Security: Fix path traversal and integer overflow vulnerabilities (44 issues fixed)#21

Open

devin-ai-integration[bot] wants to merge 41 commits into

devin/1761065602-security-fixes

Commits on Oct 21, 2025

fix: address high severity security issues and initial medium severity fixes

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path traversal validation in embed_wgsl.py

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation in Python scripts

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation in additional Python scripts

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation for remaining Python file operations (PT vulnerabilities)

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation for C++ file operations (PT vulnerabilities)

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation for gguf-split file operations (PT vulnerabilities)

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation for tokenize and quantize file operations (PT vulnerabilities)

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add path validation for remaining C++ file operations (PT vulnerabilities)

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add bounds checking for integer overflow in console.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: convert PosixPath to string before calling endswith in create_ops_docs.py

devin-ai-integration[bot]
and
jakexcosme
committed
fix: remove trailing whitespace in embed_kernel.py

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add bounds checking for integer overflow in linenoise.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add bounds checking for integer overflow in perplexity.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add bounds checking for integer overflow in convert-llama2c-to-ggml.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add overflow check for tensor memory allocation in gguf.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add null pointer checks for fopen calls in save-load-state.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add null pointer checks for fopen calls in vulkan-shaders-gen.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add null pointer check for fopen call in ggml.c

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add null pointer check for getenv call in llama-model-loader.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add URL validation to prevent SSRF attacks in run.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add URL validation to prevent SSRF attacks in tts-outetts.py

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add URL validation to prevent SSRF in json_schema_to_grammar.py

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add URL validation to prevent SSRF in bench.py and pydantic_models_to_grammar_examples.py

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for memory allocation in llama-batch.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for memory allocation in ggml-backend.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow check for memory allocation in ggml-alloc.c

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for memory allocation in ggml-vulkan.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for calloc in ggml_gallocr_new_n

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow check for calloc in ggml_build_backward_expand

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for calloc in ggml_backend_graph_copy

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for realloc in ggml_backend_sched_split_graph

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for malloc in ggml-quants.c

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow check for malloc in ggml_backend_multi_buffer_alloc_buffer

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow checks for malloc in llama-android.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add integer overflow check for realloc in linenoiseAddCompletion

devin-ai-integration[bot]
and
jakexcosme
committed
fix: replace strcpy with strncpy for safer string operations in quantize.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: replace strcpy with strncpy for safer string operations in ggml-opt.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add overflow checks for calloc operations in tokenize.cpp

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add overflow check for calloc operation in clip_log_internal_v

devin-ai-integration[bot]
and
jakexcosme
committed
fix: add overflow check for calloc operation in ggml_log_internal_v

devin-ai-integration[bot]
and
jakexcosme
committed