Re: arm64: perf test 26 rpi4 oops

From: Mike Galbraith
Date: Tue Aug 01 2023 - 04:39:42 EST


On Tue, 2023-08-01 at 08:27 +0100, Will Deacon wrote:
> On Tue, Aug 01, 2023 at 06:34:15AM +0200, Mike Galbraith wrote:
> > On Tue, 2023-08-01 at 01:48 +0200, Mike Galbraith wrote:
> > >
> > > root@rpi4:~# dmesg|tail -2
> > > [  979.003019] __vm_enough_memory: pid: 12439, comm: objdump, not enough memory for the allocation
> > > [  979.003080] __vm_enough_memory: pid: 12439, comm: objdump, not enough memory for the allocation
> >
> > Ancient bug droppings are due to VMALLOC_END-VMALLOC_START=265885319168.
>
> FWIW, I see the same thing on my x86 laptop. I'm guessing objdump tries
> to read all of the vmalloc space into memory, which is going to end
> pretty badly!

Weird, I don't see that even booting my 16G desktop box with mem=4G,
with a crashdump kernel setup and the GUI up and running...

homer:/root # echo 0 > /proc/sys/vm/overcommit_memory
homer:/root # free -h
total used free shared buff/cache available
Mem: 3.2Gi 1.6Gi 667Mi 18Mi 1.1Gi 1.6Gi
Swap: 4.0Gi 0B 4.0Gi
homer:/root # swapoff -a
homer:/root # perf test 26
26: Object code reading : Ok
homer:/root # dmesg | tail
[ 9.379431] br0: port 1(eth0) entered blocking state
[ 9.379434] br0: port 1(eth0) entered forwarding state
[ 9.379527] IPv6: ADDRCONF(NETDEV_CHANGE): br0: link becomes ready
[ 9.388643] NET: Registered PF_PACKET protocol family
[ 17.225053] usblp0: removed
[ 17.226278] usblp 2-10:1.1: usblp0: USB Bidirectional printer dev 5 if 1 alt 0 proto 2 vid 0x04E8 pid 0x342E
[ 17.911334] r8169 0000:03:00.0: invalid VPD tag 0x00 (size 0) at offset 0; assume missing optional EEPROM
[ 18.430787] NFSD: Using UMH upcall client tracking operations.
[ 18.430802] NFSD: starting 90-second grace period (net f0000000)
[ 20.578718] logitech-hidpp-device 0003:046D:401B.0006: HID++ 2.0 device connected.
homer:/root # uname -r
6.4.7-stable
homer:/root #

Same for a limited objdump -d, though unrestricted will oom instantly.

-Mike