[Letux-kernel] [PATCH 00/20] A bunch of JZ4730 fixups for letux-kernel

[H. Nikolaus Schaller]

> Am 14.01.2021 um 22:58 schrieb H. Nikolaus Schaller <hns at goldelico.com>:
>> 
>> OK. It would be good to figure that out and get it tested by CI20 users.
> 
> Yes, I want to test soon.

CI20+HDMI still works with all the ingenic/drm patches (incl. conditionally swap DMA descriptors) incl. X11.


X11 on the Mipsbook seems to hang in a syscall (LEDs continue to blink) but console doesn't respond to ctrl-C):

root at letux:~# strace /usr/bin/X :0 -nolisten tcp vt7
execve("/usr/bin/X", ["/usr/bin/X", ":0", "-nolisten", "tcp", "vt7"], [/* 12 vars */]) = 0
brk(0)                                  = 0x413000
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x77e04000
access("/etc/ld.so.nohwcap", F_OK)      = -1 ENOENT (No such file or directory)
...
open("/proc/2382/cmdline", O_RDONLY)    = 13
read(13, "/usr/bin/X\0:0\0-nolisten\0tcp\0vt7\0", 4097) = 32
close(13)                               = 0
ioctl(7, FBIOBLANK, 0)                  = 0
clock_gettime(CLOCK_MONOTONIC, {186, 840000000}) = 0
clock_gettime(CLOCK_MONOTONIC, {186, 840000000}) = 0
rt_sigprocmask(SIG_BLOCK, [IO], [], 16) = 0
rt_sigprocmask(SIG_BLOCK, [ALRM CHLD WINCH TSTP TTIN TTOU VTALRM], [IO], 16) = 0
rt_sigprocmask(SIG_SETMASK, [IO], NULL, 16) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 16) = 0
clock_gettime(CLOCK_MONOTONIC, {186, 840000000}) = 0
_newselect(256, [1 3 5], NULL, NULL, {600, 0}

What I could imagine is that there are a lot of syscalls to clock_gettime() which get
confused by our non-monotonic sched_clock behaviour.

So X11 getting stuch is very likely just another symptom of the underlaying issue.

I think I'll focus on I2C now again after a break...

BR,
Nikolaus