NetBSD Problem Report #22218

Received: (qmail 7185 invoked by uid 605); 22 Jul 2003 08:04:17 -0000
Message-Id: <200307220804.h6M84CZ11954@localhost>
Date: Tue, 22 Jul 2003 10:04:13 +0200 (CEST)
From: coto@marsrakete.de
Sender: gnats-bugs-owner@NetBSD.org
Reply-To: coto@marsrakete.de
To: gnats-bugs@gnats.netbsd.org
Subject: USB isoc locks
X-Send-Pr-Version: 3.95

>Number:         22218
>Category:       kern
>Synopsis:       using a USB isochronous endpoint freezes kernel
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Jul 22 08:05:00 +0000 2003
>Closed-Date:    
>Last-Modified:  Thu Oct 09 11:45:01 +0000 2008
>Originator:     Thomas Runge
>Release:        NetBSD 1.6U, snapshot from 07/20/03
>Organization:
>Environment:
Architecture: i386
Machine: i386
>Description:
	Using bktr2jpeg (version 2.0a9) with a USB webcam the kernel
freezes hard whenever the app reads from the isochronous endpoint. This
happens on 1.6.1_STABLE too.
>How-To-Repeat:
	Using said application available here:
http://core.de/~coto/projects/bktr2jpeg/ with a supported webcam
(ov511 based) or probably any other application reading from any isoc.
>Fix:
	There once was a suggestion to comment out the last line in
src/sys/dev/usb/ohci.c:ohci_device_isoc_done (remove xfer->hcpriv = NULL;)
as a workaround. This works (on -stable and -current), but seems to just
hide the bug.
>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->feedback
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Thu, 18 Sep 2008 04:53:01 +0000
State-Changed-Why:
Is this still an issue nowadays?


From: Thomas Runge <coto@marsrakete.de>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
Date: Tue, 30 Sep 2008 11:19:50 +0200 (CEST)

 On Thu, 18 Sep 2008, dholland@NetBSD.org wrote:

 > Date: Thu, 18 Sep 2008 04:53:03 +0000 (UTC)
 > From: dholland@NetBSD.org
 > Reply-To: gnats-bugs@NetBSD.org
 > To: kern-bug-people@NetBSD.org, netbsd-bugs@NetBSD.org,
 >     gnats-admin@NetBSD.org, dholland@NetBSD.org, coto@core.de
 > Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
 > 
 > Synopsis: using a USB isochronous endpoint freezes kernel
 >
 > State-Changed-From-To: open->feedback
 > State-Changed-By: dholland@NetBSD.org
 > State-Changed-When: Thu, 18 Sep 2008 04:53:01 +0000
 > State-Changed-Why:
 > Is this still an issue nowadays?

 Sorry for answering late. I grabbed the old laptop out of the basement
 and installed -current. My test scenario (old webcam, cambevao-software)
 works. Today I'll try to find some time to set it up to serve 24 hours
 a day. But even then we'll have to see it survive a few days or weeks.

 -- 
 Tom

From: Thomas Runge <coto@marsrakete.de>
To: gnats-bugs@NetBSD.org
Cc: dholland@NetBSD.org
Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
Date: Wed, 8 Oct 2008 18:52:32 +0200 (CEST)

   This message is in MIME format.  The first part should be readable text,
   while the remaining parts are likely unreadable without MIME-aware tools.

 --0-1917849325-1223484752=:21239
 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed

 On Thu, 18 Sep 2008, dholland@NetBSD.org wrote:

 > Synopsis: using a USB isochronous endpoint freezes kernel
 > State-Changed-Why:
 > Is this still an issue nowadays?

