm_encoder_cache_size is not None
+        ):
+            decoder_mm_positions = prompt_inputs["mm_placeholders"]
+            for modality, mm_positions in decoder_mm_positions.items():
+                for mm_position in mm_positions:
+                    embed_length = mm_position.get_num_embeds
+                    if embed_length > self.mm_encoder_cache_size:
+                        raise ValueError(
+                            f"The {prompt_type} prompt contains a(n) {modality} item "
+                            f"with length {embed_length}, which exceeds the "
+                            f"pre-allocated encoder cache size "
+                            f"{self.mm_encoder_cache_size}. Please reduce the input "
+                            f"size or increase the encoder cache size "
+                            f"by setting --limit-mm-per-prompt at startup."
+                        )
+
     def stat_mm_cache(self) -> MultiModalCacheStats | None:
         return self.input_preprocessor.stat_mm_cache()