On Mon, Nov 20, 2017 at 04:22:25PM +0100, Martin Pieuchot wrote:
> Diff below remove the KERNEL_LOCK() around all pr_input() routines.
> It's a bit rough so I'd appreciate more tests before splitting it into
> pieces.
regress/lib/libpthread runs into a deadlock:
===> socket/1
cc -O2 -pipe -I/usr/src/regress/lib/libpthread/socket/1/../../include -Wall
-DSRCDIR='"/usr/src/regress/lib/libpthread/socket/1"'
-I/usr/src/regress/lib/libpthread/socket/1/../include -MD -MP -c
/usr/src/regress/lib/libpthread/socket/1/socket1.c
cc -o socket1 socket1.o -lpthread
./socket1
A lot of processes are waiting for the net lock:
ddb{0}> ps
PID TID PPID UID S FLAGS WAIT COMMAND
92841 53206 1 0 3 0x82 thrsleep socket1
92841 34582 1 0 3 0x4000082 netcon socket1
92841 427059 1 0 3 0x4000002 netlock socket1
17869 237863 64620 0 3 0x100082 piperd gzip
64620 287694 20261 0 3 0x100082 piperd pax
20261 394098 49123 0 3 0x82 pipewr perl
49123 39681 68118 0 3 0x10008a pause ksh
68118 101874 6279 0 3 0x52 netlock sshd
71426 167931 1 0 3 0x83 netio login
47052 97105 1 0 3 0x100083 ttyin getty
4929 501560 1 0 3 0x100083 ttyin getty
17447 409940 1 0 3 0x100083 ttyin getty
3248 432204 1 0 3 0x100083 ttyin getty
92079 257598 1 0 3 0x100083 ttyin getty
29168 457419 1 0 3 0x100098 poll cron
10039 371687 1 99 3 0x100090 poll sndiod
84840 178202 1 110 3 0x100090 poll sndiod
29145 32675 41152 95 3 0x100092 kqread smtpd
64178 437103 41152 103 3 0x100092 kqread smtpd
70671 253152 41152 95 3 0x100092 kqread smtpd
58596 268547 41152 95 3 0x100092 kqread smtpd
69700 24499 41152 95 3 0x100092 kqread smtpd
92805 258698 41152 95 3 0x100092 kqread smtpd
41152 475143 1 0 3 0x100080 kqread smtpd
6279 90305 1 0 3 0x80 select sshd
92796 374585 0 0 3 0x14200 acct acct
59546 95746 0 0 3 0x14280 nfsidl nfsio
58735 375345 0 0 3 0x14280 nfsidl nfsio
6690 353929 0 0 3 0x14280 nfsidl nfsio
12672 42719 0 0 3 0x14280 nfsidl nfsio
70271 450856 1 0 3 0x100080 poll ntpd
72004 124611 70995 83 3 0x100092 poll ntpd
70995 25369 1 83 3 0x100012 netlock ntpd
75907 420740 97225 74 3 0x100092 bpf pflogd
97225 518514 1 0 3 0x80 netio pflogd
46707 301105 82852 73 3 0x100090 kqread syslogd
82852 52006 1 0 3 0x100082 netio syslogd
633 227768 1 77 3 0x100090 poll dhclient
44449 13774 1 0 3 0x80 poll dhclient
32381 23590 25453 115 3 0x100092 kqread slaacd
64762 226200 25453 115 3 0x100092 kqread slaacd
25453 18123 1 0 3 0x80 kqread slaacd
92740 239333 0 0 3 0x14200 bored ttm_swap
72784 47424 0 0 3 0x14200 pgzero zerothread
596 230658 0 0 3 0x14200 aiodoned aiodoned
34515 469546 0 0 3 0x14200 syncer update
26581 499471 0 0 3 0x14200 cleaner cleaner
81584 406555 0 0 3 0x14200 reaper reaper
66383 338329 0 0 3 0x14200 pgdaemon pagedaemon
77049 251648 0 0 3 0x14200 bored crynlk
65487 523752 0 0 3 0x14200 bored crypto
82067 365041 0 0 3 0x14200 usbtsk usbtask
67863 195584 0 0 3 0x14200 usbatsk usbatsk
4602 167235 0 0 3 0x14200 bored sensors
1303 467637 0 0 3 0x40014200 acpi0 acpi0
61494 169229 0 0 7 0x40014200 idle1
97874 376418 0 0 3 0x14200 tqbar softnet
37711 87043 0 0 3 0x14200 bored systqmp
37542 272189 0 0 3 0x14200 netlock systq
12285 95321 0 0 3 0x40014200 netlock softclock
*71200 301193 0 0 7 0x40014200 idle0
840 20751 0 0 3 0x14200 bored sbar
16875 348937 0 0 3 0x14200 kmalloc kmthread
1 172792 0 0 3 0x82 wait init
0 0 -1 0 3 0x10200 scheduler swapper
softnet is holding the net lock:
ddb{0}> x/x netlock,2
netlock: d58009bf
netlock+0x4: d0b793b4
376418 softnet 0xd58009bc 0xf52fe000 0xd0d18534
softnet is waiting for the task barrier:
ddb{0}> trace /p 0t376418
sleep_finish(f5300874,f5300878) at sleep_finish+0x62
sleep_finish(f5300878,1) at sleep_finish+0x62
taskq_barrier(d0bfe1b8) at taskq_barrier+0x9b
sofree(d54b5b98) at sofree+0x72
in_pcbdetach(d579ac40) at in_pcbdetach+0x47
tcp_close(d5619c60) at tcp_close+0x102
tcp_input(f5300a3c,f5300a38,6,2) at tcp_input+0x194b
ip_deliver(f5300a3c,f5300a38,6,2) at ip_deliver+0x21f
ip_local(f5300a3c,f5300a38,d094e188,0) at ip_local+0x139
ipintr() at ipintr+0x54
if_netisr(0) at if_netisr+0
db_enter() at db_enter+0x4
bluhm
