Issue Station
Sign In
Tools hereToolbox

vllm - 💡(How to fix) Fix [Bug]: Structured-output scheduler can keep advancing a terminated xgrammar matcher

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

I found a structured-output lifecycle bug in the xgrammar path.

In a frontend-legal structured-output workload, a structured-output request can already reach a terminated xgrammar matcher state, but later scheduler iterations still try to advance that same matcher with more generated tokens. On the re-verified Qwen3 AsyncLLM runs from May 14, 2026, this reproduces 2/2 times and leaves a stable internal error surface in the engine logs:

Failed to advance FSM
Unexpected: grammar rejected tokens
The matcher has terminated ... but is trying to accept new token

This is not a malformed-input bug. The request shape is legal. The problem is that the terminated grammar state is not consistently honored across the structured-output scheduling path.

Error Message

times and leaves a stable internal error surface in the engine logs: logger.error(

  • error.txt on child failure ERROR ... Failed to advance FSM for request r1_route-... for tokens 0. Please file an issue. ERROR ... Failed to advance FSM for request probe_seq-... for tokens 0. Please file an issue.

Root Cause

This is a structured-output state-management bug, not a bad-request rejection.

The important distinction is:

  • the request is frontend-legal
  • the matcher already reached a terminated state
  • later scheduler iterations still try to advance that same matcher

So the real problem is that the structured-output scheduling path does not consistently stop grammar advancement once the matcher has already terminated.

Fix Action

Fix / Workaround

============================== CPU Info

Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 52 bits physical, 57 bits virtual Byte Order: Little Endian CPU(s): 384 On-line CPU(s) list: 0-383 Vendor ID: AuthenticAMD Model name: AMD EPYC 9654 96-Core Processor CPU family: 25 Model: 17 Thread(s) per core: 2 Core(s) per socket: 96 Socket(s): 2 Stepping: 1 Frequency boost: enabled CPU max MHz: 3707.8120 CPU min MHz: 1500.0000 BogoMIPS: 4792.57 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good amd_lbr_v2 nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 invpcid_single hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid overflow_recov succor smca fsrm flush_l1d Virtualization: AMD-V L1d cache: 6 MiB (192 instances) L1i cache: 6 MiB (192 instances) L2 cache: 192 MiB (192 instances) L3 cache: 768 MiB (24 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0-95,192-287 NUMA node1 CPU(s): 96-191,288-383 Vulnerability Gather data sampling: Not affected Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Mitigation; Safe RET Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected

Code Example

Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 13.1.0-8ubuntu1~22.04) 13.1.0
Clang version                : 16.0.6 (++20231112100510+7cbf1a259152-1~exp1~20231112100554.106)
CMake version                : version 3.22.1
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.10.0+cu128
Is debug build               : False
CUDA used to build PyTorch   : 12.8
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.9 | packaged by Anaconda, Inc. | (main, Feb  6 2025, 18:56:27) [GCC 11.2.0] (64-bit runtime)
Python platform              : Linux-6.5.0-35-generic-x86_64-with-glibc2.35
    
==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.8.61
CUDA_MODULE_LOADING set to   : 
GPU models and configuration : 
GPU 0: NVIDIA GeForce RTX 4090
GPU 1: NVIDIA GeForce RTX 4090

