Hello Kevin, On Fri, 8 Mar 2024 at 17:38, Prasad Pandit <ppan...@redhat.com> wrote: > I'm trying to test it against the Fedora-26 kernel, which was < 4.13.0, and > did not support the AIO_FDSYNC call.
[PATCH v2] -> https://lists.nongnu.org/archive/html/qemu-devel/2024-03/msg02495.html * I've sent v2 of this patch which checks the return value from 'laio_co_submit' function and returns if it is >= zero(0). * I tested this and previous version of this patch on host kernels which support IO_CMD_FDSYNC and which don't. 1) When kernel supports IO_CMD_FDSYNC, everything works well. No issues. 2) When kernel does _not_ support IO_CMD_FDSYNC - With [PATCH v1], guest does not boot, instead it opens a rescue shell - With [PATCH v2], guest boots and seems to work fine. But after some time guest kernel threads seem to hang and show traces like === INFO: task kworker/u2:0:9 blocked for more than 245 seconds. INFO: task (tmpfiles):482 blocked for more than 123 seconds. INFO: task (tmpfiles):482 blocked for more than 368 seconds. INFO: task systemd-random-:477 blocked for more than 368 seconds. [ 492.932383] Not tainted 6.6.7-100.fc37.x86_64 #1 [ 492.935404] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. === * I'm not yet sure how to fix this. I'd appreciate if you've suggestions. Thank you. --- - Prasad
