On 09/09/2016 02:36 PM, Christoph Hellwig wrote: > This series adds support to the RDMA core to implicitly allocate the required > CQEs when creating a QP. Hello Christoph, Have you already had the chance to test the ib_srpt changes included in this patch series? I'm asking because running srp-test on top of a kernel that includes this patch series triggers an OOM message I had not yet encountered while testing ib_srp or ib_srpt: ion-dev-ib-ini:~ # ~bart/software/infiniband/srp-test/run_tests -f xfs SRP LUN /sys/class/scsi_device/19:0:0:1 / sdt: removing /dev/dm-0: done SRP LUN /sys/class/scsi_device/19:0:0:2 / sdv: removing /dev/dm-1: done Unloaded the ib_srp kernel module Unloaded the ib_srpt kernel module Configured SRP target driver Running test /home/bart/software/infiniband/srp-test/tests/01 ... Connection to 10.60.180.187 closed by remote host. Connection to 10.60.180.187 closed. The ssh connection was closed by the OOM killer. >From the netconsole output: [ 286.878805] ------------[ cut here ]------------ [ 286.878901] WARNING: CPU: 7 PID: 53 at drivers/infiniband/core/cq.c:196 ib_free_cq+0xc0/0xf0 [ib_core] [ 286.886153] CPU: 7 PID: 53 Comm: kworker/7:0 Not tainted 4.8.0-rc6-dbg+ #1 [ 286.886206] Hardware name: Dell Inc. PowerEdge R430/03XKDV, BIOS 1.0.2 11/17/2014 [ 286.886259] Workqueue: srp_remove srp_remove_work [ib_srp] [ 286.887924] Call Trace: [ 286.887988] [<ffffffff81322945>] dump_stack+0x68/0x93 [ 286.888046] [<ffffffff81062e76>] __warn+0xc6/0xe0 [ 286.888105] [<ffffffff81062f48>] warn_slowpath_null+0x18/0x20 [ 286.888177] [<ffffffffa04e5e90>] ib_free_cq+0xc0/0xf0 [ib_core] [ 286.888237] [<ffffffffa027b787>] srp_free_ch_ib.isra.39+0xa7/0x1a0 [ib_srp] [ 286.888301] [<ffffffffa027bc04>] srp_remove_work+0xd4/0x1f0 [ib_srp] [ 286.888362] [<ffffffff81080425>] process_one_work+0x1f5/0x690 [ 286.888472] [<ffffffff81080909>] worker_thread+0x49/0x490 [ 286.888591] [<ffffffff810870ea>] kthread+0xea/0x100 [ 286.888652] [<ffffffff8162b5bf>] ret_from_fork+0x1f/0x40 [ 286.889383] ---[ end trace 57bd282839f289df ]--- [ ... ] [ 295.458538] ib_srpt Received SRP_LOGIN_REQ with i_port_id 0x0:0xe41d2d03000a85b1, t_port_id 0xe41d2d03000a85b0:0xe41d2d03000a85b0 and it_iu_len 4148 on port 1 (guid=0xfe80000000000000:0xe41d2 d03000a85b1) [ 295.658698] ib_srpt Received SRP_LOGIN_REQ with i_port_id 0x516d0a00032d1de4:0x2c90300a02dc1, t_port_id 0xe41d2d03000a85b0:0xe41d2d03000a85b0 and it_iu_len 260 on port 1 (guid=0xfe80000000000 000:0xe41d2d03000a6d51) [ 295.660820] ib_srpt Received SRP_LOGIN_REQ with i_port_id 0x516d0a00032d1de4:0x2c90300a02dc2, t_port_id 0xe41d2d03000a85b0:0xe41d2d03000a85b0 and it_iu_len 260 on port 1 (guid=0xfe80000000000 000:0xe41d2d03000a6d51) [ 305.335833] systemd-udevd invoked oom-killer: gfp_mask=0x27000c0(GFP_KERNEL_ACCOUNT|__GFP_NOTRACK), order=2, oom_score_adj=-1000 [ 305.335904] systemd-udevd cpuset=/ mems_allowed=0 [ 305.336037] CPU: 10 PID: 567 Comm: systemd-udevd Tainted: G W 4.8.0-rc6-dbg+ #1 [ 305.336088] Hardware name: Dell Inc. PowerEdge R430/03XKDV, BIOS 1.0.2 11/17/2014 [ 305.336789] Call Trace: [ 305.336847] [<ffffffff81322945>] dump_stack+0x68/0x93 [ 305.336907] [<ffffffff811d47e9>] dump_header+0x57/0x229 [ 305.337022] [<ffffffff811591bd>] oom_kill_process+0x20d/0x3e0 [ 305.337078] [<ffffffff811598f2>] out_of_memory+0x512/0x530 [ 305.337192] [<ffffffff8115eeab>] __alloc_pages_nodemask+0xe5b/0x1000 [ 305.337252] [<ffffffff81060312>] copy_process.part.45+0x102/0x1ac0 [ 305.337418] [<ffffffff81061eb1>] _do_fork+0xe1/0x6d0 [ 305.337588] [<ffffffff81062524>] SyS_clone+0x14/0x20 [ 305.337642] [<ffffffff81002c93>] do_syscall_64+0x53/0x120 [ 305.337697] [<ffffffff8162b45a>] entry_SYSCALL64_slow_path+0x25/0x25 Thanks, Bart. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