Nvidia driver version        : 570.86.10
cuDNN version                : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.3.0
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                       x86_64
CPU op-mode(s):                     32-bit, 64-bit
Address sizes:                      52 bits physical, 57 bits virtual
Byte Order:                         Little Endian
CPU(s):                             384
On-line CPU(s) list:                0-383
Vendor ID:                          AuthenticAMD
Model name:                         AMD EPYC 9654 96-Core Processor
CPU family:                         25
Model:                              17
Thread(s) per core:                 2
Core(s) per socket:                 96
Socket(s):                          2
Stepping:                           1
Frequency boost:                    enabled
CPU max MHz:                        3707.8120
CPU min MHz:                        1500.0000
BogoMIPS:                           4792.57
Flags:                              fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good amd_lbr_v2 nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 invpcid_single hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid overflow_recov succor smca fsrm flush_l1d
Virtualization:                     AMD-V
L1d cache:                          6 MiB (192 instances)
L1i cache:                          6 MiB (192 instances)
L2 cache:                           192 MiB (192 instances)
L3 cache:                           768 MiB (24 instances)
NUMA node(s):                       2
NUMA node0 CPU(s):                  0-95,192-287
NUMA node1 CPU(s):                  96-191,288-383
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit:        Not affected
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      Not affected
Vulnerability Retbleed:             Not affected
Vulnerability Spec rstack overflow: Mitigation; Safe RET
Vulnerability Spec store bypass:    Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:           Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:           Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.4
[pip3] numpy==2.0.2
[pip3] nvidia-cublas-cu12==12.8.4.1
[pip3] nvidia-cuda-cupti-cu12==12.8.90
[pip3] nvidia-cuda-nvrtc-cu12==12.8.93
[pip3] nvidia-cuda-runtime-cu12==12.8.90
[pip3] nvidia-cudnn-cu12==9.10.2.21
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft-cu12==11.3.3.83
[pip3] nvidia-cufile-cu12==1.13.1.3
[pip3] nvidia-curand-cu12==10.3.9.90
[pip3] nvidia-cusolver-cu12==11.7.3.90
[pip3] nvidia-cusparse-cu12==12.5.8.93
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.4.2
[pip3] nvidia-cutlass-dsl-libs-base==4.4.2
[pip3] nvidia-ml-py==13.590.48
[pip3] nvidia-nccl-cu12==2.27.5
[pip3] nvidia-nvjitlink-cu12==12.8.93
[pip3] nvidia-nvshmem-cu12==3.4.5
[pip3] nvidia-nvtx-cu12==12.8.90
[pip3] optree==0.15.0
[pip3] pyzmq==27.1.0
[pip3] torch==2.10.0+cu128
[pip3] torch_c_dlpack_ext==0.1.5
[pip3] torchaudio==2.10.0+cu128
[pip3] torchvision==0.25.0+cu128
[pip3] transformers==4.56.1
[pip3] triton==3.6.0
[conda] flashinfer-python         0.6.4                    pypi_0    pypi
[conda] numpy                     2.0.2                    pypi_0    pypi
[conda] nvidia-cublas-cu12        12.8.4.1                 pypi_0    pypi
[conda] nvidia-cuda-cupti-cu12    12.8.90                  pypi_0    pypi
[conda] nvidia-cuda-nvrtc-cu12    12.8.93                  pypi_0    pypi
[conda] nvidia-cuda-runtime-cu12  12.8.90                  pypi_0    pypi
[conda] nvidia-cudnn-cu12         9.10.2.21                pypi_0    pypi
[conda] nvidia-cudnn-frontend     1.18.0                   pypi_0    pypi
[conda] nvidia-cufft-cu12         11.3.3.83                pypi_0    pypi
[conda] nvidia-cufile-cu12        1.13.1.3                 pypi_0    pypi
[conda] nvidia-curand-cu12        10.3.9.90                pypi_0    pypi
[conda] nvidia-cusolver-cu12      11.7.3.90                pypi_0    pypi
[conda] nvidia-cusparse-cu12      12.5.8.93                pypi_0    pypi
[conda] nvidia-cusparselt-cu12    0.7.1                    pypi_0    pypi
[conda] nvidia-cutlass-dsl        4.4.2                    pypi_0    pypi
[conda] nvidia-cutlass-dsl-libs-base 4.4.2                    pypi_0    pypi
[conda] nvidia-ml-py              13.590.48                pypi_0    pypi
[conda] nvidia-nccl-cu12          2.27.5                   pypi_0    pypi
[conda] nvidia-nvjitlink-cu12     12.8.93                  pypi_0    pypi
[conda] nvidia-nvshmem-cu12       3.4.5                    pypi_0    pypi
[conda] nvidia-nvtx-cu12          12.8.90                  pypi_0    pypi
[conda] optree                    0.15.0                   pypi_0    pypi
[conda] pyzmq                     27.1.0                   pypi_0    pypi
[conda] torch                     2.10.0+cu128             pypi_0    pypi
[conda] torch-c-dlpack-ext        0.1.5                    pypi_0    pypi
[conda] torchaudio                2.10.0+cu128             pypi_0    pypi
[conda] torchvision               0.25.0+cu128             pypi_0    pypi
[conda] transformers              4.56.1                   pypi_0    pypi
[conda] triton                    3.6.0                    pypi_0    pypi

