OCI: Cgroup Memory Corruption on UEKr4 Kernels

Applies to:

Oracle Cloud Infrastructure

Symptoms

On a system with kernel UEKr4, the out-of-memory killer ( oom-killer ) killed a process due to its memory usage

[ 315.230766] sshd invoked oom-killer: gfp_mask=0xd0, order=0, oom_score_adj=0
 [ 315.230771] sshd cpuset=/ mems_allowed=0
 [ 315.230780] CPU: 0 PID: 12149 Comm: sshd Tainted: P W OE 4.1.12-124.74.2.el7uek.x86_64 #2
 [ 315.230783] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 0.0.0 02/06/2015
 [ 315.230786] 0000000000000286 00000000e615e435 ffff88030c4afc38 ffffffff81756bb4
 [ 315.230791] ffff8802fcd45400 ffff88045703f000 ffff88030c4afce8 ffffffff817527c6
 [ 315.230795] ffff88030c4afd08 ffff8802fcd45400 ffff8802fcd45400 ffff8802fcd45eb0
 [ 315.230798] Call Trace:
 [ 315.230809] [<ffffffff81756bb4>] dump_stack+0x63/0x81
 [ 315.230814] [<ffffffff817527c6>] dump_header+0x8e/0x221
 [ 315.230819] [<ffffffff811a3b16>] ? find_lock_task_mm+0x56/0xc0
 [ 315.230822] [<ffffffff811a412e>] oom_kill_process+0x1ee/0x3e0
 [ 315.230830] [<ffffffff81214524>] mem_cgroup_oom_synchronize+0x594/0x5f0
 [ 315.230835] [<ffffffff81210490>] ? kzalloc_node.constprop.37+0x20/0x20
 [ 315.230839] [<ffffffff811a4a94>] pagefault_out_of_memory+0x24/0xe0
 [ 315.230842] [<ffffffff81750f1e>] mm_fault_error+0x67/0x140
 [ 315.230848] [<ffffffff810747f6>]