 I kept the app running for a few days. This works quite nicely, as long,
 as there is no activity on this system. When I start a cvs update, the
 kernel will panic soon. This is my latest stacktrace:

 ugen_isoc_rintr(c1079100,c6e41374,0,1,0,e500,15c0,4,3c1,4) at netbsd:ugen_isoc_rintr+0x56
 usb_transfer_complete(c1079100,15c0,20,a,3,c6e41380,c6509520,0,0,0) at netbsd:usb_transfer_complete+0xba
 ohci_softintr(c6509524,0,30,c63b0010,c63c0010,0,1fc7be0,c0e9f080,0,c63bbda0)  at netbsd:ohci_softintr+0x543
 softint_dispatch(c74adae0,4,0,0,0,0,c63bbd90,c63bbbd8,c63bbc30,28) at netbsd:softint_dispatch+0x7c
 DDB lost frame for netbsd:Xsoftintr+0x3d, trying 0xc63bbd88
 Xsoftintr() at netbsd:Xsoftintr+0x3d
 ---  interrupt  ---
 fatal page fault in supervisor mode
 trap type 6 code 0 eip c052f6f7 cs 8 eflags 10202 cr2 3c ilevel 8
 kernel: supervisor trap page fault, code=0
 Faulted in DDB;  continuing...
 db{0}>


 This happens, when I use pkgsrc/graphics/cambevao with an ov511+ based
 camera.
 I tried making a short test code. You'll find it attached. This one crashes
 the kernel as well, but with a different stack trace. The same happens,
 when I start cambevao twice. In earlier NetBSD versions I just got an
 error when I tried opening the control device twice, now I can do this
 and get the panic after a few usb commands (the USB_SET_ALTINTERFACE
 fails and then I close the control device. POFF!):
 http://coto.marsrakete.de/IMG_1514.JPG (71 KB)
 http://coto.marsrakete.de/IMG_1515.JPG (95 KB)

 This shows, that we do have serious problems with isochronous USB
 transfers, still.
 I'm willing to do more testing and debug code as good as my expertise
 goes. I can give remote access to my machine, if that would help.

 -- 
 Tom
 (ex-coto@core.de)
 --0-1917849325-1223484752=:21239
 Content-Type: TEXT/PLAIN; charset=US-ASCII; name=usb_iso_crash.c
 Content-Transfer-Encoding: BASE64
 Content-ID: <Pine.NEB.4.64.0810081852320.21239@beaver.marsrakete.de>
 Content-Description: 
 Content-Disposition: attachment; filename=usb_iso_crash.c

 I2luY2x1ZGUgPHN0ZGlvLmg+DQojaW5jbHVkZSA8c3RkbGliLmg+DQojaW5j
 bHVkZSA8ZmNudGwuaD4NCiNpbmNsdWRlIDxlcnJuby5oPg0KI2luY2x1ZGUg
 PGRldi91c2IvdXNiLmg+DQojaW5jbHVkZSA8c3lzL3R5cGVzLmg+DQojaW5j
 bHVkZSA8c3lzL2lvY3RsLmg+DQoNCmludCBtYWluKGludCBhcmdjLCBjaGFy
 ICoqYXJndikNCnsNCglvcGVuX2RldigwKTsNCglvcGVuX2RldigxKTsNCglv
 cGVuX2RldigyKTsNCn0NCg0Kdm9pZCBlcnJvcihjaGFyICptc2cpDQp7DQoJ
 ZnByaW50ZihzdGRlcnIsICIgICAgICBGYWlsZWQhICVzOiAlc1xuIiwgbXNn
 LCBzdHJlcnJvcihlcnJubykpOw0KfQ0KDQppbnQgb3Blbl9kZXYoaW50IG51
 bSkNCnsNCgljaGFyIGN0cmxkZXZpY2VbRklMRU5BTUVfTUFYXTsNCgljaGFy
 IGlzb2NkZXZpY2VbRklMRU5BTUVfTUFYXTsNCglzdHJ1Y3QgdXNiX2FsdF9p
 bnRlcmZhY2UgYWx0Ow0KCWludCBjdHJsZmQsIGlzb2NmZCwgdW5pdDsNCg0K
 CXVuaXQgPSAwOw0KDQoJZnByaW50ZihzdGRlcnIsICJydW4gIyVkXG4iLCBu
 dW0pOw0KCXNucHJpbnRmKGN0cmxkZXZpY2UsIEZJTEVOQU1FX01BWCwgIi9k
 ZXYvdWdlbiVkLjAwIiwgdW5pdCk7DQoJc25wcmludGYoaXNvY2RldmljZSwg
 RklMRU5BTUVfTUFYLCAiL2Rldi91Z2VuJWQuMDEiLCB1bml0KTsNCg0KCWZw
 cmludGYoc3RkZXJyLCAiICAgb3BlbmluZyAlc1xuIiwgY3RybGRldmljZSk7
 DQoJaWYoKGN0cmxmZCA9IG9wZW4oY3RybGRldmljZSwgT19SRFdSKSkgPCAw
 KQ0KCXsNCgkJZXJyb3IoIm9wZW4gY3RybGRldiIpOw0KCQlyZXR1cm47DQoJ
 fQ0KDQoJZnByaW50ZihzdGRlcnIsICIgICBzZXR0aW5nIGFsdCBpbnRlcmZh
 Y2VcbiIpOw0KCWFsdC51YWlfaW50ZXJmYWNlX2luZGV4ID0gMDsNCglhbHQu
 dWFpX2FsdF9ubyA9IDc7DQoJaWYoaW9jdGwoY3RybGZkLCBVU0JfU0VUX0FM
 VElOVEVSRkFDRSwgJmFsdCkgPCAwKQ0KCXsNCgkJZXJyb3IoInNldCBhbHQi
 KTsNCgkJY2xvc2UoY3RybGZkKTsNCgkJcmV0dXJuOw0KCX0NCg0KCWZwcmlu
 dGYoc3RkZXJyLCAiICAgb3BlbmluZyAlc1xuIiwgaXNvY2RldmljZSk7DQoJ
 aWYoKGlzb2NmZCA9IG9wZW4oaXNvY2RldmljZSwgT19SRE9OTFkpKSA8IDAp
 DQoJew0KCQllcnJvcigib3BlbiBpc29jZGV2Iik7DQoJCWNsb3NlKGN0cmxm
 ZCk7DQoJfQ0KDQoJLyoNCglmcHJpbnRmKHN0ZGVyciwgIiAgIGNsb3Npbmcg
 aXNvY2RldmljZVxuIik7DQoJY2xvc2UoaXNvY2ZkKTsNCglmcHJpbnRmKHN0
 ZGVyciwgIiAgIGNsb3NpbmcgY3RybGRldmljZVxuIik7DQoJY2xvc2UoY3Ry
 bGZkKTsNCgkqLw0KfQ0KDQo=