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.17.1
vLLM Build Flags:
  CUDA Archs: 8.9; ROCm: Disabled; XPU: Disabled
GPU Topology:
        GPU0    GPU1    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      NODE    96-191,288-383  1               N/A
GPU1    NODE     X      96-191,288-383  1               N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

==============================
     Environment Variables
==============================
TORCH_CUDA_ARCH_LIST=8.9
CUDA_PATH=/usr/local/cuda
LD_LIBRARY_PATH=/usr/local/cuda/lib64:/home/neil/code/llm/llama.cpp/build-cuda/bin
CUDA_HOME=/usr/local/cuda
CUDA_HOME=/usr/local/cuda
CUDAToolkit_ROOT=/usr/local/cuda
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_neil

---

Failed to advance FSM
Unexpected: grammar rejected tokens
The matcher has terminated ... but is trying to accept new token

---

# vllm/v1/structured_output/backend_xgrammar.py
   def accept_tokens(self, request_id: str, tokens: list[int]) -> bool:
       if self._is_terminated:
           return False
       for token in tokens:
           if not self.matcher.accept_token(token):
               logger.error(
                   "Failed to advance FSM for request %s "
                   "for tokens %s. Please file an issue.",
                   request_id,
                   token,
               )
               return False
           self.num_processed_tokens += 1
       self._is_terminated = self.matcher.is_terminated()
       return True

---

# vllm/v1/structured_output/__init__.py
   def should_advance(self, request: "Request") -> bool:
       if not request.use_structured_output:
           return False
       if self.reasoner is None:
           return True
       ...

---

# vllm/v1/core/sched/scheduler.py
   if new_token_ids and self.structured_output_manager.should_advance(request):
       struct_output_request = request.structured_output_request
       assert struct_output_request is not None
       assert struct_output_request.grammar is not None
       ok = struct_output_request.grammar.accept_tokens(req_id, new_token_ids)
       if not ok:
           logger.warning(
               "Unexpected: grammar rejected tokens %s for request %s.",
               new_token_ids,
               req_id,
           )

---

export POC_PY=/path/to/python3
export G5_ASYNC=/path/to/repro_g5_structured_fsm_asyncllm.py
export VLLM_POC_G5_MODEL=/path/to/qwen3

CUDA_VISIBLE_DEVICES=0 "$POC_PY" "$G5_ASYNC" \
  --model "$VLLM_POC_G5_MODEL" \
  --run-name g5_structured_fsm_asyncllm

---

ERROR ... Failed to advance FSM for request r1_route-... for tokens 0. Please file an issue.
WARNING ... Unexpected: grammar rejected tokens [0] for request r1_route-...
[grammar_matcher.cc:493] Warning: The matcher has terminated after accepting the stop token, but is trying to accept new token ...

---

ERROR ... Failed to advance FSM for request probe_seq-... for tokens 0. Please file an issue.
WARNING ... Unexpected: grammar rejected tokens [0] for request probe_seq-...

---

{
  "child_returncode": 0,
  "target_hit": true,
  "target_hits": [
    "Failed to advance FSM",
    "Unexpected: grammar rejected tokens",
    "grammar rejected tokens",
    "matcher has terminated",
    "terminated after accepting the stop token"
  ]
}
RAW_BUFFERClick to expand / collapse

Your current environment

<details> <summary>The output of <code>python collect_env.py</code></summary> 
Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 13.1.0-8ubuntu1~22.04) 13.1.0
Clang version                : 16.0.6 (++20231112100510+7cbf1a259152-1~exp1~20231112100554.106)
CMake version                : version 3.22.1
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.10.0+cu128
Is debug build               : False
CUDA used to build PyTorch   : 12.8
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.9 | packaged by Anaconda, Inc. | (main, Feb  6 2025, 18:56:27) [GCC 11.2.0] (64-bit runtime)
Python platform              : Linux-6.5.0-35-generic-x86_64-with-glibc2.35
    
==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.8.61
CUDA_MODULE_LOADING set to   : 
GPU models and configuration : 
GPU 0: NVIDIA GeForce RTX 4090
GPU 1: NVIDIA GeForce RTX 4090

