"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Fix helper writes to read-only maps\n\nLonial found an issue that despite user- and BPF-side frozen BPF map\n(like in case of .rodata), it was still possible to write into it from\na BPF program side through specific helpers having ARG_PTR_TO_{LONG,INT}\nas arguments.\n\nIn check_func_arg() when the argument is as mentioned, the meta->raw_mode\nis never set. Later, check_helper_mem_access(), under the case of\nPTR_TO_MAP_VALUE as register base type, it assumes BPF_READ for the\nsubsequent call to check_map_access_type() and given the BPF map is\nread-only it succeeds.\n\nThe helpers really need to be annotated as ARG_PTR_TO_{LONG,INT} | MEM_UNINIT\nwhen results are written into them as opposed to read out of them. The\nlatter indicates that it's okay to pass a pointer to uninitialized memory\nas the memory is written to anyway.\n\nHowever, ARG_PTR_TO_{LONG,INT} is a special case of ARG_PTR_TO_FIXED_SIZE_MEM\njust with additional alignment requirement. So it is better to just get\nrid of the ARG_PTR_TO_{LONG,INT} special cases altogether and reuse the\nfixed size memory types. For this, add MEM_ALIGNED to additionally ensure\nalignment given these helpers write directly into the args via *<ptr> = val.\nThe .arg*_size has been initialized reflecting the actual sizeof(*<ptr>).\n\nMEM_ALIGNED can only be used in combination with MEM_FIXED_SIZE annotated\nargument types, since in !MEM_FIXED_SIZE cases the verifier does not know\nthe buffer size a priori and therefore cannot blindly write *<ptr> = val."
