On Mon, 2017-01-30 at 18:25 -0800, Alexei Starovoitov wrote:
> under memory pressure 'ethtool -S' command may warn:
> [ 2374.385195] ethtool: page allocation failure: order:4, mode:0x242c0c0
> [ 2374.405573] CPU: 12 PID: 40211 Comm: ethtool Not tainted
> [ 2374.423071] Call Trace:
> [ 2374.423076] [<ffffffff8148cb29>] dump_stack+0x4d/0x64
> [ 2374.423080] [<ffffffff811667cb>] warn_alloc_failed+0xeb/0x150
> [ 2374.423082] [<ffffffff81169cd3>] ? __alloc_pages_direct_compact+0x43/0xf0
> [ 2374.423084] [<ffffffff8116a25c>] __alloc_pages_nodemask+0x4dc/0xbf0
> [ 2374.423091] [<ffffffffa0023dc2>] ? cmd_exec+0x722/0xcd0 [mlx5_core]
> [ 2374.423095] [<ffffffff811b3dcc>] alloc_pages_current+0x8c/0x110
> [ 2374.423097] [<ffffffff81168859>] alloc_kmem_pages+0x19/0x90
> [ 2374.423099] [<ffffffff81186e5e>] kmalloc_order_trace+0x2e/0xe0
> [ 2374.423101] [<ffffffff811c0084>] __kmalloc+0x204/0x220
> [ 2374.423105] [<ffffffff816c269e>] dev_ethtool+0xe4e/0x1f80
> [ 2374.423106] [<ffffffff816b967e>] ? dev_get_by_name_rcu+0x5e/0x80
> [ 2374.423108] [<ffffffff816d6926>] dev_ioctl+0x156/0x560
> [ 2374.423111] [<ffffffff811d4c68>] ? mem_cgroup_commit_charge+0x78/0x3c0
> [ 2374.423117] [<ffffffff8169d542>] sock_do_ioctl+0x42/0x50
> [ 2374.423119] [<ffffffff8169d9c3>] sock_ioctl+0x1b3/0x250
> [ 2374.423121] [<ffffffff811f0f42>] do_vfs_ioctl+0x92/0x580
> [ 2374.423123] [<ffffffff8100222b>] ? do_audit_syscall_entry+0x4b/0x70
> [ 2374.423124] [<ffffffff8100287c>] ? syscall_trace_enter_phase1+0xfc/0x120
> [ 2374.423126] [<ffffffff811f14a9>] SyS_ioctl+0x79/0x90
> [ 2374.423127] [<ffffffff81002bb0>] do_syscall_64+0x50/0xa0
> [ 2374.423129] [<ffffffff817e19bc>] entry_SYSCALL64_slow_path+0x25/0x25
>
> ~1160 mlx5 counters ~= order 4 allocation which is unlikely to succeed
> under memory pressure. Convert them to vzalloc() as ethtool_get_regs() does.
> Also take care of drivers without counters similar to
> commit 67ae7cf1eeda ("ethtool: Allow zero-length register dumps again")
> and reduce warn_on to warn_on_once.
I think this is generally not a good idea as
most uses already fit fine in a kcalloc.
Maybe use Michal Hocko's kvmalloc changes.
https://lkml.org/lkml/2017/1/30/120