Nvidia driver version        : 570.86.10
cuDNN version                : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.3.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.3.0
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                       x86_64
CPU op-mode(s):                     32-bit, 64-bit
Address sizes:                      52 bits physical, 57 bits virtual
Byte Order:                         Little Endian
CPU(s):                             384
On-line CPU(s) list:                0-383
Vendor ID:                          AuthenticAMD
Model name:                         AMD EPYC 9654 96-Core Processor
CPU family:                         25
Model:                              17
Thread(s) per core:                 2
Core(s) per socket:                 96
Socket(s):                          2
Stepping:                           1
Frequency boost:                    enabled
CPU max MHz:                        3707.8120
CPU min MHz:                        1500.0000
BogoMIPS:                           4792.57
Flags:                              fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good amd_lbr_v2 nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 invpcid_single hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid overflow_recov succor smca fsrm flush_l1d
Virtualization:                     AMD-V
L1d cache:                          6 MiB (192 instances)
L1i cache:                          6 MiB (192 instances)
L2 cache:                           192 MiB (192 instances)
L3 cache:                           768 MiB (24 instances)
NUMA node(s):                       2
NUMA node0 CPU(s):                  0-95,192-287
NUMA node1 CPU(s):                  96-191,288-383
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit:        Not affected
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      Not affected
Vulnerability Retbleed:             Not affected
Vulnerability Spec rstack overflow: Mitigation; Safe RET
Vulnerability Spec store bypass:    Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:           Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:           Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.4
[pip3] numpy==2.0.2
[pip3] nvidia-cublas-cu12==12.8.4.1
[pip3] nvidia-cuda-cupti-cu12==12.8.90
[pip3] nvidia-cuda-nvrtc-cu12==12.8.93
[pip3] nvidia-cuda-runtime-cu12==12.8.90
[pip3] nvidia-cudnn-cu12==9.10.2.21
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft-cu12==11.3.3.83
[pip3] nvidia-cufile-cu12==1.13.1.3
[pip3] nvidia-curand-cu12==10.3.9.90
[pip3] nvidia-cusolver-cu12==11.7.3.90
[pip3] nvidia-cusparse-cu12==12.5.8.93
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.4.2
[pip3] nvidia-cutlass-dsl-libs-base==4.4.2
[pip3] nvidia-ml-py==13.590.48
[pip3] nvidia-nccl-cu12==2.27.5
[pip3] nvidia-nvjitlink-cu12==12.8.93
[pip3] nvidia-nvshmem-cu12==3.4.5
[pip3] nvidia-nvtx-cu12==12.8.90
[pip3] optree==0.15.0
[pip3] pyzmq==27.1.0
[pip3] torch==2.10.0+cu128
[pip3] torch_c_dlpack_ext==0.1.5
[pip3] torchaudio==2.10.0+cu128
[pip3] torchvision==0.25.0+cu128
[pip3] transformers==4.56.1
[pip3] triton==3.6.0
[conda] flashinfer-python         0.6.4                    pypi_0    pypi
[conda] numpy                     2.0.2                    pypi_0    pypi
[conda] nvidia-cublas-cu12        12.8.4.1                 pypi_0    pypi
[conda] nvidia-cuda-cupti-cu12    12.8.90                  pypi_0    pypi
[conda] nvidia-cuda-nvrtc-cu12    12.8.93                  pypi_0    pypi
[conda] nvidia-cuda-runtime-cu12  12.8.90                  pypi_0    pypi
[conda] nvidia-cudnn-cu12         9.10.2.21                pypi_0    pypi
[conda] nvidia-cudnn-frontend     1.18.0                   pypi_0    pypi
[conda] nvidia-cufft-cu12         11.3.3.83                pypi_0    pypi
[conda] nvidia-cufile-cu12        1.13.1.3                 pypi_0    pypi
[conda] nvidia-curand-cu12        10.3.9.90                pypi_0    pypi
[conda] nvidia-cusolver-cu12      11.7.3.90                pypi_0    pypi
[conda] nvidia-cusparse-cu12      12.5.8.93                pypi_0    pypi
[conda] nvidia-cusparselt-cu12    0.7.1                    pypi_0    pypi
[conda] nvidia-cutlass-dsl        4.4.2                    pypi_0    pypi
[conda] nvidia-cutlass-dsl-libs-base 4.4.2                    pypi_0    pypi
[conda] nvidia-ml-py              13.590.48                pypi_0    pypi
[conda] nvidia-nccl-cu12          2.27.5                   pypi_0    pypi
[conda] nvidia-nvjitlink-cu12     12.8.93                  pypi_0    pypi
[conda] nvidia-nvshmem-cu12       3.4.5                    pypi_0    pypi
[conda] nvidia-nvtx-cu12          12.8.90                  pypi_0    pypi
[conda] optree                    0.15.0                   pypi_0    pypi
[conda] pyzmq                     27.1.0                   pypi_0    pypi
[conda] torch                     2.10.0+cu128             pypi_0    pypi
[conda] torch-c-dlpack-ext        0.1.5                    pypi_0    pypi
[conda] torchaudio                2.10.0+cu128             pypi_0    pypi
[conda] torchvision               0.25.0+cu128             pypi_0    pypi
[conda] transformers              4.56.1                   pypi_0    pypi
[conda] triton                    3.6.0                    pypi_0    pypi

