Re: [PATCH v3 bpf] bpf: Try harder when allocating memory for large maps

2019-03-18 Thread Daniel Borkmann
On 03/18/2019 04:39 PM, Michal Hocko wrote: > On Mon 18-03-19 16:10:26, Martynas Pumputis wrote: >> It has been observed that sometimes a higher order memory allocation >> for BPF maps fails when there is no obvious memory pressure in a system. >> >> E.g. the map (BPF_MAP_TYPE_LRU_HASH, key=38, val

Re: [PATCH v3 bpf] bpf: Try harder when allocating memory for large maps

2019-03-18 Thread Michal Hocko
On Mon 18-03-19 16:10:26, Martynas Pumputis wrote: > It has been observed that sometimes a higher order memory allocation > for BPF maps fails when there is no obvious memory pressure in a system. > > E.g. the map (BPF_MAP_TYPE_LRU_HASH, key=38, value=56, max_elems=524288) > could not be created d

[PATCH v3 bpf] bpf: Try harder when allocating memory for large maps

2019-03-18 Thread Martynas Pumputis
It has been observed that sometimes a higher order memory allocation for BPF maps fails when there is no obvious memory pressure in a system. E.g. the map (BPF_MAP_TYPE_LRU_HASH, key=38, value=56, max_elems=524288) could not be created due to vmalloc unable to allocate 75497472B, when the system's