Hi Eike, Excellent, this is the backtrace which we needed to see. This is same stall as the one I see building libpreludedb. I'm going to send a new patch with the real fix soon. The previous patches just messed with the timing of things. On 2020-07-19 3:03 a.m., Rolf Eike Beer wrote: > [835212.602791] IAOQ[0]: d_alloc_parallel+0x424/0x668 > [835212.602791] IAOQ[1]: d_alloc_parallel+0x428/0x668 > [835212.602791] RP(r2): d_alloc_parallel+0x16c/0x668 > [835212.602791] Backtrace: > [835212.602791] [<000000004034a5f0>] __lookup_slow+0xb0/0x280 > [835212.602791] [<000000004034c6a0>] walk_component+0x1d8/0x300 > [835212.602791] [<000000004034cd88>] path_lookupat+0xd0/0x228 > [835212.602791] [<000000004034fa64>] filename_lookup+0xac/0x1a8 > [835212.602791] [<000000004034fccc>] user_path_at_empty+0x74/0x90 > [835212.602791] [<000000004033c774>] vfs_statx+0xd4/0x1a0 > [835212.602791] [<000000004033cdb8>] __do_sys_stat64+0x48/0xa0 > [835212.602791] [<000000004033d5f0>] sys_stat64+0x20/0x30 > [835212.602791] [<0000000040191018>] syscall_exit+0x0/0x14 Regards, Dave -- John David Anglin dave.anglin@xxxxxxxx