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.17.1
vLLM Build Flags:
  CUDA Archs: 8.9; ROCm: Disabled; XPU: Disabled
GPU Topology:
        GPU0    GPU1    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      NODE    96-191,288-383  1               N/A
GPU1    NODE     X      96-191,288-383  1               N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

==============================
     Environment Variables
==============================
TORCH_CUDA_ARCH_LIST=8.9
CUDA_PATH=/usr/local/cuda
LD_LIBRARY_PATH=/usr/local/cuda/lib64:/home/neil/code/llm/llama.cpp/build-cuda/bin
CUDA_HOME=/usr/local/cuda
CUDA_HOME=/usr/local/cuda
CUDAToolkit_ROOT=/usr/local/cuda
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_neil
</details>

🐛 Describe the bug

Describe the bug

Version: vLLM 0.17.1
Model: Qwen/Qwen3-0.6B-GPTQ-Int8
Hardware reproduced on: NVIDIA GeForce RTX 4090, single GPU

Summary

I found a structured-output lifecycle bug in the xgrammar path.

In a frontend-legal structured-output workload, a structured-output request can already reach a terminated xgrammar matcher state, but later scheduler iterations still try to advance that same matcher with more generated tokens. On the re-verified Qwen3 AsyncLLM runs from May 14, 2026, this reproduces 2/2 times and leaves a stable internal error surface in the engine logs:

Failed to advance FSM
Unexpected: grammar rejected tokens
The matcher has terminated ... but is trying to accept new token

This is not a malformed-input bug. The request shape is legal. The problem is that the terminated grammar state is not consistently honored across the structured-output scheduling path.

Trigger chain

  1. Start a legal structured-output request with xgrammar enabled.
  2. Keep a live mixed batch running so speculative decoding and multi-step scheduling remain active around that structured-output request.
  3. The structured-output request reaches a terminated matcher state.
  4. Later scheduler iterations still call the grammar-advance path for that same request.
  5. xgrammar rejects the new token, and vLLM logs repeated internal failures.

Details

Trigger path in code

  1. In the xgrammar backend, accept_tokens() returns False once the matcher is already terminated or if accept_token() fails for a new token. 
    # vllm/v1/structured_output/backend_xgrammar.py
def accept_tokens(self, request_id: str, tokens: list[int]) -> bool:
    if self._is_terminated:
        return False
    for token in tokens:
        if not self.matcher.accept_token(token):
            logger.error(
                "Failed to advance FSM for request %s "
                "for tokens %s. Please file an issue.",
                request_id,
                token,
            )
            return False
        self.num_processed_tokens += 1
    self._is_terminated = self.matcher.is_terminated()
    return True
  2. On the scheduler side, any request that uses structured output is allowed to advance when StructuredOutputManager.should_advance() returns True. In the default non-reasoning path, it returns True unconditionally. 
    # vllm/v1/structured_output/__init__.py
def should_advance(self, request: "Request") -> bool:
    if not request.use_structured_output:
        return False
    if self.reasoner is None:
        return True
    ...
  3. During output processing, the scheduler still calls grammar.accept_tokens(...) whenever there are new tokens and should_advance(request) is true. 
    # vllm/v1/core/sched/scheduler.py
if new_token_ids and self.structured_output_manager.should_advance(request):
    struct_output_request = request.structured_output_request
    assert struct_output_request is not None
    assert struct_output_request.grammar is not None
    ok = struct_output_request.grammar.accept_tokens(req_id, new_token_ids)
    if not ok:
        logger.warning(
            "Unexpected: grammar rejected tokens %s for request %s.",
            new_token_ids,
            req_id,
        )
  4. In this issue, the request has already reached a terminated grammar state, but later iterations still reach the accept_tokens(...) call and produce the repeated FSM-advance failure.

