On Wed, Jun 22, 2022 at 8:43 AM Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx> wrote: > > On 2022-06-21 17:27:22 [+0200], Jinpu Wang wrote: > > Hi, there > Hi, > > > > > on test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16G > > > > > > > … > > > > [ 17.451787][ T1] rtrs_server L2256: Loading module rtrs_server, proto 2.0: (max_chunk_size: 131072 (pure IO 126976, headers 4096) , sess_queue_depth: 512, always_invalidate: 1) > > > > [ 17.470894][ T1] swapper: page allocation failure: order:5, mode:0xcc0(GFP_KERNEL), nodemask=(null) > > > > > > If I read this right, it allocates "512 * 10" chunks of order 5 / 128KiB > > > of memory (contiguous memory). And this appears to fail. > > > This is either a lot of memory or something that shouldn't be used on > > > i386. > > It allocates 512 * 128 KiB of memory, which is probably to big for > > this VM setup. > > why 512 * 128KiB? It is: > | chunk_pool = mempool_create_page_pool(sess_queue_depth * CHUNK_POOL_SZ, > | get_order(max_chunk_size)); > with > | static int __read_mostly max_chunk_size = DEFAULT_MAX_CHUNK_SIZE; > | static int __read_mostly sess_queue_depth = DEFAULT_SESS_QUEUE_DEPTH; > | #define CHUNK_POOL_SZ 10 > > so isn't it (512 * 10) * 128KiB? eh, you're right, I forgot we have mempool. We discussed internally in the past to remove that, we should do it. Sorry > > > Thanks! > > Sebastian
