Re: kern/22646 (Panic in ohci_add_done())

[Nick Hudson <skrll%netbsd.org@localhost>]

The following reply was made to PR kern/22646; it has been noted by GNATS.

From: Nick Hudson <skrll%netbsd.org@localhost>
To: gnats-bugs%NetBSD.org@localhost, gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost,
 Andreas Gustafsson <gson%gson.org@localhost>
Cc: 
Subject: Re: kern/22646 (Panic in ohci_add_done())
Date: Sat, 23 Apr 2016 19:35:41 +0100

 On 04/23/16 18:45, Andreas Gustafsson wrote:
 > The following reply was made to PR kern/22646; it has been noted by GNATS.
 >
 > From: Andreas Gustafsson <gson%gson.org@localhost>
 > To: skrll%NetBSD.org@localhost
 > Cc: gnats-bugs%NetBSD.org@localhost
 > Subject: Re: kern/22646 (Panic in ohci_add_done())
 > Date: Sat, 23 Apr 2016 20:41:20 +0300
 >
 >   skrll%NetBSD.org@localhost wrote:
 >   > Does this problem still exist in 7.99.28 (nick-nhusb merged)
 >   
 >   I repeated the test from the PR using the following configuration:
 >   
 >    - Desktop with Asus P5B-V motherboard
 >    - amd64 GENERIC kernel built from 2016.04.23.13.14.52 sources
 >    - Userland built from 2016.04.10.18.27.56 sources
 >    - /dev/usb9 through /dev/usb15 added using MAKEDEV
 >    - Nikon CoolPix 4500 in PTP mode connected to ALI EHCI/OHCI PCI card
 >    - gphoto2-2.5.8
 >    - libgphoto2-2.5.8nb1
 >    - libusb-0.1.12nb4 + patch to increase number of supported USB busses to 16
 >    - gphoto2 command: gphoto2 --port usb --debug --capture-image >log 2>&1
 >   
 >   It took a few runs of starting gphoto2, seeing it hang,
 
 Can you generate usbhist logs with a OHCI_DEBUG kernel and
 
      sysctl -w hw.ohci.debug=10
 
 You can extract the logs using
 
      vmstat -u usbhist
 
 
 >   and killing it
 >   after 5-30 seconds by hitting control-C twice, but the end result was
 >   that the machine locked up, instead of panicing as it had done in
 >   earlier tests.
 >   
 >   Breaking into ddb on the serial console shows:
 >   
 >   db{0}> bt
 >   breakpoint() at netbsd:breakpoint+0x5
 >   comintr() at netbsd:comintr+0x59a
 >   Xintr_ioapic_edge9() at netbsd:Xintr_ioapic_edge9+0xea
 >   --- interrupt ---
 >   ohci_softintr() at netbsd:ohci_softintr+0x454
 >   usb_soft_intr() at netbsd:usb_soft_intr+0x1f
 >   softint_dispatch() at netbsd:softint_dispatch+0xd3
 >   DDB lost frame for netbsd:Xsoftintr+0x4f, trying 0xfffffe8044f7eff0
 >   Xsoftintr() at netbsd:Xsoftintr+0x4f
 >   --- interrupt ---
 >   0:
 >   db{0}> ps
 
 
 ps/w output as well please
 
 >   PID    LID S CPU     FLAGS       STRUCT LWP *               NAME WAIT
 >   170      1 3   3         0   fffffe80aa9ee8e0               cron tstile
 >   783      1 3   1         0   fffffe80a6503500               cron tstile
 >   574      1 3   3         0   fffffe80a6503920               cron wait
 >   171      1 3   1         0   fffffe80ab6ec0c0               cron wait
 >   686      1 3   1         0   fffffe80ab6ec4e0            gphoto2 usbxfer
 
 bt/a fffffe80ab6ec4e0
 
 >   146      1 3   1        80   fffffe804642d780                 sh wait
 >   429      1 3   1        80   fffffe8046848420                 sh wait
 >   104      1 3   1        80   fffffe804642d360               sshd select
 >   486      1 3   1        80   fffffe80adb2a860               sshd select
 >   636      1 3   1         0   fffffe804642dba0                 sh biolock
 >   455      1 3   2        80   fffffe80464a3760              getty ttyraw
 >   625      1 3   3        80   fffffe80460d5320              getty ttyraw
 >   643      1 3   0        80   fffffe80464a3b80              getty ttyraw
 >   600      1 3   1        80   fffffe8045e87620              login wait
 >   516      1 3   1         0   fffffe80ae0ad460               cron tstile
 >   613      1 3   0        80   fffffe80adb2a020              inetd kqueue
 >   565      1 3   3        80   fffffe80ae0ad040               qmgr kqueue
 >   511      1 3   3        80   fffffe80adb2a440             pickup kqueue
 >   571      1 3   3        80   fffffe80aa9ee0a0             master kqueue
 >   349      1 3   0        80   fffffe80ae0ad880             powerd kqueue
 >   182      1 3   1         0   fffffe8046848000            syslogd biowait
 >   1        1 3   1        80   fffffe8045facae0               init wait
 >   0       87 3   1       200   fffffe804600c720            physiod physiod
 >   0       86 3   2       200   fffffe80460d5740           aiodoned aiodoned
 >   0       85 3   3       200   fffffe80460d5b60            ioflush genput
 >   0       84 3   2       200   fffffe804600c300           pgdaemon pgdaemon
 >   0       81 3   2       200   fffffe804600cb40          atapibus0 sccomp
 >   0       78 3   2       200   fffffe8045e87200              usb12 usbevt
 >   0       77 3   2       200   fffffe8045e49a80              usb13 usbevt
 >   0       76 3   2       200   fffffe8045ec8640              usb11 usbevt
 >   0       75 3   2       200   fffffe8045e49660               usb9 tstile
 bt/a fffffe8045e49660
 
 (in fact  backtrace from all the "tstile"d lwps is probably best)
 
 Thanks,
 Nick