AsyncLLM script breakdown

repro_g5_structured_fsm_asyncllm.py is a standalone AsyncLLM reproducer.

  • It reads the Qwen3 checkpoint path from VLLM_POC_G5_MODEL or the built-in /path/to/qwen3 placeholder.
  • It uses AsyncLLM directly.
  • It launches one live mixed batch containing:
    • warm prefix-cache requests
    • one structured-output route request
    • one non-structured speculative pressure request
    • one later structured-output probe request
    • one short tail request
  • It enables:
    • xgrammar structured output
    • speculative decoding
    • prefix caching
  • The parent process captures child stdout/stderr and checks for the target FSM-failure strings in the engine logs.
  • It writes:
    • repro_config.json
    • request_payloads.json
    • engine_args.json
    • child.stdout.log
    • child.stderr.log
    • post_run_summary.json
    • post_run_summary_parent.json
    • per-request output JSON files
    • error.txt on child failure

AsyncLLM repro

export POC_PY=/path/to/python3
export G5_ASYNC=/path/to/repro_g5_structured_fsm_asyncllm.py
export VLLM_POC_G5_MODEL=/path/to/qwen3

CUDA_VISIBLE_DEVICES=0 "$POC_PY" "$G5_ASYNC" \
  --model "$VLLM_POC_G5_MODEL" \
  --run-name g5_structured_fsm_asyncllm

Reproduce Environment

ItemValue
OSUbuntu 22.04.5 LTS
KernelLinux 6.5.0-35-generic
GPU2 x NVIDIA GeForce RTX 4090
GPU memory24564 MiB each
Driver570.86.10
CUDA runtime12.8 (nvidia-smi)
CUDA toolkit12.8.61 (nvcc)
Python3.12.9
vLLM0.17.1
PyTorch2.10.0+cu128
transformers4.56.1
tokenizers0.22.0
flash_attn2.8.3
triton3.6.0
numpy2.0.2

Observed result

The re-verified standalone AsyncLLM Qwen3 script exits successfully and records the target failure surface in the child logs. The issue can surface on either structured-output request in the live batch, for example r1_route-* or probe_seq-*:

ERROR ... Failed to advance FSM for request r1_route-... for tokens 0. Please file an issue.
WARNING ... Unexpected: grammar rejected tokens [0] for request r1_route-...
[grammar_matcher.cc:493] Warning: The matcher has terminated after accepting the stop token, but is trying to accept new token ...

In another verified rerun, the same surface appeared on probe_seq-*:

ERROR ... Failed to advance FSM for request probe_seq-... for tokens 0. Please file an issue.
WARNING ... Unexpected: grammar rejected tokens [0] for request probe_seq-...

post_run_summary_parent.json from the verified runs contains:

{
  "child_returncode": 0,
  "target_hit": true,
  "target_hits": [
    "Failed to advance FSM",
    "Unexpected: grammar rejected tokens",
    "grammar rejected tokens",
    "matcher has terminated",
    "terminated after accepting the stop token"
  ]
}

post_run_summary.json from the child remains non-failing because it only summarizes per-request outputs. The actual bug signal lives in the captured engine logs, so post_run_summary_parent.json is the correct top-level proof artifact for this repro.

Root cause

This is a structured-output state-management bug, not a bad-request rejection.

The important distinction is:

  • the request is frontend-legal
  • the matcher already reached a terminated state
  • later scheduler iterations still try to advance that same matcher

So the real problem is that the structured-output scheduling path does not consistently stop grammar advancement once the matcher has already terminated.

Attachments

The attachment bundle for this report should contain:

  • repro_g5_structured_fsm_asyncllm.py

Before submitting a new issue...

  • Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#environment variable#logging issue#authentication issue#prompt issue

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

vllm - 💡(How to fix) Fix [Bug]: Structured-output scheduler can keep advancing a terminated xgrammar matcher