 --0-1917849325-1223484752=:21239--

State-Changed-From-To: feedback->open
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Wed, 08 Oct 2008 17:36:21 +0000
State-Changed-Why:
Feedback received.


From: "Jared D. McNeill" <jmcneill@invisible.ca>
To: gnats-bugs@NetBSD.org
Cc: kern-bug-people@netbsd.org, gnats-admin@netbsd.org, netbsd-bugs@netbsd.org, 
    coto@core.de
Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
Date: Wed, 8 Oct 2008 14:18:58 -0400 (EDT)

 On Wed, 8 Oct 2008, Thomas Runge wrote:
 > I kept the app running for a few days. This works quite nicely, as long,
 > as there is no activity on this system. When I start a cvs update, the
 > kernel will panic soon. This is my latest stacktrace:
 >
 > ugen_isoc_rintr(c1079100,c6e41374,0,1,0,e500,15c0,4,3c1,4) at netbsd:ugen_isoc_rintr+0x56
 > usb_transfer_complete(c1079100,15c0,20,a,3,c6e41380,c6509520,0,0,0) at netbsd:usb_transfer_complete+0xba
 > ohci_softintr(c6509524,0,30,c63b0010,c63c0010,0,1fc7be0,c0e9f080,0,c63bbda0)  at netbsd:ohci_softintr+0x543
 > softint_dispatch(c74adae0,4,0,0,0,0,c63bbd90,c63bbbd8,c63bbc30,28) at netbsd:softint_dispatch+0x7c
 > DDB lost frame for netbsd:Xsoftintr+0x3d, trying 0xc63bbd88
 > Xsoftintr() at netbsd:Xsoftintr+0x3d
 > ---  interrupt  ---
 > fatal page fault in supervisor mode
 > trap type 6 code 0 eip c052f6f7 cs 8 eflags 10202 cr2 3c ilevel 8
 > kernel: supervisor trap page fault, code=0
 > Faulted in DDB;  continuing...
 > db{0}>

 This looks identical to an issue I'm seeing with a USB ATSC tuner and 
 ehci(4). I suspect the ehci isoc code was based on the ohci and uhci 
 implementations.. are you able to try this camera on a uhci to see if you 
 can reproduce the same crash there?

