On Fri, Jun 10, 2022 at 01:18:56PM +0300, Alex Natalsson wrote: > Hello friends. > Suspend to RAM on Amlogic A311D chip (Khadas VIM3 SBC) was broken with > 5.17.13 and with 5.18.2 it also presents. > > When I trying do this I recieve error message: > VIM3 ~ # LANG=C echo mem > /sys/power/state > [ 952.824117] PM: suspend entry (deep) > [ 952.828333] Filesystems sync: 0.003 seconds > [ 952.829473] Freezing user space processes ... > [ 972.833509] Freezing of tasks failed after 20.003 seconds (1 tasks > refusing to freeze, wq_busy=0): > [ 972.841178] task:mplayer state:D stack: 0 pid: 779 > ppid: 736 flags:0x00000205 > [ 972.849457] Call trace: > [ 972.851868] __switch_to+0xf8/0x150 > [ 972.855315] __schedule+0x1f8/0x570 > [ 972.858758] schedule+0x48/0xc0 > [ 972.861856] schedule_preempt_disabled+0x10/0x20 > [ 972.866417] __mutex_lock.constprop.0+0x158/0x590 > [ 972.871071] __mutex_lock_slowpath+0x14/0x20 > [ 972.875296] mutex_lock+0x5c/0x70 > [ 972.878573] dpcm_fe_dai_open+0x44/0x194 > [ 972.882456] snd_pcm_open_substream+0xa4/0x174 > [ 972.886857] snd_pcm_open.part.0+0xd8/0x1dc > [ 972.890994] snd_pcm_playback_open+0x64/0x94 > [ 972.895223] snd_open+0xac/0x1d0 > [ 972.898411] chrdev_open+0xdc/0x2c4 > [ 972.901861] do_dentry_open+0x12c/0x380 > [ 972.905652] vfs_open+0x30/0x3c > [ 972.908758] do_open+0x1e4/0x3a0 > [ 972.911948] path_openat+0x10c/0x280 > [ 972.915486] do_filp_open+0x80/0x130 > [ 972.919019] do_sys_openat2+0xb4/0x170 > [ 972.922731] __arm64_sys_openat+0x64/0xb0 > [ 972.926700] invoke_syscall+0x48/0x114 > [ 972.930421] el0_svc_common.constprop.0+0xd4/0xfc > [ 972.935070] do_el0_svc+0x28/0x90 > [ 972.938347] el0_svc+0x34/0xb0 > [ 972.941367] el0t_64_sync_handler+0xa4/0x130 > [ 972.945595] el0t_64_sync+0x18c/0x190 > > [ 972.950674] OOM killer enabled. > [ 972.953781] Restarting tasks ... done. > -bash: echo: write error: Device or resource busy > VIM3 ~ :( # > > With 5.16.2 kernel version suspend to RAM is working, but system very > slow after resume. > Can you use 'git bisect' to track down the offending commit? And does 5.19-rc1 also have this issue? thanks, greg k-h
