[Bug]: IMA with the prefix_prefill::context_attention_fwd kernel when capturing full cudagraphs #25985
Description
Your current environment
The output of python collect_env.py
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 22.04.5 LTS (x86_64)
GCC version : (Ubuntu 10.5.0-1ubuntu1~22.04.2) 10.5.0
Clang version : Could not collect
CMake version : version 3.31.6
Libc version : glibc-2.35
==============================
PyTorch Info
==============================
PyTorch version : 2.8.0+rocm6.3
Is debug build : False
CUDA used to build PyTorch : N/A
ROCM used to build PyTorch : 6.3.42131-fa1d09cbd
==============================
Python Environment
==============================
Python version : 3.10.12 (main, Aug 15 2025, 14:32:43) [GCC 11.4.0] (64-bit runtime)
Python platform : Linux-5.15.0-131-generic-x86_64-with-glibc2.35
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : Could not collect
CUDA_MODULE_LOADING set to : LAZY
GPU models and configuration : AMD Instinct MI300X (gfx942:sramecc+:xnack-)
Nvidia driver version : Could not collect
cuDNN version : Could not collect
HIP runtime version : 6.3.42131
MIOpen runtime version : 3.3.0
Is XNNPACK available : True
==============================
CPU Info
==============================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 57 bits virtual
Byte Order: Little Endian
CPU(s): 104
On-line CPU(s) list: 0-103
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) Platinum 8470
CPU family: 6
Model: 143
Thread(s) per core: 1
Core(s) per socket: 52
Socket(s): 2
Stepping: 8
BogoMIPS: 4000.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 invpcid_single cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
L1d cache: 4.9 MiB (104 instances)
L1i cache: 3.3 MiB (104 instances)
L2 cache: 208 MiB (104 instances)
L3 cache: 210 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-51
NUMA node1 CPU(s): 52-103
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 Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS SW sequence; BHI BHI_DIS_S
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
==============================
Versions of relevant libraries
==============================
[pip3] conch-triton-kernels==1.2.1
[pip3] numpy==2.2.6
[pip3] pytorch-triton-rocm==3.4.0
[pip3] pyzmq==27.1.0
[pip3] torch==2.8.0+rocm6.3
[pip3] torchaudio==2.8.0+rocm6.3
[pip3] torchvision==0.23.0+rocm6.3
[pip3] transformers==4.56.2
[pip3] triton==3.3.0
[conda] Could not collect
==============================
vLLM Info
==============================
ROCM Version : 6.2.41134-65d174c3e
vLLM Version : 0.10.1.dev2056+g8616300ae (git sha: 8616300ae)
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled
GPU Topology:
============================ ROCm System Management Interface ============================
================================ Weight between two GPUs =================================
GPU0 GPU1 GPU2 GPU3 GPU4 GPU5 GPU6 GPU7
GPU0 0 15 15 15 15 15 15 15
GPU1 15 0 15 15 15 15 15 15
GPU2 15 15 0 15 15 15 15 15
GPU3 15 15 15 0 15 15 15 15
GPU4 15 15 15 15 0 15 15 15
GPU5 15 15 15 15 15 0 15 15
GPU6 15 15 15 15 15 15 0 15
GPU7 15 15 15 15 15 15 15 0
================================= Hops between two GPUs ==================================
GPU0 GPU1 GPU2 GPU3 GPU4 GPU5 GPU6 GPU7
GPU0 0 1 1 1 1 1 1 1
GPU1 1 0 1 1 1 1 1 1
GPU2 1 1 0 1 1 1 1 1
GPU3 1 1 1 0 1 1 1 1
GPU4 1 1 1 1 0 1 1 1
GPU5 1 1 1 1 1 0 1 1
GPU6 1 1 1 1 1 1 0 1
GPU7 1 1 1 1 1 1 1 0
=============================== Link Type between two GPUs ===============================
GPU0 GPU1 GPU2 GPU3 GPU4 GPU5 GPU6 GPU7
GPU0 0 XGMI XGMI XGMI XGMI XGMI XGMI XGMI
GPU1 XGMI 0 XGMI XGMI XGMI XGMI XGMI XGMI
GPU2 XGMI XGMI 0 XGMI XGMI XGMI XGMI XGMI
GPU3 XGMI XGMI XGMI 0 XGMI XGMI XGMI XGMI
GPU4 XGMI XGMI XGMI XGMI 0 XGMI XGMI XGMI
GPU5 XGMI XGMI XGMI XGMI XGMI 0 XGMI XGMI
GPU6 XGMI XGMI XGMI XGMI XGMI XGMI 0 XGMI
GPU7 XGMI XGMI XGMI XGMI XGMI XGMI XGMI 0
======================================= Numa Nodes =======================================
GPU[0] : (Topology) Numa Node: 0
GPU[0] : (Topology) Numa Affinity: 0
GPU[1] : (Topology) Numa Node: 0
GPU[1] : (Topology) Numa Affinity: 0
GPU[2] : (Topology) Numa Node: 0
GPU[2] : (Topology) Numa Affinity: 0
GPU[3] : (Topology) Numa Node: 0
GPU[3] : (Topology) Numa Affinity: 0
GPU[4] : (Topology) Numa Node: 1
GPU[4] : (Topology) Numa Affinity: 1
GPU[5] : (Topology) Numa Node: 1
GPU[5] : (Topology) Numa Affinity: 1
GPU[6] : (Topology) Numa Node: 1
GPU[6] : (Topology) Numa Affinity: 1
GPU[7] : (Topology) Numa Node: 1
GPU[7] : (Topology) Numa Affinity: 1
================================== End of ROCm SMI Log ===================================
==============================
Environment Variables
==============================
CUDA_HOME=/usr/local/cuda
CUDA_HOME=/usr/local/cuda
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY
🐛 Describe the bug
In order to repro the bug, you'll have to undo the fix in vllm/v1/attention/backends/rocm_attn.py. Simply cut out the following lines in build_for_cudagraph_capture
if envs.VLLM_V1_USE_PREFILL_DECODE_ATTENTION:
# Here we set the query start locs to 0. This is to
# cover up an invalid memory access in the prefix_prefil kernel
# that we run into during graph capture
common_attn_metadata.query_start_loc.zero_()
common_attn_metadata.query_start_loc_cpu.zero_()
Once those lines are cut out, you can repro the IMA by running the following command
VLLM_V1_USE_PREFILL_DECODE_ATTENTION=1 vllm serve meta-llama/Llama-3.1-8B-Instruct --max-num-seqs 256 --compilation-config='{"cudagraph_mode":"FULL"}'
This bug was introduced in #24845 which added proper setup of query_start_locs to the _dummy_run in the GpuModelRunner. Something about this causes the prefix_prefill kernel to trigger an IMA when writing to it's output tensor. It would be good to investigate this further and fix the IMA in the kernel. Then we can get rid of the bandaid fix inside of build_for_cudagraph_capture
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.