When I try writing a large amount of data to a GPT partition on my NVM SSD using dd, after a second or so I get "kernel BUG" messages in dmesg, and the writing comes to a halt. The machine remains responsive and I can write to other disks, but if I run "sync", it hangs forever.
The bug is 100 % repeatable. To replicate, run:
dd if=/dev/zero of=/dev/nvme0n1p1
This happens only when I write more than 0.5-1.5 GiB to a GPT partition on the NVM device. The bug does not happen in the following cases:
- writing just 500 MB on a GPT partition
- writing to the raw device (/dev/nvme0n1)
- writing to a DOS partition (i.e. when the NVM device is using a DOS style partition table)
The device in question is a 1 TB Samsung 960 EVO M.2 SSD.
Ubuntu release: 17.04
I have dmesg logs available of four instances of this happening, with very similar stack traces. I will be happy to experiment and provide any additional information you may need.
When I try writing a large amount of data to a GPT partition on my NVM SSD using dd, after a second or so I get "kernel BUG" messages in dmesg, and the writing comes to a halt. The machine remains responsive and I can write to other disks, but if I run "sync", it hangs forever.
The bug is 100 % repeatable. To replicate, run:
dd if=/dev/zero of=/dev/nvme0n1p1
This happens only when I write more than 0.5-1.5 GiB to a GPT partition on the NVM device. The bug does not happen in the following cases:
- writing just 500 MB on a GPT partition
- writing to the raw device (/dev/nvme0n1)
- writing to a DOS partition (i.e. when the NVM device is using a DOS style partition table)
The device in question is a 1 TB Samsung 960 EVO M.2 SSD.
Ubuntu release: 17.04
I have dmesg logs available of four instances of this happening, with very similar stack traces. I will be happy to experiment and provide any additional information you may need.