bug#31268: 'guix system vm-image' hangs

[Ludovic Courtès]

Hello,

George myglc2 Clemmer <address@hidden> skribis:

> Thank you Eric! Actually I meant to send it to address@hidden
> before. Here is something that fails here.
>
> guix system vm-image --image-size=10G sysi19.scm

As of 37b9be5878d1694967a41d313de97f7a957df120 I see that ‘guix system
vm’ or “make check-system TESTS=basic” hang.  Specifically, the VM that
creates the image hangs like this:

--8<---------------cut here---------------start------------->8---
Welcome, this is GNU's early boot Guile.
Use '--repl' for an initrd REPL.

loading kernel modules...
[    0.618457] usbcore: registered new interface driver usb-storage
[    0.620812] usbcore: registered new interface driver uas
[    0.626976] hidraw: raw HID events driver (C) Jiri Kosina
[    0.628093] usbcore: registered new interface driver usbhid
[    0.628804] usbhid: USB HID core driver
[    0.635125] isci: Intel(R) C600 SAS Controller Driver - version 1.2.0
[    0.654740] PCI Interrupt Link [LNKC] enabled at IRQ 11
[    0.673721] PCI Interrupt Link [LNKD] enabled at IRQ 10
[    0.693204] PCI Interrupt Link [LNKA] enabled at IRQ 10
[    0.714857] PCI Interrupt Link [LNKB] enabled at IRQ 11
[    0.759651] virtio_blk virtio5: [vda] 143360 512-byte logical blocks (73.4 
MB/70.0 MiB)
[    0.765620] FS-Cache: Loaded
[    0.767868] 9pnet: Installing 9P2000 support
[    0.768899] 9p: Installing v9fs 9p2000 file system support
[    0.769556] FS-Cache: Netfs '9p' registered for caching
configuring QEMU networking...
[    0.780494] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
loading '/gnu/store/d116816xkzfaa7y1d2srvbxy5ip5cm12-linux-vm-loader'...
[    0.802603] random: fast init done
environment variable `PATH' set to 
`/gnu/store/1n3cvnxrz1m6jqc09d9m1grbfbs91kmv-qemu-minimal-2.12.0/bin:/gnu/store/h8670f6j1rawhhaz8wvdy5gv79lny0am-parted-3.2/sbin:/gnu/store/nm305rpb2mvridkyj4l3636nc9ql4lf9-e2fsprogs-1.43.6/bin:/gnu/store/nm305rpb2mvridkyj4l3636nc9ql4lf9-e2fsprogs-1.43.6/sbin:/gnu/store/0sq2nflm42x0znkv44add0gk82khkcb6-dosfstools-4.1/sbin:/gnu/store/0hl513mnpkhszm2hjai2w9cxmpxs0vgq-sed-4.4/bin:/gnu/store/1h44pkgdd7n6s3i2vjh54awsvfmc219j-grep-3.1/bin:/gnu/store/kgzvfby2ggi1xawsh5vjh4s93qk2dp9k-coreutils-8.28/bin:/gnu/store/k7r2m2wgj8x8jjhccwjsiimp0dlzxb7i-findutils-4.6.0/bin:/gnu/store/ahxc89r6npzf2bbl8yg5vdjicskzzjf3-gawk-4.1.4/bin'
creating partition table with 2 partitions (20.0 MiB, 40.0 MiB)...
[    1.344130] tsc: Refined TSC clocksource calibration: 2808.014 MHz
[    1.345186] clocksource: tsc: mask: 0xffffffffffffffff max_cycles: 
0x2879d389cc4, max_idle_ns: 440795324601 ns
[    1.792320] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
--8<---------------cut here---------------end--------------->8---

Reverting ba39d3280759fb20a4ff0092a5afac721a5fa769 (“gnu: linux-libre:
Update to 4.16.4.”) solves the problem for me.

Normally what we see is something like this:

--8<---------------cut here---------------start------------->8---
loading '/gnu/store/4i02v1lvw9nwzcfs5l1il9grhkw17iqm-linux-vm-loader'...
environment variable `PATH' set to 
`/gnu/store/1n3cvnxrz1m6jqc09d9m1grbfbs91kmv-qemu-minimal-2.12.0/bin:/gnu/store/h8670f6j1rawhhaz8wvdy5gv79lny0am-parted-3.2/sbin:/gnu/store/nm305rpb2mvridkyj4l3636nc9ql4lf9-e2fsprogs-1.43.6/bin:/gnu/store/nm305rpb2mvridkyj4l3636nc9ql4lf9-e2fsprogs-1.43.6/sbin:/gnu/store/0sq2nflm42x0znkv44add0gk82khkcb6-dosfstools-4.1/sbin:/gnu/store/0hl513mnpkhszm2hjai2w9cxmpxs0vgq-sed-4.4/bin:/gnu/store/1h44pkgdd7n6s3i2vjh54awsvfmc219j-grep-3.1/bin:/gnu/store/kgzvfby2ggi1xawsh5vjh4s93qk2dp9k-coreutils-8.28/bin:/gnu/store/k7r2m2wgj8x8jjhccwjsiimp0dlzxb7i-findutils-4.6.0/bin:/gnu/store/ahxc89r6npzf2bbl8yg5vdjicskzzjf3-gawk-4.1.4/bin'
creating partition table with 2 partitions (20.0 MiB, 40.0 MiB)...
Warning: The resulting partition is not properly aligned for best performance.
creating ext4 partition...
mke2fs 1.43.6 (29-Aug-2017)
ext2fs_check_if_mount: Can't check if filesystem is mounted due to missing mtab 
file while determining whether /dev/vda1 is mounted.
Creating filesystem with 20480 1k blocks and 5136 inodes
Filesystem UUID: bb8da7fb-04b2-437e-887a-f468da978a40
Superblock backups stored on blocks: 
        8193

Allocating group tables: done                            
Writing inode tables: done                            
Creating journal (1024 blocks): done
Writing superblocks and filesystem accounting information: done

[    0.979292] EXT4-fs (vda1): mounted filesystem with ordered data mode. Opts: 
(null)
…
--8<---------------cut here---------------end--------------->8---

So it seems that the VM image hangs when trying to run Parted.

Mark, are you aware of anything particular that could explain this?
Perhaps a regression/incompatible change in 9P or virtio?

Ludo’.