FYI. Adding if (len <= 0) return; in the scsi_disk_new_request_dump solved the compilation issue for me.
So indeed gcc thinks len == -1
I am pretty sure the build qemu is functional, as this path is only taken if the trace_event_get_state_backends(TRACE_SCSI_DISK_NEW_REQUEST) is true, which by default it is not.
BTW. Also, aarch64-softmmu/qemu-system-aarch64 takes very long time to link compared to other targets, so I recommend using -flto=16 to increase parallelism, and reduce lto link time to about 4 minutes. (But 64GB of memory recommended).
I also tested with --disable-slirp configure flag. Still same issue.
FYI. Adding if (len <= 0) return; in the scsi_disk_ new_request_ dump solved the compilation issue for me.
So indeed gcc thinks len == -1
I am pretty sure the build qemu is functional, as this path is only taken if the trace_event_ get_state_ backends( TRACE_SCSI_ DISK_NEW_ REQUEST) is true, which by default it is not.
BTW. Also, aarch64- softmmu/ qemu-system- aarch64 takes very long time to link compared to other targets, so I recommend using -flto=16 to increase parallelism, and reduce lto link time to about 4 minutes. (But 64GB of memory recommended).
I also tested with --disable-slirp configure flag. Still same issue.