 Cheers,
 Jared

From: Thomas Runge <coto@marsrakete.de>
To: gnats-bugs@NetBSD.org
Cc: kern-bug-people@NetBSD.org, gnats-admin@NetBSD.org, netbsd-bugs@NetBSD.org
Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
Date: Thu, 9 Oct 2008 08:25:00 +0200 (CEST)

 On Wed, 8 Oct 2008, Jared D. McNeill wrote:

 > Date: Wed,  8 Oct 2008 18:20:04 +0000 (UTC)
 > From: Jared D. McNeill <jmcneill@invisible.ca>
 > Reply-To: gnats-bugs@NetBSD.org
 > To: kern-bug-people@NetBSD.org, gnats-admin@NetBSD.org,
 >     netbsd-bugs@NetBSD.org, coto@marsrakete.de
 > Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
 > 
 > The following reply was made to PR kern/22218; it has been noted by GNATS.
 >
 > From: "Jared D. McNeill" <jmcneill@invisible.ca>
 > To: gnats-bugs@NetBSD.org
 > Cc: kern-bug-people@netbsd.org, gnats-admin@netbsd.org, netbsd-bugs@netbsd.org,
 >    coto@core.de
 > Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
 > Date: Wed, 8 Oct 2008 14:18:58 -0400 (EDT)
 >
 > On Wed, 8 Oct 2008, Thomas Runge wrote:
 > > I kept the app running for a few days. This works quite nicely, as long,
 > > as there is no activity on this system. When I start a cvs update, the
 > > kernel will panic soon. This is my latest stacktrace:
 > >
 > > ugen_isoc_rintr(c1079100,c6e41374,0,1,0,e500,15c0,4,3c1,4) at netbsd:ugen_isoc_rintr+0x56
 > > usb_transfer_complete(c1079100,15c0,20,a,3,c6e41380,c6509520,0,0,0) at netbsd:usb_transfer_complete+0xba
 > > ohci_softintr(c6509524,0,30,c63b0010,c63c0010,0,1fc7be0,c0e9f080,0,c63bbda0)  at netbsd:ohci_softintr+0x543
 > > softint_dispatch(c74adae0,4,0,0,0,0,c63bbd90,c63bbbd8,c63bbc30,28) at netbsd:softint_dispatch+0x7c
 > > DDB lost frame for netbsd:Xsoftintr+0x3d, trying 0xc63bbd88
 > > Xsoftintr() at netbsd:Xsoftintr+0x3d
 > > ---  interrupt  ---
 > > fatal page fault in supervisor mode
 > > trap type 6 code 0 eip c052f6f7 cs 8 eflags 10202 cr2 3c ilevel 8
 > > kernel: supervisor trap page fault, code=0
 > > Faulted in DDB;  continuing...
 > > db{0}>
 >
 > This looks identical to an issue I'm seeing with a USB ATSC tuner and
 > ehci(4). I suspect the ehci isoc code was based on the ohci and uhci
 > implementations.. are you able to try this camera on a uhci to see if you
 > can reproduce the same crash there?

 I'm afraid, I can't. I do not own any uhci hardware. This old laptop has
 ohci built-in, but I tried (on the same hardware) a USB2-PC-CARD. This
 one procides ehci and ohci. Same results.

 Do you know any hardware that is known to have uhci? Maybe I could
 try on a friends laptop.

 -- 
 Tom

From: "Jared D. McNeill" <jmcneill@invisible.ca>
To: Thomas Runge <coto@marsrakete.de>
Cc: gnats-bugs@NetBSD.org, kern-bug-people@NetBSD.org, 
 gnats-admin@NetBSD.org, netbsd-bugs@NetBSD.org
Subject: Re: kern/22218 (using a USB isochronous endpoint freezes kernel)
Date: Thu, 09 Oct 2008 07:41:57 -0400

 Thomas Runge wrote:
 > Do you know any hardware that is known to have uhci? Maybe I could
 > try on a friends laptop.

 Anything with an Intel or VIA chipset should do.

 Cheers,
 Jared

>Unformatted:

Home	PR Database Search